1 - KrishiKosh

PEENTICE-HALL PHYSICS SERD
DONALD H . MENZEL. Editor
PRACTICAL
SPECTROSCOPY
By
GEORGE R. HARRISON,
PH.D., SC.D.
Professor of Physics
RICHARt) C. LORD, PH.D.
Associate Professor of Chemistry
JOHN R. LOOFBOUROW, Sc.D.
Professor of Biophysics
OF THE
SPECTROSCOPY LABORATORY
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
K\
New York
P R E N T I C E - H A L L , INC.
1948
COPYRIGHT, 194.8, BT
PRENTICE-HALL, INC.
70 Fifth Avenue, New York
ALL RIGHTS RESERVED.
NO PART OF THIS BOOK
MAT BE REPRODUCED IN ANV FORM, B T MIMEOGRAPH OR ANT OTHER MEANS, WITHOUT PERMISJ
SION I N WRITING ESOM i b E PUBLISHERS.
PRINTED IN THE UNITED STATES OF AMERICA
Preface
In operating the Spectroscopy Laboratory of the Massachusetts
Institute of Technology, the authors have for some years felt the
need of a text and reference book that would help the worker in any
branch of science to evaluate the aid which the techniques of spectroscopy might lend to the solution of his own problems. In our
attempt to fill this need, we, as a physicist, a chemist, and a biophysicist, respectively, have tried to synthesize our three viewpoints
in a way that would be helpful to all who use or might use the techniques of experimental spectroscopy.
Since other texts are available which present effectively the history
of spectroscopy, we have avoided the historical method of approach
and have attempted rather to give a comprehensive view of the
status and possibilities of experimental spectroscopy as it exists
today. Because the subject matter to be covered is so extensive,
we have had to choose between comprehensive and exhaustive coverage and have selected the former alternative.
In Chapter 1 we view the field as from a great altitude, to enable
the reader who is unacquainted with the methods and accomplishments of spectroscopy to judge for himself which parts, if any, may be
of importance to him. In the remainder of the book we reconsider
the various topics -in considerably greater detail. We have endeavoured to include a sufficient number of appropriate specific
references to the literature to enable the reader to investigate still
more closely subjects which may directly concern him.
References to specific points are given as footnotes; at the ends of
most chapters appropriate general references are also given. While
the bibliography is not intended to be exhaustive, we have attempted
in specific references to cite both the original and the most up-to-date
treatment of the topic involved, and in the general references to
vi
PREFACE
cover the subject broadly. Since we discuss many topics from
several viewpoints, we have made liberal use of cross-references.
A book covering spectroscopy from several aspects is likely to contain a certain amount of inconsistency in terminology. Resolution
of such inconsistencies is not made simpler by the fact that the symbology of spectroscopy is far from stabilized.
We gratefully acknowledge the courtesy of the Technology Press,
John Wiley and Sons, the McGraw-Hill Book Co., Inc., PrenticeHall, Inc., and others as specified later, for permission to reproduce
figures and tables, and appreciate deeply the willing cooperation of
the various manufacturers of spectroscopic equipment who have
furnished illustrations of apparatus, as credited in each instance.
We are especially grateful for the suggestions of Messrs. W. R. Erode,
R. S. McDonald, W. F . Meggers, K. W. Meissner, R. A. Sawyer,
A. L- Schoen, and Van Zandt Williams, each of whom has read and
criticized one or more chapters dealing with his own specialty. We
also thank Professor Donald H. Menzel, editor of the series of which
the book forms a part, for his suggestions regarding the manuscript.
George R. Harrison
Richard C. Lord
John R. Loofbourow
Cambridge, Massachusetts
Contents
CHAPTER
PAGE
1. SPECTROSCOPY AS A SCIENTIFIC TOOL
T H E SPECTRUM
1
1
.1.1 Spectroscopy. 1.2 Origins of Spectroscopy. 1.3 Measurement of the Spectrum. 1.4 The Infrared Spectrum.
1.5 The Visible Spectrum. 1.6 The Ultraviolet. 1.7 Spectroscopes and Spectrographs.
TECHNIQUES OF SPECTROSCOPY
10
Emission Spectroscopy. 1.9 Qualitative Spectroscopic
Analysis of/ Materials. 1.10 Quantitative Spectroscopic
Analysis, r l l Absorption Spectroscopy. 1.12 Absorption
Spectrophotometry. 1.13 Fluorescence Spectroscopy.
USES OF SPECTROSCOPY
1.14 The Spectroscope in Chemistry. 1.15 Spectroscopy in.
Astronomy. 1.16 Spectroscopy in Physics. 1.17 Spectroscopy in Biology and Medicine. 1.18 Spectroscopy in Food
Research. 1.19 Spectroscopy in Metallurgy and Mineralogy.
1.20 Forensic Spectroscopy.
15
•
2. SELECTION OF SPECTROSCOPIC INSTRUMENTS . . .
2.1 Dispersion. 2.2 Resolving Power. 2.3 The Dispersing
Element. 2.4 Dispersing Prisms. 2.5 Diffraction Gratings.
2.6 The Slit.' 2.7 The CoUimating and Focusing Systems.
2.8 Observing and Recording Systems. 2.9 Comparison of
Prism" and Grating Spectrographs. 2.10 Speed and EiBciency.
2.11 Scq,ttered Light and False Lines. 2.12 Shapes of Spectrum Lines. 2.13 Astigmatism. 2.14 Space Requirements.
2.15 Summary of Comparison. 2.16 Monochromatic Illuminators.
27
3. PRISM SPECTROSCOPES AND SPECTROGRAPHS . . .
3.1 Materials for Dispersive Prisms. 3.2 The Simple Spectrometer. 3.3 Modern Wavelength Spectrometers. 3.4 Direct-Vision Spectroscopes. 3.5 Portable Spectrographs. 3.6
Special Glass-Prism Spectrographs. 3.7 Quartz-Prism
51
viii
CONTENTS
CHAPTEB
PAGE
Spectrographs. 3.8 The Littrow Mounting. 3.9 The Fery
Spectrograph. 3.10 The Gaertner Large Quartz Spectrograph. 3.11 Spectrographs with Interchangeable Optical
Systems. 3.12 Prism Monochromators.
4. DIFFRACTION-GRATING SPECTROGRAPHS
4.1 Plane-Grating Spectrographs. 4.2 The Rowland Concave Grating. 4.3 The Rowland Mounting. 4.4 The Abney
Mounting. 4.5 The Paschen-Runge Mounting. 4.6 The
Eagle Mounting. 4.7 The Wadsworth Stigmatie Mounting.
4.8 The Choice of a Grating Mounting. 4.9 Astigmatism of
the Concave Grating and Its Reduction. 4,10 The Testing of
Diffraction Gratings. 4.11 Grating Monochromators.
5. T H E TESTING, ADJUSTMENT, AND CARE OF SPECTROSCOPIC EQUIPMENT
T H E TESTING OF SPECTROSCOPIC COMPONENTS
5.1 The Testing of Slits. 5.2 The Testing of Prisms and
Lenses. 5.3 The Testing of Gratings. 5.4 The Testing of
Prism Spectrographs. 5.5 The Hartmann Test.
ADJUSTMENT OP PRISM AND GRATING SPECTROGRAPHS . . . .
71
96
96
^
103
5.6 Adjustments Requirjsd for Various Types of Instruments.
5.7 Adjustment of Slit Width and Length. '5.8 Adjustment
of Slit Perpendicular to the Direction of Dispersion. 5.9 Focusing the Spectrum: Commercial Prism Spectrographs.
5.10 Focusing the Spectrum: Commercial Littrow and EagleMounting Spectrographs. 5.11 Adjustment of Spectrometers.
5.12 Adjustment of Concave Gratings. 5.13 Adjustment of
Wadsworth Grating Mountings. 5.14 Adjustment of Plane
Gratings.
T H E CARE OF SPECTROSCOPIC EQUIPMENT
115
5.15 General. 5.16 Care of Mirrors, Prisms, and L ^ s e s .
5.17 Cleaning and Care of Gratings.
6. ILLUMINATION OF T H E SPECTROSCOI^E
118
6.1 Coherent and Noncoherent Radiation. 6.2 Spectral
Line Shape and the Rayleigh Criterion for Resolution. 6.3 Selection of Optimum Slit Width. 6.4 Filling the Aperture of
the Spectroscope with Light. 6.5 Use of a Condensing Lens
or Mirror. 6.6 Uniform Illumination of the Slit. 6.7 Illumination of the Slit by a Source Extended in Depth. 6.8 Elumination to Obtain Maximum Radiant Intensity or Total Radiant Power in the Spectral Image. 6.9 Factors Governing the
Radiant Power Transmission of a Spectroscopic System. 6.10
Effect of Entrance Slit Width on Spectral Purity. 6.11 Effect
CONTENTS
ix
CHAPTER
PAGE
of Exit Slit. 6.12 Expression for Radiant Power Transmission
of a Spectroscopic System.
7. PHOTOGRAPHY OF THE SPECTRUM
141
7.1 Photographic Plates and Films. 7.2 Response of the
Emulsion to Light. 7.3 Contrast. 7.4 Speed, Inertia, and
Latitude. 7.5 Resolving Power and Graininess. 7.6 Types
of Plates and FUms. 7.7 Variation of Emulsion Characteristics with Wavelength. 7.8 Storage and Handling of Photographic Materials. 7.9 The Photographic Darkroom. 7.10
Development and Processing. 7.11 Developers. 7.12 Common Defects in Spectrum Photographs. 7.13 The Eberhard
Effect. 7.14 Halation and Spreading. 7.15 Photography of
Various Regions of the Spectrum. 7.16 Selection of Spectrally Sensitive Emulsions.
8. LIGHT SOURCES FOR SPECTROSCOPY
166
GENERAL CHARACTEKISTICS OF SOURCES
167
8.1 Spectral Energy Distribution. 8.2 The Power Output of
Sources. 8.3 Practical Considerations.
THERMAL EMISSION SOURCES
170
8.4 Spectral Characteristics of Blackbody Radiation. 8.5
Blackbody Radiators. 8.6 Incandescent Electric Lamps.
8.7 Enclosed Metallic Arcs with Incandescent Electrodes.
8.8 Low-Temperature Thermal Radiators. 8.9 Other Thermal Radiators.
OPEN ARCS
8.10 Electrical
Metallic Arcs.
./
Characteristics. ^ . 1 1 Carbon
179
Arcs.
8.12
ENCLOSED ARCS
184
8.13 Electrical Characteristics. 8.14 Low-Pressure Mercury
" Arcs. 8.15 High-Pressure Mercury Arcs. 8.16 Other Enclosed Metallic-Vapor Arcs. 8.17 Enclosed Carbon Arcs.
DISCHARGE TUBES
8.18 General Characteristics.
188
8.19 Glow-Discharge Tubes.
ELECTRIC SPARKS
1^
8.20 General Characteristics. 8.21 The Spark in Air and
Other Gases. 8.22 The Hot Spark in Vacuum. 8.23 The
Underwater Spark. 8.24 The Spark as a Source in Qualitative and Quantitative Analysis.
MISCELLANEOUS SOURCES
8.25 Cathodoluminescence Devices. 8.26 Fluorescence,
Phosphorescence, Resonance Radiation, and Chemilumines-
196
X
CONTENTS
CHAPTER
PAGE
cence. 8.27 Pulsed Discharge Tubes.
Source of Radiation.
8.28 The Sun as a
9. IDENTIFICATION OF SPECTRUM LINES
199
9.1 Identification of Lines and Bands by Appearance. 9.2
Identification by Comparison Spectra. 9.3 Spectrum Charts
for Comparison. 9.4 Identification of Lines \)y Wavelength
Determination. 9.5 Measurement of Spectrogt-ams. 9.6 Use
of the Comparator. 9.7 Calculation of Wavelengths. 9.8
Standards of Wavelength. 9.9 Intensity Estimates. 9.10
Catalogues of Wavelengths. 9.11 The Harrison Automatic
Comparator. 9.12 Limitations of Wavelengtl^ Measurement
with Diffraction Gratings.
10. THE O R I G I N S ' O F ATOMIC SPECTRA ,
228
SPECTRAL. SERIES AND ATOMIC ENERGY STATES
228
10,1 The Hydrogen Atom. 10.2 Quantum Numbers in
Atomic Spectra. 10.3 Series in Atoms with Many Electrons.
10.4 Multiplicity in Atomic Spectra.
LINE INTENSITIES IN ATOMIC SPECTRA
243
10.5 Some Basic Results of Quantum Mechanics. io.6 Selection Rules and Intensities of Spectral Lines. ' lo.7 The Effect
of External Influences on Atomic Spectra,, 10.8 The Stark
and Zeeman Effects.
ATOMIC SPECTRA AND ATOMIC STRUCTURE
. .
.. '
252
10.9 The Pauli Exclusion Principle and the Periodic Table.
11. MOLECULAR SPECTRA AND MOLECULAR. STRUCTURE
ENERGY LEVELS IN MOLECULES
261
261
11.1 Rotational Energy Levels in Molecule^. 11.2' Vibrational Energy Levels. 11.3 Electronic Energy Levels.
MOLECULAR SELECTION RULES AND THE APPEARANCE OF M O LECULAR SPECTRA
'
11.4 Pure Rotational Spectra. 11.5 Vibrational Spectra.
11.6 Rotational Fine Structure in Vibrational Spectra. 11.7
Electronic Spectra of Diatomic Molecules, l i . s Vibrational
Structure of Diatomic. Spectra. 11.9 Rotational Fine Structure of an Electronic-Vibrational Band. 11.10-Electronic
Spectra of Polyatomic Molecules. 11.11 Vibrational Structure of Electronic Spectra in Polyatomic McJecules. 11.12
Rotational Fine Structure in Electronic Spectra. 11.13 The
Effects of External Influences on Molecular ijpectra. 11.14
Summary of Molecular Spectra.
276
CONTENTS
CHAPTER
xi
-
PAGE
12. T H E MEASUREMENT OF SPECTRAL INTENSITIES
.
RADIOMETRY
300
302
12.1 Bolometers. 12.2 Metal Bolometers. 12.3 Semiconductor Bolometers. 12.4 A Superconductor Bolometer. 12.5
Thermocouples and Thermopiles. 12.6 Other Thermal Detectors. 12.7 Amplification and Recording Methods in
Radiometry.
12.8 Photorelays. 12.9 Alternating-Current
Amplifiers. 12.10 A Direct-Current Amplifier. 12.11 Recorders.
PHOTOELECTRIC MEASUREMENT OF INTENSITY
314
12.12 Photoemissive Cells and Electron-Multiplier Tubes.
12.13 Photoconductive
Cells. 12.14 Photovoltaic
Cells.
12.1.5 The Incorporation of the Photocell in the Spectrograph.
12.16 Amplification and Recording of Photocurrents. 12.17
Photoelectric Spectrometers.
13. PHOTOGRAPHIC PHOTOMETRY
13.1 Photometric Characteristics of the Emulsion.
326
HOMOCHROMATIC PHOTOMETRY
13.2 Calibrating the Emulsion.
Light Intensity.
330
13.3 Methods of Varying
HETEROCHROMATIC PHOTOMETRY
13.4 Uses of Hetercjphromatic
Sources for Standardization.
342
Photometry.
13.5 Light
APPARATUS AND METHODS FOR PHOTOGRAPHIC PHOTOMETRY
.
346
13.6 Selection of the Spectrograph. 13.7 Selection of the
Emulsion. 13.8 Timing the Exposure. 13.9 The Brush
Effect. 13.10 Processing the Spectrogram. 13.11 Shortc u t Methods of Photographic Photometry.
DENSITOMETRY
350
13.12 Densitometers. 13.13 Photoelectric
Densitometers.
13.14 Thermoelectric Densitometers. 13.15 Operation of the
Densitometer. 13.16 Recording Densitometers. 13.17 Precision, of Densitometers. 13.18 Special Computing Densitometers.
14. ABSORPTION SPECTROPHOTOMETRY
LAWS OF ABSORPTION AND THEIR APPLICATION TO ABSORPTION
SPECTROPHOTOMETRY
14.1 Lambert's Law. 14.2 Beer's Law. 14.3 Variables
Measured in Absorption Spectrophotometry. 14.4 Elimination of Effects Due to Reflections and Absorption by Cell
Windows and by Solvents. 14.5 Presentation of Data.
362
362
xii
CONTENTS
CHAPTER
PAGE
SELECTION OF APPARATUS FOH ABSOBPTION SPECTROPHOTOMETHY
368
14.6 Choice of Source of Radiation. 14.7 Choice of Absorption Cells. 14.8 Choice of Spectrophotometric Method.
VISUAL ABSORPTION SPECTROPHOTOMETRY
374
14.9 General Considerations. 14.10 Advantages and Limitations of the Visual Method. 14.11 Instruments Using the
Martens Type of Polarizing Photometer. 14.12 Hilger-Nutting Polarizing Spectrophotometer. 14.13 Manipulation of
Polarizing Spectrophotometers. 14.14 Other Visual Spectrophotometers and Their Manipulation.
PHOTOGRAPHIC ABSORPTION SPECTROPHOTOMETRY
384
14.15 General Considerations. 14.16 Advantages and Limitations of the Photographic Method. 14.17 Spectrographs for
Photographic Absorption Spectrophotometry. 14.18 Photographic Materials and Processing. 14.19 Semiquantitative
and Plate-Calibration Methods. 14.20 Split-Beam Photometers. 14.21 Multiple-Beam Photometers. 14.22 Considerations Governing Alignment and* Illumination. 14.23 Choice
of Density in Specimens: Determination of Match Points. •.
14.24 Precision of Determination of Wavelengths and
Densities.
,
PHOTOELECTRIC ABSORPTION SPECTROPHOTOMETRY
399
14.25 General Considerations. 14.26 Point-by-Point Instruments for Relative Intensity Measurements. 14.27 Photoelectric Null Methods: Nonrecording. 14.28 Automatic
Recording Photoelectric Spectrophotometers. 14.29 Abbreviated Absorption Spectrophotometry and Fluorimetry.
14.30 Optical Filters.
ABSORPTION OF SUBSTANCES IN SOLUTION
417
14.31 General Remarks "Regarding Solvents. 14.^2 Absorption by Solvents: Purification. 14.33 Relation of Absorption
to Chemical Constitution. 14.34 Photochemical Effects.
Fluorescence.
- ;
15. QUALITATIVE SPECTROGRAPmC
TERIALS
ANALYSIS OF MA-
15.1 Sensitivity of Detection of Various Elements. 15.2
Sensitive Lines and Ultimate Lines. 15.8 Improvement of
Sensitivity Limits. 15.4 Identification of Elements. 15.5
Light Sources and Handling of Material. 15.6 Moving-Plate
and Fractional Distillation Methods. 15.7 Analysis for Elements Difficult to Detect. 15.8 The Qualitative Analysis.
424
CONTENTS
xiii
CHAPTER
PAGE
16. QUANTITATIVE SPECTROCHEMICAL ANALYSIS
. . .
T H E BASIC METHOD OF QUANTITATIVE ANALYSIS
^
450
451
16.1 Basic Procedure. 16.2 Sources of Excitation. 16.3
Form and Preparation of the Sample. 16.4 Standards for
Comparison. 16.5 Burning of the Sample. 16.6 Selection
and Use of the Spectrograph. 16.7 Selection of Lines for
Quantitative Analysis—The Working Curve. 16.8 The Calibration Curve. 16.9 A Typical Analysis.
SPECIAL METHODS
461
16.10 The Method of Internal Standards. 16.11 Methods
for High Concentrations. 16.12 Methods for Extremely Low
Concentrations. 16.13 High-Precision, Rapid, and Shortc u t Methods. 16.14 Methods for Special Elements. 16.15
Photoelectric Methods of Analysis.
17. SPECTROSCOPY OF T H E INFRARED REGION . . . .
472
17.1 Radiation Sources and Filters for Infrared Spectroscopy.
17.2 Prism Spectrometers for the Infrared. 17.3 Grating
Spectrometers. 17.4 The Measurement of Infrared Absorption.
APPLICATIONS OF INFRARED SPECTROSCOPY
494
17.5 Qualitative Cheniical Analysis by Infrared Absorption
Spectra. 17.6 Quantitative Chemical Analysis by Infrared
Absorption Spectra. 17.7 The Determination of Molecular
Structure from Infrared Spectra. 17.8 Astrophysical and
Biological Applications of Infrared Spectroscopy.
18. RAMAN SPECTROSCOPY
506
18.1 The , Raman ^Effect. 18.2 Technique of the Raman
Effect. 18.3 Sources for Excitation of the Raman Effect.
18.4 Filters for the Raman Effect. 18.5 Arrangement of the
Excitation Unit. 18.6 The Scattering Tube. 18.7 Spectrographs for the Raman Effect. 18.8 Measurement of Intensity
and Polarization of Raman Lines. 18.9 Applications of the
Raman Effect. 18.10 Determination of Molecular Structure.
18.11 Qualitative and Quantitative Chemical Analysis.
19. SPECTROSCOPY OF T H E VACUUM ULTRAVIOLET . .
SPECTROGRAPHS FOR THE VACUUM ULTRAVIOLET
19.1 General Considerations. 19.2 Prism Spectrographs.
19.3 Normal-Incidence Grating Spectrographs. 19.4 Grazing-Incidence Grating Spectrographs. 19.5 Housings and
Vacuum Equipment.
530
531
xiv
CONTENTS
CHAPTER
PAGE
ACCESSORY APPARATUS
540
19.6 Light Sources. 19.7 Photographic Materials. 19.8
Nonphotographic Radiation Detectors. 19.9 Accessories for
Absorption Measurements.
APPLICATIONS
543
19.10 Atomic Spectra. 19.11 Wavelength Standards. 19.12
Molecular Spectra. 19.13. Miscellaneous Applications.
20. INTERFEROMETRIC SPECTROSCOPY
547
9\\ Line Broadening and Its Causes. 20.2 Line Sources for
High-Resolution Spectroscopy. 20.3 Selection of a Spectroscope of High Resolving Power. 20.4 Limitations of Diffraction Gratings at High Resolutions. 20.5 The LummerGehrcke Plate. 20.6 The Fabry-Perot Etalon. 20.7 Operation and Design of the Etalon. 20.8 Plate Coatings for the
Etalon. 20.9 Adjustment of the Etalon. 20.10 Crossing of _
Etalon Dispersion with that of a Spectroscope. 20.11 Rediic- tion of Etalon Patterns. 20.12 Direct Determination of
Wavelengths from Fabry-Perot Patterns. 20.13 The Michelson-Williams Echelon. 20.14 Optical Systems Using the
Reflection Echelon.
APPENDIX 1. Sensitive Lines of the Elements Arranged According to Elements
574
APPENDIX 2. Sensitive Lines of the Elements Arranged Ac*
cording to Wavelengths
582
INDEX
587
PRACTICAL
SPECTROSCOPY
CHAPTER
1
Spectroscopy as a Scientific Tool
T H E ACCOMPLISHMENTS ACHIEVED BY SCIENTISTS through use of the
spectroscope form a Hst so imposing as to leave no doubt that this
instrument is one of the most powerful now available for investigating
the natural universe. But spectroscopy is valuable not only to the
research scientist; it finds everyday and increasing use in technological laboratories. Today directors of such varied enterprises as
factories, assay oflSces, arsenals, mines, crime detection bureaus,
public health departments, hospitals, museums, and technical research institutes consider access to spectroscopic equipment essential
to the proper functioning of their laboratories.
T H E SPECTRUM
A spectrum has been defined as the ordered arrangement of radiation
according to wavelength. Electromagnetic radiations have been discovered that have wavelengths of every value in the range from
thousands of kilometers to trillionths of a millimeter. A complete
electromagnetic spectrum would comprise all these radiations arranged in order from the longest to the shortest wavelengths. Since
no single instrument exists that will separate radiation containing all
these wavelength's into a spectrum, the electromagnetic spectrum has
been divided into various "regions" in accordance with the types of
instruments available to produce and detect the waves of various
lengths.
Long electromagnetic waves, upwards of a meter in length, can be
separated from each other by means of ordinary tuned radio circuits.
Shorter waves, down to a few millimeters long, can be analyzed by
microwave equipment. When absorbed by matter, all electromagnetic waves produce heat. Since waves shorter than a few
millimeters and longer than about 3 X 10~' mm can be detected by
1
2
SPECTROSCOPY AS A SCIENTIFIC TOOL
[§1.1
this effect more readily than by any other, they are often called heat
waves. The range of waves from a few millimeters to 2.5 X 10~^ mm
in length is known as the far infrared region; that from 2.5 X 10~^ to
7.5 X 10~* mm is known as the near infrared. Waves that can be
seen by the eye range in length from 7.5 X 10~* mm in the red to
4 X 10~* mm in the violet; this range is called the visible region.
Waves slightly too short to see, 4 X 10~* to 3 X 10~^ mm, lie in the
near ultraviolet; then come the far ultraviolet and the extreme ultraviolet regions, which extend from 3 X 10"'* to 2 X 10~^ mm and from
there to 2 X 10~* mm, respectively. Since air is opaque to these
shorter waves, they are studied in vacuum, and the range from
2 X 10~^ to 2 X 10~* mm is also known as the vacuum ultraviolet. We
then come to the region of soft Xrays, and below 10~' mm to the
hard Xray and gamma-ray regions, to which air is again transpa!Jent.
The names, ranges, and properties of these spectral regions are summarized in Table 1.1.
1.1. Spectroscopy. The teriQ spectroscopy as used in this book is
restricted to the study of those radiations which lie in the infrared,
visible, ultraviolet, and vacuum ultraviolet regions. . The techniques
discussed are quite distinct from those used in such fields as microwave spectroscopy. X-ray spectroscopy, gamma-ray spectroscopy,
and mass spectroscopy. We are concerned here only with those
electromagnetic waves which can readily be separated into a spectrum
by means of prisms, optical gratings, and optical interferometers.
1.2. Origins of Spectroscopy. The best-known early investigator
of the spectrum was Sir Isaac Newton, who in 1666 inserted a prism
in a beam of sunlight shining into a dark room and saw a band of
colors on the wall. By using a lens in conjunction with the prism he
was able to spread the colors out into a fairly pure spectrum 10 in.
long. He fell short of producing a spectroscope of the modern type
only because he let the light shine through a round hole instead of a
narrow slit. It was not until 1802 that W. H. Wollaston, and in 1814
Joseph Fraunhofer, independently observed spectrum lines, that is,
images of a narrow slit each containing only light of one color. The
first practical spectroscope was developed by G. R. Kirchhoff and
R. Bunsen in 1859.
Newton is responsible for the practical application of the prism and
Fraunhofer for that of the diffraction grating; these are the basic
components used in spectroscopes today to separate the wavelengths
of light. Kirchhoff and Bunsen showed that the spectroscope could
?1.2]
SPECTROSCOPY AS A SCIENTIFIC TOOL
TABLE 1.1
T H E SPECTROSCOPIC PART OP THE ELECTROMAGNETIC SPECTRUM
freWave
quency number
m sec'^ in cm~^
'
Wavelength
Characteristics oj radiation
Natural Laboratory
origin
sources
Prism
materials
cm
n AU
Detectors
10'
. 10-'
10-»
IQt
- 40-«
10-* 10»
10"-
IC* -I-
Inner
electrons
in
atoms
High
voltage
spark
10'»-
Kfi
-'
-•
10"
10'
Outer
electrons
in atoms
and
molecules
10-'
10'
-10-'
1
10»
-10-»
10'
10*
10-2. IQ!
ly.
10-'
IV
Arc and
spark;
gas discharge
10"
10^ - •
Molecular vibrations
i
10"-
Thermal
102
radiation
Only gratings
are suitable
WMolecular rotations
tO"
Microwave
generators
Microwave
receivers
10»
4
SPECTROSCOPY AS A SCIENTIFIC TOOL
[§1.3
be used as a new means of qualitative chemical analysis; with it they
discovered several new elements and were able to demonstrate the
presence of many known elements in the sun. They are in a very real
sense the founders of modern spectroscopy.
1.3. Measurement of the Spectrum. The waves with which we
are here concerned have lengths lying between 1 mm and 10~* mm,
which can be measured with a precision varying from one part in ten
thousand to one part in sixty million, depending on the spectral region
involved. Various systems of units have been developed in which to
record wavelengths conveniently; of these the following are the most
common:
1 n (micron) = 10~* cm = 10^' mm
1 mju (millimicron) = 10~'cm = IO~^mm
1 A (angstrom*) = 10~' cm = 10~' mm
I fi = 10,000 A = 1000 m^i
In the infrared region wavelengths are commonly measured in
microns, and in the range shorter than 1 /i in angstroms. Chemists
and biologists frequently use the millimicron. The mean wavelength
of the strong yellow light emitted by sodium atoms is, in the three
systems, 0.5893 ix, 589.3 mju, and 5893 A. '
*"
Spectroscopes analyze radiation in accordance with its wavelengths,
but atoms and molecules emit radiation of characteristic frequencies.
In a sense frequency is more fundamental than wavelength, for the
frequency of monochromatic light remains constant no matter in what
medium it may be traveling, whereas the wavelength varies inversely
with the velocity of light in the medium. Therefore, in addition to
the wavelength X of a beam of light, it is often useful to specify the
frequency of oscillation v. This is related to X by the formula
Xi' = C(m), where C(„) is the velocity of light in the medium. Frequencies in the optical range are, however, very large numbers
(4 to 7.5 X lO^*), and it is more convenient to use a smaller number,
the wave number a, which is the number of waves per centimeter of
path in vacuum. X and o- are related by the formula 'Ka = 10' when
X is expressed in angstroms, cr is then expressed in reciprocal centimetersi written cm"'.
^
1.4. The Infrared Spectrum. Sir William Herschel in 1800 used a
simple thermometer to measure the heating power of the various
' For a more exact definition of the angstrom see § 9.8;
§1.4]
SPECTROSCOPY AS A SCIENTIFIC TOOL
5
colors in the spectrum of sunlight. He found the greatest heating
effect entirely outside the visible portion, just beyond the red edge of
the spectrum. Thus he discovered the infrared region. His son,
Sir John Herschel, in 1840 moistened a piece of blackened paper with
alcohol and found that the alcohol evaporated faster in certain places
than in others when the infrared part of the solar spectrum was
allowed to fall on it, revealing in this way the presence of infrared
bands of absorption and transmission.
Infrared radiation can be dispersed into a spectrum by means of
coarse diffraction gratings or prisms of special material such as rock
salt, which is transparent to much longer waves than is glass, and the
rays can then be detected with a device sensitive to small heating
effects. Bolometers, thermocouples, and various other radiometers
can be used for this purpose. Globar heaters and other incandescent
sources emit infrared radiation profusely. The methods and apparatus used for infrared spectroscopy are discussed in detail in
Chapter 17.
, The infrared region, as has been indicated, is conveniently divided
into the near infrared, which extends from the edge of the visible up to
about 25 n (250,000 A) and the far infrared, which extends from the
near infrared to 1 mm (1000 fi or 10 million angstroms). The distinction between regions is usually made on the basis of the type of
spectrograph used, since prisms have not been available which transmit much beyond 25 ^l. However, the two spectral regions also
correspond roughly to those which contain the vibrational frequencies
of light molecules (near infrared) and their rotational frequencies (far
infrared). A further subdivision of the near infrared is sometimes
made when spectroscopists distinguish a region called the photographic
or photoelectric infrared, because radiation of wavelengths up to the
neighborhood of 2 M can be detected photographically and photoelectrically. Radiation in this region penetrates atmospheric haze
better than do the shorter waves of visible light, and so infrared photography is useful when long distances are involved in camera work.
The invisibility of these waves also makes them of importance in
military signaling. Contrary to a widespread impression, however,
infrared radiation in the photographic and photoelectric region will
not penetrate fog or mist to a significantly greater extent than will
visible radiation.
The near infrared spectrum has assumed great importance in
chemical and biological research because of the highly specific absorp-
6
SPECTROSCOPY AS A SCIENTIFIC TOOL
[§1.5
tion of chemical compounds at these wavelengths. The near infrared
absorption spectrum of a molecule has aptly been called the molecule's
fingerprint. Recent development of good commercial infrared prism
spectrometers has done much to make the fingerprinting process more
widely usable.
1.5. The Visible Spectrum. The spectral sensitivity of the eye of
a typical observer is shown in Fig. 1.1. The actual limits of sensitivity at the two ends differ somewhat from observer to observer,
1.00
4000
5000
6000
7000
Wavelength in Angstroms
Fig. 1.1. Spectral sensitivity of the human eye.
some people being able to see slightly farther into the infrared and
others farther into the ultraviolet. The visible region is usually
arbitrarily set between the limits 4000 A and 7500 A. The fact that
rays in this band can be seen makes possible visual spectroscopy and
the entire science of colorimetry. The eye is an excellent detector of
visible radiation and is a moderately satisfactory comparison device,
but it is not a good quantitative measuring instrument. Its use in
spectroscopy is discussed in Chapters 12 and 14.
In addition to affecting the optic nerve, waves in the visible part of
the spectrum are characteyized by their ability to pass through glasses
of various types that are readily obtainable, which can be fashioned
into prisms and lenses for the production of optical equipment.
Visible radiation can be photographed, though special sensitization of
the photographic emulsion is required for waves longer than about
5000 A, in the blue-green.
1.6. The Ultraviolet.. The ultraviolet region, which begins at ap-
§1.7]
SPECTROSCOPY AS A SCIENTIFIC TOOL
7
proximately 4000 A, was discovered in 1801 by J. W. Ritter. In his
studies of the relative efficacies of rays in different portions of the
spectrum in blackening silver chloride, he found that the most active
rays lie beyond the violet.
Many minerals and organic materials fluoresce strongly in ultraviolet light, converting the invisible radiation into visible light.
Ultraviolet light also causes numerous photochemical reactions, and
in the range of waves shorter than 2900 A it is markedly bactericidal.
Photoelectric effects are particularly pronounced in the ultraviolet.
Since ordinary glass is not transparent to ultraviolet radiation,
optical parts of quartz, fluorite, rock salt, or special modern glasses
are used. Water is highly transparent to ultraviolet waves longer
than 1900 A. The region is of great importance in absorption spectrophotometry and in the analysis of materials by the emission spectrum,
and forms one of the richest and most productive regions of the entire
spectrum.
The frequently used division into near and far ultraviolet is somewhat artificial. It arises from the fact that the solar spectrum is cut
oS below 2900 A as a result of absorption by the ozone layer in the
atmosphere. Beyond the far ultraviolet, that is, below 2000 A, lies
the extreme or vacuum ultraviolet, which is the region of radiation
absorbed markedly by the oxygen and water vapor in the air. Hydrogen and helium, and to a much lesser degree nitrogen, are transparent to these shorter waves. Victor Schumann in 1893 used
optical parts of fluorite, pumped the air out of his spectrograph, and
eliminated most of the gelatin from the photographic emulsions he
used. By these three measures he was able to extend the spectrum
from 2000 down to 1250 A. Lithium fluoride has since been found
to be somewhat transparent down to nearly 1000 A, but below this
wavelength no solid material has been discovered from which prisms
and lenses can be constructed. Diffraction gratings mounted in a
vacuum can be used to study the spectrum at shorter wavelengths.
In the hands of Lyman, of Millikan and his collaborators, and of
Siegbahn and his associates the spectrum has been studied with
gratings to 10 A, where the methods of X-ray spectroscopy become
more effective. Hence the extreme ultraviolet, discussed in Chapter 19,
covers the range between 2000 and 10 A.
1.7. Spectroscopes and Spectrographs. Any instrument that can
be used to produce a spectrum, visible or invisible, is called a spectroscope. Under this general heading instruments are classified accord-
^8
SPECTROSCOPY AS A SCIENTIFIC TOOL
[§1.7
ing to the means by which the spectrum is observed. A spectrograph
produces a photographic record of the spectrum called a spectrogram.
The word spectroscope is sometimes used in a restricted sense to
designate an instrument arranged so that the spectrum can be viewed
by eye. I t will be used in this book only in the broad sense; the
term visual spectroscope will be used to designate instruments arranged
for direct eye observation of the spectrum. Spectrometers are so
built that an observer can determine wavelengths by reading a scale,
which may or may not be calibrated to read directly in microns,
millimicrons, or angstroms.
Most spectroscopes contain three main elements: a slit; a dispersing device such as a prism or a diffraction grating to separate radiation
Fig. 1.2. Optical system of a simple spectroscope. S, slit; C, collimator lens;
P, prism; T, telescope lens; F, curve along which the various parts of the spectrum are in focus; B, blue or short wavelength part; R, red or long wavelength
part.
according to wavelength; and a suitable optical system to produce
the spectrum lines, which are monochromatic images of the slit. A
simple spectroscope optical" system is shown in Fig. 1.2. The spectrum lines are arrayed along a focal curve where, they may "be photographed, observed with an eyepiece if visible,, or isolated from their
neighbors by a second slit. The first method is used in spectrographs,
the second in visual spectroscopes, and the third in monochromators.
Spectrum lines are detected or recorded by various means. Infrared spectroscopes are usually equipped with radiometers, which produce variations in current through a galvanometer and hence vary its
deflection. These variations of deflection may ;be recorded in curves
of the type, shown in Fig. 1.3. The spectrum can be recorded by this
means at any wavelength, but more sensitive methods are used in
SPECTROSCOPY AS A S C I E N T I F I C TOOL
§1.7]
Deflection •
of
Galvanometer
Wavelength
Fig 1.3.
»—
Record of galvanometer deflections produced by an infrared spectrometer. See also Fig. 11.7.
(b)
J
IA-^IMA
4150
.
60
Li J.
lk.Mh
70
80
90
u JLAA-
4200
_JdlluL
10
20
-M.
30
(c)
UMH
MMWIMHM
i^
UMH
HWCWL
MMII*
m[uMmk
Fig. 1.4. Photographic and photoelectric records of the same spectrum.
(a) Photograph of the' spectrum of iron in the violet region; (b) photoelectric
record of the same spectrum; (c) densitometer record of the spectrogram shown
in (a). (Courtesy Prof. G. H. Dieke.)
10
SPECTROSCOPY AS A SCIENTIFIC TOOL
[§1.8
spectral regions where they are available. Photography is feasible
between 15,000 and 10 A. Though sensitive and convenient, photography requires careful control if quantitative results are to be obtained. Fluorescence and phosphorescence methods, combined with
visual observation or photography, can also be used between 15,000
and 10 A, with some loss in sharpness of narrow lines. Photoelectric
recording has been used between 33,000 A and the short vacuum
ultraviolet. In all these cases the 10 A limit is purely arbitrary, since
the sensitivity extends on into the region of X-ray spectroscopy.
Records obtained by means of the two principal detection and recording methods are shown in Fig. 1.4.
TECHNIQUES OF SPECTROSCOPY
In using the techniques of spectroscopy, one is concerned eitk^r
with studying the wavelengths and intensities of the radiations
emitted by atoms and molecules under various physical conditions or
with the radiations absorbed on passing through matter Jn various
forms. Thus it is useful to distinguish hetvfeen emission spectroscopy
and absorption spectroscopy.
- •
1.8. Emission Spectroscopy. Three kinds of emission spectra can
be distinguished, known respectively as line, band, and continuous
spectra. Typical examples of each of these are shown in Fig. 1.5.
Line spectra originate from atoms or atomic ions which are separated
^9 «o
4S
:
h^-iH^ 1 ! 1 i i ! I h !! i 11
Ja)
(b)
[c)
'.
1II9IK
!'
. ilillli I H I
Fig. 1.5. Typical emission spectra taken with a low-dispersion spectrograph.
(a) Continuous spectrum of an incandescent filament; (b) line spectrum of the
iron arc; (c) band spectrum of molecular, nitrogen XNz).
by such distances from their neighbors that between collisions they
can radiate as individuals. Hence line spectra are obtained from
incandescent gases and vapors.
Band spectra originate from molecules composed of two or more
atoms, either ionized or un-ionized, when these molecules are sufficiently separated to be fairly independent of their neighbors. Band
spectra are emitted from polyatomic incandescent gases and vapors
§1.9]
SPECTROSCOPY AS A SCIENTIFIC TOOL
11
which are cool enough that not all the molecules are dissociated into
atoms and ions. Many band spectra, such as those emitted by the
hydrogen molecule H2, have the appearance of line spectra, because
the individual lines in the bands are of wide separation, and the intermingling of neighboring band lines is extensive.
Continuous spectra result when light is radiated from incandescent
solids or liquids, or under certain special circumstances from individual atoms or molecules. A continuous spectrum may be regarded as
the equivalent of an infinite number of spectrum lines forming a dense
array of overlapping monochromatic images of the slit.
When the light from an electric arc is sent through a spectroscope,
all three types of spectra—line, band, and continuous—are likely to be
observed together, since the arc stream contains atoms, ions, molecules, and gross incandescent particles.
1.9. Qualitative Spectroscopic Analysis of Materials. Each type
of atom or molecule can be made to produce a characteristic set of
spectrum lines or bands which serve to indicate the presence of that
atom or molecule as a radiating center, whether in a sample of metal
in the laboratory or in a star or nebula. More than half a million
different atomic spectrum lines and countless bands have been observed. Most of the more intense atomic lines have been assigned to
their atom or ion of origin. Since it is possible to measure their wavelengths to a precision of 1 part in several million if necessary, most
spectrum lines can be identified as to parent atom without possibility
of error; and since each atom emits many characteristic spectrum
lines, the presence or absence of a particular type of atom can be
determined quite readily by spectroscopic means.
The spectroscope provides one of the most highly specific of all
methods of qualitative analysis; it is in addition direct, rapid, and
simple. A small piece of the material to be analyzed can be burned
in an electric arc or spark and its spectrum recorded within a few
seconds. A simple inspection of the resulting pattern of spectrum
lines serves to identify the presence or absence of some 70 of the
chemical elements. All metallic elements present are revealed in a
single operation without requiring a guess on the part of the operator
as to which will be found.
The emission method is not directly applicable to the detection of
molecules except in certain special instances, because most molecules
are dissociated in the electric arc or spark. Nor does the method
detect negative radicals. Such elements as sulfur, selenium, the
12
SPECTROSCOPY AS A SCIENTIFIC TOOL
[§1.10
halogens, and the gases require special spectroscopic techniques, which
are frequently more complicated than alternative methods of chemical
detection. Analytical techniques based on emission spectra are
discussed in Chapters 15 and 16.
1.10. Quantitative Spectroscopic Analysis. At very low concentrations of an element in a sample, the amount of light emitted by
that element is directly proportional to the number of its atoms
present, if all other factors are kept constant. This linearity provides
a very convenient basis for quantitative analysis by the emission
spectrum. The number of factors other than concentration that
affect intensity is great, however, so that only a null method of analysis is satisfactory. A sample can be analyzed if one duplicates it '
fairly closely with a mixture of known content, which when burned in
the arc emits lines of similar relative intensities. This procedure is
not difficult when concentrations of 1 per cent or less of each important
element are involved, because then no constituent in a sample influences the light emission of the others, and it is possible to determine
simultaneously the concentration of a large number of. impurities in
a given sample.
The spectrographic method can be applied quantitatively to the
determination of any element that can be detected qualitatively. As
a result more than 70 elements of the periodic table are susceptible to
a method that is much more rapid than chemical wet methods and
can be carried out on much smaller samples, 10 mg usually being
sufficient for a determination. The method is also extremely sensitive, being effective in some cases down to concentrations of 1 part
in 100 million.
Spectrographic quantitative analysis provides fairly uniform precision at all concentrations. Thus it is as easy to measdre the difference
between 0.0010 and 0.0011 per cent content of an impurity as that between 1.00 and 1.10 per cent. At low concentrations the precision
of the spectrographic method is superior t o ' that of chemical wet
methods, but it becomes inferior at concentrations of about 5 per cent
and over. At concentrations below 5 per cent it is possible to reduce
the average deviation among successive determinations on the same
sample to less than 2 per cent.
Analysis by emission spectra, commonly called spectrochemical analysis, is now widely used in industry, especially for the analysis of impurities in metals, for the determination of constituents in alloys, and for
the examination and testing of biological, medical, and food products.
§1-11]
SPECTROSCOPY AS A SCIENTIFIC TOOL
13
l . l l . Absorption Spectroscopy. When a beam of light passes
through a piece of colored glass, certain wavelengths are reduced in
intensity by absorption. Even glass that appears colorless will show
absorption in the infrared and ultraviolet regions. Pure liquids and
solids in solution exhibit similar absorption. Each band of wavelengths removed by a solid or liquid is usually fairly wide and may
extend over many hundreds of angstroms, as shown in Fig. 1.6.
The spectroscopic study of the absorption of radiation has three
broad objectives: to learn which wavelengths of radiation are absorbed; to learn how much radiation is absorbed under specific conditions; and to learn tvhy the radiation is absorbed. The first of these
objectives is of value because it furnishes information that serves as a
basis for the qualitative analysis of the absorbing material, as in the
"fingerprinting" of chemical substances by their infrared absorption
Fig. 1.6. Absorption spectrum of tlie cytosine molecule in the solid state at
liquid-hydrogen temperature.
spectra. Such information is also useful for the production of radiation filters required for the removal of certain waves from a beam of
light, for example in the removal of infrared rays by a heat filter from
the beam in a motion-picture projector. The second objective permits extending chemical analysis, by means of absorption, from the
qualitative to the quantitative. The use of ultraviolet absorption
spectra, for instance, is widespread in the quantitative analysis of
vitamins and of many other important substances that are difficult to
analyze by other means. In the achievement of the third objective,
one seeks to understand the absorption of radiation in terms of the
atoms and molecules responsible for it. No feature of the absorption
spectrum can be overlooked in this process, and in the course of
14
SPECTROSCOPY AS A SCIENTIFIC TOOL
[§1.12
understanding details of the spectrum, one can obtain detailed knowledge of how atoms are held together in molecules. Thilk knowledge
is of great value to the physicist and the chemist.
\
1.12. Absorption Spectrophotometry. Absorption spectrophotometry is quantitative absorption spectroscopy. It can be used for the
quantitative determination of organic molecules, for example in the
estimation of dye concentrations and in the analysis of vitamins,
hormones, and other complex organic molecules. By means of an
absorption photograph of the type shown in Fig. 1.6, a curve may be
determined as shown in Fig. 1.7, which gives the relative absorption
of light at various wavelengths
through a given thickness of the
sample. Absorption can be measured by means of a spectrophotometer. Once the specific absorption at all wavelengths is known
for a material, the resulting curve
enables one to determine the
actual amounts of absorbing
material present. Thus, though
the absorption method is not
highly specific for qualitative an3100 3000 2900 Z800 2700 2600 2500
2400
alysis, it is extremely precise in a
Wavelength in A
quantitative sense and can be
Fig. 1.7. Absorption curve deter- made* even more sensitive than
mined from the spectrogram shown in emission analysis jn certain cases.
Fig. 1.6, with wavelength scale reAbsorption
spectrpphotometry
versed.
forms the subject of Chapter 14.
1.13. Fluorescence Spectroscopy., A branch of spectroscopy capable of much further development is that involving the use of fluorescence. Fluorescence spectroscopy is widely applied in mineralogy,
biochemistry, biology, medicine, and the food industries.
When one uses fluorescence spectroscopy, the object to be studied,
for example a mineral specimen, is shielded from extraneous light and
is then illuminated with ultraviolet light, usually from a quartz mercury lamp covered with a filter that removes visible radiation. The
specimen is likely to glow brightly; if it does, its type can often be
determined from the color of its fluorescence.
Many organic materials fluoresce. Certain species of bacteria have
characteristic fluorescence, and different strains of the same species
SI.14]
SPECTROSCOPY AS A SCIENTIFIC TOOL
15
may show different shades of color. Molds and bacterial growth on
meat samples can be detected, and in some cases identified, by means
of their fluorescence. Various parts of plant or animal cells often
fluoresce with different colors, so details that cannot be seen in ordinary light are sometimes revealed by a fluorescence microscope.
Mineral oil has a characteristic blue fluorescence not found in most
organic oils, and it is thus possible to detect contamination or dilution
of organic oils by mineral oil. Oleomargarine can be detected in
butter by means of fluorescence; as little as 5 per cent of artificial fat
in butter can be shown up in the same way. Flour of one type may
show a bluish fluorescence, whereas another glows white and a third
exhibits a pinkish glow. Many other food products can be tested for
quality by fluorescence analysis. Real and artificial gems can also be
distinguished by this means. The spectroscope makes fluorescence
analysis more specific than does visual observation of fluorescence.
USES OF SPECTROSCOPY
A complete catalogue of the uses of spectroscopy would be too
lengthy for the present chapter, but we may list (a) the study of the
absorption and emission of light by matter in all forms; (b) the
analysis of the atomic and molecular varieties present in a given sample of matter and determination of their relative numbers; (c) the
investigation of the structures of atoms and molecules, and (d) the
determination of the size, mass, temperature, speed of motion, and
many other characteristics of the heavenly bodies. Spectroscopy has
thus contributed materially to all the natural sciences, particularly
to astronomy, physics, chemistry, and biology.
1.14. The Spectroscope in Chemistry. When the spectroscope
Was first developed in practical form (1859), it was used immediately
by chemists as a powerful tool for qualitative analysis. As a byproduct of this use came the discovery of many chemical elements,
among them cesium and rubidium by Bunsen and Kirchhoff, and
later helium, gallium, indium, and thallium by various other chemists.
In addition, spectroscopic analysis was such a powerful aid in the
separation of the various'rare-earth elements that the discovery of
many of these may properly be credited to the spectroscope. The
spectroscopically discovered elements are listed in Table 1.2. In
later years the same application of spectroscopic methods, grown
more powerful with the passage of time, led to the discovery of rare
16
SPECTROSCOPY AS A SCIENTIFIC TOOL
[§1.14
TABLE 1.2
LIST OF CHEMICAL ELEMENTS DISCOVERED SPBCTBOSCOPICALLY
Element
Cesium
Rubidium
Thallium
Indium
Helium
Gallium
Thulium
Praseodymium
Neodymium
Samarium
HoJmium
Ytterbium
Lutecium
Discoverer
Date
Bunsen and Kirchhoff
Bunsen and KirchboB
Crookes
Beich and Richter
Lockyer
de Boisbaudran
Cleve
von Welsbach
von Welsbach
de Boisbaudran
de Boisbaudran
Urbain; von Welsbach
Urbain; von Welsbach
1860
1861
1861
1863
1868
1875
1879
1885
1885
1886
1886
1907
1907
isotopes of t h e common elements hydrogen (Urey, 1932), carbon,
nitrogen, and oxygen. I m p r o v e m e n t s in techniques of measuring
emission and absorption intensities also p e r m i t t e d t h e extension of'
these methods to q u a n t i t a t i v e chemical analysis, as has been mentioned above. T h u s t h e spectroscope, in one form or another, h a s
jjsp
31
ZZ 33 34 35 36 37 38 39 40
45.'
50
55 60 65
|iiiJ)tiil(i)ilit(i!iiiihMiimifiiii)i,;i)wi!tii«i(Sj!itoitoili**«ij i l tTt't D i lrlj"iffilihli!<iil(„M!!,.i>i,ai;iiilllB
ir
nil
, i!
it
; K
I
"i ! 1 •
•r
1 1 1 : -! ;.1i-:.,---
.
:L,J
Fig. 1.8. Emission spectra of the alkali metals. Top to bottom: lithium,
sodium, potassium, rubidium, cesium.
become a foremost instrument in all branches of chemistry because of
its analytical power.
T h e contributions of spectroscopy t o chemistry are b y no means
limited t o the field of analysis, however, because fio chemical substance can emit or absorb radiation without revealing much a b o u t its
fundamental n a t u r e . By study of t h e emission of radiation from
§1.14]
SPECTROSCOPY AS A SCIENTIFIC TOOL
1-7
isolated atoms and ions, enough information about the electronic
structure of atoms has been obtained to explain completely the
arrangement of the chemical elements in the periodic table (see
Chapter 10). Similarly, the structures of many molecules, including
some of great complexity, have been revealed by their spectra. The
chemist has profited both in the determination of the actual geometry
of molecules whose structural formulas were previously known and
in the elucidation of unknown structural formulas. The structures of
such molecules as penicillin and vitamin K, for example, were worked
out with extensive help from the spectroscope.
The chemist is somewhat more interested in the reactions that molecules undergo than in their structures while at rest between reactions.
Since both the speed of a reaction and the extent to which it takes
place are dependent on the forces between atoms, that is, the chemical
bond forces, any information about these is of chemical importance.
The contribution of spectroscopy here has been a double one. The
actual strengths of the chemical bonds (dissociation energies) have
been nieasured spectroscopically for many diatomic molecules, frequently with an accuracy that could not be attained by other means.
In addition, spectroscopic data have been used to make highly precise
calculations of chemical equilibrium constants, which determine how
far reactions proceed. For example, the equilibrium of the reaction
of hydrogen and chlorine to form hydrogen chloride gas at any temperature up to 5000°C can be calculated entirely from spectroscopic
data, with precision far greater than that with which it can be determined otherwise.
One branch of chemistry that should be expected to make extensive
use of spectroscopy is the highly complex field of photochemistry.
The existence and nature of molecular fragments which cannot be
chemically isolated but which are important links in a chain of steps
making up photochemical reactions have been demonstrated spectroscopically. As examples of such fragmentary molecules one can
cite scores of diatomic hydrides, such as the OH radical, which have
only a fleeting existence in an arc or flame or in the high-temperature
areas of a reaction vessel, but whose properties, such as interatomic
distances, vibration and rotation frequencies, and electronic states,
have been determined spectroscopically in minute detail. At present,
photosynthesis, the great problem of photochemistry, is being
attacked in many ways, but the spectroscope is never absent from the
laboratories of those concerned with the problem.
18
SPECTROSCOPY AS A SCIENTIFIC TOOL
[§1.14
li^ntaiMiWHBfWaM
1
K^*"iiJi'i • -w*.
I
J2
o
.-
^•U..-.:^
.
--1.'
- > . ! •
3
O
'3m::ssa?*i--" u
_;- - ' ~--—"
:
: • - .
o
ii^Mb j , A »
F
3
-
is
....— "'
r
<D
C><
U3
.. .
-.A
!L..
3
I . -,-^
C!
a,
<
F^'-^.
•tir"HlLfiti'
•.
- » & * * * » « ! • •
«.
§1.15]
SPECTROSCOPY AS A SCIENTIFIC TOOL
19
T h e kinds of spectroscopy in which chemists are most interested a t
present are quaHtative and q u a n t i t a t i v e emission spectroscopy, ultraviolet and infrared absorption spectrophotometry, and R a m a n spectroscopy. These subjects and some chemical problems to which t h e y
can be applied are discussed in C h a p t e r s 14 t h r o u g h 19.
1.15. Spectroscopy in Astronomy. Modern astronomers h a v e
completely refuted the dictum of Auguste Comte, " T h e r e are some
things of which the h u m a n race must remain forever in ignorance, for
example, the chemical constitution of the heavenly bodies." By
using the spectroscope, astronomers have been able t o make qualitative analyses of m a n y of the stars and also a q u a n t i t a t i v e analysis of
Fig. 1.9c.
Spectrum of the star 0 Aurigae, showing Doppler effect.
(Courtesy Harvard Observatory.)
the surface of the sun t h a t is perhaps more complete t h a n a n y yet
made of the earth. Astrophysicists have t h u s been able to demonstrate t h a t many of the same chemical elements which we find on
e a r t h compose the sun, stars, and the most d i s t a n t nebulae. Sixtysix elements have been found spectroscopically in the sun. A n u m b e r
of spectrum lines, which were for m a n y years unidenti6ed, are found
in the spectra of t h e solar corona and of various nebulae; these were
once thought to arise from atoms not found on earth. I t has since
been proved thp-t every one of these lines originates from a familiar
a t o m found in the stars in some special s t a t e of excitation n o t ordinarily produced in the laboratory.
A large proportion of the information t h a t the astronomer possesses
20
SPECTROSCOPY AS A SCIENTIFIC TOOL
[§1-16
regarding the constitution of the heavenly bodies has come via
spectroscopy. Spectra obtained when astronomers collect light from
a star with a large telescope and then separate it by means of an
attached spectroscope serve for the detection and estimation of
molecules in the stars. The Doppler effect of spectral lines can be
used to measure the approach and recession velocities of stars and
nebulae (see Fig. 1.9c) and the rotation of the sun; a spectral line
from a calcium atom, for example, is shifted toward shorter wavelengths when emitted by a star that is approaching and toward longer
wavelengths if from one that is receding.
The spectroscope has been used to measure the sun's motion among
the stars and the distances of hundreds of individual stars, and it has
revealed the double character of many stars that are too close together to be resolved as double by the telescope.
1.16. Spectroscopy in Physics. In physics the trail of spectroscopic discoveries is longer than in any other science. Spectroscopic
data give the most precise standards of length, and it has been suggested that the standard meter bar might well be replaced as a
standard of length by the wavelength of a sharp red line emitted by
cadmium atoms, or of a sharp green line emitted by mercury atoms of
the isotope 198. The lengths of such waves may well be expected to
remain more constant over the passing ages than that of any manmade standard.
Most of our precise information about the electronic structures of
atoms has come via spectroscopy. The physicist is interested in
studying the production of light iij a magnetic field, and through the
Zeeman and related effects he has been able to determine the quantum
numbers and the locations of the electrons in the various kinds of
atoms. Many apparently sharp lines can be resolved into a number of
still sharper monochromatic lines packed close together. ' This socalled hyperfine structure reveals information regarding the spin of
the nucleus of the emitting atom. Magnetic susceptibilities and the
electronic configurations of some atoms in the solid state can also be
determined by spectroscopic means.
Much of the attention of the spectroscopist has been given to
determining electronic energy levels in atoms. I t has been found
possible to remove electrons one by one from atoms and to study the
new array of spectrum lines that results each time another electron
is removedi Thus normal uranium atoms, with 92 electrons around
the nucleus of each, emit many thousands of spectrum lines when
§1.17]
SPECTROSCOPY AS A SCIENTIFIC TOOL
21
excited. When an electron is removed from such an atom, an entirely
new array of thousands of spectrum lines will result on excitation.
When two electrons are removed, still another new spectrum can be
produced. Thus it is theoretically possible to have 92 different
characteristic spectra of uranium. Analogous series of ionization
spectra may be obtained with the other chemical elements, so that
more than 4000 different kinds of atoms and atomic ions can be
expected to emit independent spectra, and many millions of atomic
spectrum lines will thus be produced. To date only some 350,000
of these spectrum lines have been assigned to their parent atoms,
and a much smaller number have been classified in terms of the
energy levels and the state of ionization of the atom from which
they originate.
1.17. Spectroscopy in Biology and Medicine. Qualitative and
quantitative analyses made by means of the emission spectrum have
been employed extensively for determining the presence of metallic
elements in biochemical substances, cells, and tissues. Such studies
provide the biologist with a means of determining the various trace
elements needed by cells for life and growth. In medicine, their
greatest application has been to the investigation of toxicological
problems, as for example, in determining the accumulation of lead,
copper, and so on, in the blood and tissues of persons exposed to 'such
substances in the course of their daily work. Attempts have also
been made to apply emission spectroscopy to diagnostic procedures
in certain pathological conditions characterized by changes in the
heavy-metal content of the body fluids.
Biochemists and biologists are confronted by unusually difficult
problems of molecular structure, in which the molecules involved are
large, complex, and often ill-defined entities. Infrared and Raman
spectroscopy, and visible and ultraviolet absorption spectrophotometry, have been invaluable in the study of such problems. Such
spectroscopic methods have, for example, contributed to the elucidation of the structures of many of the known vitamins, enzymes, and
coenzymes.
Absorption spectrophotometry, in particular, is useful in controlling
the preliminary isolation of biochemical substances and in their later
characterization and qualitative or quantitative assay after they have
been isolated and identified. The characteristic absorption band of
vitamin A, with a maximum at 3280 A, is used, for example, in the
quantitative determination of this substance and in tests of the purity
22
SPECTROSCOPY AS A SCIENTIFIC TOOL
[§1-18
of vitamin A preparations. The kinetics of enzyme activity are
readily studied by absorption spectrophotometric methods, both in
the visible and ultraviolet regions.
!
Fluorescence spectroscopy is similarly useful in the isolation and
identification of those biochemical substances which exhibit characteristic fluorescence. One of the most striking examples of the
application of this technique has been the identification of a coal-tar
constituent that is responsible for causing "coal-tar cancer," as a
complex polycyclic hydrocarbon, 3 : 4-benzpyrene. Fluorescence is
employed routinely in the quantitative analysis of various biochemical substances, such as vitamins Bi and B2.
Microabsorption spectrophotometry in the ultraviolet is finding
increasing application in the study of the distribution of substances
like nucleic acids in cells and tissues in relation to fundamental biological and medical problems, such as those of cell division, growth,
differentiation, normal physiology, and pathology. The use of this
technique in the visible region (in connection with specific color reactions) and in the infrared promises to lead to the solution of many
hitherto abstruse biological and medical problems. Microfluorescence
spectroscopy, though not yet used on a wide scale, has similar applications.
Spectroscopy of the extreme ultraviolet has thus far been used but
little in the study of biochemical substances. It may be expected to
receive increasing application to the study of biochemical structure
problems involving saturated organic compounds.
1.18. Spectroscopy in Food Research. Spectroscopic methods are
being used increasingly in testing and controlling the production of
foods and their containers as advances in knowledge indicate the
importance of substances that may be beneficial or injurious in very
minute quantities. The great sensitivity of spectroscopic methods
and the smallness of the samples required for analysis are of considerable importance in food technology.
An example of the use of the spectrograph for determining small
but important concentrations of metals in foods is given by the measurement of copper in cranberries. The cranberry bogs in the Cape
Cod district had for many years been sprayed with copper sulfate
solution as an aid in pest control, and there was some fear that copper
salts might be accumulating in the soil to a degree that would result in
undesirably large quantities of copper in the cranberries. Spectrographic analysis was used to determine the concentration of copper in
§1.19]
SPECTROSCOPY
AS A S C I E N T I F I C
TOOL
23
soil samples, in various parts of the cranberry plant, and in the berries
themselves, with the result that it was found that the danger level had
not been remotely approached.
Lead determinations in condensed milk are another example of
control procedures. By means of the spectroscope, comparison runs
of different brands can easily be made whenever desired. The lead
content is usually found to lie between 3 and 20 parts per ten million.
Chocolate manufacturers who wish to know the metallic content of
Fig. 1.10.
m
Corner of an industrial spectroscopy laboratory.
inum Company of .\merica.)
(Courtesy Alum-
their product find the spectrograph indispensable. The metallic
content of beer kept in cans and in bottles is determined in a routine
manner with the spectrograph.
Absorption spectrophotometry is useful in testing comestible oils,
liquors, and other liquids. Absorption and fluorescence methods are
widely used in the food industries in the assay of foods for their
vitamin content.
1.19. Spectroscopy in Metallurgy and Mineralogy. For many
years archeologists were unable to solve the secret of the origin of the
I
24
SPECTflOSCOPY AS A SCIENTIFIC TOOL
[§1.U>
purple color of the gold sequins found on mummies in Egyptian
tombs. An eminent spectroscopist was called to their aid, and found
t h a t the color was due to certain impurities present naturally in some
samples of gold. By identifying these impurities he m a d e it possible
to fix definitely the place where the gold had been mined and t o
duplicate the purple color. Since only a minute a m o u n t of material
was available for analysis, the spectrograph was particularly useful;
not a single sequin had to be destroyed in order to give the answer.
In metallurgy t h e spectrograph is especially valuable because of t h e
speefl with which analyses can be made. T w o hundred tons of copper
Fig. 1.11. A spectroscopy laboratory of the Federal Bureau of Investigation.
(Courtesy F. B. I.)
ore may be smelted a t one time in a furnace, and it is of great importance to stop the smelting process a t the exact instant when the concentrations of various impurities are a t the optimum. A spectroscope
can be used to give the desired result within five minutes or less. For
this reason, spectroscopy has come into wide application in the
various metals industries, particularly in iron and steel foundries, in
copper smelting, and in the production of the more valuable metals.
Much steel is made by the addition of industrial scrap to new
-
V
' i\
§1.20]
SPECTROSCOPY AS A SCIENTIFIC TOOL
25
material obtained from iron ore. Over the years the percentage of
impurities introduced with scrap into American steel h a d slowly been
rising, until in the 1930's a level had been reached a t which the tin
content was becoming so high as to affect the rolling properties of t h e
steel. T h e tin concentration was still low enough, however, so
t h a t it could be determined readily only by means of spectrographic
analysis.
Minerals can be identified with t h e aid of the spectrograph, a n d the
more t h a n 2700 recognized varieties h a v e been analyzed for their constituent elements a t least partially by this means. T h e important
role t h a t fluorescence spectroscopy plays in mineralogy has already
been mentioned.
1.20. Forensic Spectroscopy. Most up-to-date police laboratories
use the spectrograph as an aid in criminal investigations. An expert
of the Massachusetts S t a t e Police has given the following as an example of the nature of the problems t h a t a police laboratory m a y be
called upon to solve: " I n the course of a typical day one m a y receive
a half dozen quart jars filled with h u m a n organs t o be analyzed for
small traces of poison, from another case a chisel having on its surface
an almost invisible a m o u n t of a foreign metal, while again one m a y
receive a pair of old trousers with traces of dust on the knees, or paint
from a car involved in a hit-and-run accident, or blood stains on a
knife, or a pellet of shot removed from a body, or ashes a n d debris
from an incendiary fire, or broken pieces of a burglar's tool, or residues
from the scene of an explosion. Unusual and varied subject m a t t e r
requires versatile methods a n d instruments. I n m a n y of these cases
spectrographic analysis can be advantageously applied a n d in a few
it is essential."
T h e principal advantages of using spectrographic techniques in
police work appear to result from the rapidity with which the results
of an analysis can be obtained, a n d from the fact t h a t the samples
to be studied are often very small. I n one case the lead pellets available from a shotgun homicide case weighed but 65 mg, yet 10 m g
sufficed for spectrographic analyses adequate for identification.
From the above examples, which h a v e been selected a t r a n d o m , it
is apparent t h a t spectroscopic methods m a y be applied to t h e solution
of many diverse problems. T h e chapters t h a t follow contain detailed
information designed to aid in selection of the most suitable spectroscopic equipment for solving a n y specific problem, and in its effective
26
SPECTROSCOPY AS A SCIENTIFIC TOOL
[§1.20
GENERAL REFERENCES
1.1.
B,. A. Ssmyer, Experimental Spectroscopy. New York: Prentice-Hall,
Inc., 1944.
1.2.
"W. H. Tirode, Chemwal Spectroscopy. New York: John Wiley & Sons,
Inc., 1943.
1.3.
G. R. Harrison (Ed.),Proceedings of the M.I.T. Conferences on Spectroscopy and Its Applications. New York: John Wiley & Sons, Inc.,
5th, 6th, and 7th Conferences, 1938, 1939, 1940.
1.4.
T. R. P. Gibb, Jr., Optical Methods of Chemical Analysis. New York::
McGraw-Hill Book Company, Inc., 1942.
1.5.
S. Judd Lewis, Spectroscopy in Science and Industry. London: Adam
Hilger, Ltd., 1933.
1.6.
F. T-wyraan, Spectrochemical Analysis of Metals and Alloys. London:
Adam Hilger, Ltd., 1941.
1.7.
Walter Gerlach and Werner Gerlach, Clinical and Pathological Applications of Spectrum Analysis. London: Adam Hilger, Ltd., 1934.
1.8.
R. A. Morton, The Application of Absorption Spectrophotometry to the
Study of Vitamins, Hormones, ami Enzymes.
liOndoa: Adam
Hilger, Ltd., 1942.
1.9.
Jj.JleiluieyeT, Spectrophotometry in Medicine. London: Adam Hilger,
Ltd., 1943.
1.10. R. B. Barnes, R. C. Gore, U. Liddel, and V. Z. Williams, Infrared
Spectroscopy. New York: Reinhold Publishing Corporation, 1944.
1.11. George Glockler, "The Raman Effect." Rev. Mod. Phys^lS,
111
(1943).
CHAPTER
2
Selection of Spectroscopic Instruments
SPECTROSCOPES ARE AVAILABLE IN MANY DIFFERENT TYPES,
and
selection of even an all-purpose instrument allows of a certain degree
of choice. To be considered in making such a choice are (a) the range
of wavelengths over which the spectroscope can be used, (b) the
extent to which it disperses light, (c) the variation of this dispersion
with wavelength, (d) the resolving power of the instrument, and
(e) the brightness of the spectrum that it produces. Important
secondary considerations are freedom from scattered light and false
lines, suitable shape and size of the spectrum lines produced, and ease
of adjustment. The relative bulkiness and portability, as well as the
availability of different instruments, will often influence the final
choice. The relative importance of the factors enumerated may well
determine whether a prism or a grating spectroscope will be chosen.
2.1. Dispersion. The dispersion of a spectroscope is a measuie of
the' way it distributes light in space according to wavelength. This
property can be expressed as angular or as linear dispersion. Angular
dispersion (dd/dX in Fig. 2.1) is fundamental. It depends on the
dispersing element used, and measures the variation with wavelength X
Fig. 2.\. Angular dispersion, dB/dX,'and linear dispersion, dl/dx.
of the angle of deviation d of the emergent light beam. Of more
frequent practical use is the linear dispersion dl/d\, which gives the
actual distance dl of separation in the spectrum of two close lines
differing in wavelength by d\. The reciprocal of this ratio, called
27
28
SELECTION OF SPECTROSCOPIC INSTRUMENTS
[§2.2
the plate factor or sometimes the reciprocal dispersion, is almost
always used in common practice t o measure the dispersion; t h u s
30 A / m m indicates a low dispersion, whereas 1 A / m m corresponds to
a relatively high value. T h e linear dispersion obtained with most
dispersing elements can be controlled, since it depends not only on
the angular dispersion b u t also on the distance between t h e dispersing
element P a n d t h e focal curve. T h e inclination of this curve t o t h e
optic axis of the o u t p u t side of the spectroscope also affects the linear
dispersion (Fig. 2.1).
2.2. Resolving Power. T h e resolving power Pr of a spectroscope is
defined as \/dX, dX being the wavelength interval between two close
lines of similar intensity t h a t can just
be resolved with the instrument a t
wavelength X. E a c h spectrum line is
an image of the slit. E v e n if t h e slit
is extremely narrow, its image is of
appreciable width, in the form of a
diffraction p a t t e r n consisting of a
bright central m a x i m u m on each side
of which are lesser maxima. T w o
Fig. 2.2. Intensity distribu- spectrum lines of. equal intensity are
tion of the diffraction patterns of considered as being just resolved
two spectrum lines of wave- -syhen the diffraction m a x i m u m of one
lengths Xi and X2, that are just re- « ,,
iU c j.
. . j- n
solved, plotted ks a function of fa"« «« ^^^ first m m i m u m of the
linear position, I in the spec- other.i T h e resolving powers of most
trum. The resolving power is spectroscopes lie between 5000 and
X/dX, and the linear dispersion is
dl/d\.
_
/UU,UUU.
Dispersion and resolving power are
often confused. Figure 2.2 illustrates their difference. Dispersion
determines the approximate position in the spectrum a t which light
of a given wavelength will fall; resolving power determines how well
t h a t light can be separated from light of other wavelengths falling
near by. Resolving power and dispersion are closely related, since
the resolving power of any spectroscopic device is equal to its dispersing power multiplied by its effective linear aperture A. I n a
prism spectroscope, A is as shown in Fig. 2.3. Resolving power
thus depends fundamentally on the material, shape, a n d quality of
the dispersing element. However, the actual power of an instrument
' Lord Rayleigh, "Wave Theory." Encyc. Britt, 9th ed., XXIV, (1888).
§2.4]
SELECTION OF SPECTROSCOPIC INSTRUMENTS
29
to resolve two close spectrum lines m a y be reduced by its auxiliary
parts. I n a visual spectroscope an inferior telescope lens or eyepiece
m a y reduce the resolution to a value below t h a t to be expected from
the resolving power of the prism or grating used. I n a spectrograph
the same loss m a y result from a t t e m p t i n g to photograph the spectrum
on a plate whose graininess is too coarse to separate t h e lines definitely.
2.3. The Dispersing Element. I n general, prism instruments are
more readily portable a n d somewhat more rugged t h a n grating instruments. T h e y suffer from the disadvantage t h a t their dispersion changes markedly
with wavelength; as a consequence, any
particular instrument is useful over only
a comparatively limited band of wavelengths. Their principal uses are for measuring simple emission a n d absorption
°
„
.-•
•
1 •
Fig. 2.3. Effective linear
spectra, for special purposes mvolvmg
aperture, 4 , of a prism for a
relatively high light-gathering power, and beam of radiation emerging
as small portable instruments.
^""""^ ^^^ ^^** ^^^^ ** ^"^
T-..fl. iI1
•
angletoa the perpendicular.
DiflFraction-gratmg spectroscopes give if / i^ the length of the exit
more uniform dispersion t h a n prism instru- face, A = / cos a.
ments, and a single grating can be used t o
cover a very wide spectral range. A good grating gives, in general,
greater dispersion and resolving power t h a n a prism of similar aperture
or cost. F o r t h a t reason grating spectrographs are widely used in research laboratories, especially for emission spectroscopy. T h e principal disadvantage of most diffraction-grating instruments is t h a t t h e y
suffer from astigmatism; this defect can, however, be eliminated b y
methods discussed below. T h e shortcomings sometimes n o t e d in t h e
past, t h a t gratings are fragile, wasteful of light, a n d produce so m a n y
false lines as to be useless for analytical purposes, have been elimin a t e d . Overlapping of different orders of grating spectra in certain
regions need cause little difficulty.
2.4. Dispersing Prisms. T h e dispersive action of a prism arises
from the variation with wavelength of the refractive index of t h e
material of which it is composed and from the angle between its
entrance a n d exit faces. I n regions of normal dispersion, rays of
short wavelength are bent more in passing through a prism t h a n are
those of longer wavelength. T h e variation with wavelength of t h e
refractive index of a suitable prism material, and hence the dispersion
of a prism formed from it, tends t o increase rapidly as an absorption
30
SELECTION OF SPECTROSCOPIC INSTRUMENTS
[§2.4
band is approached from either side. T h e usefulness of t h e material
therefore becomes greater from the standpoint of dispersion as it
becomes less from the standpoint of light transmission. This relation
is illustrated in Fig. 2.4, where curves showing t h e variation of refrac1.68
1
-Visible • — I R - *
1.66
1.64
\ D I inse flint glas 5
1.62
1.60
\
1.58
1
•
X 1.56
0}
"O
•S 1.54
a>
>yst ] |
3uar
Drdinary
^R
\s
I 1.52
a
ay
3Ck S
alt
^ Cr )wn cliass
^
"5 1.50
a:
\\
^'
\ ^ < ^
1.48
1.46
*
\
1.44
jylvir e
jsed quo tz
Ru xite
•
f.42
1.40
0
.2
.4
.e
.8
1.0
1.2
1.4
1.6
IB
2.0
2.2
2.4
2.6
2,8
3.0
Wavelength,^
Fig. 2.4. Refractive index as a function-of wavelength for
several prism materials.
t i v e index a n d a b s o r p t i o n ' w i t h wavelength are given for several of
t h e materials most commonly used in prisms. T h e dispersion a t
a n y wavelength is measured b y the slope of t h e refractive index curve,
a n d is greater as this is steeper.
T h e dispersive power of a prism is often 20 times as great a t one
end of its useful spectral range as at the other, with a resultant
equivalent variation in resolving power. Quartz prisms, though
t r a n s p a r e n t a t wavelengths longer t h a n 4500 A, h a v e diminished
usefulness there because of their low dispersion, a n d a t wavelengths
shorter t h a n 2000 A their transparency is low. M o s t glass prisms
are effective only in t h e spectral range 10,000 to 3500 A.
§2.4]
SELECTION OF SPECTROSCOPIC INSTRUMENTS
31
Modern dispersing prisms are usually cut with 60-deg refracting
angles. This choice is a compromise between smaller angles, which
^ive less dispersion, and larger angles, which require more material,
produce a greater loss of light, a n d result in decreased aperture. W i t h
material of low average refractive index, it would be advantageous to
use a prism of larger refracting angle t h a n 60 deg, b u t t h e 60-deg
compromise is close enough to t h e optimum in most cases t o m a k e
special angles unnecessary.
T h e angular dispersion d9/d\ of a prism is given approximately
b y the formula
do
dn % t a n i
' d\
d\
n
where 6 (the angle of deviation) is t h e angle between t h e rays incident
upon and emergent from the prism, n is the refractive index of t h e
prism material, and i is the angle of incidence of the ray on t h e prism,
as shown in Fig. 2.5. This formula is strictly true only for t h e case
of minimum deviation, in which the rays pass through t h e prism symmetrically, the incident and emergent rays making equal angles with
t h e normals to the prism faces a t which the rays enter a n d emerge.
T h e resolving power Pr, defined as \/d\, is given for a prism by
t h e formula
-p _ rpdn
1.22
d\
n
where T is the thickness of t h e prism base.
I t can be shown t h a t this value is equal to
t h e dispersion, as given above, times the
linear width A of the beam entering the
Fig. 2.5. Ray passing
-jgjjj
through a prisni at mini. . .
i.1 1
7 • •!! mum deviation, under
A prism is ordinarily shaped so t h a t it will which conditions Z i = Z rB,
a c c e p t a beam of circular cross section falling angle of deviation; T,
o n its front face at approximately t h e angle P"®"* thickness,
of minimum deviation. This condition leads to a s t a n d a r d set of
dimension ratios for a n y material, the length of face of a 60-deg prism
being for most substances roughly 1.6 times its height. F o r good
definition the prism height should be a t least three times as great as
t h e maximum length of slit t h a t is to be used with it, and preferably
t h e ratio should be even greater. Spectrum lines produced with
pcjsms are curved, a n d definition m a y be lost when the prism-slit
height ratio is too small.
32
SELECTION OF SPECTROSCOPIC
LNSTllUMIiXTS
[§2.5
Both dispersion and resolving power can be increased if a number
of prisms are used in train, or if a beam of light is sent through the
(a) 1 slit
(d) 5 slits
1^1
IH ly^H
• ^'^^H^^^^^l
(b) 2 slits
(e) 6 slits
(c) 3 slits
(f) 20 slits
Fig. 2.6. Diffraction patterns produced by various numbers of equidistant
slits illuminated by parallel Monochromatic light. (From 1-". A. Jenkins and
H. E. White, Fiindamcninis of Phyxirdl Opfirs, Mcnraw-Hill Book Company. Inc.,
New York (1937), page 147. Courtesy authors and publisher.)
same prism several times. Attempts to bring about a large increase
in these quantities by either means usually result in difficulties due to
scattered light or to undue loss of light by reflection and absorption,
long before resolving powers are reached that are equivalent to those
obtainable with diffraction gratings.
2.5. Diffraction Gratings. The diffraction grating lias been in use
for more than a hundred years, but its full potentialities have not yet
been realized on account of the difficulty of ruling and reproducing in
quantity gratings of high quality. Even so, many modern gratings
§2.5]
SELECTION OF SPECTROSCOPIC INSTRUMENTS
33
are superior in a variety of ways to modern prisms. A diffraction
grating consists essentially of a large number of close equidistant slits
or diffracting lines. The greater the number N of these slits, the
greater the theoretical resolving power of the grating. The more
closely the lines are packed together, the greater the dispersion of the
grating.
Figure 2.6 illustrates the diffraction patterns that result when a
collimated beam of monochromatic light from a slit falls on various
numbers of equidistant slits and is then brought to focus by a telescope lens. The grating space, or distance between the slits, has been
assumed to remain constant in the various diagrams. The successive
maxima, which correspond to the different orders of a diffraction
grating, are seen to become sharper and more definite as the number
*of slits (grating rulings) is increased. Multiple patterns are produced
by a grating when a beam of light containing several wavelengths
Wavelength
Ao
"I
^ 2 "t
3
3
2 2
\ Aj X.i ^2
II
0
II
2 2
Order of Spectrum
3
3
Fig. 2.7. Distribution of various orders of spectra produced by a diffraction grating for radiation of two wavelengths, Xi and Xj.
falls on it, each pattern to a scale proportional to the wavelength of
light involved in its production. The resulting composite pattern
forms a group of spectra in various orders. This condition is illustrated in Fig. 2.7, drawn for a beam containing waves of two lengths.
In small grating spectrometers and spectroscopes, transmission
gratings are commonly used. These consist of transparent plates on
which there are thousands of diffracting lines. The lines may be as
few as 500 per inch in a very coarse grating or as many as 30,000 per
inch in a fine high-dispersion grating.
Large spectrographs generally employ reflection gratings. Original gratings of this type are ruled on highly polished mirrors, of
aluminum-coated glass or of other materials, and may be either plane
or concave. Concave mirrors up to seven inches in diameter have
been successfully ruled with as many as 180,000 lines, and even
larger plane gratings have been ruled. The ruling of a large diffraction grating presents considerable difficulty, since the lines, engraved
34
SELECTION OF SPECTROSCOPIC INSTRUMENTS
[§2.5
on the polished surface by a sharp diamond, must be straight, parallel,
and equally spaced. Standard ruling spacings are approximately
5000, 7500, 10,000, 15,000, 25,000, and 30,000 lines to the inch.
Ruling a large grating may require two weeks or more, during which
time the ruling engine must be kept operating uniformly and at
constant temperature.
A reflection grating has the great advantage that the light does not
traverse material which will inevitably vary in transparency in different regions of the spectrum. If necessary, a single grating can be
used to cover the range 100 to 10,000 A. When a grating is ruled on
a concave mirror, a concave grating is obtained, which requires no
lenses for coUimating or focusing. The concave reflection grating,
developed by H. A. Rowland in 1882,^ is one of the most powerful
dispersing devices available.
The angular dispersion of a diffraction grating
with fixed slit is given by the formula
dl _
Nm
d\
A cos B
where A'^ is the number of rulings on the grating, '
m is the order used, and A is the linear aperture
Fig. 2.8.
Linear
of the grating, in this case the distance from the
aperture ^ of a diffraction grating. The
first ruled line to the last, as shown in Fig. 2.8.
width W of a beam of
The direction B in which any particular waveradiation
diffracted
length X will be thrown by the grating is given
at an angle 0 is related to the linear
by the formula
aperture A by the exA
pression W = A cos d.
mX = -^ (sin i ± sin B)
where i is the angle of incidence of the light on the grating and 6 is
the angle of emergence, both measured from the normal. Differentiating this formula gives the dispersion formula previously cited.
Near the normal to the grating, the dispersion is almost uniform,
and a so-called normal spectrum is obtained. This spectrum is most
convenient for the identification of sp&trum lines. By varying i,
any desired range of wavelengths can be .thrown to the vicinity of
the normal.
The resolving power P, of a grating is given by the formula
Pr = ir = Nm
dh
2 H. A. Rowland, Phil. Mag., 13, 469 (1882).
§2.5]
SELECTION OF SPECTROSCOPIC INSTRUMENTS
35
where A'^, as before, is the total number of rulings and m the order
used, X is the average wavelength of the lines, and d\ is their separation in wavelength. Two spectrum lines that are just distinguishable
in a given order of a grating containing, say, 10,000 rulings, will
appear as a single line if a grating with a smaller number of rulings is
used, owing to the decrease in sharpness of the diffraction patterns
produced and the resulting lower resolution.
As an example of the use of the above formulas, we may consider
the case of the first-order spectrum of a 30,000 line/in. grating with
6 in. of ruled width, for which m = 1 and A^ = 180,000. Changing all
units of length into angstroms, A = 6 in. X 25.4 mm/in. X 10^ A/mm
= 152 X 10' A; cos e = 1; dd/d\ = 0.000118 radian per A for the
dispersion on the normal in the first order. If this grating were used
• in a spectrograph that focused the spectrum at a distance r of 21 ft
(6300 mm) from the grating, the linear dispersion dl/d\ would be
r • dd/dk or 0.743 mm/A. Inverting this to get the plate factor, we
obtain 1.35 A/mm. The theoretical resolving power of this grating
Pr = Nm = \/d\ would be 180,000 in the first order, 360,000 in the
second order, 540,000 in the third order, and so on. A perfect grating
of this type could then be expected to separate, at 6000 A, two lines
of equal intensity not closer than 0.033 A in the first order, 0.0167 A
in the second, and 0.011 A in the third. A grating capable of such
performance has not yet been ruled. Although in a good grating the
first order is likely to give nearly the theoretical resolving power, the
second will perhaps give only half again as much as the first, and the
third perhaps only double the first. In any actual grating some
orders come much closer to perfection than others, for reasons discussed in Chapter 4.
The ruled area on a grating is rectangular in shape, the length of a
ruling usually not exceeding 2 in. on a concave grating and 4 in. on
a plane grating. Factors having to do with astigmatism make short
rulings desirable in the case of a concave grating.
Most of the diffraction gratings used in small spectroscopes up to
1947 were not original gratings but were replicas made by a process
devised by Thorp' and perfected by Wallace,^ in which a thin film is
formed when dissolved collodion or gelatin is poured on the surface
of a grating. When this film hardens, it forms a cast of the rulings.
' T. Thorp, Manchester Lit. and Phil. Soc. Mem., 44, 1 (1900).
' R. J. Wallace, Astrophys. Jour., 22, 123 (1905).
36
SELECTION OF SPECTROSCOPIC INSTRUMENTS
[§2.6
T h e film is stripped oS and is then carefully m o u n t e d on a plate of
optical glass. Such a grating cannot be expected t o show the full
resolution of its parent b u t may give slightly greater dispersion as a
result of shrinkage of the grating space. I t s use is of course limited
to the transmission region of glass.
I n 1947 White and Frazer^ of the Perkin-Elmer Corporation developed a process of m a k i n g fairly good concave grating reflection
replicas by casting a t h i n plastic model, fronted with a n evaporated
aluminum coating a n d backed with flexible plate glass, of a convex
master grating ruled by R. W. Wood a t Johns Hopkins. Although
the replicas duplicated the master grating closely, t h e thinness of the
flexible backing appeared t o limit the resolution obtainable. I t
seems probable t h a t improved replica gratings will soon become
widely available.
T h e distribution of light among a number of grating orders n a t u rally results in a loss of light in any one order. This loss can be
reduced considerably by shaping the point of the ruling diamond so
t h a t more light will be thrown in one direction t h a n in another. I t
is desirable, in any case, to have most of the light thrown into t h e
orders on one side of t h e central image. When very high resolving
power is desired, a grating can sometimes be found in which most of
the energy is thrown into the higher orders. I n general, if a grating
shows high intensity in one order in a given direction, lines of all
other orders lying in t h a t same direction will tend t o be strong. T h u s
a grating t h a t is found t o give high intensity on one side in the secondorder green (5500 A) m a y be expected t o be bright.also in t h e infrared
near 11,000 A in the first order and in the ultraviolet near 3660 A in
the third order. This effect cannot always be counted on, however,
since target p a t t e r n m a y change the distribution of light, as discussed
in Chapter 4.
Before proceeding t o the detailed comparison of prism and grating
spectroscopes, it is convenient t o consider other p a r t s of the instruments t h a t are essentially identical in the two cases.
2.6. The Slit. Since a spectrum line is merely a monochromatic
image of the slit, the slit is one of the most i m p o r t a n t parts of a
spectroscope. T h e accuracy with which it is made and can be
adjusted governs t h e character of the spectrum lines produced.
Therefore the slit m a y have an important effect on resolution.
^See R. W. Wood, Jovr. Opt. Soc Am.. 36, 71.5 (1946).
r
§2.6]
SELECTION OF SPECTROSCOPIC INSTRUMENTS
37
The slit width should be variable and in fine instruments should be
capable of adjustment between 1 mm and 0.005 mm. The slit jaws
are usually separated by a calibrated screw acting in opposition to
a spring that tends to move them together. A typical spectrograph
slit is shown in Fig. 2.9.
Fig. 2.9.
Typical small-spectograph slit. (Courtes
Ash Company, Boston.)
Jarre II-
In order that the space between the jaws shall form a suitable line,
it is necessary that the edges of the jaws be accurately ground to
straightness and mounted truly parallel, and that the front faces of
the jaws lie in the same plane. The jaw edges are ordinarily beveled
so that light reflected from them will not enter the spectroscope, and
the beveled side is turned away from the entering beam.
Slits in which only one jaw is movable, the so-called unilateral type,
are cheaper than the symmetrically opening type but have the disadvantage that the centers of spectrum lines produced with them
move when the slit width is changed. Although the bilateral slit is
necessarily more complicated to construct, a number of very satisfactory forms have been devised. A slit of this type is desirable on
38
SELECTION OF SPECTROSCOPIC INSTRUMENTS
[§2.7
any good spectroscope. The best adjustable slits are made to close
at the ends only, so that their sharp jaw edges will not be marred by
careless closure. The jaws should be made of some hard and durable
material, such as stellite or stainless steel, which can be ground to a
sharp edge and polished.
A simple slit can be made by coating a plate of quartz or other
transparent material with a thin opaque coating of metal or lacquer
and by engraving lines of the desired width in this. Slits of s^eral
widths may thus be provided, the proper one being set into the slit
holder as needed. For certain purposes such slits are more useful
than the adjustable type, since a definite slit width can be reproduced
more accurately than by setting a screw for which backlash and zero
position may change.
The slit is usually mounted in a drawtube in such a way that it can
be moved into or out of the spectroscope for focusing purposes and
rotated about a horizontal axis so as to be brought accurately parallel
to the edges of the dispersing element. Diaphragms should be provided with which the effective length of the slit or the portions of it
through which light passes can be varied.
A slit that is almost closed may cause horizontal streaks to appear
in the spectrum, because of dust particles which close the slit entirely
Fig. 2.10. Simple spectroscope system. S, slit; Lu collimator
lens; P, prism; La, telescope lens; Xi — Xi, spectrum.
at the spots where they occur. Any adjustable slit should occasionally be cleaned by opening it and carefully stroking its edges in one
direction with a freshly sharpened stick of soft wood.
2.7. The CoUimating and Focusing Systems. To give the greatest
resolution, dispersing devices must be illuminated with a collimated
beam of light, usually one in which the rays are parallel. The light
that has passed through a slit is divergent. I t niay be made parallel
by a positive lens called the collimating lens, shown as Jj\ in Fig. 2.10.
After passing through the dispersing system (prism P in Fig. 2.10)
the beam must be brought to a focus to give sharply defined images
§2.8]
SELECTION OF SPECTROSCOPIC INSTRUMENTS
39
of the slit. The lens that fulfills this function is called the telescope
lens, camera lens, or focusing lens, depending on whether a visual
spectroscope- or spectrometer, a spectrograph, or a monochromator is
involved. The functions of these lenses can, of course, be.carried out
by mirrors, which are sometimes used, especially in the infrared region.
Lenses have the advantages over mirrors of giving greater light transmissivity at most wavelengths except the infrared, of fitting somewhat
better into the geometry of the spectroscope, and of being easier to
correct for certain aberrations.
^
In a good spectroscope, all lenses or mirrors are corrected for
spherical aberration, and so as to focus the spectrum on a fairly flat
focal surface. This surface is often curved to some extent, but the
curvature is made as small and as even as possible. The focal surface
in a prism spectrograph is sometimes sharply inclined to the light rays
that strike it, being closer to the prism at short wavelengths, owing to
the higher index of refraction of the lens material for these waves.
This condition is not undesirable, since the increased apparent linear
dispersion produced may in some cases be an advantage; spectrograph lenses are therefore usually left uncorrected for chromatic
aberration.
2.8. Observing and Recording Systems. When the spectrum is to
be examined by eye, an eyepiece is provided that magnifies by from
3 to 10 times the spectrum imaged by the telescope lens. This eyepiece, together with the telescope lens, forms a telescope for observing
the beams of monochromatic light that leave the dispersing element.
This telescope may be arranged to swing on an arm about the prism
so that it can be pointed at the prism from different angles to observe
the various parts of the spectrum. Alternatively, the telescope may
be fixed in position and a special prism may be used which, when
rotated, sends successive spectral regions into the eyepiece. The
eyepiece is usually provided with a fine cross hair, or with an illuminated scale or pointer, to serve as a reference mark.
The collimating and telescope lenses of good visual spectroscopes
are always corrected for chromatic aberration, to avoid the necessity
of changing the focus of the eyepiece with wavelength.
In a spectrograph, no eyepiece is needed. Instead, provision is
made for holding a photographic plate or film so that the spectrum is
in focus on it throughout the range to be recorded at one time. One
criterion by which a good spectrograph may be distinguished from a
poor one is the ability to focus sharply, at one time, all the lines
40
SELECTION OF SPECTROSCOPIC INSTRUMENTS
[§2.9
within the spectral range incident on the plate. The» plateholder
(see the following paragraph) should be such as to bend the plate to
fit the focal curve exactly. I t is desirable to keep this curvature as
small as possible to avoid the necessity of using films or very thin
glass plates. In some instruments there is so little curvature of the
focal surface that the entire spectrum can be focused sharply on a
flat plate.
The plate or film is held in a plateholder or cassette, which, if necessary, is provided with templates to bend the plate to the proper curvature. The plateholder is often provided with a dark slide that can be
opened after the plateholder is in place, and closed when the plate is
to be carried to the darkroom for development. The plateholder
mounting usually has provision for moving the plateholder vertically,
so that a number of different spectra can be photographed on the
same plate.
Prism instruments are sometimes provided with wavelength scales,
which can be impressed on the spectrogram by swinging the engraved
transparent scale into position before the plate and making a brief
exposure to a small incandescent lamp. These scales are apt to shift
with use and can be relied on to a few angstroms only.
2.9. Comparison of Prism and Grating Spectrographs. Discussion of detailed designs of prism and grating spectrographs will be
reserved for Chapters 3 and 4, but it is convenient now to compare
the general characteristics of instruments of the two types. Most
important, perhaps, is the range of spectrum covered. As mentioned
previously, when a concave diffraction grating is used, only one
spectrograph is required for working in the ultraviolet, visible, and
near infrared regions of the spectrum. With a prism spectrograph,
hbwever, at least two sets of optical parts are required to cover these,
regions satisfactorily. Quartz prisms and lenses are ordinarSy used
for the ultraviolet region, glass for the visible, and rock salt, fluorite,
lithium fluoride, or potassium bromide for the near infrared. Some
prism spectroscopes are provided with interchangeable optical trains.
To compare the relative dispersions of prism and grating instruments, it is convenient to express the angular dispersion of a prism in
terms of the number of lines per inch required on a grating of equivalent dispersion (and if of the same linear aperture, a grating having
equivalent theoretical resolving power). Such data are given in
Table 2.1, where the first column lists the types of prisms considered,
the second column gives the wavelengths to which the data apply, and
§2.9]
SELECTION OF SPECTROSCOPIC INSTRUMENTS
41
the third column shows the n u m b e r of lines per inch in a grating
having t h e same angular dispersion as the prism referred to. I t is
evident t h a t a grating c a n ' b e ruled which will give dispersion in any
spectral region as great as t h a t given b y a standard prism. Since
most gratings are ruled with from 15,000 to 30,000 lines per inch.
Table 2.1 shows t h a t practical gratings produce dispersions superior
to those of single prisms for all wavelengths longer t h a n a b o u t 2000 A,
even in the first order.
TABLE 2.1
APPROXIMATE NTJMBBRS OF LINES PER INCH ON A CONCAVE GRATING
NEEDED TO MATCH DISPERSIONS OF 60-DEOREE PRISMS
Prism type
Wavelength,
in angstroms
Angular
dispersion
Linear dispersion
with maximum
plate tilt
Glass
8000
0000
4000
500
1250
5000
1500
3500
14,000
Quartz
4000
3000
2000
2000
5000
30,000
6000
15,000
90,000
I t is much easier t o obtain a diffraction grating having a ruled surface 6 in. wide t h a n it is t o obtain a transparent prism of equivalent
aperture. Gommonly used gratings have three times t h e linear aperture of the most commonly used prisms, so t h a t t h e grating h a s , in
general, a considerable a d v a n t a g e in resolving power.
I n prism spectrographs, the plate is often tilted because of variation
of t h e focal length of the camera lens with wavelength, and t h e linear
dispersion along the plate is greater t h a n the value obtained by
multiplying the angular dispersion by the focal distance. With
s t a n d a r d quartz spectrographs, the a p p a r e n t magnification of dispersion which results m a y be as great as threefold, as shown in column 4
of Table 2 . 1 . Even when this additional apparent dispersion of the
prism spectrograph is t a k e n into account, practical gratings give dispersions exceeding those of single prisms a t all wavelengths longer
t h a n a b o u t 2500 A.
If we compare spectrographs having the same linear a p e r t u r e and
focal length, a 30.000-line-per-inch grating instrument will excel a
42
SELECTION OF SPECTROSCOPIC INSTRUMENTS
[§2.10
glass-prism instrument in dispersion and resolving power at all wavelengths, and a quartz-prism instrument at all wavelengths longer than
2500 A, even in the first order. The second order of the grating is
also available for the short wavelengths, giving double the dispersion
and increased resolution.
That the various orders of a grating overlap is sometimes cited as a
disadvantage of the grating spectrograph. Undesired orders can
usually be removed, however, by filters or by crossing with another
spectroscope, which may be one of low dispersion.
Figure 2.H shows the distribution of wavelengths in typical spectrograms taken with prism and grating spectrographs that give
2000
0. I
2000
2100
I
I
2200
2200
I
2400
I
2300 2400 2500 2600
2800 3000
3500
4000
I I I I I I I I I I I hlllihlihlihlllilllil
2600
2800
3000
3200
3400
3600
3800
4000
b. I I I I h I I i I I I h I I I 1 h I I h I I I I h h I I I I 1) I I I
Fig. 2.11. Comparison of a prism scale, a, for a quartz prism, and a grating
scale, b, for normal dispersion, in the case of two spectra of equal length from
2000 to 4000 A.
spectra of the same over-all length from 2000 to 4000 A. The crowding of the prism spectrum at long wavelengths is obvious. It should
be emphasized, however, that this crowding is of importance only
where wavelengths are being considered. In terms of frequencies or
wave numbers, the prism scale is more uniform than that of the
grating.
2.10. Speed and EfSciency. The principal purpose of a spectroscope is to separate light in accordance with its wavelengths. The
greater the resolution of the instrument used, the greater the purity
of the spectrum produced, other factors remaining constant. However, the purity of the spectrum depends also on the width of the slit
used, the focal lengths of the coUimating and camera lenses, and the
freedom from scattered light and false lines.
The spectral purity of light that can be isolated by a slit of given
width can be increased by increasing the focal length of the camera
lens. This procedure will cut down the intensity of the spectrum,
however. For a given prisni or grating the intensity of light at a
given position in the spectrum and the purity of the spectrum at that
point can each be altered only at the expense of the other. The
efficiency of a spectroscope is defined as the product of intensity times
purity. A small prism instrument of high light-gathering power and
§2.10]
SELECTION OF SPECTROSCOPIC INSTRUMEISfTS
43
low dispersion may be as efficient as a large concave-grating spectrograph of low light-gathering power and high dispersion.
The speed of a spectrograph is a measure of the intensity of the
light it transmits at any wavelength. The speed at each wavelength
varies directly with the transmission factor of the instrument and
inversely with its aperture ratio, which is the ratio'of the focal length
of the spectrograph to its linear aperture. The aperture ratio is
equivalent to the / number as used with ordinary camera lenses.
Aperture ratios of prism spectrographs usually lie between 7 and 15,
whereas large quartz Littrow instruments usually have an aperture
ratio of about 23. A standard 21-ft concave-grating spectrograph
ordinarily has an aperture ratio of 42 horizontally and 120 vertically.
Concave gratings are ordinarily not used at high-aperture ratios
because the greater the curvature of a grating blank, the more difficult
is it to rule, since the ruling diamond will cut on different portions of
its edge at different stages of the stroke. The aperture ratio of a
given grating can be doubled by usin^ the stigmatic mounting discussed in § 4.7, which cuts the effective focal distance of the grating
in half, making readily available a horizontal aperture ratio of 21.
The diffraction grating has a fundamental advantage over the
prism in separating energy in accordance with wavelength. This
advantage was at one time offset more often than not by the low
transmission factor resulting from the low reflecting power of gratingmirror materials, which resulted in greater loss of light than that
caused by the absorption and reflections in prism instruments. Since
the introduction of gratings ruled on aluminum-covered glass surfaces, grating spectrographs have been made with transmission
factors larger than those of prism instruments equivalent in size and
in purity of spectrum. Grating spectrographs have the reputation of
low light transmission because in the past they were ruled on speculum metal, which has a reflecting power of 10 per cent or less for
wavelengths in the far ultraviolet. Aluminum-on-glass gratings
having reflection factors of 65 per cent even in the ultraviolet are not
uncommon today.
Formerly, another limitation of the grating was its waste of light,
since as much as 50 per cent of the light sometimes went into the
undispersed central image and 60 per cent of the remainder into orders
not being used. This effect has been overcome by selection of diamond points so shaped that they engrave rulings which throw much
of the light in one general direction. An aluminum-on-glass grating
44
SELECTION OF SPECTROSCOPIC INSTRUMENTS
[§2.11
ruled by B. W. Wood ^ has been found by measurement to throw
80 per cent of its reflected green light into one first order. This
directional effect, coupled with an 80 per cent reflection coefficient for
the grating material, results in the appearance in one first order of
more than half of the green light sent into the instrument. Such a
transmission factor is somewhat greater than that of most prism
spectrographs, where the presence of from 6 to 14 quartz-to-air or
glass-to-air surfaces results in large losses by reflection and scattering.
Prism spectroscopes have seldom been found to transmit more than
30 per cent of the light sent into them when their transmission factors
have been measured precisely.
2.11. Scattered Light and False Lines. Most concave-grating
spectrographs using original ruled gratings, as opposed to replicas,
show less scattered light than prism instruments, because of the
smaller number of optical surfaces they contain. Usually the light
scattered from a grating ruled on an aluminum surface is far less than
that from a speculum metal gating, because of the grainy character
of the latter. Coating a speculum-metal grating with aluminum will
increase its reflecting power and hence its speed, but this is of little
aid in reducing scattered light, since the latter is increased in intensity
proportionately to the increase in intensity of the spectrum.
All gratings show false lines of a type known as ghosts.'' Rowland
ghosts are produced by periodic errors in the screw that moves the
diamond forward a definite amount between strokes while the grating
is being ruled. These ghosts, though annoying if intense, cause much
less trouble than the so-called Lyman ghosts, which are produced by
a different type of irregularity in the drive of the ruling engine.
Lyman ghosts are usually widely separated from their parent lines.
Thus a line of wavelength X may be found to have Lyman ghosts at
positions corresponding to X/a, 2X/a, 3X/a, . . . n\/a, where a is an
integer. Although these ghosts have the color of the parent line and
so can often be distinguished visually, in spectrographs they may
cause great confusion and have led to many errors in the past. Fortunately, Lyman ghosts are seldom strongly present in gratings ruled
on modern engines. A grating that shows them in intensity greater
than about 0.01 per cent of that of the parent line should be considered
«R. W. Wood, Jonr. Opt. Soc. Am., 34, 509 (19J.4).
' See W. F. Meggers, C. C. Kiess, C. Runge, and J. A. Anderson, Jour. Opt. Soc. Am.,
6, 417 (1922).
§2.13]
SELECTION OF SPECTROSCOPIC INSTRUMENTS
45
us unsuitable for ordinary work. T h e presence of L y m a n ghosts can
readily be detected by methods described in § 5.3 and in Reference 7.
Rowland ghosts, if present, are easy to detect. Usually two or more
pairs of equally intense ghosts will be found symmetrically placed
about every very strong line, as shown in Fig. 2.12. T h e user, once
familiar with a given grating, has no difficulty in identifying Rowland
ghosts or in knowing when their presence can be
neglected. I n a good grating such ghosts will
have intensity less t h a n 0.1 per cent of the parent line and can usually be neglected entirely.
2.12. Shapes of Spectrum Lines. The lines
produced in a prism spectrograph are not
straight b u t are curved as a result of t h e inFig. 2.12.
Photocreased deviation of rays t h a t pass through graph of Rowland
ghosts of a diffraction
the prism obliquely. This curvature sets a grating (mercury l!ne
limit to t h e length of slit t h a t can be used with at 5461 A).
a given collimator lens.
Gratings produce very straight spectrum lines and give fairly uniform magnification of the slit images, an important consideration
in m a n y types of photographic photometry, including those using
logarithmic spiral disks or step sectors.
T h e contour of a spectrum line is likely to be of the form shown in
Fig. 2.13a. A prism spectrogram or one from a good grating will
usually show smooth and symmetrical line contours, b u t in imperfect
gratings the lines are apt to appear irregular or asymmetrical, as
shown in Fig. 2.13b. This effect is usually a p p a r e n t only in gratings
of high dispersion and resolving
power. I t will be discussed further
in §§ 9.12 and 20.4.
2.13. Astigmatism. A lens or
mirror, unless anastigraatically designed, produces a t r u e image of an
(a)
lb)
object only when both lie close t o
Fig. 2.13. Contours of spectrum
the optic axis. As t h e angle of a
lines, (a) Regular contour, a_s probeam of light departs from t h e opduced by a good grating or prism;
tic axis more and more, greater
(b) irregular contour, as produced
by a defective grating.
amounts of astigmatism are introduced, the rays being brought t o a
line focus at one distance and t o a second line focus perpendicular to
the first a t a greater distance, as shown in Fig. 2.14. I n ordinary
46
SELECTION OF SPECTROSCOPIC INSTRUMENTS
[§2.13
prism spectrographs, most of the rays passing through the camera
lens deviate very little from the optic axis, and the astigmatism
can usually be neglected, since extremely fine focus is needed only
in the horizontal direction to resolve close spectrum lines, and a
Circle of
.^
least Confusion
Fig. 2.14. Astigmatism produced by a lens. Off-axis rays from the point Q
come to focus in two line images, Q\ and Qt'. The position of minimum beam
cross section is called the circle of least confusion. (From A. C. Hardy and F. H.
Perrin, Principles of Optics, McGraw-Hill Book Company, Inc., New York
page 100. Courtesy authors and publishers.)
focus only one-tenth as sharp will serve in thq vertical direction.
Also, the camera lens can be figured so as to reduce astigmatism.
The spectrum lines produced by a concave diffraction grating, as
ordinarily used, are astigmatic images of the slit, each illuminated
§2.14]
SELECTION OF SPECTROSCOPIC INSTRUMENTS
47
point on the slit being imaged as a vertical line in the spectrum rather'
t h a n as a point. N o decrease in the purity of the spectrum results so
long as t h e slit is accurately parallel to t h e rulings of t h e grating and
neither t h e slit nor t h e astigmatic images are curved. W i t h most
gratings a very slight line c u r v a t u r e does exist, and it is advisable to
keep the illuminated portion of the slit or grating as short as possible
in the vertical plane when high resolving power is required.
Since each astigmatic line image is longer t h a n t h e slit t h a t produced it, astigmatism m a y result in a decrease of speed, which becomes very serious a t large angles of incidence and reflection. I n the
high orders of a grating, astigmatism m a y produce a twentyfold
decrease in intensity, when a short slit is used, as discussed in § 4.9.
A more important disadvantage of astigmatism is t h a t the long,
even spectrum lines produced b y an astigmatic spectrograph mask the
variation of illumination along t h e slit, which might otherwise reveal
important information about t h e source of light being studied. A
prism or a diaphragm for producing comparison spectra cannot be
placed at the slit of such an instrument, nor can rotating photometric
disks be used at the slit, as t h e y can with a stigmatic instrument,
unless special compensation is introduced.
Astigmatism is occasionlly useful, as with certain types of intensity
measurements, and for producing spectrograms t h a t are neat in
appearance. T h e astigmatism of concave gratings a n d m e t h o d s for
overcoming this, some of which are applicable t o all astigmatic
spectrographs, are further discussed in § 4.9.
2.14. Space R e q u i r e m e n t s . A grating spectrograph requires no
more space t h a n an equivalent prism spectrograph. M a n y grating
installations are large because bulk can be tolerated and much can be
accomplished with the increased dispersion and range t h u s m a d e
available. Prism instruments are not made large principally because
large prisms are very expensive and are a p t to absorb much light a t
short wavelengths, and because of optical limitations on t h e lenses
used. T h e standard 6-in. grating usually has a 21-ft focal length.
This focal distance results in a large spectrograph. If the size m u s t
be held t o t h a t of a standard Littrow prism spectrograph, a smaller
grating can be used and the instrument will t h e n occupy no more
space t h a n _the equivalent prism instrument. Various designs of
prism instruments are discussed in Chapter 3, and in Chapter 4 t h e
different mountings in which concave gratings have been used are
described in detail.
48
SELECTION OF SPECTROSCOPIC INSTRUMENTS
[§2.15
A diffraction grating can be arranged to give a t high dispersion with
a single setting t h a t p a r t of the spectrum most commonly photo-^
graphed, from 2000 t o 5000 A. When one uses the stigmatic mounting discussed in C h a p t e r 4, it is often convenient t o cover 2500 to
5000 A in the first order, with 2000 t o 2500 A overlapping in the
second order. This overlapping ordinarily causes little inconvenience
when line emission spectra are involved. T h r e e 10-in. plates placed
end to end can be used t o cover this range with a dispersion corresponding to 3.3 A / m m , which is sufficient for m u c h routine work.
Photographing 30 in. of spectrum a t one time requires t h a t the
spectrograph be somewhat wider t h a n the s t a n d a r d prism instrument.
I n the prism spectrograph these three ranges would be photographed
separately, one after t h e other, adjustments of t h e prism, plate, and •
optical system being required between exposures. W h e n space is
more i m p o r t a n t t h a n time, a narrow mounting can be used for the
grating spectrograph, enabling a single plate t o be used with adjustments similar t o those required b y the prism instrument.
2.15. S u m m a r y of Comparison. T h e results of t h e foregoing discussion are summed u p in Table 2.2 where -|- indicates t h a t the device
so marked is superior t o the other in the quality indicated.
TABLE 2.2
SUMMARY OP COMPARISON BETWEEN PRISM AND GRATING SPECTROGRAPHS
Characteristics
Spectral range
Linear aperture
Resolving power
Relative dispersion
Uniformity of dispersion:
By wavelength
By frequency
Speed
Freedom from:
Astigmatism
Stray light
Spurious lines
Line curvature
Prism
instruments
+
+
+
+
Grating
instruments
+
+
+
+
+
f
+
+
I n the past, grating spectrographs have been used less often t h a n
prism spectrographs for a very simple reason: the difficulty of con-
§2.16]
SELECTION OF SPECTROSCOPIC INSTRUMENTS
49
structing and operating satisfactory ruling engines. A second problem to be solved has been a metallurgical one: to obtain a metal hard
enough to be figured accurately and polished like glass and soft
enough to avoid wearing out the ruling diamond too rapidly. A
practical solution has been to figure a glass mirror and to evaporate
on this a coating of chromium, and then on this a coating of aluminum. The ruling is then done on the soft aluminum surface, which
gradually protects itself with a thin transparent film of oxide, leaving
a surface that reflects quite well all wavelengths between 1000 and
10,000 A. Such gratings are not so sturdy as prisms but last well
under proper care.
Original diffraction gratings cost from $200 to $1500 apiece, although the price varies greatly with the source of supply, and the
size; replicas are somewhat cheaper. A quartz prism costs about
the same per unit area of aperture, and a glass prism somewhat,
though not much, less. Since both quartz and glass prisms must be
provided to give the desirable spectrum coverage, to say nothing of
the auxiliary lenses and prisms that are required, it is justifiable to
conclude that a grating spectrograph is fundamentally a less expensive piece of apparatus to build than a prism spectrograph. A concave grating costing $800 can be made to do the work of a quartz
prism costing $10,000, if mere size is a criterion. Gratings seem
destined to come into much wider use as their availability increases.
A partial list of manufacturers of spectroscopic equipment is given
at the end of this chapter. Most of the firms listed are glad to furnish
material giving detailed descriptions of their apparatus, and are
usually willing to supply sample spectrograms. The most widely
used types of prism spectrographs are described in the next chapter,
and grating instruments are discussed in Chapter 4.
2.16. Monochromatic Illuminators. Isolation of a comparatively
narrow band of spectral wavelengths can be produced by means of
absorption filters, refraction filters, interference filters, or especially
adapted spectroscopes called monochromators. The various types of
filters are described in Chapter 14. Monochromators may be of
special design, to give convenience in changing the wavelength of the
isolated beam or extreme freedom from scattered light, or may consist
merely of spectroscopes in which an exit slit has been provided in place
of a viewing eyepiece. The most satisfactory monochromators are
those designed with fixed entrance and exit slits, so that the emergent
beam has a fixed direction no matter what its wavelength. Prism
50
SELECTION OF SPECTROSCOPIC INSTRUMENTS
[§2.16^
monochromators are described in Chapter 3 a n d those using gratings
in Chapter 4. Almost all spectrometers for use in the infrared r e g i o n . ' '
are monochromators of types described in detail in Chapter 17.
'
REPRESENTATIVE
MANUPACTUBERa O F S P E C T R O S C O P I C
EQUIPMENT
The specialties listed for each manufacturer are intended to be indicative
only, and do not imply that products are limited to those mentioned.
*
Applied Research Laboratories, Glendale, Calif, (grating spectrographs and
equipment).
Applied Physics Corporation, Pasadena, Calif, (automatic recording spectrophotometers).
Baird Associates, Cambridge, Mass. (grating spectrographs, infrared equipment) .
Bausch & Lomb Optical Company, Rochester, N. Y. (prism apparatus).
Ch. Beaudoin, Paris (prism apparatus).
R. and J. Beck, Ltd., London (small prism apparatus). Agents in U.S.A.:
Jarrell-Ash Company, Boston, and Pfaltz and Bauer, Inc., New York.
Bellingham and Stanley, Ltd., London (prism apparatus).
Central Scientific Company, Chicago (small grating and prism apparatus).
Gaertner Scientific Corporation, Chicago (prism spectroscopic instruments).
General Electric Company, Schenectady, N . Y. (recording spectrophotom-,
eters for the visible region).
Hilger and Watts, Ltd. (formerly Adam Hilger, Ltd.), London (spectroscopic
equipment of all types). Agents in U.S.A.: Jarrell-Ash Company, Boston.
Huet (Societe G^nerale d'Optique), Paris (prism apparatus).
Jarrell-Ash Company, Boston (grating spectrographs, dealers in and importers '
of spectroscopic equipment of all types).
Kipp and Zonen, Delft, Holland (prism apparatus and equipment). Agents'
in U.S.A.: James G. Biddle Co., Philadelphia. '
Lane-Wells Company, Pasadena, Calif. (Prism Spectrographs for Raman
effect).
,
^<^
National Technical Laboratories, South Pasadena, Calif, (spectrophotometers, infrared equipment, flame photometers).
Perkin-Elmer Corporation, Glenbrook, Conn, (infrared equipment, flame
photometers).
CHAPTER 3
Prism spectroscopes and Spectrographs
R E P R E S E N T A T I V E MODELS O F PRISM SPECTROSCOPES will he described
in this chapter, t o illustrate t h e forms commonly used. Where a p p a r a t u s of a particular maker is depicted, t h e basis of selection Is t o
some extent arbitrary, since equally good instruments of a similar t y p e
are often obtainable from other manufacturers.
3.1. Materials for Dispersive P r i s m s . T h e chart in Fig. 3.1
shows t h e relative dispersive powers a n d ranges of transmission of
0.58 9/U
034 0.38
1.0
Flint Gloss
0.185
0.26
0.185
0.38 1.0
Quartz
0.26 03*.0 35
3.5
9.0
Calcium Fluoride (Fluorite)
0.20
0.26
0.38
1.0 35 9.0
17.0
Rock Salt
0.38
1.0 9.0
17.0
2;.0
3.5
Sylvine
Fig. 3.1. Useful regions of transmission and relative spectral dispersions of
several materials employed for prisms, on wavelength scales centered on
XS896 A.
several optical materials found useful in t h e various regions of t h e
spectrum commonly studied. Optical glasses, on account of their
narrow transmission ranges, can b e used for prisms only in t h e visible
region, t h e near infrared, a n d t h e near ultraviolet. Since most glasses
produce greater dispersion in these regions t h a n do other optical
materials, glasses are commonly used in visual spectroscopes. T h e
51
52
PRISM SPECTROSCOPES \ N D SPECTROGRAPHS
[§3.1
various kinds of optical glass differ greatly in dispersion, in refractive,
index in a given wavelength region, and in permanence of surface.
Flint glasses are likely t o be transparent at longer wavelengths t h a n
are crown glasses. Special glasses such as Uviol a n d Corex transmit
somewhat farther into the ultraviolet t h a n do ordinary glasses.
Quartz in the crystalline form found in nature, as distinct from so-;
called fused quartz, is doubly refracting. If a prism is cut from a
quartz crystal, its optic axis should be made to coincide with the optic
axis of the crystal, to avoid doubling of the spectrum. Even when
this condition is fulfilled, a slight doubling of the spectrum lines results from the circular polarization produced by t h e quartz. This
defect can be eliminated by a method devised by Cornu, in which half
of the prism is cut from a crystal producing right-handed rotation
and half from a crystal producing left-handed rotation. Fused or
vitreous q u a r t z does not produce birefringence or circular polarization,
but it has less dispersive power t h a n crystalline q u a r t z , absorbs wavelengths shorter t h a n 2800 A more strongly, a n d is seldom produced in
sufficient homogeneity to give high optical quality. Even natural
q u a r t z crystals v a r y somewhat in transmission from sample to sample,
especially for wavelengths shorter t h a n 2500 A. Manufacturers will,
on occasion, furnish optical parts especially selected for high transparency in t h e region near 2000 A.
Because rock salt a n d potassium bromide are very hygroscopic,
optical p a r t s m a d e of these materials m u s t be carefully protected
from moisture. Despite this limitation, rock salt is so transparent
between 1800 a n d 160,000 A t h a t it is occasionally used in commercial spectrographs for the visible and ultraviolet. I t is widely
used in instruments for the infrared a n d will be discussed further in
this connection in C h a p t e r 17. Sylvine (KCl) and potassium bromide
extend even farther into t h e infrared t h a n rock salt (to 230,000 a n d
270,000 A, respectively), and are often used for prisms.
Fluorite (calcium fluoride) held for m a n y years the unique position
of being t h e only material suitable for optical p a r t s transparent to
wavelengths as short as 1250 A, but pieces larger t h a n an inch in
diameter were prohibitive in cost. During World W a r I I , considerable progress was m a d e in growing large crystals of this material ^ and
of lithium fluoride, which is transparent to about 1050 A.^ T h e latter,
material has the disadvantage of being brittle and difficult to work
»D. C. Stockbarger. O.S.R.D. Report No. 4690.
2 D. C. Stockbarger. Rev. Sei. Imf.. 7, 133 fl936).
PRISM SPECTROSCOPES AND SPECTROGRAPHS
53
without cleaving, and it is fortunate t h a t synthetic calcium fluoride
crystals' have become obtainable in diameters as large as m a y be
desired for spectroscopic equipment.
-Reflecting surfacesFig.'3.2. Multiple prism of the Littrow type.
Prisms of extreme size are sometimes needed, not so much for the
increased resolving power given by greater base thickness as for the
larger prism aperture, which will transmit more radiant flux. In very
large prisms this advantage m a y be offset by increased transmission
Fig. 3.3. Divided-circle spectrometer with prism removed.
Gaertner Scientific Corp., Chicago.)
(Courtesy
losses through absorption and scattering in the thicker material. It
m a y then be desirable to use instead a multiple prism of lower resolving power, as shown in Fig. 3.2.
From the Harshaw Chemical Co., Ea.st 97th Street, ClevelanH fi, Ohio.
i
54
'T.
PRISM SPECTROSCOPES AND SPECTROGRAPHS
C§3.2
Numerous special types of dispersing prisms have been designed.,
Most of these are not in common use; certain especially useful forms
are described below in connection with spectroscopes designed
around them.
3.2. The Simple Spectrometer. A tj^pe of prism spectrometer
extensively used in teaching and research laboratories is illustrated in«
Fig. 3.3. The slit is affixed to the collimator tube in such a manner
as to slide in and out for focusing adjustment, and the glass collimator
lens is rigidly mounted at the end of this tube. The prism is mounted
on a table that rotates about a vertical axis, and the telescope tube
Fig. 3.4. Spekker Steeloscope, a two-prism spectroscope for analysis of
ferrous alloys by visual observation of spectra. (Courtesy Adam Hilger, Ltd.,
London.)
•
rotates about the same axis, carrying an objective lens and an eye-,
piece. Graduated circles with vernier scales are provided to read
the angles through which the prism and the telescope arm are turned.
Such spectrometers are used extensively for instructional purposes,
for measuring the indices of refraction of various solid materials in
prism form, for testing prisms, and for observing simple spectra.
A small transmission grating can be used in place of the prism in
such a spectrometer. It is then desirable that ample swing be provided for the telescope tube, to permit observation of orders on both
sides of the normal.
>»= *
-ll
§3.2]
P R I S M SPECTROSCOPES AND S P E C T R O G R A P H S
^''Axis of rotation
(a)
Fig. 3.5. Arrangements to obtain constant deviation for rays traversing the
prism at minimum deviation, (a) Pellin-Broca prism for 90-deg. deviation.
(b) Wadsworth mounting, with which the over-all deviation depends upon the
angle a.
For routine examination-of selected spectra, prism spectrometers
can be obtained in wiiich collimator tube, prism, and telescope tube
are all fixed in position, as in Fig. 3.4, where the Hilger Spekker
Steeloscope is depicted. These spectrometers are usually provided
with scales on which the positions of important lines are marked. If
the prism and the lenses of such an instrument are made of quartz, a
Fig. 3.6.
Constant-deviation wavelength spectrometer mounted with source
on optical bench. (Courtesy Adam Hilger, Ltd., London.)
56
PRISM SPECTROSCOPES AND SPECTROGRAPHS
[§3.3
fluorescent eyepiece can be used which may make possible observation .
of ultraviolet lines as far down as 1850 A.
3.3. Modem Wavelength Spectrometers. A very convenient
type of spectrometer is one using the constant-deviation prism of
Abbe* as modified by Pellin-Broca.'" As shown by the dotted lines iux
Fig. 3.5a, the single prism of peculiar shape is equivalent to twb>
30-deg dispersing prisms connected by a 45-deg total reflecting prism.
When one wishes to scan the spectrum, the prism is rotated, and
various wavelengths are sent successively in the direction OA at right
angles to the incoming rays of light. All lenses are corrected for
chromatic aberration, to avoid the necessity of refocusing for each
wavelength. A calibrated drum is provided from which wavelengths
can be read directly to within a few angstroms (Fig. 3.6).
The resolving power of this type of constant-deviation spectroscope
as manufactured by Gaertner, Hilger, and others is usually less than
Fig. 3.7. Direct-vision spectroscope employing an Amici prism (two prisms of
crown glass and one of flint glass). (Courtesy Bausch & Lomb Optical Company, Rochester, N. Y.)
5000. An eyepiece provided with a pointer can be obtained. By
means of a small mirror this pointer can be illuminated from the
source through screens of various colors, so that a color that contrasts
with any part of the spectrum can be chosen. Small cameras can
also be obtained which, when fastened in place after removal of the
observing telescope, transform the instrument into a spectrograph for
photographing the visible region at low dispersion.
* E. Abbe, Jena ZeiUchr. f. Med. u. NaUirwisa.. 5, 459 (1870).
'" P. Pellin and A. Broca, Jour, de Phys., 8, 314 (1899).
5]
PRISM SPECTROSCOPES AND SPECTROGRAPHS
3.4. Direct-Visioa Spectroscopes. The simplest type of spectroscope, and one that can be made small enough to carry in the pocket,
is a small replica diffraction grating, mounted in a flat container that
has a hole passing through it. If a source of small extent is looked at
through this device, spectra in various orders will be visible; and if
the source is very small, monochromatic images will be seen.
A somewhat more elegant device is the direct-vision spectroscope
which makes use of a nondeviating prism. A Bausch & Lomb model
of this instrument is illustrated in Fig. 3.7. The dispersing system
Fig. 3.8. Zeiss three-prism spectrograph, with cover removed.
consists of several prisms alternately of dense flint and orown glass, so
arranged that the mean deviation of the light beam by one set of
prisms is neutralized by that of the other set, while a certain amount
of dispersion remains. This prism is mounted in a convenient
tube with an adjustable slit and a magnifying eyepiece. A wellconstructed instrument of this type will resolve many of the Fraunhofer lines in the solar spectrum.
3.5. Portable Spectrographs. A spectrograph is a spectroscope
ptovided with a camera. This camera usually consists of a pair of
ways on which slides a cassette or plateholder. The latter holds the
58
PRISM SPECTROSCOPES AND SPECTROGRAPHS
[§3.6
photographic plate or film and can be moved up and down in order
to photograph a number of spectra on the same spectrogram.
In a small spectrograph the entire transmitted spectrum can usually
be photographed at a single setting, so that no motion of the prism
or varying focusing adjustment is necessary. In larger instruments, '
where the spectrum cannot be recorded on a plate of reasonable size,
some provision must be made for turning the prism and changing the
focal distance of the lenses and the tilt of the plate when various
regions of the spectrum are to be photographed.
Fig. 3.9. Small quartz spectrograph with source mounted on optical bench.
(Courtesy Adam Ililger, Ltd., London.)
3.6. Special Glass-Prism Spectrographs. For many year.s Zeiss
manufactured a three-prism spectrograph having the optical system illustrated in Fig. 3.8. The prism train is that designed by
Forsterling, which gives constant deviation with dispersion equivalent
to that of three 60-deg prisms. All three prisms are arranged on a
mounting that turns about an axis, above which the central constantdeviation prism is placed. The other prisms are kept at minimum
deviation by steel bands which communicate to their mountings the
proper rotations as the wavelength drum is turned.
The instrument covers the range 3700 to 10,000 A and is provided
_t:
§3.7]
PRISM SPECTROSCOPES AND SPECTROGRAPHS
59
with three interchangeable cameras having 6-cm-diaineter lenses of
85,27, and 11 cm focus, respectively. A collimating lens of 30 cm focus
and 6 cm diameter is used. W i t h the long camera the plate factors
are 58, 27, and 6 A m m at 8000, 6000, and 4000 A, respectively.
T h e cameras of shorter focal length give proportionately smaller
dispersions but concentrate the same a m o u n t of light into smaller
and hence brighter spectra. F o u r sets of adjustments of the optical
p a r t s are needed to photograph the whole visible spectrum with the
long camera, whereas with the others single settings suffice. T h e long
camera takes plates of 4.5 X 12 c m ; those of the shorter cameras
are 6 X 9 cm.
H^^'
Fig. 3.10.
Medium quartz spectrograph. (Courtesy Bausch & Lomb
Optical Company, Rochester, N. Y.)
Hilger manufactures a large-aperture glass spectrograph in which
t w o dense 55-deg prisms are used. T h e camera objective is 3.5 in.
in diameter and is used a t / ' 5 . 7 . T h e spectrum from 8000 to 3500 A
is 4 in. long and is photographed on a 3 j - X 4|-in. plate.
3.7. Quartz-Prism Spectrographs.
Spectrographs containing
q u a r t z optical parts are widely used. These can be obtained in t h r e e
standard sizes, the small and medium models covering the entire
spectral range with a single setting of the focusing adjustments.
A small spectrograph covering the range 8000 t o 1850 A, with a
spectrum length of about 85 m m , as manufactured by Hilger, Steinheil, Gaertner, and others, is illustrated in Fig. 3.9, where the Hilger
PRISM SPECTROSCOPES AND SPECTROGR.VPIIS
60
[P.7
model is depicted. This t y p e finds its greatest usefulness a t wavelengths shorter t h a n 2500 A, where its relatively high light transmission aids in photography of a region somewhat difficult t o record,
since prism absorption a n d lack of plate sensitivity in this region
conspire t o reduce t h e density of spectrograms. T h e spectrograph
can be obtained fitted with a transparent wavelength scale. A
fluorescent screen c a n be used t o make t h e ultraviolet spectrum
visible and thus aid in the preliminary focusing adjustments.
Probably t h e most commonly used of all spectrographs are t h ^
medium-sized quartz instruments, types of which are manufactured^
by several firms. T h e Bausch & Lomb instrument is shown in
Sb
«
! Si
FeAl SnCw
SbJ%(%Sn
P^SB
Sb
Sr, J!
Al
Pb
Pt
Pb
II? IIMI III ifi=ttilM-.l»
II
iiji
C«
11II. iiiiiii apiipi
QtPh
CU M
Oi Mi
( T H H h U lfuHMlH7HUUi4lTt)lHMn'lMl!MtlTmtll!ttlml»»lllllhwllAlilll*l
Fig. 3.11. Spectra and wavelength scale taken with medium quartz spectrograph of the type illustrated in Fig. 3.10. (Courtesy IJausch & Lomb Optical
Company, Rochester, N. Y.)
Fig. 3.10. Lenses of 600 m m focus and 51 m m diameter are used,
giving a spectrum extending from 2100 t o 8000 A which is about
200 m m long. T h e prism, of the Cornu type, is 41 m m high by 65 m m
length of face. A s t a n d a r d 4- X 10-in. photographic plate is used.
This is the largest s t a n d a r d size of quartz instrument t h a t will give
the entire ultraviolet region in air a t a single setting of t h e prism and
camera.
T h e medium quartz spectrograph can be obtained with or without
a transparent scale of wavelengths or frequencies. T h e variation of
dispersion with wavelength is illustrated b y the scale shown in Fig.
3.11. When such a scale is purchased, care should be taken t o see
t h a t each division is properly spaced, because some manufacturers
§3.7]
PRISM SPECTROSCOPES AND SPECTROGRAPHS
61
have used a uniform and hence incorrect spacing between correctly
spaced 100-A divisions.
Hilger also manufactures a spectrograph of this type which extends
to 1850 A the range of the spectrum covered. In this case the
quartz optical system gives the range 3700 to 1850 A in a spectrum
225 mm long, whereas when the spectrum is photographed with a
corresponding glass system it extends from beyond the red end of the
visible to 3650 A. The quartz instrument is rather unusual in that
the camera lens produces an image field in which the residual curvature is reduced to a fraction of a millimeter, so that ordinary thick
and rigid photographic plates can be used if desired.
3J
•ViliV
I
111
1 1 M 11!
1 < 1n
1 I I I M 11
1 I 11111111II i i T i i i i i i i i i i i l l l l l | : l | l i ! ' l i M l . i
! 1 lliffl
11 V i l i - i SIS m 1!
1 III in • ill ill li
1
m n ' i "11! 11". ' iSi5Iliill! 1111
'1 f l 1 i« H! • 1; III 1!
'. •11 m
H - f p i ii} i
f LEAD
f MASMEIU
fTIN
1 ZINC
3i
lU.
i
;••':
1 In
Mi..
1
1 saMlNW
Icopres
Fig. 3.12. Spectrograms of two die-casting alloys taken with the quartz spectrograph illustrated in Fig. 3,10. (Courtesy Bausch & Lomb Optical Company,
Rochester, N. Y.)
Typical spectrograms taken with the instrument depicted in
Fig. 3.10 are shown in Figs. 3.11 and 3.12. The aperture of the
spectrograph i s / 12, which is sufficient to give satisfactory exposures
in a few seconds with most ordinary arc and spark sources. The
resolution and dispersion in the ultraviolet are ample for use with
simple emission spectra and for absorption spectroscopy of solutions.
In purchasing any single-setting instrument, care should be taken
to see that the manufacturer has mounted prisms and lenses with
such rigidity that they will not readily get out of focus when once
adjusted, and that the plateholder, if of wood, is constructed so that
it will not warp. The slit should be of high quality, since very fine
lines can be obtained with the resolving power available. It is of
advantage to use 4- X 10-in. plates when possible, since this size is
readily obtainable in almost all emulsions. Manufacturers should be
asked to submit sample spectrograms taken on the instrument to
be purchased.
In cases where higher dispersion is needed than is available from
62
P R I S M S P E C T R O S C O P E S AND S P E C T R O G R A P H S
[§3.8
the medium quartz instrument, recourse is usually had to the Littrow
type of mounting, in order to save space and improve rigidity.
Some types of quartz spectrographs can be obtained with auxiliary
glass optical parts, which render them more suitable for use in the
visible. Though quartz is also transparent in this region, its disp'ersion is so low that quartz spectrographs are not especially useful at
wavelengths longer than about 5200 A, where the ordinary photographic plate becomes insensitive, and special plates must be used in
any case.
3.8. The Littrow Mounting. The device of autocollimation developed by Littrow is widely used with both prisms and plane gratings
in spectrographs designed to give high linear dispersion with a camera
lens of long focus. The principle of the method is illustrated in
Focal
plane
Reflecting surface
Fig. 3.13. Diagram of the optical system of a Littrow spectrograph.
The lens has been reversed to reduce scattered light.
Fig. 3.13. The beam of radiation diverging from the slit is made
parallel by the collimator lens and enters the dispersing system, which
in this case is a 30-deg prism mirror-coated on its back face. The
radiation is reflected from this, passes back through the prism, and
retraverses the collimator, which behaves now as a camera lens and
brings the spectrum to a focus. Special advantages accrue when
quartz is used, for the passage through the 30-deg prism in the reverse
direction compensates for any optical rotation produced in the initial
passage, and crystalline quartz of a single type will suffice.
Two inherent defects keep the Littrow mounting from replacing
other types of prism mounting to the extent that its simplicity and
rigidity would lead one to expect. The proximity of slit and plateholder requires the introduction of a reflecting prism or other device i
§3.8]
PRISM SPECTROSCOPES AND SPECTROGRAPHS
63
to separate the incoming and outgoing beams, and the reflection and
scattering of light from the front face of the colUraator directly back
t o t h e photographic plate is liable to cause objectionable fogging t h a t
is h a r d to eliminate. This false light can often be thrown off the
plate by tipping the lens slightly, thus introducing a certain a m o u n t
of astigmatism, or much of it can be trapped by introducing stops a n d
diaphragms at strategic points. I n any event the inside of t h e case
surrounding a Littrow mount should be thoroughly blackened, and
numerous baffles should be used t o cut down stray light.
Fig. ^.14. Large quartz spectrograph of the Littrow type.
(Courtesy Adam Hilger, Ltd., London.)
A widely used type of Littrow instrument having q u a r t z optical
parts, made by several manufacturers, is shown in Fig. 3.14, where
the Hilger E492 model is illustrated. T h e lens and prism system of
a corresponding model by Bausch & Lomb is shown in Fig. 3.15. T h e
length of the case is slightly more t h a n 6 ft, but since the optical
system m a y be considered as having been folded together in t h e
middle by use of the autocollimation principle, the dispersion is
equivalent t o t h a t of an instrument of the ordinary t y p e almost twice
as long.
T h e quartz prism and lens are mounted on a carriage t h a t moves
along a slide, their position on this being determined by means of a
64
PRISM SPECTROSCOPES AND SPECTROGRAPHS
[§;5.8
scale and index. T h e prism can be rotated t o throw various regions
of the spectrum on the 4- X 10-in. plate. T h e plateholder can be
rotated t o bring it into coincidence with the focal curve for any spectral region between 2000 and 8000 A. In the model illustrated, all
necessary adjustments for any spectral region can be carried out from
the operator's position at the slit end of the instrument.
Spectrographs of this t y p e are found highly satisfactory in analytical
work for which the dispersion of the medium-size quartz spectrograph
is not sufficiently great. In the region 2500 to 2000 A, the linear
Fig. 3.15. Lens and prism system of large Littrow quartz spectrograph.
(Courtesy Bausch & Lomb Optical Company, Rochester. N. Y.)
dispersion given by a large Littrow instrument is as great as t h a t of
a large concave-grating spectrograph (see Table 2.1).
Glass optical p a r t s can be obtained to fit the s t a n d a r d large Littrow
.spectrographs. W i t h glass parts in the Hilger model the spectrum
from 9000 to 4000 A is about 34 cm long, and can be photographed in
two settings on 4- X 10-in, plates.
Hilger manufactures a very large glass-prism I^ittrow spectrograph,
working at aperture j 11, which contains one 'fiO-deg a n d one 30-deg
prism. The.se prisms are 6 in. on a side and 4.6 in. high. A .5-in.diameter camera objective is used, the spectrum from 38o0 to 8000 .\
being 9 in. long so t h a t it can be photographed in one exposure on a
4- X 10-in. plate. T h e lens has been figured to reduce astigmatism
§3.9]
PRISM SPECTROSCOPES AND SPECTROGRAPHS
65
to a negligible amount over a 10-in. region. In order to retain the
advantages of glass and yet to extend to shorter wavelengths the
range of the spectrum covered, ultraviolet transmitting glass is sometimes used in this model, extending its transparency to 2900 A. This
type of glass is especially useful in astronomical work, where the
atmosphere absorbs wavelengths shorter than 2900 A. A glass that
will transmit down to this limit is as satisfactory as the more expensive
quartz and is more dispersive in this region.
A small Littrow spectrograph has been developed by Bausch &
Lomb to obtain moderate dispersion in the ultraviolet at relatively
low cost. This instrument has an optical system of crystalline
quartz and covers the range 2100 to 7000 A, giving a 150-mm spectrum
on a 5- X 7-in. plate. Four standard fixed slits are provided, ranging
from 0.002 to 0.02 mm width, each cut on a protected metallic coating
deposited on a single quartz slide. The manufacturers point out
Photographic
plate
Fig. 3.16. Diagram of Fery spectrograph.
that this type of slit, though inexpensive, has the advantage of
mechanical stability, parallelism of edges, and ease of cleaning, since
the exposed quartz side can be readily wiped with a cloth.
Other applications of the Littrow mounting are described in § 3.11,
where certain instruments having interchangeable optical parts are
discussed.
3.9. The Fery Spectrograph. An ingenious application of the
autocollimation principle was made by Fery, who designed a single
quartz optical unit that combines the properties of prism, mirror, and
lenses. The front face of the prism is usually a cylindrical surface
with axis vertical, so figured that diverging rays from the slit all strike
it at the proper angle for minimum deviation. The rear surface, also
a vertical cylinder, is backed by a metallic coating chosen to give high
reflection in the ultraviolet. The spectrum is brought into horizontal
focus on a surface of fairly great curvature, as shown in Fig. 3.16.
66
PRISM SPECTROSCOPES AND SPECTROGRAPHS
[§3.10
T h e Fery spectrograph presents its greatest a d v a n t a g e in the far
ultraviolet, where loss of light is minimized by the small number of
air-quartz surfaces and the small thickness of optical material used.
These virtues are to some extent offset by the high degree of astigm a t i s m involved, since t h e light is not focused at all in the vertical
direction. Each point on the slit is spread out in t h e spectrum into
a line 2 in. or more in height. N o provision need be made for moving
t h e plateholder u p and down, since various spectrograms can be taken
Fig. 3.17.
Large two-lens quartz spectrograph.
Scientific Corp., Cliicago.)
(Courtesy Gaertner
'
«'
by means of a diaphragm t h a t moves vertically in front of the plate,
covering ail of each spectrum line except a region of the desired height. ,
T h e astigmatism produces very straight and even lines, so t h a t Fery
spectrograms are usually of excellent appearance. T h e instrument is
compact and readily portable.
3.10. The G a e r t n e r Large Quartz Spectrograph. Gaertner has
introduced a large quartz-prism sj^ectrograph designed t o eliminate
the disadvantages of t h e Littrow mounting while retaining its compactness. I n this instrument no a t t e m p t is made t o use a single lens
§3.10]
PRISM SPECTROSCOPES AND SPECTROGRAPHS
67
for both coUimating and camera lenses, b u t the length of t h e instrument is cut in half by the introduction of a large first-surface mirror.
In addition, the right-angle prism just behind the slit has been
eliminated. I n this way it has been found possible to reduce greatly
the scattered and stray light customarily found with the Littrow
mounting.
A view of the instrument is shown in Fig. 3.17, and Fig. 3.18 shows
its optical system. T h e light entering the slit falls directly on a collimator lens and passes through a quartz prism, after which it is
focused by a camera lens in the usual manner. After leaving this lens
Fig. 3.18. Optical arrangement of spectrograph illustrated in Fig. 3.17.
(Courtesy Gaertner Scientific Corp., Chicago.)
the light path is reversed by a plane first-surface mirror so t h a t t h e
camera can be placed immediately beside the slit. This arrangement
gives the advantage possessed b y the Littrow mounting of bringing
all adjustment controls to one end of the spectrograph.
T h e plateholder will take a 14-in. plate which at the dispersion used
will cover the spectrum from 2500 to 5900 A with a single exposure.
A handwheel control adjusts the wavelength region, and brings t h e
lenses to the proper positions and the plateholder t o the proper tilt,
giving accurate focus of any part of the spectrum from 2000 to 8000 A.
An arrangement is provided whereby the wavelength at the center of
the plate and the wavelengths a t each end are projected on a large
grovmd-glass screen on top of t h e spectrograph, where they can
readily be observed.
PRISM SPECTROSCOPES AND SPECTROGRAPHS
68
[§3.11
3.11. Spectrographs with Interchangeable Optical Systems. Some
of the prism spectrographs described above can be obtained with
interchangeable optical p a r t s . T h e aiitocoUimation principle (§ 3.8)
lends itself particularly to interchangeabiUty, since the focusingdispersing unit m a y consist of a lens and one or more prisms, a lens
Fig. 3.19. Quartz monochromator employing a Cornu prism in a Wadsworth mounting. (Courtesy Gaertner ScientiSc Corp., Chicago.)
«
and a plane grating (§ 4.1), a Fery prism (§ 3.9), or a concave grating
(§4.6).
Hilger manufactures an interchangeable mounting of the Littrow
type t h a t can be obtained fitted with glass or q u a r t z lenses of 100,
1.50, or 300 cm focal length. Behind the lens can be placed a single
30-deg reflecting prism, a combination of one of these with a single
60-deg jjrism, or a plane grating. Alternatively, a concave grating
can be used without t h e lens, forming an Eagle mounting of the t y p e
discussed in § 4.6. T h e design of this line of instruments has been
\
§3.12]
P R I S M SPECTROSCOPES A N D S P E C T R O G R A P H S
69
standardized to ensure interchangeability of the optical systems.
Plates 4 X 10 in. in size are used, and two models are made—one in
which all parts are adjusted by hand, the other in which automatic
adjustments are provided. The Bausch & Lomb large Littrow instrument with interchangeable optical parts is also provided with automatic focusing for predetermined regions of the spectrum.
Intermediate
slit
':
:^
Fig. 3.20. The optical system of the van Cittert zero dispersion monochromator. The position and width of the intermediate slit determine the spectral
range transmitted by the instrument. The dash lines show that the first
camera lens is imaged on the second collimator lens.
3.12. Prism Monochromators. For use in the visible region,
almost any constant-deviation spectroscope can be converted into a
monochromator, if an exit slit is substituted for the eyepiece. For
the ultraviolet region a constant-deviation method due to Wadsworth"
is frequently used, in which a Cornu prism and reflecting mirror are
Fig. 3.21. Young-ThoUon arrangement of two 30-deg dispersing prisms
with lenses.
rotated to vary the spectral band emerging through the exit slit. The
optical system of this device is shown in Fig. 3.5b, and a quartz
monochromator employing the principle is illustrated in Fig. 3.19.
Since uncorrected quartz lenses have strong chromatic aberration,
it is necessary in ultraviolet instruments to refocus the coUimating
and focusing lenses for each new wavelength region, as well as to turn
« F. L. O. Wadsworth, Astrophji.t. Jour.. 1, 232 (1895).
. ,
70
PRISM SPECTROSCOPES AND SPECTROGRAPHS
[§3.12
the prism table. Infrared monochromators (Chapter 17) use mirrors
instead of lenses and thus eliminate the need for refocusing. A
monochromator principle which is very effective for use in that region
involves a single mirror which serves as both collimator and focusing
element. Parallel light from the collimator traverses the prism in
one direction, and then is reflected by a plane mirror (or a backed
30-deg prism) again through the prism to be focused on the exit slit
by the original collimating mirror. Improved surfaces for ultraviolet
reflection are making possible the utilization of this simple system
in the visible and ultraviolet regions, but scattered light must be
Fig. 3.22. Monochromator employing Young-Thollon prism arrangement and
achromatic collimator and telescope lenses. (Courtesy Farrand Optical Company, New York.)
reduced by other means when this mirror analogue of the Littrow
mounting is used.
On account of the frequent importance of reducing scattered light
to a minimum, double monochromators of various types have been
designed, of which outstanding models are those of van Cittert,^
manufactured by Ivipp and Zonen, and of Miiller, manufactured by
Hilger. A diagram of the optical system of the former is shown in*
Fig. 3.20. Several sizes of ultraviolet monochromators that use
quartz-lithium fluoride achromatic lenses and double prisms OT the
Young-Thollon type (Fig. 3.21) are manufactured by the Farrand
Optical Company. These instruments, one of which is illustrated inFig. 3.22, may also be obtained with glass optics for the visible regionGrating monochromators are discussed in S 4.11.
' P . H. van Cittert, Rev. d'Optique, 5, 393 (1920); Physica, 3, 181 (1933).
CHAPTER
4
Diffraction-Grating Spectrographs
A G E N E R A L COMPARISON OF P R I S M AND GRATING
SPECTROGRAPHS
was made in Chapter 2. The present chapter deals with the methods
of mounting diffraction gratings that have been found most useful
and with descriptions of commercial grating spectrographs.
The advantages of the diffraction-grating spectrograph over the
prism instrument may again be summarized as follows: broader
spectral coverage, greater available dispersion and resolving power
per unit cost, greater uniformity of dispersion, greater light transmission in certain cases, and the possibility of greater freedom from
scattered light. The relative disadvantages are greater astigmatism
(except as discussed below); more rapid deterioration with age; and
until the late 1940's, comparative scarcity. Costs of the two types
of instruments are not greatly dissimilar.
Large gratings of long focal distance are used mainly in physics
research laboratories. Such gratings involve mountings that may be
from 20 to 35 ft long, filling an entire room, and under these circumstances the various parts of the spectrograph are usually mounted
separately. Most commercial grating spectrographs, on the other
hand, are small or medium-sized instruments ranging in length from
3 to 17 ft, which are built to be handled as a single unit.
Small diffraction gratings are tested and guaranteed by the spectrograph manufacturer, but large gratings must usually be obtained
directly from laboratories that operate ruling engines. Although
gratings are commonly sold on a guarantee basis, the user should be
prepared to test thein. Methods for the selection and testing of
diffraction gratings are discussed in § 5.3.
4.1. Plane-Grating Spectrographs. Transmission gratings, as discussed in.§ 2.5, are seldom used in any but small spectrometers and
in instruments for student use. Most instruments in which plane
gratings are used for the general purposes of spectroscopy contain
71
72
DIFFRACTION-GRATING SPECTROGRAPHS
[§4.1
gratings of the reflection type and are ordinarily used in a Littrow
mounting similar to that of the autocollimation prism spectrograph
(§ 3.8). A typical mounting of a plane diffraction grating is shown
in Fig. 4.1.
^
Several large plane-grating installations are in existence, the largest
being probably an autocollimating instrument at -Mount Wilson
Observatory, which has a focal length of 75 ft. Plane gratings having '
8 in. of ruling and giving nearly theoretical resolving power in the
second or third orders have been produced occasionally.
A grating so mounted is somewhat similar in its behavior to a
concave grating in the Eagle mounting (§ 4.6). It has the advantage/
over the latter that it gives stigmatic images over the narrow spectral
range ordinarily used at any one setting, with a resulting gain in
brightness and resolution in the higher orders. To change from one"
Fig. 4.1. Diagram of a Littrow mounting of a plane diffraction grating. S, slit;
M, reflecting mirror or prism; L, collimator and telescope lens; G, grating;
A B, spectrum.
spectral region to another with such a spectrograph, it is necessary to
rotate the grating, refocus the lens, L of Fig. 4.1, and rotate the
plateholder so that it will lie in the focal curve determined by the
color correction of the lens. A lens carefully corrected for chromatic
aberration is required if the plane grating is to be used in orders higher
than the first; otherwise, lines of overlapping orders will not be
brought to a focus on the same curve.
The lens used should have the same aperture as the grating. Since
chromatically corrected lenses of large sizes have in the past been
obtainable only of glass, plane-grating spectrographs have ordinarily
been used only for the visible region and the near ultraviolet and,
infrared. If a large quartz lens is available^ even without color
correction, the ultraviolet first order can be used, but special arrangements must be made to throw out the overlapping second order at
wavelengths longer than 4000 A. A right-angle prism itiust be used
behind the slit to separate slit frojn plateholder, except in instruments
§4.2]
DIFFRACTION-GRATING SPECTROGRAPHS
73
of very long focal length, where the two can be separated without
introducing a large angle of incidence that would increase astigmatism
and coma.
A mirror can be used for collimating the light on a plane grating,
and this type of mounting is found especially effective with monochromators. It is discussed further in § 4.11, and for the infrared
in Chapter 17.
4.2. The Rowland Concave Grating. One of the most important
advances in the history of spectroscopy occurred in 1882, when
Rowland ^ showed that a spherical concave mirror, ruled with parallel
lines equally spaced along the chord of its arc, will produce spectrum
lines in sharp focus on a circle whose diameter is equal to the radius
of curvature of the mirror. This "Rowland circle" is shown in
Fig. 4.2. Eliminating the need for
any transparent materials as it does,
the concave grating has become one
of the most powerful tools of spectroscopy. I t can be made to provide
greater dispersion and resolution than
are obtainable with prisms and can
be used at any wavelengths for which
its rulings are properly spaced. A
single grating has been used to cover
the range from 100 to 11,000 A.
In the ruling of concave gratings,
Fig. 4.2. Diagram of the
spacings of approximately 7500,
Rowland circle.
S, slit:
6,
10,000, 13,000, 25,000, or 30,000 lines grating; AB, spectrum; /J/2.
per inch are commonly used. Certain radius of Rowland circle; R,
of the largest concave gratings thus radius of curvature of grating.
far successfully ruled, the so-called
"seven-inch" gratings, contain about 180,000 lines on a ruled area
about 15 X 5 cm. In the higher orders, resolving powers of
400,000 have been attained on occasion. The maximum useful radius
of curvature of a "seven-inch" grating for photographic purposes is
about 10 meters, as may be demonstrated by calculating the plate
factor required to match the resolving power of a grating to that of
the photographic emulsions commonly used with it. Emulsions of
suitable speed for large grating spectrographs are capable of resolving about 30 lines/mm (§ 7.5). The resolving power of an
» H. A. Rowland, Phil. Mag., 13, 469 (1882); 16, 197, 210 (1883).
74
DIFFRACTION-GRy^TING SPECTROGRAPHS
[§4.2
excellent grating with 6 in. of ruling is not likely to be greater than
300,000 (§ 2.5). At wavelength 5000 A, substituting in the formula
Pr = \/d\, we obtain d\ = 0.016 A. If this spectral range is to
cover not more than 0.033 mm of emulsion, a plate factor of 0.5 A/mm
is required. This would be obtained in the third order of a 15,000line-per-inch grating of 10-meter radius (§ 2.5).
A concave grating is somewhat more difficult to rule than a plane/
grating of equal size and spacing, and in general the longer the radius!
of curvature, the easier it is to produce a good grating. Standard
radii of curvature are approximately 1, 2, and 3 meters, 10, 15, and
21 ft, and 10 meters. The actual radius obtained may vary as much
as ± 5 per cent from the value ordered; hence it is usually wise to
build a spectrograph with ample flexibility of adjustment.
Up to about 1932, most concave gratings designed for use in the
visible and ultraviolet regions were ruled on speculum metal, which
has moderately high reflecting power in the visible region but much
less at shorter wavelengths, reaching a low of 10 per cent or less at
normal incidence in the extreme ultraviolet. R. W. Wood has ruled
concave gratings on glass for the vacuum region, and many experiments have been made on coating these with evaporated or sputtered'
metals to increase their reflecting power.^ The most satisfactory
gratings at present are those ruled on an aluminum surface that has
been evaporated on glass. The only question in regard to these" is
that of permanence—any damage to the aluminum surface »may
damage the -rulings irreparably. For this reason, experiments have
been conducted with a new technique of ruling the grating, on a goldon-chromium surface evaporated on glass, which is then coated with
evaporated alurninum.'
Concave gratings can sometimes be obtained from university and
other laboratories which operate ruling engines. The physics department of the Johns Hopkins University has in the past supplied many
gratings. Gratings are being increasingly supplied mounted in
spectrographs by the firms listed on page^O as dealing in grating
equipment.
In mounting a concave grating the slit may be placed anywhere on
the Rowland circle. The location chosen will depend on the type of
work to be done. The resulting positions of the various orders' and ;
wavelengths can be quickly determined bV drawing a diagram like
2 J. Strong, Astrophys. Jour., 83, 401 (1936).
* J. Strong, unpublished communication.
f
§4.2]
DIFFRACTION-GRATING SPECTROGRAPHS
75
Fig. 4.3, after Beutler/ which shows the distribution of wavelengths
for various positions of the sUt as calculated from the formula of § 2.5,
X = —^ (sin i ± sin 6)
where the symbols have the meanings given there.
The five mountings of concave gratings most commonly used are
described in succeeding sections. In selecting a mounting, one should
Diffracted wave-lenjfhs by a 2C(X0 line/inch grating in first order
lapOOA
10,OOM
-90" -80" -70* -6tf
-40° -30" -20' -10' (f
10* 20* 30* 40' 50" 60*
Angle of diffraction
70* 80* 90"
Fig. 4.3. Angles of incidence and diffraction for various wavelengths for a
30,000-line-per-inch diffraction grating. (From H. G. Beutler, by permission
Jour. Opt. Soc. Am.)
consider the grating orders to be used, the wavelength range to be
covered, the degree of astigmatism that can be tolerated/ the brightness of the resulting spectra, the freedom from spurious lines (which
may depend on angle), and the departure from uniform dispersion
(smallest on the normal). In general, the most satisfactory spectra
are obtained in the direction of the normal to the grating.
A large concave grating should be mounted only in a room that
can be kept clean and dry. The inside of the room is itself the
camera and must be kept dark during an exposure. The former
custom was to paint the inside of a grating-room black, but modern
' H. G. BeuUer, Jour. Opt. Soc. Am., 35, 318 (1945).
• G. H. Dieke, Jour. Opt. Soc. Am., 23, 274 (1933).
76
DIFFRACTION-GRATING
SPECTROGRAPHS
[§4.2
practice inclines toward light colors for the ceiling and walls, with
a black strip extending up perhaps 5 ft from the floor. All developing
work should be carried out in a separate room from that in which the
gating is mounted. Excessive vibration of a large-grating mounting
will of course make it impossible to obtain sharp spectrum lines even
with short exposures, and with long exposures it is also necessary to
guard against temperature changes of the grating itself. As th^
grating expands, the distance between its rulings increases and the
dispersion correspondingly decreases, so the relative positions of the
lines in the spectrum are shifted. For this reason most
rooms 'containing large gratings are thermostated, and the
temperature is held constant
to 0.1°F or better. To test
whether the constancy of temperature is sufficient in any
given case, three brief superposed exposures may be taken
at 10-hour intervals; if the resulting spectrum lines are as
sharp as those taken in a single
equivalent exposure, no loss of
resolution from this cause is
to be expected in a 20-hour
exposure to weak light.
Fig. 4.4. Rowland mounting. S, slit;
Ou Gi, Oa, various positions of grating along
A convenient height above
bar OY; AB = B where R = radius of the floor should" be chosen for
grating; Bi, Bj, Bj, various intercepts of
the plane containing the slit,'
grating-normal bar with bar OX.
grating, and focal curve. Fortysix inches is a standard value that gives suitable clearances for mounting optical benches and similar parts on ordinary tables. Slit, grating,,
and plate carriers should all be based on heavy concrete or brick pierS '
resting on a solid foundation. Wherever any part passes through a
wall, sufficient clearance should be provided around it to ensure
against vibrations being communicated to any member involved
in the optical system.
."
Beutler* has given a very thorough discus.sion of the theory of the
'' H. G. Beutler, Jo2tr. Opt. Soc. Am., 35, 311 (1945).
54.3]
DIFFRACTION-GRATING
SPECTROGRAPHS
77
concave grating and has prepared many charts that show at a glance
its properties in various mountings. His paper should be consulted
by anyone contemplating the construction of a grating spectrograph.
4.3. The Rowland Mounting. The classical mounting for the
concave grating, used less nowadays than formerly, is that originally
described by Rowland ' and illustrated in Fig. 4.4. The plateholder
and grating are rigidly mounted at opposite ends of a stiff beam
whose ends are held on carriages that run on tracks placed at right
angles to one another. The slit is placed at the junction of these
two tracks, and the light passing through it falls on the grating and
is spread by this around the Rowland circle. Only that portion of
the spectrum is used which falls on a plateholder placed at the normal
to the grating. This arrangement
gives a spectrum of almost uniform
dispersion, a property of great
advantage when wavelength measurements are to be made in terms of
only a few standard lines. Nowadays, however, "so many standards
of wavelength are accurately known
that wavelength interpolations need
be made over only short distances,
and a normal spectrum is not so
necessary as it once was.
The disadvantages of the RowFig. 4.5. Optical system of modland mounting are that only a ified
Abney mounting illustrated in
limited region of the spectrum can Fig. 4.6. R, Rowland circle; Si
be photographed at one setting; and S2, fixed slits so placed t h a t
that it has a high degree of astigma- either of two different regions of the
spectrum may be photographed by
tism, so that much intensity may illuminating the appropriate slit;
be lost, especially in the higher P, plateholder; C, light-tight ease.
orders; and that the highest orders
of the grating cannot be reached. Also, the mounting is somewhat
cumbersome, and the fact that the grating and plateholder both move
is a disadvantage. The Rowland mounting can be used most satisfactorily with gratings relatively free from error of run; if this is
present, the focal curve tends to depart from the expected circle.
In order to secure good temperature control, necessary for research
• H. A. Rowland, Phil. Mag., 16, 197, 210 (1883).
78
DIFFRACTION-GRATING SPECTROGRAPHS
[§4.4
purposes, the Rowland mounting is sometimes arranged vertically,
the grating being placed at the bottom of a pit, and the end of the arm
that carries the plateholder moving on a horizontal track at the floor
level. This mounting results in saving of space but has disadvantages, such as a horizontal slit and a vertical illuminating beam;
also, the grating, lying face up near the bottom of the pit, is liable to
injury from falling objects and to coating with dust, and is rather
inaccessible. The chief advantage lies in the constancy of temperature obtainable at the bottom of a deep pit.
4.4. The Abney Mounting. This mounting,^ shown in Fig. 4.5, is
Fig. 4.6. Commercial spectrograph employing a 1.5-meter concave grating in
a modified Abney mounting. (Courtesy Applied Research Laboratories, Pasadena, Calif.J
a variant of the Rowland mounting but has the property that both
grating and plateholder are kept stationary while the slit is moved
when different regions of the spectrum are to be photographed. The
Abney mounting has never come into wide use with large gratings
because of the difficulty of moving the source, condensing len.ses, and
other external equipment to keep pace with the slit whenever the slit
is moved, so that a different spectrum region will be thrown onto the
plate. It is more convenient to have a number of slits for one
instrument, from two to ten sometimes being provided. This is
practicable because the same fairly long plateholder, capable of
photographing several feet of spectrum at one time, can be used
for all slits.
r
"
' W. de \Y. Abney, Phil. Trans., 177, 11, 457 (1886).
- ' I
J
A
/
,§4.5]
DIFFRACTION-GRATING SPECTROGRAPHS
79
A commercial form of the Abney mounting, widely used for spectrographic analysis of materials, is the instrument rbanufactured b y
the Applied Research Laboratories and shown in Fig. 4.6. T h i s
mounting uses a small grating having a ruled width of a b o u t 2 in. a n d
height of about 1 in., with 150 cm radius of curvature and 24,000 lines
per inch. This combination results in a plate factor in the first order
of about 6 A m m . T w o slit positions are provided, each covering a
spectral range of about 2200 A in t h e first order, so t h a t the spectrum
from 2130 to 6570 A can be photographed in two exposures.
Because of the high degree of curvature of so small a Rowland circle,
film is used -instead of plates for photographing the spectrum. A
convenient holder is provided so
t h a t motion-picture film of s t a n d a r d
35-mm width can be used. Special
equipment is furnished for use w i t h
the instrument to simplify the p r o cedures of handling, developing, a n d
drying film.
T h e Abney mounting sufl'ers from
t h e same limitations of bulk a n d
astigmatism as the Rowland m o u n t ing.
4.5. The Paschen-Runge Mounting.
T h e mounting most commonly used at present for large
Fig. 4.7. Diagram of Paschenconcave gratings of the research Runge mounting of the concave
t y p e is t h a t originally described b y grating. S, slit; G, grating; AB,
spectrum.
Paschen and Runge.^ I t has t h e
great advantage t h a t slit, grating, a n d plateholder are all fixed, so t h a t
all parts of the spectrum are in focus at all times and its entire extent
can be photographed with a single exposure on many plates. T h e
m o u n t i n g can be arranged so t h a t almost the complete Rowland
circle is available, or the entire spectrum can be covered b y using one
or more orders on only one side. A diagram of a typical arrangement
of t h e Paschen-Runge rtiounting is shown i n l j i g , . . 4 ^ Figure 4.8
reproduces a photograph of a portion of a 10-meter P a s c h e n - R u n g e
mounting in the Spectroscopy Laboratory of the M a s s a c h u s e t t s
^ I n s t i t u t e of Technology.
' C. R. Runge and F. Paschen, Abh. d. K. Akad. d.Wiss. z. Berlin, Anhang 1 (1902).
•<%<»..
^
[§4.5
DIFFRACTION-GR\TING SPECTROGRAPHS
80
One of the first problems in setting up a grating in tiie PaschenRunge mounting is the angle of incidence to be chosen for its illumination. Although the whole circle m a y be available, the region at the
normal to the grating is the most valuable because there are found'
n'*^l^
1Y
1 '•
__-«.* ^
1 "
^-^' , .
S
II
^^9H
'
%
w rWxJ'w'^
^^^HH
'•
"^^9R
-
- •
•'
\^^
x^.'..
•
%
'"';\
0
Fig. 4.8. Paschen-Rimge mounting in the M.I.T. Spectroscopy
Laboratory.
. jp
,\
t h e most uniform dispersion) the least astigmatism (for a chosen
angle of incidence, b u t not the least attainable—see Fig. 4.18, page 91).
a n d in some cases the highest resolving power. T h e slit is placed on
§4.5]
DIFFRACTION-GRATING SPECTROGRAPHS
81
the normal only when it is necessary to obtain the low orders on both
sides of the grating, as for certain types of intensity measurements.
Where higher orders are to be used, illumination at 12- to 60-deg
incidence is common.
It is not unusual to provide Paschen-Runge mountings with two
slits, one placed for illumination at about 13 deg and the other at
40 deg. The small angle of incidence is used when low orders are to
be studied, and the large angle for higher orders. The two slits give
the added advantage of providing space for two complete source
setups; thus cumbersome apparatus for the Zeeman effect, for
example, can be left in place at one slit while the other is available
for general use.
Where the main use of a Paschen-Runge mounting is to be in
making wavelength measurements, a fixed track is usually provided
To grating
-Latch to hold plate against trock
-Upper track
-Emulsion
- Plate
- Lower track
Table top
|
Fig. 4.9. Detail of plateholder track for Paschen-Runge mounting in
the M.I.T. Spectroscopy Laboratory.
to hold a series of plates bent to the Rowland circle. One method of
holding plates on such a track is shown in Fig. 4.9, the long (2- X 20in.) plates being clipped with their emulsion side against the back of
the track, to avoid any displacement due to varying glass thickness.
The posts holding the track are usually bolted to heavy slabs of
Alberene stone or slate mounted on concrete piers to give great
rigidity and are arranged so as to permit a certain amount of adjustment for focusing before being finally bolted in place.
For routine photography of selected spectral regions, and for
intensity work where a number of spectra are to be photographed on
the same plate, it is desirable to provide cassettes (Fig. 4.10) that
slide along a horizontal track following the Rowland circle and carry
plateholders taking 4- X 10-in. or similar plates. Both types of track
are visible in Fig. 4.8. Each has the spectrum brought into proper
82
DIFFRACTION-GRATING SPECTROGRAPHS
[§4.5
Cassette
Sleeper
Sleeper
Fig. 4.10. Detail of cassette for Paschen-Runge mounting in
the M.I.T. Spectroscopy Laboratory.
Fig. 4.U.
Small Paschen-Runge spectrograph employing a concave grating
replica. (Courtesy Central Scientific Company, Chicago.)
w
§4.6]
83
D I F F R A C T I O N - G R A T I NG S P E C T R O G R A P H S
focus on it when its own slit is used. Calculations indicate that by
properly displacing the slit the focal curve of a 10-meter grating may
be caused to depart as much as a foot from the Rowland circle without
introducing objectionable aberrations.
A small model of the Paschen-Runge mounting appears in commercial form in the Cenco grating spectrograph produced by the
il
Fig. 4.12.
I
Spectrogram of a copper arc in the region 5400-4400 A, taken with
the Cenco grating spectrograph.
Central Scientific Company (Fig. 4.11). A replica grating having a
ruled surface about 1 in. square with a radius of curvature of about
1 meter produces a spectrum covering the range 2300 to 8000 A in
one setting. The entire spectrum is covered with a plate factor of
16 A/mm on a film 10 in. long, so the instrument can be kept in
permanent adjustment. A section of a
spectrogram taken with this spectrograph is shown in Fig. 4.12.
4.6. The Eagle Mounting. A mounting described in detail by Eagle'" but
used long before in vacuum spectrographs" is not only economical of space
but also keeps astigmatism as low as
is possible withont increasing the
complexity of the optical system, and
simplifies the control of the grating
temperature.
The Eagle mounting is
Fig. 4.13. Eagle mounting
illustrated in Fig. 4.13. This mounting
of the concave grating. R,
Rowland circle; S, slit; r, reoccupies a long narrow space, a characflecting prism; G, concave
teristic that led to its extensive use in
grating; P, plate; C, lightvacuum spectrographs of the normal-intight case.
cidence type in the manner originated by
Lyman. Higher orders can be reached than in the Rowland Mounting
and the astigmatism is less. In changing from one wavelength range to
i» A. Eagle, Axtrophys. Jour., 31, 120 (1910).
" T. Lyman, Spectroscopy of the Extreme Ultraviolet. New York: Longmans, Green
and Company, 1928.
;
84
DIFFRACTION-GRATING SPECTROGRAPHS
[§4.^
another, it is necessary to turn the grating, change its distance f r o m /
the plate, and rotate the plateholder. The Eagle mounting of t^e
concave grating is similar to the Littrow mounting of the plane
grating, but it is superior in that no lens is needed, so that it can be
used in all spectral regions. It does not suffer greatly from the
Littrow's defect of scattering light directly back onto the photo-
Fig. 4.14. Optical bench and plateholder end of commercial Eagle spectrograph, showing power unit for arc and spark sources. (Courtesy Baird Associates, Cambridge, Mass.)
graphic plate. I t has the disadvantage of not being as stigmatic as
the Littrow.
i
The Eagle mounting is used in a commercial instrument manufactured by the Baird Associates, shown in Fig. 4.14. In the standard
model a grating with 4 in. of ruling, having 15,000 lines to the inch
and a 3-meter radius of curvature, is used, which gives 5.2 A/mm
plate factor in the first order. The spectrum from 2000 to 10,000 A
can be covered in several orders in a series of exposures, with a range
.1.
§4.7]
DIFFRACTION-GRATING SPECTROGRAPHS
85
of 1200 A recorded at one exposure in the first order. The three necessary focusing adjustments are controlled by electric motors operated
by push buttons, and it is a matter of but a few moments to bring
the spectrograph to focus in a new spectral region. One motor drives
the screw that moves the grating forward or backward on stainless
steel ways, a second motor turns the grating to the proper angle, and
a third racks the plateholder. All adjustments are controlled from
the front panel, and automatic end stops are provided. Each motion
is controlled by two switches, one of which provides high-speed
adjustment forward and reverse, and the other a one-tenth-speed
motion for accurate setting. Revolution counters are connected
through flexible cable to the part controlled so that accurate setting
is easy. Sufficient travel is provided to allow the red of the fourth
order to be covered, in which order the plate factor is 1.3 A/mm.
The gratings used have a high concentration of visible light in the
first order and are sometimes even faster than prism instruments.
Four- by ten-inch plates or films can be used, but the plateholder
is arranged to hold any plate 10 in. long and narrower than 4 in.
Models of shorter focus, having correspondingly greater compactness
and less dispersion, are also manufactured by the Baird Associates.
ANGULAR
RANGE
ANOLC
AS COVEREO BV D I F F E R E N T
htOUNTlNGS
O F OIFFBACTIOM
Pig. 4.15. Angular ranges of various grating mountings. (From H. G.
Beutler,"^ by permission Jour. Opt. Soc. Am.)
Figure 4.15, after Beutler,'' shows the angular range covered by the
various grating mountings.
4.7. The Wadsworth Stigmatic Mounting. A disadvantage for
many purposes of all mountings of the concave grating previously
discussed is that they are astigmatic to a greater or less degree (see
'2 H. G. Beutler, Jour. Opt. Soc. Am., 35, 318 (1935).
DIFFR.VCTION-GRATING SPECTROGRA.PHS
86
[§4:
§ 4.9).
inated,
making
parallel
Wadsworth^' noted that this astigmatism might be elimat least over any desired short region of the spectrum, by
use of the fact that if a grating is illuminated with a beam of
light, a normal stigmatic spectrum is produced at the normal
to the grating. The dispersion is then
,s 0,
cut in half, however, since at the normal
the new focal curve lies halfway out to the
Rowland circle. Various workers used a
large convex lens to make parallel the light
on the grating, and Meggers and Burns"
originated a mounting of this type in
which a concave mirror is used. The loss
in light from the extra reflection is more
than compensated by the increased angular aperture of the system and by the
elimination of astigmatism.
Fig. 4.16.
Wadsworth
A Wadsworth mounting of the Meggersmounting of the concave
Burns
type is illustrated in Fig. 4.16.
grating. S, slit; M, concave
mirror; G, concave grating; Grating and plateholder are connected by
P. plateholder; R, bar along a rigid bar, but provision must be made
which plateholder slides; 0,
axis of rotation of grating and for adjustment of their distance apart,
of bar R; C, light-tight case. since the focal curve is a parabola. The
grating turns with the bar so that the
plateholder is always on its normal, whereas the slit and concave
mirror remain fixed. The curvature of the plate must be changed
slightly from one region of the spectrum to another. Though
truly stigmatic images are obtained only on the normal, for practical
purposes a range of many hundred angstroms can be arranged for use
at one setting. In general, a length of spectrum equal to about
one-sixth of the distance from plateholder to grating will be found
in sufficiently stigmatic focus to permit use of a rotating-sector disk
or other photometric device at the slit.
When designing a Wadsworth mounting one should select a grating
that does not have more than 15,000 or 20,000 lines per inch. In the
formula for wavelength position,
•
mX =
A
N
, .
.
.
.
(sm I ± sm 6)
" F . L. O. Wadsworth. Astrophys. Join.. 3, 34 (1896).
" W. F. Meggers and K. Burns, Bur. Standards Sci. Paper 411, 18, 185 (1922).
§4.7]
DIFFRACTION-GRATING SPECTRO GRAPHS
87
B is always close to 0 deg for the Wadsworth mounting, so the formula
reduces to
rnK =
TT
sm i
N
Thus it will be seen that with a 30,000-line-per-inch grating the longest
wavelength that can be reached at any reasonable angle of incidence i
(not over 43 deg) is about 8000 A. The wavelengths that will appear
on the normal for a 15,000-line-per-inch grating at any given angle
for a stigmatic mounting are given in Table 4.1.
TABLE 4.1
WAVELENGTHS AND FOCAL DISTANCES ON THE NORMAL IN THE WADSWORTH
MOUNTING FOR VARIOUS ANGLES OF INCIDENCE
Angle of
incidence
8°
10°
12°
14°
16°
18°
20°
24°
28°
32°
36°
Wavelength
on normal
2359 A
2943
3524
4100
4672
5238
5797
6894
7958
8981
9963
Focal distance
(fraction of R)
0.5024
0.5038
0.5055
0.5075
0.5099
0.5126
0.5156
0.5226
0.5311
0.5411
0.5528
The figures in Table 4.1 hold for a 15,000-line-per-inch grating in the first order.
If other orders are used, the wavelengths are to be divided by the order number.
The wavelengths for a 30,000-line-per-inch grating are obtained by dividing those
in the table by 2. The focal distance in any unit is given by R, the radius of curvature
of the grating in that unit, multiplied by the fraction in column 3.
A very satisfactory Wadsworth stigmatic mounting in operation at
the Massachusetts Institute of Technology, where two instruments of
this type have been in frequent use for many years, consists of a
35-ft concave grating used in conjunction with a 7-in. aluminized
glass mirror of equal radius of curvature. A plateholder 30 in. long
is provided that will hold one, two, or three 4- X 10-in. spectrum
plates, or one centered 2- X 20-in. plate, at one time. Arrangement
is made for moving this plateholder up and down so that as many as
40 spectra can be photographed on a single plate. The fixed bar
88
[§4.7
DIFFRACTION-GRATING SPECTROGRAPHS
on which rod R in Fig. 4.16 slides is calibrated with a wavelength
scale. In changing from one spectral range to another, the movable
rod R is first put in the desired position, so that the proper spectral
region will be centered on the plateholder P, and then the plateholder
is moved along the rod to the position given by a numerical table of
focal positions. The only other adjustment required is a slight
change in curvature of the plateholder. This can be made flexible,
or separate plateholders can be provided for the various regions. The
curvature adjustment can be made by means of calibrated screws
at each end of the plateholder. In one instrument, all other adjustments are made with electric motors; and the racking adjustment,
Fig. 4.17. Commercial 21-ft concave-grating spectrograph of the Wadsworth
type.
(Courtesy Jarrell-Ash Company, Boston.)
which is also done electrically, can be controlled from outside the
room in which the grating is mounted.
The plate factor obtained with the Massachusetts Institute of
Technology instruments is 3.3 A/mm in the first order and 1.65
A/mm in the second order. Any spectral region from 2000 to 10,000
A is readily available, and in the first order a region 2500 A long can
be photographed at one time on three 10-in. plates placed end to end.
Each instrument occupies a space about 18 ft long by 12 ft wide.
One of these spectrographs, having a grating that throws most of
its light into one first order, is very fast, a 3-second exposure being
sufficient to give a strong spectrogram of an iron arc throughout most
of the visible and ultraviolet regions.
The Jarrell-Ash Company manufactures a spectrograph of the
Wadsworth type,'^ shown in Fig. 4.17.
.
•
—
^
15 R. F. Jarrell, Jovr. Opt. Soc. Am., 32, 666 (1942).
'•
' •
/
/
'
§4.8]
DIFFRACTION-GRATING SPECTROGRAPHS
89
F o r m a n y routine purposes, such as qualitative and q u a n t i t a t i v e
spectrographic analysis of materials, it is convenient to have a
Wadsworth mounting in fixed a n d permanent focus. For analyzing
ferrous materials and other spectra rich in lines, it is desirable to h a v e
a plate factor of 5 A / m m or less. M o s t routine analyses can be m a d e
using lines lying in the region 2400 t o 4400 A. T o meet the requirem e n t t h a t this 2000-A range should be photographed on a single plate
with a plate factor of 5 A / m m , a plate 400 m m , or about 16 in. long,
is needed. Or one can photograph the entire range from 2000 t o
5000 A on a plate only 20 in. long, covering 2500 to 5000 A in the first
order and 2000 to 2500 A in the second. Such overlapping is m u c h
less objectionable in emission analysis t h a n is often supposed, because
of t h e high dispersion available with such gratings.
T h e criterion t h a t the stigmatic range in a Wadsworth m o u n t i n g
is a b o u t one-sixth of the grating-plate distance indicates t h a t we
should make this distance at least 10 ft in designing a fixed-focus
spectrograph. A plate factor of 5 iV'mm will be obtained a t a focal
distance of 10 ft with a grating having about 15,000 lines per inch.
T h u s a 6-in. concave grating with 15,000 lines per inch and having
a 21-ft radius of curvature could serve as the heart of such an instrument.
4.8. T h e Choice of a Grating Mounting. Small commercial concave-grating instruments, which are of necessity portable, usually
are m a d e with the Eagle mounting, or in some cases with the AbneyW h e n one is faced with the necessity of choosing the most suitable
m o u n t i n g for a large grating, the first consideration must be t h a t of
space. Where only a long narrow corridor or vertical shaft can be
used, the Eagle mounting is suitable. T h e cross section of its containing box is determined only by the length of spectrum t o be
photographed a t one setting and by the baffles needed to cut down
s t r a y light.
If a room of medium size (say 12 X 15 ft) is available, a 21-ft
grating in the Wadsworth m o u n t i n g will probably be found more
useful t h a n a shorter-focus grating in any other mounting. T h e
grating of longer radius will cost no more t h a n a shorter one for t h e
same area of ruling, and the advantages of a stigmatic m o u n t i n g are
o b t a i n e d ; moreover, the grating can also be used at its full dispersion
later if a larger space becomes available. When space is available
a n d astigmatism is unobjectionable, the Paschen-Runge m o u n t i n g is
so m u c h more flexible t h a n the others t h a t its use is advantageous.
90
DIFFRACTION-GRATING SPECTROGRAPHS
[§4.9
As to the choice of grating spacing and number of lines, much depends on the t y p e of problem to be attacked. I n general it is advantageous to have a large number of lines and a close line spacing,
provided ghost intensity is not thereby increased a n d high orders
need not be reached. N o instrument has ever suffered from too
much resolving power, and dispersion can always be decreased if
necessary by using auxiliary mirrors as in the W a d s w o r t h mounting.
For constant angular aperture and slit width, increased dispersion
decreases the intensity of continuous spectra b u t not of m o n o ,
chromatic line spectra. Increasing the dispersion, of course, decreases the range of spectrum t h a t can be photographed on a plate
of given length.
Modern mirror coatings having high reflecting power in the ultra-'
violet region as well as in the visible make practicable a number of
modifications of the s t a n d a r d mountings. T h u s by use of a single
plane mirror and a slit t h a t can be rotated a b o u t a vertical axis, the
grating can be illuminated from any angle, in a Paschen mounting,
which can then be used also as an Abney mounting. T w o movable
mirrors and a fixed slit can also be used. One of t h e systems described in § 4.9 for eliminating astigmatism with the help of a cylindrical quartz lens can sometimes be introduced to i m p a r t t o any m o u n t ing some of the stigmatic advantages of the W a d s w o r t h mounting.
4.9. Astigmatism of the Concave Grating and Its Reduction. T h e
astigmatism of a concave-grating spectrograph can be measured in
terms of the length of t h e line into which a point on the slit is focused
on the Rowland circle. This length depends on the grating and how
it is illuminated, and is proportional to the length of the rulings.
Calculations of astigmatism are usually given in t e r m s of the quant i t y f, which measures the astigmatism per unit length of ruling: '
• 2 a I ^in^ ^ cos d
f = sm^ 6 H
-.—
cos I
Figure 4.18, from a chart due to B e u t l e r , " shows the astigmatisni t o
be expected when various angles of incidence and diffraction are used.
T h e Eagle mounting has much less astigmatism t h a n the Rowland
a n d is somewhat superior in this respect to t h e P a s c h e n . I n general,
t h e least astigmatism is found a t the normal to the grating and when
t h e angle of incidence of t h e light on the grating is kept small. However, t o get m i n i m u m astigmatism a t a given wavelength, i should be
set equal t o d, aS in t h e Eagle mounting.
§4.9]
. DIFFRACTION-GRATING SPECTROGRAPHS
91
Astigmatism reduces the intensity of a line if light is sent t h r o u g h
only a short portion of the slit, because the lengthening of t h e image
spreads the light over a longer line. T h e intensity of a line produced
in a n astigmatic spectrograph can, therefore, be increased b y illumin a t i n g a longer portion of the slit. If a sufficient length of t h e slit
is illuminated, the intensity in the middle of the line will be as great
as t h a t in a stigraatic image. T h e source used must of course be of
sufficient extent t h a t the cone of light from it will completely fill the
grating.
Astigmatism reduces resolution slightly, since the astigmatic line
images are slightly curved. This effect is very small, however, a n d is
of importance only when extremely high resolution is required.
Astigmatism
In units of tlie lenjth of tite grating grooves for point source on slit
•<
^
<
<
<
<
<
<
<
-Wf -70r -6Cf -50° MO" -30° -20° -10° 0°
10° 20° Stf 4Cf 5Cf 6(J" 1&
Angle of diffraction
80° 90°
Fig. 4.18. Astigmatism in imits of the length of grating grooves for a point
source at the slit for various angles of incidence and diffraction. (From H. G.
Beutler'', by permission Jour. Opt. Soc. Am.)
T h e principal drawback of astigmatism in a concave-grating spectrograph is t h a t it prevents use of several of the more effective m e t h ods of photographic p h o t o m e t r y t h a t are valuable in q u a n t i t a t i v e
spectrographic analysis. Astigmatism has been discussed in some
detail by Dieke^^ and by B e u t l e r . " Oldenberg^' has critically disi« G. H. Dieke, Jour. Opt. Soc. Am., 23, 274 (1933).
" H. G. Beutler, Joiir. Opt. Soc. Am., 35, 324 (194.5).
18 O. Oldenberg, Jour. Opt. Soc. Am., 22, 441 (1932); also G. H. Dieke, Proc. Sixth
Summer Conf. on Spectrography, p. 71. New York: John Wiley & Sons, Inc., 1939.
92
DIFFRACTION-GRATING SPECTROGRAPHS
[§4.9
cussed several m e t h o d s t h a t have been suggested for eliminating the
effects of astigmatism in t h e concave grating. A m e t h o d due to
Sirks'^ is widely used, a n d is here reproduced.
T h e R o w l a n d circle represents t h e horizontal focus of a grating,
b u t somewhere outside of this circle t h e light from t h e slit passes
through a vertical focus, where each point on t h e slit is imaged as a
horizontal line. A n y point on the Rowland circle m a r k s also the
vertical focus of some point outside of the slit. This positioning is
easier t o visualize if one thinks of a point source of light placed a t t h e
position of a spectrum line, which
will be brought t o a horizontal
focus (vertical line) on the slit a n d
t o a vertical focus beyond this.
T h u s in Fig. 4.19 an object placed
a t the outer focus 0 will be brought
to a vertical focus on the circle
a t the grating normal, provided n o
lenses intervene, a n d a diaphragm,
step weakener, or logarithmic'sector (§ 13.3) can be placed there.
T h e point 0 is located in the general
case by extending t h e straight line
A
0'
Fig. 4.19. Sirks' construction, for which connects t h e position of the
the special case in which a grating plateholder on t h e Rowland circle
is used on the nonnal, to find the with the intersection of the normal
position, O, at which horizontal
stops may be placed so as to be to the grating and t h e circle, until
imaged as points along the spectrum it intersects t h e line connecting slit
lines. R, Rowland circle; G, grat- and grating. I t s distance d from' the
ing; S, slit; AO, tangent to Rowland circle at intersection of grating slit is given for t h e Eagle mounting
normal, with the circle; 0, position b y the formula
at which horizontal stops should be
placed.
d = Rsmi
t a n 2i
where R is t h e radius of curvature of the grating. Obviously, only
a relatively short region of t h e spectrum can be covered in this way
a t one setting of 0 .
Runge a n d M a n n k o p f ,^'' have devised a means of producing a
stigmatic image of t h e source on the plate by using a combination of
a spherical lens a n d a cylindrical lens, or a concave mirror, t o form
" J. L. Sirks, Astron. and Astrophys., 13, 763 (1894).
'» C. R. Runge and E. Mannkopf, Z. f. Phys., 45, 13 (1927).
§4.10]
DIFFRACTION-GRATING SPECTROGRAPHS
93
an astigmatic bundle of rays that has its horizontal focus at the slit.
Using uncorrected quartz lenses, they were able to produce stigmatic
images of the source over as much as 1000 A with a single setting of
the lenses, the natural dispersion of the quartz serving to extend the
normal range.
We may obtain a stigmatic spectrum without sacrificing dispersion
by placing a small cylindrical quartz lens with its axis horizontal
between slit and grating.^' In this position the lens throws a virtual
vertical image of the slit back to the outer focus of the grating; and
stigmatic, though enlarged, images of the slit are produced on some
part of the Rowland circle, the exact position of the lens being chosen
to fit the particular region of the spectrum being studied. I t is
usually possible to bring two regions of the spectrum into stigmatic
focus simultaneously, and the natural dispersion of the quartz may
again be used to extend the range. A 5- X 5-cm plano-convex
cylindrical lens of 150 cm focal length in the vertical plane will be
found suita,ble for a 21-ft or 10-meter grating.
In all the above arrangements, the gain in light intensity is only
that incidental to the production of a stigmatic image, and in some
cases there may be loss of light if care is not taken to see that the full
length of the slit and the full aperture of the grating are filled with
light. We may, however, obtain a very great increase in line brightness, as shown by Humphreys and by Gehrcke,"^ by placing a shortfocus cylindrical lens in front of the plate. All of the light that
formerly covered a line say 5 cm long is now focused down into a
length of perhaps 5 mm, with a resulting tenfold increase in intensity.
For general work an accurate 5- X 5-cm cylindrical quartz lens with
a focal length of about 20 cm is recommended. This should not be
placed parallel to the plate but normal to the beam incident on the
lens; to avoid a decrease in resolution, great care must be taken to see
that its axis is truly horizontal. Only a short portion of the spectrum
at a time can be covered with a small lens, of course, and the arrangement finds its greatest use in studying hyperfine structure patterns
or very faint band heads.
4.10. The Testing of Diffraction Gratings. When a spectrograph
is purchased from a manufacturer, he is of course responsible for the
satisfactory performance of its optical parts. Probably as many
imperfect prisms as imperfect gratings have been produced in the past,
21 See O. Oldenberg, Jour. Opt. Soc. Am., 22, 447 (1932).
22 W. J. Humphreys, Astrophys. Jour., 18, 324 (1903); E. Gehrcke, Z.f. Inslr. Kde.,
31, 87, 217 (1911).
94
DIFFRACTION-GRATING SPECTROGRAPHS
[§4.11
b u t the former have been weeded out by the manufacturers so t h a t
the user seldom sees one. Gratings, on the other hand, are in m a n y
cases purchased directly from those who rule them, a n d since a perfect
diffraction grating has never been ruled, it is desirable t h a t the user
should be familiar with tests t h a t indicate the merit of any grating
t h a t m a y come into his hands. These tests should cover spectrum
intensity, L y m a n ghost intensity, Rowland ghost intensity, line shape,
target pattern, resolving power, scattered light intensity, satellite
intensity, and the variation of these with angle of illumination and
with wavelength. T h e details of such tests are discussed in §§ 3 . 3 5.5.
4.11. Grating Monochromators. In the past, most commercial
monochromators (§§ 2.16* 3.12, 17.2) have been constructed with
prisms as dispersing agents, b u t the introduction of the aluminumcoated grating, with its increase in intensity, decrease in scattered
light, a n d simple optical system, has m a d e possible design, of excellent
grating monochromators. Harrison-' has described a simple monochromator using a 3-meter concave grating t h a t makes beams of high
radiant flux and great purity available. As discussed in Chapter 6,
the larger a given t y p e of monochromator, the greater t h e r a d i a n t flux
of a given degree of purity t h a t can be isolated by it. For certain
applications great gains result, when numerical aperture can no longer
be increased, by merely increasing the actual sizes of t h e source, slit,
monochromator, a n d receiver. Other things being equal, a hundred
times as much monochroniatic flux can be obtained with an instrument built around a 10-ft concave grating as with one of ordinary t y p e
having a 12-in. collimator. T h e transmission of a good grating instrument is fully as great as t h a t of an equivalent prism t y p e .
Plane gratings lend themselves t o u s e in monochromators, with an
off-axis parabolic mirror for both collimating and focusing, as described in § 17.2 for infrared prism instruments. T h e improved
reflecting powers in the ultraviolet now available for mirrors make
possible the design of new types of monochromators t h a t cover t h e
entire range 2000 t o 10,000 A a t fairly uniform dispersion. Since' no
refocusing of chromatically uncorrected lenses is needed, a sirnple
wavelength control with a direct-reading dial can be provided.
Speculum-metal gratings scatter more light t h a n is desirable for
use in monochromators, even when coated with aluminum. A grating
>
^' G. R. Harrison, Proc. Sixth Conf. on Spec, p. 91, New York: John AVilej; & Sons,
Inc., 1934; also G. R. Harrison and E. P. Bentley, Jovr. Opt. Soc. Am., 30, 290 (1940).
§4.11]
DIFFRACTION-GRATING SPECTROGRAPHS
95
ruled directly on an evaporated metal surface is likely to give such
freedom from scattered light that a monochromator constructed with
it will be intermediate in spectral purity between an ordinary singleprism and a double-prism monochromator. A very simple predispersion instrument then suffices for use with it to make scattering
effects negligible.
CHAPTER 5
The Testing, Adjustment, and Care of Spectroscopic Equipment
THE ADJUSTMENT OF GENERAL TYPES OF SPECTROSCOPIC EQUIPMENT
IS discussed in this chapter. Details of adjustment of apparatus for
special applications, such as infrared spectroscopy, absorption spectrophotometry, Raman spectroscopy, and spectroscopy of the
vacuum ultraviolet, will be found in the chapters dealing specifically
with those techniques.
THE TESTING OF SPECTROSCOPIC COMPONENTS
5.1. The Testing of Slits. The jaws of an adjustable slit should
be polished to almost mirror smoothness, and should be closely
parallel to each other and in the same plane. To determine whether
these requirements are fulfilled satisfactorily, a demounted slit having
adjustable jaws may be tested by laying it on an illuminated opal or
ground glass, or supporting it so that a sheet of white paper may be
viewed through it. The slit opening may be examined with a lowpower magnifier, while the slit is closed slowly. At the instant of
complete closure, if the jaws are straight and parallel to each other,
the light will be extinguished simultaneously at all points along.the
slit; and if the jaws are in the same plane, this simultaneity will be
observed no niatter at what angle the slit is viewed. If the width
of the slit appears to vary with the aspect from which it is viewed, the
two jaws are not accurately in the same plane. Adjustments are
usually provided whereby the jaws may be made parallel if they are
found to be improperly aligned. Jaws of a slit in which the beveled
edges do not lie in the same plane should be reground for thickness.
Another method of testing is to photograph an image of the slit
under actual working conditions, with the slit mounted as it is to be
used. If the slit belongs to a visual spectroscope with a viewing
96
§5.2]
SPECTROSCOPIC EQUIPMENT
97
eyepiece or to a monochrom<ator with an exit aperture, the eyepiece
or exit aperture shouKl be removed so t h a t the spectral image can be
photographed in the position at which it is most sharply in focus. A
source comparatively rich in fine lines, such as an iron arc or a copper
or tungsten spark, m a y be used. W i t h a slit opening of 0.01 to
0.02 m m , the spectrum should be focused carefully until sharp images
of the slit are obtained in the portion of t h e spectrum to be used for
testing. T h e slit should then be opened to a width of about 1 m m
and its edges cleaned carefully by stroking t h e m with a sharpened
stick of clean wood in one direction. A series of test spectra should
then be photographed with slit openings varying from a b o u t 0.04 m m
to full closure. D u r i n g these exposures the full length of the slit
should be illuminated, any diaphragms or aoertures t h a t might shield
portions of it from illumination being removed. If the p h o t o g r a p h e d
spectrum lines appear wedge-shaped, lack of parallelism of the jaws
is indicated. If the spectrum lines appear ragged or uneven (Fig. 5.1)
the cause is probably dirt on the jaws, imperfections in their edges,
or bluntness of the jaws.
T h e second method of testing slits is applicable only to
spectroscopic systems in which
stigmatic images of the slit
are formed in the focal plane.
I n astigmatic spectroscopes,
the slit can usually be dem o u n t e d and tested by the
first method described above.
Figure 5.1 shows spectra taken
with dirty and defective slits.
Fig. S.l. Out-of-focus spectra photographed with a dirty and defective
spectrograph slit.
5.2. T h e Testing of P r i s m s and L e n s e s . T h e spectroscopist has
little occasion to test prisms or lenses except in instances in which he
is concerned with the design and con.struction of special equipment.
Inspection of such components for striae, bubbles, surfaces scratches,
a n d other gross imperfections can be accomplished easily with the aid
of a low-power magnifier. There is usually no need to go beyond
such cursory examination, since components are tested for performance in the optical shops where they are fabricated.
Instructions
regarding the measurement of indices of refraction, dispersion, prism
angles, focal lengths of lenses and mirrors, and the various aberrations
of lenses and mirrors are given in s t a n d a r d treatises on optics. In-
THE TESTING, ADJUSTMENT, AND CARE
[§5.3
formation regarding the testing of mirrors, lenses, and prisms in
combination is given in §§5.4 and 5.5.
5.3. The Testing of Gratings. No two diffraction gratings are
precisely alike. The performance of most gratings can be improved
by masking areas in which the rulings are imperfect, especially at
the edges. Increasing commercial production of grating spectrographs is helping to relieve the individual spectroscopist of the necessity of subjecting new gratings to test, but it is still often important
to be able to determine the performance of a grating in the laboratory.
a. The Foucault Knife-edge Test. This test is useful in locating
imperfect areas in concave gratings (Fig. 5.2). A knife-edge (for
example, a razor blade) is mounted on a carriage in such a manner
that it can be moved along the focal plane, through the image of an
intense, isolated spectrum line such as one emitted by a mercury arc.
Fig. 5.2. Foucault knife-edge test as used with a concave grating, a. Slit;
b, grating; c, knife-edge on carriage movable by a transverse screw; d, position
of observers eye.
When viewed from close behind the image, the grating will appear
to be filled with light, since light from all parts of the grating passes
through the image into the eye pupil. If each part contributes its
proportionate share of light to the image, and if the image is sharp
and free from effects of aberrations, moving the knife-edge into the
image will cause the light to be cut off uniformly and simultaneously
from all parts of the grating. If, however, large bright areas are
observed on the grating when the knife-edge is moved into the image,
these areas are usually due to imperfect rulings, which should be
masked. The effect of masking such areas on the quality of the
spectra produced by the grating should then be tried.
*
h. Target-Pattern Test. Another method of determining defe'ctive
areas of concave gratings is to view target patterns of strong mercury
-^ »'
§5.3]
OF SPECTROSCOPIC EQUIPMENT
99
lines (or other distinct, intense spectrum lines) on screens placed
about one-third the distance from the image plane to the grating,'
with the grating otherwise set up as for normal use. Typical target
patterns are shown in Fig. 5.3. The images so obtained are broad
areas corresponding to bundles of rays converging toward each
spectrum line. If the various portions of the grating all contribute
Figs. 5.3 a, b. Typical target patterns, (a) Pattern of a single wavelength
from a good grating, (b) Patterns of three wavelengths from an inferior grating.
Reproduced from Proceedingn of the Seventh Summer Conference on Spectroscopy
and Its Applications, by permission of the publishers (the Technology Press and
John Wiley & Sons, Inc., New York.)
j
their proportionate share of light to the formation of a sharp image,
the target pattern will have a uniform or a channeled structure.
Otherwise, the pattern will show irregular patches. If the pattern
is irregular, the effects of masking can be tried, the optimum masking
being determined by trial and error.
1 G. R. Harrison, Proceedings of the Seventh Summer Conference on Spectroscopy
and Its Applications. New York: John Wiley & Sons, Inc., 1940.
i^
THE TESTING, ADJUSTMENT, AND CARE
[§5.3
c. Determination of Defects That Give Rise to Satellites. A third
method, originated by R. W. Wood,^ is apphcable to determining
defective areas in plane or concave gratings which give rise to prominent satelhtes. A slit is placed in the focal plane in such a position
as to pass the light from a satellite but to obscure the principal line
and other lines. The light that passes through the slit is sent through
a lens onto a photographic plate in such a manner as to form an
image of the grating on the plate. If only certain portions of the
grating contribute to formation of the satellite, these portions will
appear most prominent in the photograph of the grating. Once
their position has been determined by this means, they may be
masked. It is sometimes found that the position of a satellite varies
with reference to the principal line as the spectrum is traversed, the
satellite coinciding with the line at the grating normal.
d. Observation of Ghosts. Although Rowland and Lyman ghosts
cannot be eliminated by masking, since all portions of the ruled surfaces contribute to them, it is of importance to determine their
prominence. To observe Lyman ghosts, the grating should be set up
as for normal use. If the slit is illuminated with a source having only
a few strong lines in the visible, such as a mercury arc, the Lyman
ghosts will be observed most easily as visible lines in the region between the violet end of the first-order visible spectrum and the central
image. Visible Lyman ghosts, have the same colors and appearance
as their parent lines. Hence if one employs a source having doublets
or triplets in the visible spectrum, the corresponding Lyman ghosts
will be doublets or triplets of the same color. Almost any grating
will show Lyman ghosts if a wide slit and intense source are used in
the search for them. A grating having Lyman ghosts more intense
than 1/10,000 of their parent lines should be viewed with suspicion
for all but a few spectroscopic purposes. In case of doubt, the
relative intensities of ghosts and lines should be determined photographically.
Rowland ghosts may be observed by using a narrow entrance slit
and viewing the first- or second-order visible spectrum lines with an
eyepiece. The ghosts appear as additional weak lines regularly
spaced on either side of all strong lines, as shown in Fig. 2.12. By
taking a series of photographs of the spectrum with different times of
exposure and determining the time ratios for which the ghosts produce
the same blackening in one photograph as the parent lines in another,
2R. W. Wood, Physical Optif^s. New York: The Macmjllan Company, 1934.
§5.5]
OF SPECTROSCOPIC EQUIPMENT
' '
101
one may determine the approximate relative intensities of the ghosts.
Alternatively, a rotating logarithmic sector may be used. In the
first-order spectrum of a good grating, the intensities of Rowland
ghosts should be less than I/IOOO of those of their parent lines. The
relative intensity of ghost to line can be expected to increase approximately as the square of the order; ghosts in the second order will be
about four times as strong relative to their parent lines as in the first.
This rule is only approximate, because some orders may show inordinately strong ghosts. In general, Rowland ghosts produce less
confusion in the spectrum than Lyman ghosts.
5.4. The Testing of Prism Spectrographs. An easy qualitative
test is to photograph the spectrum of a source rich in sharp lines,
such as an iron arc, and then to compare
G
this spectrogram with one made on an
•^
^r~~r?
X
instrument known to perform satisfac±0
\
torily. Prints of iron-arc spectra and other
pQ T
/
spectrograms made with good instruments
E]
^
of high dispersion may be obtained from
spectrograph manufacturers. A narrow
Fig. 5.4. The Hartmann ^u,. ^^^^^^^ ^^ ^^^^ ^^^ ^^le focus of the
Diaphragm,
t h e diaphragm
D contains several small
openings, any one of which
can be brought into juxtaposition with the slit » by
sliding the diaphragm sideways in the grooves 6.
j
-
^
. n ,,
Spectrograph should be adjusted carefully
before the test is made (see §5.9). The
^^^^ spectrograms should be taken on
°
plates oi moderate contrast, developed to
give average tonal gradations. This procedure is essential in order that relatively
weak imperfections will not be obliterated when the exposure and
development are such as to result in appropriate blackening of
the stronger lines.
Spectrograms taken with the full slit illuminated should be compared with those taken with various portions of the slit covered by
means of a Hartmann diaphragm (Fig. 5.4). The line sharpness
should be essentially the same whether the total slit length or only
portions of it are used. If this is not the case, either the focus is not
properly adjusted or defects in the components are indicated. If,
with the spectrograph in proper adjustment, the lines are abnormally
fuzzy or irregular or are found to be accompanied by satellites when
compared with standard spectrograms, defects in the components
may be suspected.
5.5. The Hartmann Test. If component defects are indicated by
102
THE TESTING, ADJUSTMENT, AND CARE
[§5.5
the above procedure, a simple test method devised by Hartmann' may
be used to locate the defects. In this test a small light source,
such as a condensed spark between tungsten-steel electrodes or a
Western Union concentrated arc lamp, is used. The source is
mounted at a distance of about 25 cm from the slit, on a carriage that
will permit it to be moved laterally in a plane parallel to the plane
of the slit and along a line perpendicular to the slit (Fig. 5.5). With
the source placed approximately on the optic axis of the collimator
lens, the slit is narrowed until its central diffraction maximum covers
only a small portion of the collimator lens. By moving the source
laterally, it is then possible to direct light in the horizontal plane
through different portions of the spectrograph optics. By ,using
different openings in a Hartmann diaphragm over the slit, it is further
possible to send light through different portions of the optics in the
vertical plane.
For each setting of the Hartmann diaphragm, a series of five or
Fig. 5.5.
The Hartmann Test, a. Point source mounted on carriage with transverse screw motion; fc, slit of spectrograph; c, collimator lens.
more spectrograms is taken with the source positioned so as to
illuminate different portions of the optics. If there are no appreciable
aberrations or other defects in the system and if the spectrograph isi
in proper adjustment, the spectrum lines in all these test spectrai
should be in good alignment with each other. If the plate is in front
of or behind the true focal plane, there will be'a progressive shifting
of the lines with change in position of the source or of the diaphragm;
openings. The focus should then be readjusted until the shifting of
the .lines is reduced to a minimum. If there are serious lens aberrations or inhomogeneities in the prism or lens materials, irregular
shifting of the lines will be observed even after the best focus has been
attained. When such defects occur, their effect may often be reduced
materially, or eliminated, by masking portions of the prism or lenses.
This masking may be accomplished by trial and error, those portions
^ J. Hartmann, Zeits.f. Instrumentenkunde. 20, 47 (1900). See also U. S. Burean of
Standards Scientific Papers, numbers 311 and 494.
,
§5.6]
OF SPECTROSCOPIC EQUIPMENT
103
of the optics which the test spectrograms show to be the worst
offenders being masked first.
The above test shows only whether light beams suffer irregular
deviations in traveling through particular portions of the lens and
prism optics. For any particular beam, the test does not indicate
whether the collimator lens, the prism, or the telescope lens is the
offending component. For Littrow-type spectrographs, the lens may
be tested separately by the Hartmann method if the prism is removed
and a mirror substituted in such a manner as to reflect the rays back
through the lens along their ordinary path. In this test, the radiation
is not spread into a spectrum; hence chromatic aberration, ordinarily
compensated by tilting of the spectrum plate, may cause appreciable
broadening of the slit image. This difficulty may be overcome by
using a source with widely spaced lines, such as a mercury arc, and
isolating one line by a suitable filter so that the slit is illuminated
with approximately monochromatic radiation. In general, the best
focus under such conditions will not correspond to the position of
sharp focus with the prism in place, since a different light path is
followed by the rays in traveling from the mirror to the plate when
no prism is interposed. Accordingly, the focus must be adjusted
before a final test plate is made.
If defects are found by the Hartmann test, it may be advisable to
remove the lenses and to test them individually, either by the knifeedge test,* as described for gratings in § 5.3, or by other standard
methods (see General References 5.1-5.3). It is well to look for
defects in the prism first. This search is made most simply by substituting another prism and repeating the Hartmann test for the
complete system.
ADJUSTMENT OF PRISM AND
GRATING SPECTROGRAPHS
5.6. Adjustments Required for Various Types of Instruments.
Spectroscopic equipment is of many different types, each of which
involves particular adjustments for best performance. In general,
such adjustments involve (a) setting the slit to an appropriate opening, and to parallelism with respect to a prism apex or to the ruled
lines of a grating, (b) optically aligning the component parts of the
* A pinhole source is used, mounted on the optic axis in the case of a lens and as
near the optic axis as feasible in the case of a mirror.
104
THE TESTING, ADJUSTMENT, AND CARE
[§5.7
system, and (c) focusing the components. These operations should
be carried out in the order given to achieve approximate adjustment,
after which it may be necessary to repeat the various procedures
until the best adjustment has been reached as a result of successive
approximations.
Commercial spectrographs, spectrophotometers, and monocKromators are usually designed so that the necessary adjustments are few
in number. The optical components are sometimes permanently
aligned and fixed in position by the manufacturer, with no provision
for further adjustment. In almost all instruments, however, adjustment of slits and focusing are required.
5.7. Adjustment of Slit Width and Length. Methods of testing
and adjusting slits for parallelism of the jaws were described in § 5.1.
. Sonie„slits are made without provision for adjusting the jaws for
parallelism. If they are found to be seriously nonparallel, a more
perfect slit must be substituted.
Slits should be kept free from dust particles or dirt, which may
interfere with the uniformity of the slit aperture. A method of
cleaning the edges of open slits was described in § 5.1. Some adjustable slits and most fixed slits are covered by plates of quartz or
glass on the side toward the light source. Cover plates should be
cleaned by wiping them with lint-free cloth or lens tissues, moistened
with ether or alcohol. Special care should be taken to avoid finger
marks on quartz cover plates. Any fingerprints present should be
removed carefully with a grease solvent before exposure to ultraviolet
radiation, because if such radiation falls on fats or other organic
substances, it inay cause permanent etching of quartz.
When open arcs or sparks are used, danger of pitting or fouling
of the slit edges by material from the source may be reduced by placing
a transparent protecting plate between the source and the slit, or
by using a condensing lens so that the source can be moved farther
away.
Protecting plates, or plates used for the mounting of fixed slit.s,
may, as the result of interference arising from multiple reflections, give
rise to variations in the intensity of the radiation transmitted through
different portions of the slit. The errors that such effects may introduce in quantitative emission analysis, and their elimination, have
been discussed by Stern.^
«
The width of an adjustable spectroscope slit is usually set at be< Joshua Stern, / . Opt. Soc. Am., 36, 634 (1946).
§5.8]
OF SPECTROSCOPIC EQUIPMENT
105
tween 0.01 and 0.06 mm. The optimum opening depends on the
instrument and its appHcation. Details of the calculation of optimum widths are given in Chapter 6.
A diaphragm with V-shaped opening is usually provided for adjusting the length of slit to be illuminated. Ordinarily only a portion of
the total length is used. In stigmatic instruments, increasing the
slit length usually merely increases the length of the spectrum lines
produced; in astigmatic instruments it may greatly increase intensity.
Lenses of short focal length may be used in lieu of slits. A pseudo
slit of this type has been described by King.^
5.8. Adjustment of Slit Perpendicular to the Direction of Dispersion. The slit should be parallel to the apex of the prism or to the
rulings of the grating used. This adjustment is accomplished by
rotating the slit in its own plane, after the prism or the grating
and the other optical components of the system are in proper alignment.
Often it is sufficient to set the slit so that the spectrum lines from
a source such as a mercury arc or neon discharge tube appear perpendicular to the line along which the various wavelengths are dispersed, as viewed on a ground-glass screen. Reference lines marked
on the screen, or a piece of translucent coordinate paper, may be used
for checking perpendicularity. In stigmatic instruments the slit
length may be closed down almost to a point, so that the resulting
spectrum, if continuous, appears as a line; if discontinuous, as a
series of points along a line. One of the coordinates of the screen or
graph paper may then be set parallel to this line, after which the
entire slit length may be exposed to light from a line source, the slit
being rotated until the spectrum lines are properly aligned with respect to the perpendicular coordinates.
For the most precise adjustment, the foregoing method may be
used merely for preliminary alignment. Then a series of trial spectrograms, taken with the instrument properly focused and with slight
changes in the rotation of the slit for each new exposure, may be used
to determine the slit angle that gives the sharpest spectrum lines
perpendicular to the line of dispersion.
In the case of prism instruments, the spectrum lines are appreciably curved. Therefore only a short middle portion of each line
will appear strictly perpendicular to the spectrum, but it is usually
possible to judge fairly well when the curved lines are symmetrically
' C. M. King, J. Opt. Soc. Am., 36, 164 (1946).
106
THE TESTING, ADJUSTMENT, AND CARE
[§5.9
disposed with reference to the perpendicular coordinates. Lines from
gratings are only slightly curved. This curvature, and any inclination of the slit, decreases resolution in astigmatic instruments.
5.9. Focusing the Spectrum: Commercial Prism Spectrographs.
It will be assumed in this section that the slit has been adjusted
properly and that the optical components are suitably aligned.
Procedures for aligning various components are given in subsequent
sections.
Procedures for focusing vary considerably for different types of
instruments. In the simplest instruments, such as the usual hand
spectroscopes or spectrographs, only one focusing adjustment is pro-
t^ l ^ ^ t l l f H I l
1 BHB
.'-A-i^^^^k-
^^^^^^^^^^^^^^^^^^mTt:
'
It
1*
I
^•i
. : . • ! • • ••
1 II I I I 1 1
1 iiill 1 1
1
Mil 1
••
mi
1 ii 1
I t ' 1 I M II 1 1
I I
Fig. 5.6.
Mil'
I
• - 1 1
1
1M
I I
11
1 >« 1
1 1 1
1 1
(1
Trial spectra photographed with different focus
settings of the spectrograph collimator.
vided, and there is no provision for change of plate tilt. The focusing
adjustment may move either the slit or the collimator lens, so as to
change the distance between these components. Alternatively, the
adjustment may move the telescope lens toward or away from the
eyepiece or plateholder. In visual instruments equipped with fixed
cross hairs or a fixed aperture viewed by an eyepiece, the focus should
be adjusted so that the spectrum is sharply imaged in the plane of the
cross hairs or aperture. When the adjustment has been made properly, no parallax should be observed as the eye is moved from side to
§5.10]
OF SPECTROSCOPIC EQUIPMENT
107
side. If movable, the eyepiece lens should first be focused so that
the cross hairs or aperture appear sharp and clear.
In photographic instruments, preliminary focusing may be accomplished by observing on a ground-glass screen the images of widely
spaced spectrum lines, such as those from a neon tube or mercury arc.
Several spectrograms should then be taken, corresponding to slight
differences in focus (Fig. 3.6). The final setting to be chosen is the
one that yields the sharpest lines in the portion of the spectrum to be
observed. In small or medium quartz spectrographs, for example,
the best setting for lines in the visible and near ultraviolet may be
somewhat different from that for lines in the far ultraviolet.
A second case is that of spectrographs on which there is only one
focusing adjustment but on which the tilt of the plate may also be
changed. Preliminary focusing and plate-tilt adjustments may be
accomplished by observation of visible lines on a ground-glass screen.
A test plate is then taken in the manner described above, and that
position of focus is selected which gives the sharpest line images near
the center of rotation of the plate. In large spectrographs other
than those of the Littrow type (discussed below) it may be necessary
as a preliminary measure to shift the plate holder along the spectrum
to a position corresponding to the spectral range to be photographed.
After the spectrum has been focused sharply for the center of rotation
of the plate, a second test plate is taken on which is photographed a
series of spectrograms corresponding to different plate tilts. The tilt
finally chosen should be that which gives the sharpest line images
throughout the range of the spectrum to be observed. If a line drawn
on the spectrogram connecting the positions of best focus in successive,
uniformly spaced exposures is straight but inclined, only a change
in angle is required. If it is curved, the plate must be bent to achieve
the best focus.
5.10. Focusing the Spectrum: Commercial Littrow and EagleMounting Spectrographs. Spectrographs of the Littrow type require
successive adjustment of the angle of the prism or grating, the focus
of the lens, and the angle of the plateholder. Commercial instruments of this type are usually supplied with calibration tables showing
the proper settings of these three adjustments for various wavelength
regions. Often two or more of the adjustments are geared together
so as to be accomplished automatically when the setting is shifted
from one spectral region to another.
If automatic coupled adjustments are not used, it is first necessary
to set the dispersing element to cause the desired spectral region to
108
THE TESTING, ADJUSTMENT, AND CARE
[§5.11
fall on the plate. If calibration data are lacking, this is accomplished
easily for the visible region by visual inspection, using a ground-glass
screen and a source providing widely spaced, easily identified spectrum lines. In the ultraviolet region a fluorescent screen may be
substituted for the ground glass, or a series of test spectrograms may
be taken with approximately correct plate angle-and focusing adjustments, and with different prism inclinations. After the angle that
gives the proper spectral range has been chosen, precise focusing and
plate-angle adjustments are accomplished as outlined in § 5.9. When
all the procedures have been carried out and the proper settings determined, these should be recorded to facilitate the making of future
adjustments.
In concave-grating instruments using the Eagle mounting, the
adjustments are analogous to those described and are carried out in
a similar manner. The concave grating corresponds to the Littrow
prism and lens system combined.
Precise focusing in all the above cases may be facilitated by use of
the Hartmann test method (§ 5.5) or by a simple modification thereof
in which aperture stops are placed on the collimator lens.
5.11. Adjustment of Spectrometers. Collimators and telescopes
with multiple movements and adjustments are provided on ^ many
spectrometers. The following adjustments must, in general, be made:
(a) The eyepiece of the telescope must be set so that the cross hairs
appear sharp to the observer; (b) the telescope must be focused so
that parallel light is brought to a sharp image in the plane of the
cross hairs; (c) the collimator must be focused so that the light from
each portion of the slit is sent into a parallel beam; (d) both the
telescope and collimator must be adjusted so that their optic axes are
perpendicular to the'axis about which the prism or grating rotates;
(e) the prism or grating must be leveled so that the prism apex.or
grating rulings lie parallel to the axis of rotation of the prism table;
(f) the collimator and telescope must be set for an angle of deviation
approximately correct for the wavelength region to be observed;
(g) the slit must be adjusted for proper width and for parallelism, to
the prism apex or grating rulings; and (h) the prism must be adjusted
for approximately minimum deviation at the wavelength to be
observed. Thus the use of a spectrometer often involves a complicated series of adjustments. Nevertheless, these adjustments are not
difiicult to make if undertaken systematically in the sequence
indicated.
§5.12]
OF SPECTROSCOPIC EQUIPMENT
109
5.12. Adjustment of Concave Gratings. The testing of gratings is
described in § 5.3. Methods of adjusting commercial grating instruments are discussed in §5.10 for Littrow and Eagle spectrographs,
and in § 5.11 for spectrometers. Here we will consider certain more
general problems that arise when the spectroscopist is not using a
commercial spectrograph previously assembled by a manufacturer,
but is confronted with the necessity of installing a grating and of
undertaking all the adjustments himself. Such situations arise particularly in the use of concave gratings of long focal length. One or
more grating holders may be set up in a room that can be darkened,
in which there is a permanently installed Paschen-Runge or Wadsworth mounting, or both. The slits to be used are usually mounted
in openings in a wall of the room, so that the light sources to be studied
may be placed outside the room. When the original gratings are set
up or when different ones are substituted or added, a series of careful
adjustments must be made if full advantage is to be taken of available
resolving power.
The adjustments will be described here only qualitatively. Discussions of the quantitative effects of each of the several adjustments
are given in various publications,^ including General Reference 5.3.
Five principal types of mountings, described in Chapter 4, are
used with concave gratings: Rowland, Paschen, Abney, Eagle, ,and
Wadsworth. In all of these except the Wadsworth, the grating and
plateholder are mounted on the Rowland circle. In the first three
mountings the slit is also on the Rowland circle; this would be true
also of the Eagle mounting were it not for the fact that light is introduced from a slit at the side and reflected toward the grating by a
prism or mirror. Accordingly, grating adjustments will be discussed
first with particular reference to the Rowland circle, and the special
problems that arise with respect to the Wadsworth mounting will be
considered separately.
a. The Grating Holder. I t is essential that the grating holder be
rigidly constructed and mounted. Adjustments should be provided
for rotating by accurately controlled amounts around each of three
mutually perpendicular axes (Fig. 5.7). One of these
lies in
the plane of the Rowland circle and is perpendicular to the grating
at the center of the ruled space. A second, y, also lies in the plane of
the Rowland circle, is perpendicular to x, and passes through x at its
point of intersection with the grating. The third, z, is perpendicular
s H. G. Beutler, J. Opt. Soc. Am., 35, 311 (1945).
110
OENCS PYTHIL'M AND SOME CHTTRIDIACE^:.
noticed t h a t the spore wall i.s no longer in direct contact with the
host, but is separated fron) it by a distinct space across which the
fine joining process extends (plate V I I I , fig. 9). This process
bores into the interor of the filament and becomes the infection
tube. I t is only visible "when the spore is exactl}^ on the optical
mai'gin of a hypha. In many cases the infection does not go any
further, and the pai-asite fails to gain an entrance. This does not
appear to depend entirely on the age of the filament. It is true
i h a t the young hyphae are usually selected by the wandering
zoospores for attack. B u t on these, as on older ones, it is connnon
to find zoospores which have failed to effect an entrance, and on the
other hand infection of the older hyplne has sometimes been
directly observed. Tlie ditference in infective power is probably
inherent in the zoospore in the niajorit}' of cases, for the numbei•of parasites which a hypha may harbour is very variable, ranging
from one to thirtj' or forty in the same culture and on hyplue
of about the same age. I t is indeed evident that the zoospores
s t a r t life with diftei-ent reserves of energy, since their period of
swarming varies considerably ; thus some maj^ have onl}' enough
•energy to enable them to reach the liost but not to enter it.
The contents of the encysted spore are eventually emptied,
in successful cases, through the infection tube into the protoplasm
of the host. The tube of entry is extremely fine and I have
failed to obtain a double contour clearly, even on high magnification. It is, however, eei-tainly a tube, remaining distinctly visible
after the contents of the spore have passed over into the filament.
T h e spore cyst also remains visible for some time, attached to
the tube. The case figured in plate X , fig. 1, of an emptied
Saprolegnia capsule with parasitic spore-capsule attached shows
t h a t the tube penetrates some distance inside the wall before it
opens. The passage of the parasite across from its ej^st into the
filament is slow, occupjnng about seven minutes in .some cases
observed.
Under favourable conditions the contents may be watched ac•cumulating in the hypha as they leave the cyst. The two plasmas
are sometimes of diftierent degrees of refractivity, which enabhs
§5.12]
OF SPECTROSCOPIC EQUIPMENT
111
When not in use, gratings should be protected from dust, fumes,
and other deleterious substances. This objective can be conveniently
accomplished by designing the holder so that closure of a cover shuts
the grating into a tightly enclosed compartment. It should be possible to open and close this cover without disturbing the grating
adjustment, and provision should be made for mounting occulting
masks in front of the grating, to shield portions of the ruled surface
from illumination if necessary.
b. Approximate Adjustments. After the grating has been mounted
in its holder, approximate adjustments should be made. These may
be accomplished by observations, measurements, and tests with light
beams as in the setting up and alignment of any optical system.
Since most gratings throw the light unequally into equivalent orders
of the two sides of the central image, the more desirable side should
first be found. Ordinarily one chooses the first or second order that
is most intense in the visible region. This choice can often be made
by holding the grating in the hand in a darkened room, illuminating it
with the full light of a mercury lamp, and observing the spectra
thrown on walls or ceiling. The eye is not a particularly good judge
of relative intensities, however, so if this test does not show one order
to be much more intense than its counterpart, photographic tests
should be made.
The combination of direct-intensity tests and target-pattern tests
will ordinarily settle the question of which orders are most useful in
a given grating for a given spectral region. Gratings tend to throw
their light in a given direction rather than into a given order, so if
the first order on one side is particularly bright at say 5500 A, the
second order at 2750 A can be expected to be bright on that side.
Target-pattern changes or decreased reflecting power of the grating
in the ultraviolet may intervene to alter this tendency.
Once the side of the grating to be used has been chosen, the face of
the grating should be set so that it lies approximately in a vertical
plane tangent to the Rowland circle and at such a height that the
xy plane is coincident with that of the Rowland circle. The grating
should be rotated so that its rulings are approximately vertical. The
slit should be set on the Rowland circle, with its aperture approximately vertical, and at a height such that the xy plane normal to the
grating (that is, the plane of the Rowland circle) bisects the slit
length.*
* An exception occurs in mountings of the Eagle type, in which the slit is sometimes
placed slightly above or below the plane of the Rowland circle.
112
THE TESTING, ADJUSTMENT, AND CARE
[§5.12
I t is assumed that the disposition of the grating with respect to
the slit along the circumference of the Rowland circle has been determined by the type of mounting to be used. The approximate position
in which spectra will be formed should be known from this choice
and from the constants of the grating (see Chapter 4).
c. Rotation About the y Axis. The central portion of the widened
slit should be illuminated with light from a mercury arc, and'the
visible spectra should be observed on white screens or the walls of
the darkened room. If the spectra are thrown above or below the
Rowland circle, the grating should be rotated about its y axis until
the spectra are coincident with the Rowland circle. After this
adjustment has been made, rotation of the grating about the 2 axis
should not raise or lower the spectra with respect to the Rowland
circle if the center of the slit is accurately in the plane of the Rowland
circle. If raising or lowering occurs, the slit height and rotation of
the grating about the y axis should be adjusted until rotation of the
grating about the z axis no longer causes such motion.
While this adjustment is being undertaken, it may be necessary to
rotate the grating slightly in its own plane (about the x axis) in order
to bring the line of dispersion of the spectrum into parallelism with
the plane of the Rowland circle, since the grating lines must be
perpendicular to this plane if the spectra are to be parallel to. it.
d. Focusing the Grating. The grating is next focused approximately
for the portion of the spectrum to be photographed. This focusing
is accomplished by motion of the holder along a line normal to the
grating. For the visible region, preliminary adjustment' may be
made by visual observation of the spectrum lines on a viewing screen
in the plateholder or with an eyepiece held in the hand, but a series
of test spectra should always be photographed to determine the best'
setting.
Often it will be found that it is not possible to focus sharply all
lines within the desired range along the curve of the plate (Rowland
circle curve), because errors of run usually cause gratings to focus
their spectrum lines along a curve slightly different from that of the
true Rowland circle. This effect jcan:be overcome to a considerable
extent by rotating the grating slightly about its z axis, so that the
plane of the grating is no longer strictly tangent to the Rowland circle.
This adjustment is known as rolling the grating, or as setting the
grating in aberration, since a perfect grating so rotated away from
the true tangential position would exhibit considerable aberration.
§5.12]
OF SPECTROSCOPIC EQUIPMENT
113
. Many gratings have a false focus that is easy to confuse with the
real focus. As the target pattern narrows down to form the spectrum
line, it splits into a constantly decreasing number of fairly sharp
fringes of varying intensity, and in some gratings a strong fringe and
a weak fringe pass through each other to form the spectrum line. If
too faint an exposure is made, the strong fringe may be mistaken for
the spectrum line itself, and for this reason typical lines should always
be exposed up to high densities. Such exposure is also necessary to
show the presence of satellites.
e. Setting Jke Grating in Aberration. Before one starts the rolling
adjustment (rotation about the z axis), an approximate determination
should be made of the effect of rotating the grating through 180 deg
in its own plane (about the x axis) on the position of the focal plane
of the spectrum. The grating holder should be moved along the
normal of the grating until such 180-deg x-axis rotation throws the
spectrum equal distances behind and in front of the Rowland circle.
The grating' may then be rolled on its z axis until trial indicates that
rotation through ISO deg on the x axis causes no appreciable motion
of the spectrum and until the spectrum is sharp along the entire
length of the Rowland circle encompassed by the plateholder.
f. Setting the Slit Aperture Parallel to the Grating Rulings. Finally,
to obtain lines of the greatest sharpness, it is necessary to adjust the
slit so that it is strictly parallel to the grating rulings. For this purpose it is desirable that the slit be provided with a tangential adjusting
screw whereby small angles of rotation may be accomplished accurately and reproducibly.
The effect of lack of parallelism of the slit and the rulings is to
broaden out the spectrum lines. By using a source that yields many
fine spectrum lines, and a narrow slit, and then photographing a series
of test spectra with the slit set at slightly different angles, one may
readily determine the optimum adjustment.
g. Final Tests. The resolving power \/d\ of a large grating can
readily be estimated by observing the hyperfine structure of the green
line at 5461 A as emitted by a cool mercury arc (see § 20.1), or by
photographing close and fine lines from an arc containing rare-earth
salts, a source giving nitrogen bands in emission, or an iodine absorption tube. Great care should be taken to see that the slit is accurately
parallel to the rulings on the grating when resolving power is being
tested.
A point that has been insufficiently emphasized in the past is that
114
THE TESTING, ADJUSTMENT, AND CARE
[§5.13
the effective resolving power obtainable in practice is definitely a*
function of the density of the lines produced, and that line shape may
change markedly with density. Almost all gratings are tested by the
makers for resolution. A common procedure is to photograph-some
line that shows hyperfine structure, such as mercury 5461 A, with
enough exposure so that this line is brought to low or medium density.
If close-lying components of the line are resolved, the grating is said
to be good. However, almost all gratings produce lines not of simple
shapes, but with many satellites grouped fairly close together. Owing
to the nonlinearity of the curve that connects light intensity with
plate density, the apparent shape of the line is altered as the density
increases, and its center of gravity may be shifted. For this reason
a grating should be tested for resolving power at all useful densities,
some lines being overexposed. At extreme overexposures, of course,
any grating will show false lines.
5.13. Adjustments of Wadsworth Grating Mountings. In the
case of the Wadsworth mounting, it is necessary first to adjust the
distance from the mirror to the slit, so that light is rendered parallel
by the mirror. This adjustment may be accomplished approximately
by inspection of the reflected light beam and somewhat more accurately by an autocollimating method. In the latter method, a plane
mirror is set up so as to return the reflected light to the collimating
mirror. The latter is turned on a vertical'axis until it forms an image
of the slit directly to one side of the slit. The collimating mirror is
then focused until the slit image is sharp.
After having been focused, the collimating mirror should be rotated
about its X, y, and z axes until the parallel beam of light from it is
centered on the aperture of the grating.
The grating adjustments are accomplished as described in the preceding sections. Strict coincidence of the plate with the focal plane
of the spectrum is obtained by the use of specially curved plateholders.
The rotation of the grating with respect to its 2 axis is usually governed
by the fact that it is desirable to photograph the spectrum as near to
the normal of the grating as is feasible. Since the focal curve of the
Wadsworth mounting is not a circle, it is necessary to change adjustments of plate curvature and plate angle when moving from one
region of the spectrum to another.
5.14. Adjustment of Plane Gratings. The foregoing discussion
applies to concave gratings. In the case of a plane gratings an
auxiliary lens system is used, which has a single combined collimator
§5.16]
OF SPECTROSCOPIC EQUIPMENT
115
and telescope lens in Littrow mountings, and separate collimator and.
telescope lenses in other mountings. The characteristics of the lenses
(chromatic aberration, flatness of field, and so on) and their positions
with respect to the grating, as well as the constants of the grating,
determine the position and shape of the focal plane of the spectrum.
The choice of grating and lenses, and of their mechanical disposition
with respect to one another, depends on the intended application
of the system, as discussed in Chapter 4. For any given system,
all adjustments of the grating except with respect to rotation about
the y axis are analogous to those described for concave gratings. The
2/-axis setting must be such as to obtain the desired angle of incident
illumination on the grating, and to send the appropriate spectral
region through the telescope lens.
THE CARE OF SPECTROSCOPIC EQUIPMENT
5.15. General. Instruments should be protected from dirt and
corrosive vapors or fumes. The spectroscopic laboratory should be
clean and dust-free, and preferably air-conditioned. Plateholders and
other parts that are subject to wear and possible damage during use
should be inspected periodically. Oiling or greasing of sliding or
rolling surfaces may occasionally be necessary. When lubricants are
used, the surfaces should be cleaned and a small amount of oil or
vaseline of good quality should be applied.
Enclosures designed to be light-tight may require testing for light
leaks. Visual inspection in conjunction with a bright source of light
is usually the most satisfactory, but photographic tests may be helpful
when small leaks are involved.
5.16. Care of Mirrors, Prisms, and Lenses. The best way to keep
mirrors, prisms, and lenses clean is to protect them from accumulations of dust and dirt by housing them in tight enclosures that are
opened to the outside air only during use. Such components should
be cleaned when necessary, but this operation must be performed with
care, especially with first-surface mirrors and gratings. Dust may be
removed from glass and quartz surfaces by wiping with lens paper,
or with soft, lint-free cloth. Fingerprints should be removed with a
grease solvent such as pure ethyl alcohol. Rock-salt lenses and
prisms must be protected from moisture by keeping them in a dry
atmosphere. Their surfaces should be cleaned only with dry lens
paper or lint-free cloth.
116
THE TESTING, ADJUSTMENT, AND CARE
[§5-17
The greatest care should be used in cleaning first-surface mirrors in
order to avoid abrading or scratching the surface. Dust particles
may be removed by careful stroking with a clean camel's-hair brush.
Light polishing with chamois or lens paper may sometimes be accomplished without damage to the surface, but considerable risk is involved. Fingerprints on aluminized surfaces may sometimes be removed satisfactorily with a very weak ammonia solution, followed
quickly by rinsing with distilled water. Caustic solutions dissolve
aluminum rapidly and offer the greatest danger to aluminized surfaces.
V
5.17. Cleaning and Care of Gratings. The surfaces of gratings are
so easily damaged by cleaning that every precaution should be. taken
to protect them from dust, dirt, corrosive fumes, and especially from
fingerprints. The ruled surface of a grating should never be touched
with the fingers under any circumstances. If a grating is used exposed in a room, it is desirable that the air supply to the roon> be
filtered to free it from dust. In any event, the grating should be
covered by a tight enclosure when not in use.
A properly housed and protected grating should not have to be
cleaned more than once in every two or three years. When cleaning
is necessary, the collodion or gelatin stripping technique is very
effective for speculum gratings, but it must be used with great caution
on gratings ruled on aluminum or other metal films deposited on glass.
A thin film of dissolved collodion or gelatin is poured over the rulings
and the surrounding area and allowed to harden. The film is then
carefully lifted at one edge and pulled off. Speculum or steel gratings
may be immersed in distilled water to facilitate removal of the
collodion film. Dirt particles remain embedded in the film and are
removed with, it. Before applying this method to metal-on-glass
gratings, the stripping technique should be tested on a small portion
of the unruled area to make sure that the metal film will not be
stripped from the glass. This method should not be used to clean
replicas.
Dust particles are sometimes removed from gratings by stroking
the grating gently in the direction of the rulings with a clean caniel'shair brush or tuft of cotton. This procedure should be applied with
caution, especially in the case of metal-on-glass rulings or firstsurface replicas.
Metal-on-glass gratings may be swabbed lightly with distilled
water in the direction of the rulings. Alternatively, a very dilute
§5.17]
OF SPECTROSCOPIC EQUIPMENT
117
solution of ammonia m a y be used to remove stains, grease, and other
accretions, followed quickly by thorough rinsing with distilled water
to remove the ammonia solution. W e t swabbing should n o t be
applied t o metal-on-glass gratings or first-surface reflection replicas
unless trial on an unruled portion has indicated t h a t t h e procedure
is safe.
Second-surface replicas, in which a collodion cast of t h e master is
m o u n t e d with rulings adjacent to the supporting surface, m a y be
cleaned by wiping them carefully with cotton or lens paper. Since
the rulings are protected, t h e principal reason for caution is t o avoid
scratching the collodion surface.
T h e reflecting power of metal-surfaced reflection gratings deteriorates with age. I n the case of speculum-metal gratings, this
deterioration is due to the formation of copper oxide coatings, which
m a y be removed rather effectively b y dilute ammonia, as described
above. After the first year, deterioration of the reflecting power of
metal-surfaced gratings is not m a r k e d over a period of 10 t o 15 years,
provided the gratings are properly protected and cleaned periodically.
T h e greatest change is usually found t o be a decrease in ultraviolet
reflection.
GENERAL REFERENCES
5.1.
A. C. Hardy and F. H. Perrin, The Principles of Optics. New York:
McGraw-Hill Book Company, Inc., 1932.
5.2. G. F. C. Searle, Experimental Optics. London: Cambridge University
Press, 1925.
5.3. R. A, Sawyer, Experimental Spectroscopy. New York: Prentice-Hall,
I n c , 1944,
CHAPTER 6
Illumination of the Spectroscope
WHEN ONE ARRANGES THE ILLUMINATION of a. spectroscopic
system, it is important to remember the following points:
1. To take advantage of the theoretical resolving power ,of the
system, the full width of the prism, ruled grating surface, orv other
dispersing device should be filled with light (except insofar as optical
defects in the system need to be masked out).
2. To.achieve minimum times of exposure in using spectrographs or
to obtain maximum radiant power in the spectra formed by spectroscopic instruments, it is essential (a) that the full aperture of the
system be filled with light and (b) that the source employed have the
highest practicable radiance (radiant power output per unit solid angle
per unit area of emitting surface) within the desired spectral range.
3. To achieve accuracy in spectrophotometric procedures, quantitative emission analysis, and various other spectroscopic applications, it is frequently necessary that the slit of the spectroscopic
system be illuminated uniformly throughout its length.
In particular situations, it is sometimes necessary to effect a compromise in which one or more of these conditions are not fulfilled as
completely as might be desired.
,
6.1. Coherent and Noncoherent Radiation. When one considers
the illumination of spectroscopic systems, it is convenient to distinguish between coherent and noncoherent radiation. In coherent
radiation, there are definite phase relationships between radiation 'at
different positions in a cross section of the radiant energy beam,
whereas in noncoherent radiation these relationships are random.
For example, a slit is filled with approximately coherent radiation
when it receives light from a small distant source, because every
portion of the slit is then illuminated by the light from each radiating
atom or molecule of the source.. On the other hand, a slit which has
an image of the source formed on it is filled with approximately non118
§6.2]
ILLUMINATION OF THE SPECTROSCOPE
119
coherent radiation, since each point of the slit is then illuminated
essentially by a single point of the source.
In practice, the illumination of any surface or aperture is never
completely coherent or noncoherent, but the approximation to one
of these two extremes may be very close. The importance of distinguishing between the extremes arises from the fact that radiation
from coherently illuminated apertures or surfaces may give rise to
interference phenomena, whereas such phenomena do not occur if
the illumination is noncoherent.
11
The computation of the theoretical resolving power of dispersing
components, for example, prisms and lenses, is usually based on the
assumption that they are illuminated with radiation that is coherent
throughout the entire width of their dispersing surfaces (see Chapters
3 and 4). This assumption is strictly true only if the dispersing
component is illuminated by radiation originating from a point source
or an infinitely narrow slit. In practice, the slit always has finite
size. If the illumination of the slit itself is noncoherent, the illumination of the dispersing component will be nearly coherent if the
optical paths from the two edges of the slit to the dispersmg component do not differ by more than about one-fourth the wavelength
of the radiation used. If the illumination of the slit is coherent, the
same approximation to coherence at the dispersing component is
achieved with path-length differences from the two edges of the slit
of about one-half wavelength. Thus a coherently illuminated slit
may be made approximately twice as wide as a noncoherently illuminated one for equally effective achievement of the maximum
resolving power of a dispersing system illuminated by the slit.
6.2. Spectral Line Shape and the Rayleigh Criterion for Resolution. The expressions for the theoretical resolving power of spectroscopic systems given in Chapters 3 and 4 were based on the
assumption that the entrance slit is equivalent to an infinitely narrow
light source and that the broadening of the slit images in the spectrum, which limits the resolving power X/d\, is due entirely to
diffraction effects. In addition, aberrations and imperfections in
optical components may cause diffuseness, broadening, or irregularity
in the slit images and may thereby reduce the realizable resolving
power. Moreover, since the slit is never infinitely narrow, it is
necessary to take into account the effect of slit width on effective
resolving power and spectral purity.
The intensity distribution of a spectrum line, as produced by a
120
ILLUMINATION OF THE SPECTROSCOPE
[§6.2
spectrograph of high resolving power, is somewhat like that shown
in Fig. 6.1. No definite width can be set for such a line, so it is
customary to use instead the half-intensity breadth b, or breadth at
the intensity which is half, that at the maximum. The observed
intensity distribution of the line is governed by two factors: the
distribution of energy in the line as a function of wavelength, which
is determined by the emitter, and the distribution of monochromatic
radiation along the spectrum, which is determined by the spectroscopic apparatus. Each approximately monochromatic section of the
true line shape is subject to the "apparatus broadening," and the
resulting spectrum line is the sum of all
these sections taken together. Of considerable effect on this line shape are
max.
the width of the slit being used and the
mode by which it is illuminated.
max./2
Any definition of what constitutes
Frequency — the limit of resolution between partially
Fig. 6.1. Intensity distribu- overlapping spectrum lines is arbitrary.
tion of a typical spectrum line at The Rayleigh criterion, previously
high resolution; b, half-inten- referred to, states that two imtiges of
sity breadth.
infinitely narrow line sources, or point
sources, are to be considered resolvable when they are separa,ted by
such an amount that the central diffraction maximum of one falls on
the first diffraction minimum of the other (Fig. 6.2). If the maxima
are of equal intensity and if the intensity curves as a function of
distance from the central maxima are of such shape as would be expected from elementary diffraction theory, the intensity midway
between the maxima is about 82 per cent of that at either maximum. .
The Rayleigh criterion provides a good working rule as to what
may be expected in the ability of the eye or of the photographic plate
to separate two neighboring diffraction maxima.* The actual separation of the maxima required in order that they may be distinguished
as separate depends upon a variety of factors, including (a) the/response of the light receptor, such as the eye, a photographic plate, or
a photoelectric device, to variations in intensity; (b) the shapes of
the intensity curves in the diffraction patterns; (c) the .relative intensities of the diffraction maxima; (d) the effects of imperfections in
optical-image formation as a result of mechanical vibrations, optical
aberrations, and defects in the optical system; and (e) the effects of
grain size or other resolution-limiting characteristics of the receptor.
§6.3]
I L L U M I N A T I O N OF T H E
SPECTROSCOPE
121
The general problem of criteria of resolving power has been considered in detail by several workers.^ For our purposes it will be
sufficient to use the Rayleigh criterion as one that is convenient and
closely in agreement with average practical experience.
0.3. Selection of Optimiun Slit Width. Suppose the slit to be
illuminated by light from a source that yields two monochromatic
spectrum lines of equal intensity and of just sufficient difference in
wavelength, d\, to be resolvable by the spectroscopic system, according to the Rayleigh criterion. Suppose the slit to be opened very
slightly, say to a width such that its optical image in the plane of the
spectrum (as determined by geometrical optics) is one-twentieth the
width of its observable pattern.
Two diffraction patterns, corresponding to the two spectrum
lines, will appear side by side in
the spectrum, each with an
intensity-distribution curve approximately that shown in Fig.
6.2. The central diffraction
Fig. 6.2. Rayleigh criterion for the
maximum of each pattern will
lie over the first minimum of resolution of spectrum lines. The central diffraction maximum of line A falls
the other pattern, as in Fig. 6.2, upon the first diffraction minimum of line
since it has been assumed that B at the limit of resolution as defined by
the separation of the patterns is the Rayleigh criterion.
just such as to satisfy the Rayleigh criterion of resolution.
Let us now consider one of these diffraction patterns only (say A
in Fig. 6.2) and examine the effect on it of increasing the slit width.
Suppose ^he slit is opened symmetrically to three times its original
width—that is, to a width such that- its optical image is threetwentieths the width of/the original diffraction pattern. If the slit
is illuminated with noncoherent radiation—that is, if there are no
definite and continuing phase relationships between radiation coming
from different portions of the slit—the effect will be as though a new
slit were placed on each side of the original slit, all being of equal
width. We can consider each of these new slits as contributing its
own diffraction pattern but displaced to the right and to the left of
the original pattern by one-twentieth the distance between the
' T h e following papers include references to earlier work: B. P. Ramsey, E. L.
Cleveland, and W. A. Bowen, Jr., Jour. Opt. Soc. Am., 32, 288 (1912); B. P. Ramsey,
O. T. Koppius, and E. L. Cleveland, Jour. Opt. Soc. Am., 31, 202 (1941).
I L L U M I N A T I O N OF T H E
122
SPECTROSCOPE
[§6.3
minima. The new intensity distribution niay be obtained by summing up the three distribution curves. If this is done it will be
found that the effect of increasing the slit width by three times has
been primarily to increase the intensity of the central maximum by
approximately three times without broadening the intensity distribution curve materially and, therefore, without substantjally
influencing the separation required for resolution.
As the slit is widened further, the central maximum continues to^
increase in intensity but at a proportionately lesser rate. Finally,
1.5
10
when the slit is sufficiently" wide to
r\
give a geometrical image of about
ri
half
the width of the original diffrac1.0
• 1 \
L/Af
' 1
tion
pattern, as measured between
1
/ ^
I
Ic^i 1
the two minima lying at either side
i i
of the central maximum, further
Lnc-/ -'
0.5 •;. 1
1 (
increase in slit width contributes
l 1 •''
negligibly to the intensity of the
/
central maximum. The principal
effect of further opening of the slit
A? = WM
is to increase the effective width of
the
diffraction pattern and thus
Fig. 6.3. Variation of intensity,
/, of the central diffraction maxi- to increase the wavelength differmum, and of the half-intensity line ence necessary in order that two
width, L, as a function of the slitwidth factor /3. j3 = D/X/, where neighboring spectrum lines may be
D is the slit width, X is the wave- separated.
length and / is the aperture ratio of
Figure 6.3 shows the variation of
the collimator lens. The subscripts
the
intensity of the central maxic and nc refer respectively to coherent and noncoherent illumination mum and of the line breadth with
of the slit, values for which are from slit width, in accordance with the
van Cittert^.
theoretical results of van Cittert.^
Curves are given for cases in which
the slit is illuminated by noncoherent and coherent radiation,
and results for any actual case will usually lie somewhere between
the two sets of curves.
'
In particular, one should note that for any spectrograph there is a
critical slit width, corresponding to an optical image about one-fourth
to one-half the width of the central diffraction maximum, beyond
•
•
•
1
1
1
2 P. H. van Cittert, Zntschr. f. Phys., 65, 547 (1930); 69, 298 (1931).
§6.3]
ILLUMINATION OF THE SPECTROSCOPE
123
which the peak intensity of a spectrum line is influenced but little
as the slit is widened, whereas its breadth increases greatly. Under
circumstances where maximum resolution is desired without undue
increase in time of exposure, one should use this critica slit width.
This width can be calculated from the formula D = /JfX, where D is
the width of the slit, X is the wavelength considered, and / is the
numerical aperture of the system, usually taken as the focal length
of the collimator lens divided by its diameter. The factor ^ lies
between 1 and 2, being approximately 2 for coherent radiation and
1 for noncoherent (see Fig. 6.3). Thus, for a prism spectrograph
having a collimator lens of 5 em diameter and of 40 cm focal length
(/ = 40/5 = 8), the critical slit width for X 5000 A is 40,000 A or
0.004 mm for noncoherent radiation and 0.008 mm for coherent.
Because a wide range of wavelengths is usually involved and the
radiation is likely to be a mixture of coherent and noncoherent, in
any actual case it is best to calculate the approximate critical slit
width, set the slit at some value less than this, and then gradually
widen the slit while observing the illumination behind the slit on a
white card. As the critical slit width is reached, a considerable
increase in brightness is observed, beyond which the. brightness
increases very slowly.
Another procedure makes use of the diffraction pattern produced
by the slit itself upon illumination with coherent light. The central
maximum of this pattern just fills the collimator with light when the
slit is sufficiently narrow to produce a geometrical image half as wide
as the half breadth b of the spectral diffraction patterns.* The slit
is illuminated by an approximately point source placed on the optic
axis at a distance of 25 cm or more. Starting with a slit width too
great to fill the collimator with light by diffraction, the observer
narrows the slit until, on looking through the system from the position
of the spectrum, he sees the central diffraction maximum fill about
two-thirds of the collimator.
The above adjustments must be made with visible light. Since
the appropriate slit width for the condition desired depends upon the
wavelength, this factor should be taken into account in making
proportionate adjustments for the wavelength region to be employed.
When maximum resolution is desired, it is customary to try first
a slit width corresponding to an optical image about one-fourth to
* This statement is true only when the collimator and the dispersing element have
the same effective aperture, which is usually the case.
124
ILLUMINATION OF THE SPECTROSCOPE
[§6.4
one-half the width of the diffraction pattern. Theoretically, this
width should give a good compromise between maximum line intensity and best possible resolution. One then experiments with slightly
different slit widths until the best results are obtained.
The effects of slit width on spectral purity for the noncoherent and
coherent modes of illumination have been considered theoretically
and experimentally by several workers, including Schuster,' van
Cittert,^ Miinster,^ and Stockbarger and Burns.^ Precise analyses
from the standpoint of diffraction theory are of limited applicability
to practical cases, both because the illumination of the slit is in
practice never completely coherent or incoherent and because effects
other than diffraction, such as aberrations, influence the practical
results.
6.4. Filling the Aperture of the Spectroscope with Light. When
the slit is wide enough so that diffraction effects may be neglected
(slit width > 2|8/X), light may be considered as traveling in straight
lines through the slit to the collimator, and the conditions for filling
the aperture of the spectroscope with light may be determined by
simple geometrical considerations. It will be assumed (a) that the
entire effective width of the collimator must be filled with radiation
to achieve maximum resolution, and (b) that the entire apertures of
both slit and collimator must be filled to achieve maximum transmission of radiant power into the spectrum. Sometimes the design
of the system is such that some component other than the collimator,
such as a prism or telescope lens, acts as a limiting aperture stop,
thereby reducing the area of the collimator aperture that can contribute effective radiation to the system. In such cases it is the
effective aperture of the collimator, as limited by the smallest aperture
stop or dispersing element in the system, that must be illuminated.
If both slit and collimator are filled with light, no additional systeni
of lenses or mirrors will increase the total radiant flux through the
system.
The entire spectroscope aperture will be" filled with radiation when
lines drawn from any position within the effective area of the collimating component through any portion of the slit intersect an
emitting area of the source. The angle that must ,be filled with
' A . Schuster, Theory of Optics. London: Edward Arnold, 1909. Also Astrophys.
Jour., 21, 197 (1905).
* C. Munster, Ann. d. Physik, 15, 619 (1932).
5 D. C. Stockbarger and L. Burns, Jour. Opt. Soc. Am., 23, .STO (1933).
§6.4]
ILLUMINATION OF T H E
SPECTROSCOPE
125
radiation in any plane passing through the optic axis of the collimator
is defined by the most oblique rays that can be drawn from the edges
of the effective aperture of the collimator past edges of the slit (for
example the solid lines in Fig. 6.4).
Source or
condensing
lens
mi"
Fig. 6.4. Geometrical condition for filling the spectroscope aperture with light.
(a) Section of coUimating system in plane perpendicular to slit, (b) Section of
coUimating system in plane parallel to slit.
Many sources are too small to fill the limiting angles with radiation
even when placed close to the slit, or they cannot be placed close
enough to the slit to fill the collimator. Condensing lenses of sufficient aperture (Fig. 6.5) are then used.
Source
(a)
Slit
Source
(b)
Fig. 6.5. Use of condensing lenses to fill the spectroscope aperture with light
when the source is too small to accomplish this purpose, (a) Spherical condensing
lens, (b) Two cylindrical condensing lenses.
126
ILLUMINATION OF THE SPECTROSCOPE
[§6.5
6.5. Use of a Condensing Lens or Mirror. In many uses of the
spectroscope, an image of the source is formed on the slit with a
condensing lens. If the lens is to be used throughout the visible and
ultraviolet, it may conveniently be an achromatic triplet with two
outer components of quartz and an inner component of fluorite. A
simple quartz lens, which has small chromatic aberration in the
visible, or a glass achromatic lens, will serve for the visible. The
problem of chromatic aberration is eliminated if a spherical concave
mirror is used (preferably one coated with aluminum for the visible
and ultraviolet), but the geometry of the condensing system is not so
convenient as with a lens.
When an image of the source is formed on the slit of a stigmatic
spectroscope, the spectrum lines will not be of uniform intensity
throughout, since different parts of the source will contribute to
different parts of each line. An astigmatic spectroscope will tend to
diminish these differences and produce more uniform and better
appearing lines. These do not, however, reveal so much information
regarding variation in radiation from different parts of the source.
The approximate linear dimensions of the source or condensing
lens required to fill the effective aperture of the collimator may be
computed as follows. Let the focal length of the coUimating component be F and its aperture ratio / = F/d, where d is the width of
this component. Let the length of the slit be / and the distance from
the source—or from the condensing lens if one is used—to the slit be x.
The width of the slit may be neglected in this computation. If the
required dimension of the source or condensing lens in the direction
parallel to the slit is called the height H, then, to a first approximation,
H = :c| + i ( | + l )
(6.1)
If X is small in proportion to F, the second term in this equation.is
approximately equal to the slit length, I. If the required dimension
of the source or condensing lens in the direction perpendicular to the
slit is called W, then
/
, W = xj
• (6.2)
r
Thus for a collimator lens of aperture ratio//10 and of 25 cm focal
length, a slit length of 0.4 cm, and a source-to-slit distance of 10 cm,
H = 2.72 cm and W = 2.0 cm.
If maximum resolution is the primary consideration, coherent
§6.6]
ILLUMINATION OF THE SPECTROSCOPE
127
radiation should fill the aperture of the dispersing element in t h e
plane perpendicular to the slit. Filling the aperture in the plane
parallel t o the slit also is essential if maximum radiant power is t o be
t r a n s m i t t e d into the spectrum, b u t coherence is not so i m p o r t a n t
in this direction.
T h e aperture required t o fill a collimator with light thrown onto t h e
slit b y a condensing lens or mirror m a y be computed most readily
for t h e simple case in which the distance from the source t o the condensing lens is the same as t h a t from t h e condensing lens t o the slit.
T h e n , in the horizontal plane, if a broad slit is used, the / n u m b e r or
numerical aperture of the condensing lens must be half t h a t of t h e
collimator lens to fill the collimator with light. If complete filling
w i t h light is desired in the vertical plane, the / number of the condenser m u s t be still less t h a n one-half t h a t of the collimator. T h i s
s e t u p , giving unit magnification, is frequently used with sources a b o u t
equal t o the slit in length. Often, however, it is desired to form an
enlarged irriage of the source on the slit because the length of t h e
useful emitting area of the source is less t h a n t h a t of the slit
In
such instances, the / number of t h e condensmg lens for complete
filling of the spectroscope aperture m u s t be smaller t h a n for unit
magnification. Conversely, if less t h a n unit magnification is permissible, complete filling of the spectroscope aperture with light m a y
be accomplished with a condenser of g r e a t e r / n u m b e r t h a n is required
for unit magnification. Application of the simple lens formula enables one t o determine the slit-to-condenser distance, x, for these
various cases, after which the size of lens required m a y be computed
from E q s . (6.1) and (6.2).
'With arc and spark sources, a magnification of about fourfold is
ordinarily found convenient t o keep the collimator a n d dispersing
element filled with light while the arc a n d spark wander.
T h e r e is no objection t o using a condenser lens of more t h a n sufficient aperture t o fill t h e collimator, provided the light sent into t h e
spectroscope a t an angle greater t h a n t h a t necessary t o fill the collimator is not scattered into t h e spectrum. Where scattered light is
a p t t o be troublesome, it is i m p o r t a n t t o stop the condenser down
until it just fills t h e collimator with light.
6.6. Uniform Illumination of the Slit. Uniformity of illumination
of t h e slit along its length is sometimes important, as in certain
m e t h o d s of photographic p h o t o m e t r y and in quantitative emission
analysis. I n other applications, for instance in studying with a stig-
128
ILLUMINATION OF THE SPECTROSCOPE
[§6.6
matic spectrograph the spectral emission from different areas of a
source, it is important t h a t t h e slit illumination be not uniform b u t
rather t h a t an image of t h e source be formed on, t h e slit.
Nearly uniform slit illumination m a y be attained by using a small
source, without a condensing lens, placed a t a considerable distance
from the slit (for example, a 0.5-cm source a t 25 cm source-to-slit
distance). With this arrangement, each point on the source illuminates every point on the slit, with negligible variation in intensity
as a result of variation in angle a t which different rays are e m i t t e d
from the source or of differences in p a t h lengths along different rays.
This m e t h o d does n o t t a k e full advantage of the resolving power of
the spectroscope system unless the slit is narrow enough t o fill t h e
collimator with diffracted light. Also, since only t h a t p a r t of t h e
slit length is used which subtends the angle of the collimator a t t h e
source, the uniform lines produced are a p t to be short.
Uniform slit illumination m a y be obtained more satisfactorily by
the use of a condensing lens placed immediately in front of t h e slit
(Fig. 6.6). T h e source is placed a t
Condensing \ens
__,—-^'TA
such a distance from t h e condens~ '
ing lens t h a t an image of t h e source
is formed on the collimator. E a c h
Source
Slit
Cniiimntnr
portiou of the source illuminates
the. entire condensing lens, a n d
Fig. 6.6. Use of a spherical con- hence the entire length of t h e slit,
densing lens immediately in front of Such illumination is not precisely
the slit to achieve uniform slit illumi-c
•
iu
ii. i
i.u
.
uniform, since t h e p a t h length
and angle of emission v a r y slightly
for rays to different points on t h e lens, b u t this effect is almost completely averaged out in the summing u p of illumination from all
portions of the source. T h e method is less wasteful of light t h a n the
previous one, and the full aperture of t h e spectroscope will be filled
with radiation if t h e source is magnified to fill t h e collimator.
A third method of uniform slit illumination involves t h e use o^ a
cylindrical condensing lens placed with its axis parallel t o the slit
(Fig. 6.7) at such a position t h a t rays in the horizontal plane focus
an image of the source on t h e slit b u t rays in the vertical plane pass
through the lens approximately undeviated. If the diameter of t h e
cylindrical lens is sufficient, this arrangement fills the width of t h e
collimator with radiation and, thereby enables t h e a t t a i n m e n t of
o p t i m u m resolution.
§6.6]
ILLUMINATION OF THE SPECTROSCOPE
129
One can obtain uniform slit illumination and a t the same time fill
b o t h slit and collimator with radiation by using two cylindrical
lenses, as in Fig. 6.8. This method is due to G. Hansen. T h e first
lens is so chosen as t o throw a vertical line of light on the slit, focused
only in the horizontal direction. T h e second lens, placed a t the slit
Source
Cylindrical gm
condensing
lens
Collimator
lens
Source
Fig. 6.7. Use of cylindrical condensing lens immediately in front of the
slit to achieve uniform slit illumination.
with its axis horizontal, throws a n image of the source on the collimator as a horizontal band of light focused in the vertical direction
only. W i t h such a system, the m a x i m u m resolution of t h e prism or
grating is available, since it is filled with coherent light across its
b r e a d t h and both slit and collimator are filled with radiation.
Source
Fig. 6.8. Use of two cylindrical lenses to obtain uniform slit illumination and maximum total illumination.
Desirable focal lengths for the two cylindrical lenses m a y be comp u t e d as follows. T h e horizontal lens a t the slit should fill t h e collimator vertically. T o obtain this result when an arc or spark is used
often requires a t least fivefold magnification (tenfold might be b e t t e r
b u t is less practicable). Substituting the slit-collimator distance as v
in t h e lens formula 1/F = 1/u + l A and in the relation v = 5M, we
h a v e , in t h e case of a 21-ft grating, v = ^1 ft. (approximately).
130
ILLUMINATION OF THE SPECTROSCOPE
[§6.7
u = 50 in., and /^ = 42 in. Here u is the distance from source to
slit and equals u' + v' for the vertical lens which is to focus the
source on the slit horizontally. Since it is convenient to magnify
an arc or spark about four times on the slit, v' = 4M'; and since
?)' + M' = 50 in., u' = 10 in. and v' = 40 in. Substituting again in
the lens equation, we have F', the focal length of the first cylinder, as
8 in. If quartz cylindrical lenses are used, they will serve without
lengths shorter than 3500 A they may be refocused or different lenses
may be used, but ordinarily one set of lenses will serve from 2000 to
10,000 A, since the focusing on the collimator need be only approximate.
6.7. Illumination of the Slit by a Source Extended in Depth. The
illumination of the slit by a source of extended depth occurs in Raman
spectroscopy and in the use of such sources as a hydrogen discharge
tube viewed end on. It is important to obtain the most effective
slit illumination under these conditions, since the radiance of any
particular section of such a source is usually low. This problem has
been considered by Wood^ and by Nielsen.^
Usually, so little absorption or scattering of the radiation by successive layers along the axis of the source occurs that a considerable
depth of the source may be made to contribute to the illumination
of the slit. If the end of the source is placed close to the slit and if
the various sections are extended enough so that each fills the full
aperture of the spectroscope system with light, the available radiation
will be used effectively. The length of the source is often so great in
proportion to its width that layers some distance from the slit are
not of sufficient area to fill the spectroscope aperture with light.
Under such conditions, a condensing lens may be used and the
optimum conditions of illumination determined by trial. One should
start with a setup in which the farther end of the source is focused on
the slit and then move the condensing lens to bring various sections
of the source into focus until maximum brightness of the spectrum
is obtained.
'
,
6.8. Illumination to Obtain Maximum Radiant Intensity/ or
Total Radiant Power in the Spectral Image. It is often desirable to
obtain either (a) the greatest possible radiant intensity (r-adiant
power per unit area) or (b) the greatest possible radiant power ,in the
spectral image. These two requirements should be distinguished
* R. W. Wood, Physical Optics. New York: The Macmillan Company, 1934.
' J. R. Nielsen, Jour. Opt. Soc. Am., 20, 701 (1930); 37, 494 (1947).
§6.8]
ILLUMINATION OF THE SPECTROSCOPE
131
clearly from each other, since they may be accomplished in different
ways.
The need for maximum radiant intensity in the spectral image
occurs most frequently in spectrography, and to a somewhat more
limited extent in photochemical and photobiological investigations by
means of monochromatic radiation. In spectrography, the required
exposured time is an inverse function of the radiant power per unit
area incident upon the photographic plate. Obviously, intensity is
what is wanted in this case, since the area of plate illuminated by a
spectrum line is likely to be of secondary importance. If the receptor is a photocell, on the other hand, and if all of the incident beam
is intercepted by the cell no matter what the optical arrangement,
the radiant power rather than the intensity is of importance. In the
first case it might be of advantage to use a camera lens of shorter
focal length than the collimator lens, to reduce the area of each
spectrum line and hence increase its intensity, whereas in the second
case this lens would merely increase the difficulty of separating lines,
the total radiant power for a given spectral range remaining the same.
Four factors determine the total radiant power available within a
given wavelength range of the spectral image: (1) the radiant power
per unit solid angle per unit projected area of the source, a factor
known as the steradiancy; (2) the area of the source effective in
illuminating the spectrum; (3) the solid angle of radiation from the
source effective in illuminating the spectrum; and (4) the transmission factor, B, of the spectroscopic system as determined by
absorption and reflection losses, etc. These factors will be considred in the above order.
For any particular wavelength, the limit to the brightness (radiant
power per unit solid angle per unit area) of the spectral image is
determined by the steradiancy of the source at that wavelength.
This limit arises from the fact that it is impossible with any optical
system to form an image of a source that is brighter, or of greater
steradiancy, than the source itself.
Lenses or mirrors may, of course, be used to form enlarged or
reduced images of the source. The ratio of the area of the image, a^,
to that of the object, Oi, is the same, however, as the ratios of the solid
angles, aj2 and coi, within which the radiation leaves the source and
enters the image. Thus if a^ is less than a\, more radiant power per
unit area is delivered into the image than is collected from the source,
but this increased power per unit area of the image is accompanied
132
ILLUMINATION OF THE SPECTROSCOPE
[§6.8
by a proportionate increase in the solid angle through which the
radiation illuminates the image (see Fig. 6.9). Insofar as geometrical
considerations are concerned, the radiant power per unit area per
unit solid angle entering the image therefore remains constant, and
no gain in brightness or steradiancy would be accomplished by forming a reduced image. In practice, some loss in steradiancy is always
to be expected when an image is formed, as a result of reflection "or
absorption losses in the image-forming components. These considerations show that for the greatest brightness of the spectral image
it is essential to select from otherwise suitable sources the one having
highest steradiancy in the wavelength region of interest (see Chapter 8).
Once a source of the highest practicable steradiancy has been
chosen, the next step in achieving maximum useful radiant power in
the spectral image is to make use of cones of radiation, each of which
will fill the entire aperture of the spectroscopic system, from as large
Fig. 6.9. To illustrate that the steradiancy is not increased by forming
a reduced image of a source (see text).
an area of the source as feasible. The simplest case to consider is
one where the|source is placed in juxtaposition to the slit, so that the
slit itself effectively acts as the primary source of radiation. Every
point on an area of the source corresponding to the slit opening will
then illuminate the entire aperture of the spectroscope system (provided the source radiates through a sufficiently wide angle, which is
usually the case). Gain in illumination can then be achieved by
widening the slit up to the point at which the slit width equals/the
source width, but this gain is obtained at the expense of loss'in
spectral purity. In general, there will be some limiting slit width,
determined by the required spectral purity, Ijeyond ,which it is not
feasible to go.
If the source cannot be placed in juxtaposition to the slit, a condensing lens or mirror may bfe used, as described in § 6.5, to form an
image of the source on the slit. If we assume that the slit has been
§6.9]
ILLUMINATION OF THE SPECTROSCOPE
133
opened to the maximum width allowable, the greatest radiant power
is then transmitted into the system when (a) the source image is just
sufficiently large to cover the slit opening and (b) the condensing lens
'is large enough to fill the aperture of the spectroscope system with
radiation. The proper focal length and size of condensing lens may
be computed as described previously.
The fourth factor mentioned above, namely the transmission
factor B, is fixed by the choice of the spectroscopic system to be used
and cannot be altered greatly by the method of illumination.
6.9. Factors Governing the Radiant Power Transmission of a
Spectroscopic System. In the preceding section, it was assumed that
a particular spectroscopic system was to be used, and the illumination
conditions necessary to transmit maximum radiant power into the
spectral image wertf considered. If some choice between spectroscopic systems is possible or if a new system is to be designed, it may
be of considerable importance to consider the factors that determine
the total radiant power which can be transmitted by various types
of spectroscopic systems under optimum conditions of illumination.
As is discussed in § 6.12, for a given spectral purity the amount of
radiant power that can be transmitted by any spectroscopic system is
directly proportional to the product of the effective area of the dispersing component. A, the angular dispersion, dd/d\, and the transmission factor B (determined by absorption and reflection losses, as
stated in § 6.8), provided the open area of the slit and the entire
aperture of the system are filled with radiation. Thus, if dO/d\ and
B are constant, the radiant power transmission of a spectroscopic
system can be increased roughly as the square of its linear dimensions
by scaling it up. Theoretically, a size can be reached, of course,
beyond which further increase results in such large slit openings and
apertures that it is impracticable to fill these with light with usual
sources and condensing lenses, but such limits are seldom reached
in practice.
It is important to note again the difference between total radiant
power and intensity in the spectral image. For a spectroscopic
system of given power transmission, the intensity of the spectral
image may be increased, within limits, by concentrating the spectrum
on a smaller area through the use of a telescope of shorter focal length
than the collimator. In a spectrograph for photographing weak
sources, this effect may be advantageous, but the extent to which the
ratio of telescope to collimator focal lengths can be decreased is limited
134
ILLUMINATION OF THE SPECTROSCOPE
[§6.10
either (a) by the smallness of the spectral image which is acceptable or
(b) by the maximum obtainable relative aperture of the telescope.*
Before considering in greater detail the factors involved in t h e
radiant power transmission of spectroscopic systems, the. effect of
slit width on spectral purity will be discussed.
6.10. Effect of Entrance Slit Width on Spectral Purity. In several
types of photographic photometry, it is convenient t o uSfeva slit wide
enough t o produce a flkt t o p on
the widest spectrum line to be
measured. Figure 6.10 shows
the contours of a spectrum-line
^j ly\
/
V
image, first with a narrow slit
(a)
(b)
and then with a wide slit. If
Fig. 6.10. Spectrum-line contours, (a)
For narrow slit, (b) For wide slit.
two lines are t o be compared in
intensity, the m a r k e d variation
in intensity t h r o u g h o u t the
width of the line image requires careful measurement in t h e first
case. I n the second case, t h e flat-topped portion of each line contains
a contribution from every p a r t of its contour, a n d hence a single
measurement on this flat-topped portion gives a value t h a t is p r o portional t o the intensity of the line. Widening t h e slit so as t o
obtain a flat-topped image results, of course, in some loss of effeptive
resolving power, b u t this can usually be tolerated without serious
difficulty in such applications as quantitative emission analysis.
Slits wider t h a n those which give optimum resolution m u s t also be
used with m a n y spectrometers and monochromators employed for
spectrophotometry, photochemical investigations, and similar uses.
I n such applications, the transmission of large a m o u n t s of r a d i a n t ,
power into the spectrum is often essential, a n d it is customary t o use
entrance slit widths m a n y times those required for m a x i m u m resolution.
I n most applications involving wide slits, t h e slit width is such
t h a t its image is m a n y times the width of t h e central diffraction
maximum. Under these circumstances, the contributions of diffraction t o the slit image m a y be neglected, and it m a y be assumed t h a t
the spectrum consists of optical images of the slit formed according
t o the ordinary laws of geometrical optics. If t h e plane of the,'spec* As the focal, length of the telescope is decreased, its diameter must be kept'constant
if it is to accept the full light beam from the collimator, and thus its relative aperture
must be increased.
^6.10]
ILLUMINATION OF THE SPECTROSCOPE
135
truin is perpendicular to the axis from it to the collimator (that is, if
no plate tilt is necessary), the size of the optical image is equal to
that of the slit multiplied by the magnification of the optical system.
This magnification equals F2/F1, where Fi and F^ are, respectively, the
focal length of the collimator and telescope. Thus if w and w' are,
respectively, the width of the entrance slit and its image, and I and
V are, respectively, the lengths of the slit and slit image, then
w' = v> ^^
l' = ljr
(6.3)
(6.4)
Or, since the slit area a and the image area a' are, respectively, equal
to wl and w'l',
(6.5)
If the spectral plane is appreciably inclined to the axis of the collimatori the slit images will be broadened as a result, the broadening
factor being equal to the cosecant of the angle of inclination.
When a wide slit is used to form images of widths Wi and W2'
of two spectrum lines of wavelengths Xi and X2, the line images
will just be separated in the spectrum, without overlapping, when
|wi' + IW2' = AZ, where Al is the distance between the positions in
the spectrum corresponding to Xi and X2 as observed with narrow slits
corresponding to maximum resolution. In instances such as a
mercury-arc spectrum in which the spectrum lines are widely separated in wavelength, exceedingly wide entrance slits may therefori^
be employed without causing overlap of the slit images, and this fact
is often made use of when monochromators are required to supply
large amounts of nearly monochromatic radiant power.
If wide slits are used with a continuous spectrum, there will be an
overlapping of slit images corresponding to an appreciable range of
wavelengths at each position in the spectrum. The simplest case is
that in which the linear dispersion is approximately independent of X
and the spectral plane is approximately perpendicular to the rays
from the telescope, as in the case of normal spectra formed by diffraction gratings. The wavelength interval, AX, in the spectrum covered
by a single slit image may then be computed as follows: If the angular
dispersion is dd/d\ and the focal length of the telescope is F2, the
linear distance in the spectrum corresponding to the wavelength
136
ILLUMINATION OF THE SPECTROSCOPE
[§6.11
difference AX is Ak(dd/d\F)2- This value, however, also represents
the width, w', of the slit image. Hence
w' = AXFi—^
(6.6,
But from Eq. (6.3), w' = w{F-i/'Fi), where w is the slit width, and
therefore
,F,
^^„
F,de
rift
w = AXf 1 ^
(6.7)
^'d\
Within the width, w', of the slit image, there is therefore partial
overlapping of other slit images corresponding to a range of wavelengths MX; for if X^ is the wavelength corresponding to the middle
of the slit image, images w^ and Ws corresponding to wavelengths
Xm + AX and X„ — AX will be displaced just sufficiently so as not to
overlap the X„ image at all. For intermediate wavelengths, there
will be intermediate amounts of overlap of the corresponding images
with that due to Xm6.11. Effect of Exit Slit. In monochromators, a portion of the
spectrum is isolated by an exit slit so that this portion only passes out
of the system. If a line source is used and the entrance slit is narrow
enough so that the images of the lines do not overlap (see § 6.10), the
exit slit may be made the full size of the line image in the spectrum
plane without passing radiation of other wavelengths except insofar
as scattered radiation is concerned. As pointed out in § 6.10, if a
continuous source is used, there is always considerable overlapping of
slit images corresponding to different wavelengths, provided the
entrance slit is of appreciable width, and the radiation passing out of
the exit sht will therefore always be somewhat impure no matter
how narrow the exit slit is made.
The type of spectral distribution curve to be expected in the
radiation from the exit slit when a continuous source is used may be
deduced by extending the considerations of § 6.10. As in that
section, it will be assumed that the spectrum is uniformly dispersed
§6.11]
ILLUMINATION OF THE SPECTROSCOPE
137
a n d lies in a plane perpendicular to rays from the telescope. Let t h e
width of the exit slit be E and that" of the image of the entrance slit
formed in the spectrum be w'. If these are equal, the exit slit will
pass a range of wavelengths from X™ + AX t o Xm — AX, where AX is
t h e wavelength range covered b y w', for w' is partially overlapped
by slit images corresponding t o wavelengths within this range, as
explained in § 6.10. If the spectrum is of constant intensity as a
function of wavelength, t h e intensity-distribution curve of radiation
passing through the exit slit will be a triangle rising from zero a t
Xm + AX t o a maximum a t Xm a n d falling t o zero again a t X^ — AX
(Fig. 6.11). T h e half-intensity b a n d width of the exit radiation is,
accordingly, AX, and the full band width is 2AX. If the exit slit E is
wider or narrower t h a n the entrance slit image w', let A'X represent
l i A - AXI
Im
Im - -
i
2
/
A.m-il
A,
^\\
^m ^m
r-1
/ i A>°'i\
/iSr\
1 '^ ;
2
(a)
''"' '
2
(b)
Fig. 6.11. Intensity distribution curves for radiation transmitted by a monochromator with exit-slit and entrance-slit image (a) of equal width and (b) of
different width.
'
the wavelength interval covered in the spectrum by the exit slit,
and AX equal t h a t covered b y the entrance slit image, as above. T h e
intensity-distribution curve of the exit radiation will t h e n be a
trapezoid with a flat top and symmetrically sloping sides (Fig. 6.11b).
T h e full band width of the exit radiation will be A'X -{- AX, the halfintensity band width will be the larger of the two values A'X a n d AX,
and t h e band width of t h e flat t o p will be j A'X — AX |.
F o r a given half-intensity b a n d width, the maximum r a d i a n t power
is t r a n s m i t t e d by a spectroscopic system when w' = E. However, if
a n intensity-distribution curve with a broad flat top is desired, w'
m u s t be considerably larger or smaller t h a n E.
M o r e complicated cases, in which t h e dispersion is not linear, t h e
intensity is not a constant function of wavelength, a n d so on, m a y
138
ILLUMINATION OF THE SPECTROSCOPE
[§6.12
be analyzed by extensions of the above considerations. A further
discussion of the effect of slit widths in using monochromators 'isN
given by Hogness et al.^ a n d in General Reference 6.3.
6.12. Expression for R a d i a n t P o w e r Transmission of a Spectroscopic System. T o obtain an expression for the r a d i a n t power t r a n s mission of a spectroscopic system, it m a y be assumed t h a t sources are
available for illuminating t h e system in such a way t h a t t h e entire
entrance slit and collimator aperture are filled with light, as described in § 6.8. U n d e r these circumstances, t h e r a d i a n t power
within a given wavelength range AX t h a t enters t h e system is equal
to Ri,\wlco, where RA\ is the steradiancy (for example, in microwatts
per steradian per cm^) of t h e source image on t h e slit within t h e
wavelength range AX; w and I are, respectively, t h e slit width and
slit length; and a; is t h e solid angle (in steradians) subtended by
t h e collimator a t t h e slit. T h e a m o u n t of r a d i a n t power PAX of
wavelength range AX t r a n s m i t t e d through t h e system into t h e spectral
image is influenced by reflection and absorption losses, and in some
instances by vignetting of t h e beam by apertures farther along in t h e
system t h a n the collimator. If all these factors are lumped together
into a single fractional coefBcient B, then
PAX = BRAXWIW
(6.9)
T h e solid angle w is approximately A/Fi^, where A is t h e area of t h e
collimator and i^i is its focal length. T h e width w of the entrance
slit t h a t can be employed depends on the spectral purity desired.
If AX is t h e wavelength interval in the spectrum t h a t it is permissible
for the slit image t o cover, t h e n from E q . (6.7), w =
A'\Fi{dd/dX),
where {dd/d\) is t h e angular dispersion of the system. F u r t h e r m o r e ,
t h e length of slit t h a t it is feasible t o use is approximately proportional
to the focal length F\ of t h e collimator, other factors being c o n s t a n t ,
and hence it is possible t o write I = KFi, where K is the allowable
slit length per unit collimator focal length. W e substitute these
values for w, w, and I in E q . (6.9):
df)
PAX = BRAXA^^KA
—
-
(6.10)
' T . R. Hogness, F. P. Zscheile, and A. E. Sidwell, Jr., Jour. Phys. Chem., 41, 379
(1937).
§6.12]
ILLUMINATION OF THE SPECTROSCOPE
139
T h u s t h e relative radiant power transmission, T, of two spectroscopic systems, assuming t h a t AX, K; and Rw are the same in the
two instances, is
B,A^
(d6\
\d\Ji
(6.11)
-••^S).
T h e relative power transmission for given spectral p u r i t y depends
only on the transmission factor B, t h e area A of the collimator (or of
the dispersing component if t h a t is smaller), and the angular dispersion dd/d\, provided the entrance slit a n d collimator are filled with
light in all instances.
T a b l e 6.1 shows the relative power transmission of several typical
spectroscopic systems as computed from E q . (6.11). I n determining
t h e transmission factors B for t h e various instruments, we h a v e
assumed t h a t the transmission of each lens is 0.90 and of each prism
0.85, allowing both for reflection a n d absorption losses, and t h a t a
grating is used which sends 40 per cent of the incident light into t h e
particular spectrum order under consideration. T h e values of T in
t h e table are based on unity for the small q u a r t z spectrograph, a n d
it is assumed t h a t the dispersing characteristics of the q u a r t z used
in the various prisms are the same in all instances.
TABLE 6.1
RELATIVE RADIANT POWER TRANSMISSION OP DIFFERENT SPECTROSCOPIC
SYSTEMS AT 3000 A FOB CONSTANT SPECTBAL PURITY
System
Small quartz spectograph
0.69
Medium quartz spectrograph
0.69
de/dK,
rad./A
A,
sq cm
6.69
X 10-5
6.69
2.5
C-A
d\
Relative
power
transmission, T
1
6.9
16.8
xio-5
21-ft grating, Wadsworth mounting, 15,000 lines/inch
Quartz monochromator (YoungThoUon prisms)
0.40
6.0
75.0
180
15.0
9.0
37
3.1
xio-5
0.59
6.69
xio-=
140
ILLUMINATION OF THE SPECTROSCOPE
GENERAL, REFERENCES
e.I.
6.2.
6.3.
6.4.
6.5.
6.6.
6.7.
R. A. Sawyer, Experimental Spectroscopy. New York: Prentice-Hall,
Inc., 1944.
C. F. Meyer, The Diffraction of Light, X-Rays, and Material Particles.
Chicago: University of Chicago Press, 1934.
W. E. Forsythe (Ed.), The Measurement of Radiant Energy. New York:
McGraw-Hill Book Company, Inc., 1937.
A. C. Hardy and F. H. Perrin, The Principles of Optics. New York:
McGraw-Hill Book Company, Inc., 1932.
M. von Rohr, The Formation of Images in Optical Instruments. London:
His Majesty's Stationery Office, 1920.
R. W. Wood, Physical Optics. New York: The Macmillan Company,
1934.
F. A, Jenkins and H. E. White, Fundamentals of Physical Optics. New
York: McGraw-Hill Book Company, Inc., 1937.
CHAPTER
7
Photography of the Spectrum
T H E VARIOUS REGIONS OF THE OPTICAL SPECTRUM can be observed
and measured by one or more of four principal methods: photographic, thermoelectric, photoelectric, and visual. Of these, the
photographic method is by far the most important in emission spectroscopy and is useful in many applications of absorption spectroscopy. Photographing the spectrum, which can be done for all
wavelengths shorter than 13,000 A, results in a permanent record
that can be studied at leisure, and makes possible the simultaneous
recording of all lines lying in broad regions of the spectrum. It can
also be used to integrate over a period of time the light from a source
of varying brightness.
In this chapter we consider what may be termed "qualitative
photography," as distinguished from quantitative photography, or
photographic photometry, which will be discussed in detail in Chapter 13. To produce satisfactory spectrograms requires some knowledge of the properties of photographic materials and the most
satisfactory methods of handling them.
The photographic emulsions most commonly used in spectroscopic
work are those classed by the manufacturers as negative materials,
since these are more sensitive to light than positive materials and
have more useful response characteristics. A few positive emulsions
such as cine positive have recently come into wide use for spectrochemical analysis, however. The light intensities ordinarily involved
in spectrum photography are much fainter than those used in portrait
and landscape photography, and exposure times are likely to range
from a few seconds to several hours. Spectrum photography requires techniques slightly different from those used in ordinary
photography.
7.1. Photographic Plates and Films. Spectrograms from which
precise wavelength determinations are to be made, or with which
permanence and ease of handling are desirable, are ordinarily made
141
142
PHOTOGRAPHY OF THE SPECTRUM
[§7.1
on photographic plates. These consist of moderately flat pieces of
glass coated with a thin layer of gelatin containing an emulsion of
silver halide salts. Photographic films, which consist of a similar
emulsion coated on thin sheets of cellulose nitrate or acetate, are
likely to be more uniform in sensitivity over their surface area than
plates, and are therefore preferable for making precise intensity
measurements. Films may be used with spectrographs that produce
a spectrum lying on a steeply curved focal surface to which glass
plates could not easily be fitted without breaking. Films and plates
each have their own advantages and disadvantages, and their selection
in a particular case, apart from the considerations given above, is
largely a matter of convenience in cutting and handling.
The negative photographic emulsion is a suspension in gelatin of
a mixture of silver bromide with a little silver iodide which has been
treated by removal of soluble salts and which has been ripened by
carefully adjusted processes to control the sizes of the multitude of
silver halide crystals that it contains. These crystals or "grains"
vary from 5 /i in diameter down to a size too small to be seen through
a microscope, a common diameter being 1 /z. Most of the useful
properties of the photographic emulsion depend on the size and
size distribution of these grains. The larger the grains, in general,
the more sensitive the emulsion.
When a photographic emulsion is prepared in the dark and is then
exposed to radiation containing wavelengths to which it is sensitive,
a latent image is formed that can be made visible by development.
This latent photographic image consists of an aggregation of. grains
of silver halide that have been altered, presumably in some photoelectric manner, by absorption of incident radiation. Development
consists of treating the emulsion with chemicals that reduce toi
metallic silver the silver salts in those grains which have been affected,
by light, and do not affect those which have not. The resulting
clusters of developed silver grains darken the emulsion locally. To
make this darkening permanent on exposure, of the whole plate to
light, the unreduced silver halide is dissolved, after development, in'a
suitable chemical solution, usually one of sodium thiosulphate (commonly called hypo). This'process is known &s fixing. Also incorporated in the fixer solution is a hardening agent that toughens the
gelatin and makes it less sensitive to temperature changes. After
being well fixed, the emulsion jis washed thoroughly to remove all
remaining hypo.
§7.2]
PHOTOGRAPHY OF THE SPECTRUM
143
The modern photographic emulsion makes use of chemical amplification, because development involves amplification quite analogous
to that produced by vacuum tubes in electronic amplification.
Many photochemical processes are known in which absorption of
light will produce a change in the color of the chemicals involved, but
such reactions are low in sensitivity, since the absorption of one
quantum ordinarily does not change more than one molecule. The
photographic emulsion consists of chemicals so arranged that the
absorption of a few quanta of light will alter a whole grain of silver
halide containing many billions of molecules. The energy that produces a simple photochemical change comes from the incident light,
whereas that which alters a grain in the photographic emulsion comes
from the chemicals used in development, the incident light furnishing
only enough energy to trigger off the reaction. Thus a photographic
plate or film can be made thousands of times more sensitive than
blueprint paper or similar materials which record light and shade
by simple photochemical processes.
The latent image, which is detected and made permanent by
development, probably consists of minute particles of silver resulting
from a change in the positions of electrons in the crystal lattice of the
grain. I t is very stable; photographs have been successfully developed years after exposure.
7.2. Response of the Emulsion to Light. The degree of blackening of a given spot on a photographic emulsion is usually expressed in
terms of its density d, a quantity closely proportional to the amount
of metallic silver in a unit area of the image. Density is easily determined by sending a beam of light through the image with a densitometer (§ 13.12) and measuring the fraction of this light that
emerges on the opposite side. The ratio of the transmitted light to
the incident light is called the transmission T of the image. The
reciprocal of transmission is called the opacity 0. Density is the
logarithm of opacity to the base 10; thus
,
d = logio 0 = logio y
A spot -that transmits one-tenth of the light sent through it has a
transmission of one-tenth or 10 per cent, an opacity of 10, and a
density of 1.
The word blackening is sometimes loosely used for opacity, for
144
[§7.2
PHOTOGRAPHY OF THE SPECTRUM
density, or even for other functions of the transmission, but its commonly accepted definition is as given in § 13.15.
The determination of the response of an emulsion to light is known
as sensitometry, and this will be discussed further in Chapter 13.
Ordinary emulsions are found to respond as shown in Fig. 7.1, which
depicts the densities produced in a given plate by various intensities
of light, when the time of exposure and the development conditions
were kept constant. A simple S-shaped curve of the type shown is
characteristic,' though some emulsions show straight-line portions in
the center which may be shorter or longer. The curve relating
Density
LoQiQ Intensity
^
'
Fig. 7.1. The characteristic curve of the photographic emulsion. This curve,
sometimes called the Hurler and Driffield or " H & D " curve, shows how the
density of the developed image changes with Hght intensity when exposure time
and development conditions are kept constant.
density to logio intensity is called the characteristic curve of the emulsion under the conditions used.
When qualitative considerations only are involved, it is usually
immaterial whether density is plotted against log I or log t, where t is
the time of exposure. For this reason it is customary to plot density
against log E, where E, the exposure, is I X t- Either the intensity
or the time of exposure, whichever is most convenient,- can then be
varied, when determining characteristic curves like Fig. 7.1. Since
time is more easily varied than intensity, a time scale is commonly
1 r . Hurler and V. C. Driffield, Jour. Soc. Chem. Ind., 9, 455 (1890).
§7.3]
PHOTOGRAPHY OF THE SPECTRUM
145
involved when log E is plotted, b u t in quantitative photometry it is
usually essential t o vary the intensity rather t h a n the time.
T h e approximate interchangeability of time and intensity, discovered b y Bunsen and Roscoe^ in 1862, is known as the reciprocity
law. Although this law holds fairly accurately for certain types of
emulsions over a moderately broad range of exposure times, for most
emulsions it is only approximate, a n d for very long or very short
exposures to weak or strong light it is a p t t o be far from exact. Exposure t o light of unit intensity for 10,000 seconds, or t o light of
intensity 10,000 for 1 second, is likely t o give a lower density t h a n
exposure t o light of intensity 100 for 100 seconds. This effect will
be discussed further in Chapter 13.
T h e curve shown in Fig. 7.1 consists of an underexposure region, or
" t o e , " from A t o B; a straight-line portion (which m a y be very short
or even nonexistent), which is the so-called normal exposure region,
from B to C; and a "shoulder," or overexposure region, from C to D.
Beyond D t h e curve m a y begin t o fall again, a phenomenon k n o w n
as reversal or solarization.
I t is n a t u r a l to suppose t h a t the most important characteristic of
a photographic emulsion is its sensitivity to light. Closer examination shows t h a t this vague concept of sensitivity has only qualitative
meaning and t h a t an exact definition of sensitivity is best m a d e after
consideration of contrast, speed, and resolving power, which will be
discussed individually.
Density
LogiQ Intensity
Fig. ?.2. The contrast, or gamma, of the photographic emulsion.
7.3. Contrast. T h e contrast of a plate or film, usually written y,
is defined as the tangent of the angle 6 between the straight-line portion of the characteristic curve and t h e intensity axis. (Fig.7.2).
2 R. W. Bunsen and H. E. Roscoe, Ann. Physik, 108, 193 (1876).
146
PHOTOGRAPHY OF THE SPECTRUM
[§7.4
The contrast in an emulsion can be controlled when it is made,
by controlling the distribution of grain size. Emulsions in which the
grains are more nearly all of the same size have higher contrast than
those in which the size variation is great. In emulsions of medium
'type the contrast is usually near 1, the straight-line portion of the'
characteristic curve having a slope of 45 deg. In emulsions of very
high contrast 7 may be as great as 6 or 7, whereas in other emulsions
it may be as low as 0.6. An emulsion of extremely high contrast is
selected when it is desired to make every part of the picture either
black or white, a condition seldom desired in spectrum photography.
Use of an emulsion of low contrast gives a response varying only
slightly with light intensity, and the resulting spectrogram does not
differentiate satisfactorily between weak and strong spectrum lines.
Therefore, emulsions of intermediate contrast are most used for
spectrum photography.
7.4. Speed, Inertia, and Latitude. The speed of an emulsion has
been defined in various ways, but is usually related to the reciprocal
3. Low speed
Density
2. Moderate
speed
Fig. 7.3. Characteristic curves for typical emulsions of high, moderate, and low
speeds.
/
of the inertia (marked i in Fig. 7.2), which measures the distance
between the intercept of the straight-line portion of the characteristic
curve with the line of zero density, and the density axis.. Most speed
ratings have been developed, however, by workers interested in the
fields of commercial and artistic photography rather than spectrum
spectrography, and hence they -involve use of integrated sunlight or
artificial light. For our present purpose it is sufficient to remember
§7.4]
PHOTOGRAPHY OF THE SPECTRUM
147
t h a t speed is an approximate measure of the minimum a m o u n t of
light required t o produce a useful image.
Figure 7.3 shows characteristic curves for typical emulsions of low,
moderate, a n d high speeds. Usually, high speed is associated with
low contrast, and low speed with high contrast. High speed arises
from t h e presence of large grains in t h e emulsion. Since a wide
variety of grain sizes usually results if very large grains are present,
conditions for producing low contrast often arise.
W e t h i n k of one emulsion as being piore sensitive t h a n another if
it will produce a higher density from a given input of light, b u t it is
obvious from Fig. 7.3 t h a t this concept is ambiguous. T h e highspeed emulsion is seen t o be more sensitive t o low intensities of light
t h a n t h e others, whereas the low-speed emulsion is more sensitive t o
high intensities. T h e concept of sensitivity can be m a d e more
rigorous by specifying it in terms of some particular density, for
example, d = 1. T h u s in Fig. 7.4, plotting against wavelength t h e
Log,o
Energy
2000
3000
4000
5000
6000
7000
\ in Angstroms
Fig. 7.4. Curves showing tlie energy required at various wavelengtlis to produce a fixed^emulsion density; curve A, density 1.7; curve B, density 1.0; curve
C, density 6.3.
energy required to produce a given density value, we show inverse
response curves for an emulsion a t t h r e e density levels. T h e concept
of sensitivity is useful to indicate the varying response of a n emulsion
to light of different wavelengths.
T h e distance along the log E axis subtended by the straight-line
portion of t h e characteristic curve measures the " l a t i t u d e " of t h e
plate. Strictly speaking, t h e latitude is defined as the ratio of t h e
exposure a t t h e upper end of the straight portion of the characteristic
curve t o t h e exposure a t the lower end. Usually, the thicker a n
emulsion the greater its latitude, a n d emulsions of low contrast h a v e
148
PHOTOGRAPHY OF THE SPECTRUM
L^i
greater latitude than those of high contrast. In terms of spectrum
photography tliis statement means that selection of an emulsion of *
too high contrast should be avoided if both weak and strong spectrum lines are to be rendered in densities that are to be directly
representative of the log of their intensities.
Latitude depends not merely on the photographic material itself
but also on the degree of development used and to some extent on
subsequent handling of the plate. Together with speed and contrast,
it varies greatly with the wavelength of the light to which the plate
is exposed.
7.5. Resolving Power and Graininess. A property of the photographic emulsion that is most important to the spectroscopist is
resolving power. This property, which measures the ability of a
plate to record separately lines that lie close together, depends to a
considerable extent on granularity, which also sets a limit to the
useful magnification to which a spectrum line can be subjected. If a
spectrograph produces two very fine spectrum lines whose intensity
maxima are only 0.02 mm apart, this high resolution can be realized
photographically only by use of a plate that will resolve 50 lines
per millimeter.
The resolving power of an emulsion, ordinarily measured by photographing a series of line gratings, is expressed in terms of the number
per millimeter of black and white lines of equal width which can be
resolved under suitable magnification by visual observation of the
emulsion.
Table 7.1 gives the resolving powers for white light of a number of
typical emulsions manufactured by the Eastman Kodak Company.
The resolving power ^ of a photographic material is controlled
largely by the contrast of the emulsion and its turbidity. Turbidity,
in turn, depends on the light absorption of the emulsion and its
scattering power. Resolving power depends greatly on the density
that the image attains and is greatest for intermediate densities^
Both granularity and resolving power can be improved by using
fine-grain developers.
'
,
,
When light in the ultraviolet is photographed, the resolving power
of an emulsion is found to be greater than at l&nger Wavelengths.
' This effect results from the low penetration of the emulsion by short
waves, which are strongly absorbed by the silver halide, while wavelengths shorter than 2500 A. are also absorbed by the gelatin.
The Eastman Kodak Company has developed emulsions in which
graininess has been almost eliminated, but these emulsions are very
1
§7.5]
PHOTOGRAPHY OF THE SPECTRUM
149
sldw. W h e n plenty of light is available a n d a suitable optical system
is used t h e y can be made to resolve 500 lines per millimeter or more.
T h e individual grains can be seen only under a microscope of high
resolving power, if a t all.
I n selecting a plate or film on which t o photograph t h e
spectrum, one ordinarily decides first whether a fast emulsion is
needed, on t h e basis of the light intensity available and t h e permissible time of exposure. If the problem is one of detecting very faint
spectrum lines, high sensitivity a t low light intensities is needed,
which suggests use of a fast plate. T o reproduce both weak and
strong spectrum lines on the same spectrogram with correct indication
TABLE 7.1
RESOLVING POWER TO WHITE LIGHT OF TYPICAL EMULSIONS
IN LINES PER MILLIMETER*
Material
Kodak 50
Kodak 40
Kodak 33
Kodak Panatomic-X
Kodak Process
TypeI-0
Type 103-E
Type Il-C
Type III-C
Type IV-C
Type V-C
Type 548
Type 649
P^
70
65
80
90
100
60
60
75
95
120
160
Approx. 500
Approx. 1000
^
* Copied by permission of the Eastman Kodak Co. from
Photographic Plates for Scientific and Technical Use, Rochester,
N. Y., 1948. The values given apply to an optical image
contrast of 20:1, and to the density values for which resolving
power is a maximum. In the case of the last two entries resolution is usually limited by the optical system used rather than
by the properties of the emulsion.'
of their relative intensities, high latitude and medium contrast are
needed. When a clean, clear background, free from fog, with crisp,
sharp spectrum lines is desired, a plate of high contrast is used.
Characteristic curves of emulsions suited for these three purposes are
given in Fig. 7,5.
Slow contrasty plates show high resolving power. Such plates are
of greatest value for use with spectrographs having low dispersion
combined with high resolving power, as is often the case in prism
i n s t r u m e n t s of short focus. T h e resolving power of a photographic
150
PHOTOGRAPHY OF THE SPECTRUM
[§7.6,
plate should be approximately m a t c h e d with t h a t of the spectrograph,
in which it is to be used.
7.6. Types of Plates and Films. Most manufacturers of photographic materials produce plates a n d films covering a broad range of
LogiQ Exposure
—
Fig. 7.5. Characteristic curves of emulsions for different spectroscopic purposes. Curve 1 is that of a fast emulsion (Eastman 103-O); curve 2, an emulsion
of medium speed and high contrast (Eastman III-O); curve 3, high contrast,
fine-grain emulsion (Eastman IV-0).
f
Speed, contrast, and resolving power. T h u s a t t h e spectroscopic
level of intensity (taken as 0.1 meter candles of sunlight) t h e E a s t m a n
K o d a k Company t y p e I - O plate has a speed of 23, and their V - 0
p l a t e a speed of 0.64. T h e respective contrast values are 1.3 and 4.5,
Sensitivity
2000
3000
4000
5000
60^0'
X in A
Fig. 7.6. The qualitative variation of the sensitivity of the photographic
emulsion through the spectrum. Thfe exact shape of this curve depends on,the
density level at which the sensitivity is measured.
§7.7]
151
PHOTOGRAPHY OF THE SPECTRUM
and the respective resolving powers are 60 and 160 lines per millimeter.
7.7. Variation of Emulsion Characteristics with Wavelength.
The unsensitized photographic emulsion shows wide variations in
sensitivity throughout the spectrum. Sensitivity becomes negligible
at long wavelengths ranging from 4400 to 5200 A, depending on the
composition of the silver halide used. Sensitivity gradually increases
toward shorter wavelengths, reaching a maximum in the violet or
near ultraviolet between 4100 and 3900 A, and then slowly falling off,
as shown in Fig. 7.6, to negligible values near 2000 A. The broad
region of insensitivity between 2000 A and the X-ray region, where
the ordinary emulsion again becornes sensitive, results from the
absorption of light by the gelatin of the emulsion.
3.00 r
E.0O
+
1.00
~3
2000
3000
4000
5000
X in A
Fig. 7.7. The variation of contrast of several spectrographic emulsion types
as a function of wavelength. Curve 1, typical slow emulsion; 2, medium-speed
emulsion; 3, fast emulsion.
The variation of sensitivity with wavelength is usually associated
even more closely with variation in contrast than with variation in
speed. Figure 7.7 shows how the contrasts of a number of different
photographic emulsions vary with wavelength, and Fig. 7.8 shows
the variation in speed of the same emulsions over the same wavelength region. Sensitivity at longer wavelengths can be greatly
increased by adding special dyes to the emulsion as explained in
§ 7.15. At shorter wavelengths the Schumann type of emulsion can
be used, or fluorescent substances can be coated on an ordinary plate
or film (§ 19.7).
C§7.8
PHOTOGRAPHY OF THE SPECTRUM
152
7.8. Storage and Handling of Photographic Materials. A photographic latent image can be produced, not merely by exposing the
emulsion to light, X-rays, or high-speed subatomic particles, but also
by applying pressure, heat, or certain chemical fumes such as ammonia. The emulsion deteriorates with age, and unexposed plates
that have been stored for a long time are apt to appear dark on development, especially around the edges. The useful life of an unexposed plate can be extended by storing it on edge in a cool, dry place.
If plates are stored flat, the weight of other plates may eventually
produce a latent image. Plates should never be stored where they
are exposed to heat, chemicals, dampness, or burning gases. If they
are stored in an electrical refrigerator, some provision should be
made for prevention of spoilage by wetting should the power fail and
the frost in the refrigerator melt.
i
Speed
2000
3000
4000
5000
X in A
Fig. 7.8. The variation of speed with wavelength for the emulsions of Fig. 7.7.
When possible, plates and films should be obtained in sizes that fit
the spectrographic cassettes used, to avoid the/necessity of cutting in
the darkroom. Standard sizes obtainable for spectrographic work are
IJ X 10 in., 2 X 7 in., 2 X 10 in., 2 X 20 in., 2 | X 10 in., 4 X 10 in.,
3i X 4i in., 3J X 18 in., 4 X 5 in., 5 X 7 in., 8 X 10 in., 50 X 250
mm, and 65 X 180 mm. Special emulsions designed for spectroscopic use are ordinarily coated on glass, but can be obtained coated
on film base on special order. ^^
The flatness of the glass on which photographic plates are cOated
§7.9]
PHOTOGRAPHY OF THE SPECTRUM
153
is of importance, because the sensitivity of the emulsion can be expected to vary over the surface if its thickness varies. Glass plates
are coated by spreading the liquid emulsion over them, chilling to set
the gelatin, and drying. Unless perfectly flat plates are used, the
emulsion will vary in thickness.
For accurate measurement of wavelengths in spectrographs in which
plates must be bent to high curvature, special glass can be selected
which is less than 1 mm in thickness, and plates can be obtained up
to 22 in. long coated on such glass. Even thinner glass plates can be
obtained, but they are very fragile.
Photographic plates are usually packed in pairs, with the emulsion
sides face to face. Plates keep better in this way than when the
emulsion side is exposed, but it is necessary to prevent adjacent
surfaces of emulsion from rubbing together. For this reason each
pair of plates is separated by thin sheets of cardboard at the edges
and is then wrapped in black paper. The emulsion side of the plate
can be determined in the darkroom by observing the diffuse reflection
in the coated side of a safe light, or in total darkness by applying the
edge of the plate to the lips or tongue or biting it gently.
When it is necessary to cut a plate into smaller sections, this can
be done in the darkroom by providing a flat board with a rule held
in the proper position by pins, so that drawing a glass cutter along
the uncoated side of the plate will produce a sharp scratch. The
plates should be laid face down on a piece of paper placed on the
board, and only a single cut should be made, the plate then being
broken by pressure against the edge of the board. A sharp scratch
rather than a deep one is what is wanted, since the object is not to cut
the glass but merely to start the fracture by a high concentration of
stresses. For this purpose a sharp diamond point serves best.
Films may be cut in thin bundles by using a large paper cutter
provided with a template so that suitable widths can be selected in
the dark. A number of films can safely be wrapped together without
separation if not subjected to abrasion, heat, or humidity.
7.9. The Photographic Darkroom. The darkroom of a spectroscopy laboratory merits considerable attention. I t should be constructed so that no external light will be admitted unless desired. If
possible, it should be arranged so that it can be entered by means of
a double door or lightproof maze. The former is simpler than the
latter, and if provided with an alarm bell that rings when one of the
doors is opeii, safer. The old-fashioned darkroom, with its walls
154
PHOTOGRAPHY OF THE SPECTRUM
[§7.9
painted a depressing black, is out of date. If safelight illumination ,
is provided, the walls may be painted light green, cream, or any other
pleasing color. When the room lights are on, ample light for good
visibility should be available, and when they are off no light should
enter the room from outside to be reflected from the walls.
Ordinary plates or films can be developed under a red light, though
this should be a safelight, since common red incandescent lamp globes
often transmit light in the blue. Panchromatic and other colorsensitive plates must be developed either with a green safelight or
in total darkness. A good spectroscopist usually prides himself on
developing all spectrum plates in total darkness as a matter of habit.
An Eastman No. 1 red safelight can be used with ordinary plates
such as Eastman 33 or type O, if it is desired to observe development,
though it is more satisfactory to develop for a predetermined time
in developer of a given strength and temperature than to depend on
visual observation to determine when sufficient development has
taken place. Various filters can be obtained for insertion in the
safelights, and green filters can be used with panchromatic erriulsions,
though total darkness is to be preferred. In developing bromide and
other papers, a much greater amount of red light can safely be perniitted than with negative materials, and under these circumstances
it is desirable to be able to watch development, since it proceeds so
rapidly. The sSfelight can be tested by exposing a section of the
emulsion to be developed under the light for twice as long as it would
be exposed in development, and comparing it on development with
an unexposed portion.
By adopting a definite routine procedure it is quite easy to carry
out the standard; series of loading and developing operations in the
dark. A typical routine for loading a spectrograph plateholder is to '
place it face down on a dry-table, with a closed box of plates on the
left. With the back of the plateholder open, turn off the lights and,
uncover the box of plates, removing two plates in their paper wrap^
ping. Take one of these plates, determine its emulsion side, ancl
place it in the plateholder with this side down. Immediately wrap
the remaining plate carefully in its black paper, reinsert it in the box,
and close the box. Then close the plateholder, turn on the lights, and
load the spectrograph. This procedure eliminates the danger of
turning on the darkroom lights to find an open box of plates awaiting
disposal.
W^hen possible, it is desirable to provide a "dry" dark room as
§7.10]
PHOTOGRAPHY OF THE SPECTRUM
155
well as a "wet" darkroom. Plates should be stored in the dry darkroom, and printing and enlarging equipment can also be installed
there, thus removing it from the corrosive action of chemicals that
may be spilled in the wet darkroom during development or fixing.
An enlarger, properly modified to take long spectrograms, is very
convenient for making reproductions of spectra. Large spectrum
prints are usually made on glossy bromide paper, and contact prints
on ordinary glossy positive paper. When it is very important to
avoid any uneven shrinkage of the print during copying, special types
of photographic paper can be obtained, such as Eastman Aeromapping
paper or Agfa Mapping Special, in which shrinkage has been greatly
reduced.
7.10. Development and Processing. After exposure, the spectrum plates or films are developed, fixed, washed, and dried. Tray
development is ordinarily used for spectrograms. The tray should be
an inch or two longer than the longest plate to be developed in it, and
not much wider than a single plate. I t may be necessary to build
special trays to fit plates longer than 10 in. Fresh developer should
be used for each important batch of plates, the plates being placed in
the bottom of the tray, emulsion side up, and the developer being
poured gently over them, the tray then being rocked from end to end
to remove air bubbles on the surfaces of the plates. The developer
should have previously been brought to the correct temperature, as
measured with a thermometer and a time clock should be set to give
an alarm at the end of the selected development period. Provision
should be made for cooling the water to be mixed with the developer
in warm weather, since the temperature of the developer must be
carefully controlled, and the washing water should not be much
warmer than 65° F to avoid peeling or reticulation of the emulsion.
Spectrographic films are not quite so simple to handle in development and processing as are plates, and manufacturers of spectrographs
that require use of films ordinarily furnish holding devices or special
developing tanks with these as part of the processing equipment.
When quantitative spectrochemical analysis and other work involving photographic photometry is carried on, it is customary to
brush the surface of the emulsion gently back and forth with a
camel's-hair brush to bring fresh developer constantly in contact with
the emulsion and thus reduce the Eberhard effect (§ 7.13).
When the development period is concluded, the plate is removed
from the developer, is held by its edges and dipped once or twice in
156
PHOTOGRAPHY OF THE SPECTRUM
[§7.10
the rinsing water (and if desired in a short-stop solution), and then is .
deposited emulsion side up in the fixing bath. After it has been in
the hypo for a minute or two, the lights can be turned on, but the plate
should remain in the fixing bath for at least 10 minutes after the last
trace of unexposed emulsion has cleared. With fresh hypo this should
require immersion of not more than 15 min. Hypo that has become
foamy and does not clear up the plate within 20 min should be replaced, but until this foaming occurs the hypo can be used for many
batches of plates.
After a plate has been fixed it should be washed in running water.
This water should not, however, be allowed to fall on it directly.
Fresh water should well up around the surface of the plate in a tray
for at least half an hour; then the plate should be "stripped"" and
placed in a rack for drying.
Stripping consists of holding the plate under the running tap and
carefully rubbing it on both sides with the palm of the hand, or with
a tuft of moistened cotton, to remove the bloom, a filmy layer deposited
on it from the tap water. Remaining droplets of water should then
be removed by careful shaking, and the plate should be rinsed on
both sides with distilled water and stood in a cool place to dry slowly.
When extreme speed is necessary, plates may be measured in the
densitometer while wet, but this is usually inconvenient.
When speed is of the greatest importai^ce, the plate may be dried
in front of a fan or over a heater, but this should be done carefully.
A preliminary rinsing in alcohol to remove some of the surface water
and speed up drying is undesirable. When alcohol is used, the plate
should be immersed in it for from 3 to 5 min. The plate will then dry
in 15 sec in a rapid current of air. However, for most spectrograms
it is desirable to use slow drying in a good circulation of dry dust-free '
air at 75 to 90°F to get uniformity of results. After drying in racks,
in the case of plates, or when hung from clips on a stretched wire, inj
the case of films, the spectrograms are ready for marking and storing
in individual envelopes.
/
I t is especially important that the darkroom be so arranged that
hypo will not spatter into the developer tank or get onto the table on
which plates are cut and plateholders are loaded, since hypo dust will
produce black marks on the emulsion. H^po in the developer is
much more serious than developer in the hypo, though the plates
should be carefully rinsed between development and fixing. yiThe
developing and fixing trays should be so positioned that they can
§7.11]
PHOTOGRAPHY OF THE SPECTRUM
157
readily be located in the dark, to avoid the mistake sometimes m^ade
of depositing a spectrogram first in the hypo bath. Since the developing and fixing processes should be kept rigidly separated, it is common
procedure to provide a sink with a developing tray on one side and a
fixing tray-on the other, with washing and rinsing trays between.
The shelves and floor of the darkroom, as well as its walls, should
be kept scrupulously clean, especially from spilled hypo, and the
operator should be able to rinse his hands quickly and dry them on a
towel before he touches anything after his hands have been in the
hypo bath.
7.11. Developers. A standard developer consists of a solution of
chemical agents, the most important of which is designed to reduce
to metallic silver the silver halide in the latent image. Common
developers are amidol, metol or elon, pyrogallic acid, rodinol, and
hydroquinone. In addition, the developer solution contains an
accelerator such as sodium carbonate or sodium hydroxide, a preservative such as sodium sulfite or bisulfite, and a restrainer, usually
potassium bromide. The purpose of this restrainer is to minimize
the development of grains which have not been exposed to light.
Probably the most satisfactory type of developer for routine spectroscopic purposes is one that can be kept mixed in large quantities,
so that all the user need do is to dilute it with water just, before use.
Dilution is desirable, especially in warm weather, so that the temperature of the diluting water can be chosen to give the proper final
tempe'rature of the developer after pouring into the tray. A typical
developer of this variety is given in Table 7.2. This can be mixed
in large quantities but should be kept in stoppered bottles, since
contact with the air will oxidize and darken it. In large laboratories
developer is often mixed in 10-gal lots, stored in 5-gal bottles, and
when one of these is opened gallon jugs are filled from it. These, in
turn, are used as needed to fill 8-oz bottles, which can be mixed on
use with 8-oz bottles of water of the proper temperature. When
three or four spectrograms are to be developed at the same time, a
single batch of developer will suffice, but developer should not be
re-used after it has stood in the tray for a few hours.
When only an occasional plate is to be developed, the standard
commercial M-Q tubes, a metol-quinone developer, will be found
useful.
Most commercial boxes of plates or films contain manufacturers'
formulas for developers recommended for use with the plates. Many
158
PHOTOGRAPHY OF THE SPECTRUM
[§7.12
formulas require that the various solutions be made up in two or
three separate parts, which are mixed just prior to use.
For some spectrographic purposes, particularly in high-speed
quantitative analysis, rapid development is needed. Under these
circumstances it may be worth while to use a two-bath developer in
which the degree of development can be more carefully controlled.
Neblette' gives such a developer, said to produce satisfactory results
in 10 sec, which consists of a 5 per cent solution of hydroquinone with
2 per cent sodium sulfite, followed by a 30 per cent solution of potassium hydroxide. The plates are immersed for 8 sec in the first bath
and 2 sec in the second bath at a temperature of 80°F. Neblette also
gives a high-speed developer which acts within 30 sec to 1 min.* ,
TABLE 7.2
DEVELOPER WITH GOOD KEEPING PROPERTIES*
Distilled water
Elon (or metol)
Sodium sulphite
Hydroquinone
Potassium bromide
Sodium carbonate
10
14
212
56
150
300
liters
grams
grams
grams
grams
grams
* This developer can be mixed in large quantities and kept
in stoppered bottles for a number of months. When used, it
should be diluted with two parts of water to one of developer.
Develop in trays for about 5 min at 18° C.
High-speed development or processing of an5^ kind is undesirable
unless speed is more important than uniformity, since results will be
more precise and more uniform if processing is allowed to proceed
at a normal rate.
,
7.12. Common Defects in Spectrum Photographs. A beginner
who is sent into a darkroom to develop a spectrogram is likely to
emerge with a sorry-looking product. However, after a few hours of
experience and with very little conscious change in procedures, he
will produce acceptable spectrograms that are clear, unmarred, and.
free from fog and dirt. That this change takes place so rapidly;
emphasizes the great effect of small variations in procedure. Com,mon experiences of the beginner range from putting his carefully
taken exposures into the hypo instead of the developer, loading the
plateholder backward, or leaving the box of plates open after removing
' C. ]}. Neblette, General Reference 7.1, page 540.
* C. B. Neblette, General Reference 7.1, page 540.
§7.12]
PHOTOGRAPHY OF THE SPECTRUM
159
one for loading, to manipulations that lead to scratching, frilling, and
reticulation of the emulsion.
Specks and dark streaks across the plate are likely to have been
caused by particles of hypo coming in contact with the emulsion
either before or during development. Fingerprints, scratches, and
abrasions can be reduced if a plate or film is held by its edges. Blisters
are usually caused by the separation of the emulsion from the glass
backing under the pressure of water falling directly against the
emulsion. Reticulation, or drying of the emulsion in irregular ridges,
results when the wash water is too warm or the hardener affects only
the surface of the emulsion; thus frilling and reticulation result when
there is insufficient hardener in the hypo bath. Transparent spots
are usually produced by air bubbles or grease on the surface of the
emulsion that keep the developer from coming in contact with it.
Insufficient development gives a spectrogram that lacks contrast,
even the densest lines having a gray look rather than appearing black.
A similar effect can be produced by underexposure, and experience is
needed to differentiate between the two causes.
Fogging, one of the most common defects in spectrum plates, should
be particularly watched for in spectrochemical quantitative analysis
and photographic photometry. Fogging is produced by a deposit of
developable silver grains in addition to those that are included in
the latent image. Fogging arises from various causes, and some
experience is needed to distinguish among them. Chemical fog is
ordinarily quite uniform over the surface of the plate. It may be
caused by improper development or by improperly mixed or spoiled
developer. The spoilage is usually from contact with metal. However, general fog all over the plate may be caused by exposure to
light, either through use of too strong a safelight during development
or through a leak in the plate box, spectrograph, or darkroom. Plates
that have been exposed to fairly high temperatures become fogged
very readily, and may even fog spontaneously. Fog caused by light
leaks can usually be detected if it occurs during exposure or during
the time the plate was in the plateholder, because it appears strongest
near one edge of the plate or in one corner. Fast plates are in general
more susceptible to fog than slow plates, and a good contrasty plate
makes a much clearer and finer-appearing spectrogram.
In spectroscopic work the finished picture is usually a negative, and
the most important consideration is not the accurate rendition of tone
values in a positive print to be made from it, but rather the deter-
160
PHOTOGRAPHY OF THE SPECTRUM
[§7-13
mination of the positions of spectrum lines, and of their densities. ,
Secondary considerations are high resolving power to separate close
lines, freedom from fog to increase the accuracy of photometry,
sensitivity to weak light so as to bring up faint spectrum lines, and
large latitude so as to make possible the recording of both weak and
strong lines on the same plate. For this last purpose it is desirable
that the emulsion reach a high density before the contrast falls to a
value too low to be useful.
7.13. The Eberhard Effect. During the development process,
soluble bromide is formed along with the reduction to silver bf the
silver bromide in the emulsion, so that the concentration of bromide
in the developer increases with use. In parts of the emulsion where a
particularly dense latent image is being developed, there may be a
>y^-=*<r\ ,
Density
jj
I
• A
B
V
Plate Distance inmnns-^
^
' '•
Fig. 7.9. The Eberhard effect in spectra. Curve A, measured density in the
contour of a strong line; curve B, true density in the absence of the Eberhard
effect.
high concentration of bromide, which weakens the developer and
accelerates the restraining process, causing the so-called Eberhard,
effect. This effect is of particular importance in quantitative spectrochemical analysis or other processes involving accurate photor
graphic photometry.
The Eberhard effect reduces the density in the center of dettise
spectrum lines and decreases the background density in the neighborhood of strong lines, as illustrated in Fig./7.9. The effect can be
avoided to a considerable extent by constant brushing^ of the emulsion
surface during development. Motion of the developer by rocking the
plate or streaming developer across the emulsion surface ordinarily
will not reduce the effect, since it is produced in an extremely thin
5 W. Clark, Phot. Jour., 65, 76 (1925).
§7.-15]
PHOTOGRAPHY OF THE SPECTRUM
161
layer of developer that clings to the surface of the emulsion. Some
developers are worse offenders than others in this regard.
7.14. Halation and Spreading. Halation, an effect commonly
noticed with spectrum lines that have been greatly overexposed)
arises from the penetration of light through the emulsion to the glass
backing of the plate from which it is reflected and returned back
through the emulsion, causing a ring to appear around a dark spot or
a rectangle to appear around a line. Halation effects are usually
more marked in plates than in films, since in a film the small thickness
of the celluloid causes the reflected light to be thrown back directly
on the main image. Formerly some plates were painted black on the
glass side to reduce halation. Many plates are now backed, the
backing being designed to absorb the radiation to which the plate is
sensitive, and to be bleached in ordinary processing. Colored dyes
on the side of the emulsion next to the glass or film support are also
used as antihalation backings. When necessary, halation can be
reduced by using a complex developer formula and relatively short
development, which confines the image largely to the upper layers
of the emulsion.
Spreading is another photographic phenomenon that is important
in spectroscopy. The image on a photographic plate will not be
confined entirely to the area on which the light strikes, because of
halation and irradiation from light scattered in the emulsion so that it
strikes grains which otherwise would not be exposed. Spreading and
halation both reduce the resolving power of an emulsion and may be
of great importance when intense close spectrum lines are studied,
since these will be greatly broadened by these effects. To obtain the
highest resolving power with a plate it is therefore useful to try
surface development. In the ultraviolet region, surface exposure
accomplishes the same effect, on account of the decreased penetration
of the emulsion by the shorter waves.
7.15. Photography of Various Regions of the Spectrum. Unsensitized photographic plates or films can be used satisfactorily only
in the range 2200-5000 A. At longer wavelengths it is necessary to
use optically sensitized emulsions. A great number of special dyes
have been made for this purpose, providing sensitization as far as
13,000 A. By proper choice of these dyes sensitivities ranging from
50-100 per cent of the sensitivity in the blue may be obtained anywhere in the visible spectrum. Sensitizations for the infrared are
weaker, the effective sensitivity generally becoming progressively less
162
PHOTOGRAPHY OF THE SPECTRUM
C§7il6
with increasing wavelength. The plates sensitized for the infrared ^
are generally improved by hypersensitizing by bathing with water or
ammonia shortly before use. A discussion of the chemistry and
application of these dyes is given by Mees.^
Contrast, and in some cases sensitivity, of ordinary emulsions can
be improved in the region 2500 to 2000 A by coating the emulsion
with a fluorescent material, such as a mineral oil or one of 5/number
of substances developed for the purpose.' Ordinary emulsions can
be sensitized to wavelengths as short as 200 A by similar methods.
Ultraviolet sensitizing solution can be obtained from the Eastman
Kodak Company, who also furnish plates especially sensitized for the
short-wave ultraviolet. Ilford, Ltd., of London manufactures " Q "
plates that also have excellent characteristics in this region. Schumann emulsions, almost free from gelatin, can also be used. The
problem of ultra-short-wave photography is discussed further in
Chapter 1,9.
7.16. Selection of Spectrally Sensitive Emulsions. The sensitivity of an emulsion in various spectral regions can be qualitatively
determined by so-called wedge spectrograms, which are made by
placing an absorbing optical wedge in front of the slit of a stigmatic
sp>ectrograph, thus subjecting a plate to a regular variation in illumination from a continuous source, from top to bottom. Typical
wedge spectrograms, obtained for emulsions manufactured by the
Eastman Kodak Company, are shown in Fig. 7.10. Approximate
wavelengths in angstroms divided by 100 are marked on the spectrograms horizontally, and the height of the light portion at any wavelength gives a qualitative indication of the relative sensitivity of the
emulsion at that wavelength. Since such spectrograms are almost
always made with glass apparatus, and with incandescent lamps as '
light sources, wavelengths shorter than about 3800 A are not utilized,
and the indication of relative sensitivity at wavelengths less than
4500 A is poor and should be taken as only approximate. The emulsion retains much of its sensitivity from 4200 to 2300 A with alm|st
all dyes.
" C. E. K. Mees, General Reference 7.2, page 968 et seq.
' J. Duclaux and P. Jeantet, Jour, de Phys. et le Radium, 2, 156 (1921).
G. R. Harrison, Jour. Opt. Soc. Am., 11, 113 (1925).
'
G. R. Harrison and P. A. Leighton, Jour. Opt. Soc. Ant., 20, 313 (1930); Phys. Rev.,
36, 779 (1930).General Reference 7.5, 5th ed., page 21.
§7.16]
PHOTOGRAPHY OF THE SPECTRUM
163
The Eastman Kodak Company manufactures plates sensitive to
the various regions of the spectrum, as shown in Table 7.3. From
such a table it is possible to choose a suitable emulsion for photographing any desired spectral region at high dispersion. For use in
low-dispersion instruments the L type of sensitization covers the
entire range from 9000 to 2000 A, but it has rather low sensitivity.
Each of the sensitizations listed is available in various emulsion
types, having high, medium, and low sensitivity and medium, high,
and very high contrast. The selection of suitable emulsions for a
given purpose can be greatly facilitated by consulting the appropriate
(a)
(b)
(c)
Fig. 7.10. Wedge spectrograms for several spectrographic emulsions, (a)
Eastman green-sensitized emulsion G. (b) Eastman Panchromatic emulsion B.
(c) Eastman extreme-red-sensitized emulsion N. (Courtesy Eastman Kodak
Company, Rochester, New York.)
Eastman Kodak Company publication (General Reference 7.5) and
similar publications of other manufacturers.
Most of the dyes now used for sensitization up to 10,000 A are quite
stable, but all plates sensitized for the infrared should be kept in a
refrigerator as much of the time as is possible. Plates that have been
stored thus should be removed some hours before use so that they will
reach room temperature before exposure; in this way, surface condensation of moisture will be avoided.
[§7.16
PHOTOGRAPHY OF THE SPECTRUM
164
It is sometimes desirable to hypersensitize infrared-sensitive plafts,
with ammonia, particularly those being used for the longer wavelengths. Sensitivity can be increased by bathing the plates for one
minute in a 4 per cent solution of 28 per cent ammonia at a temperature of 55°F or less. The plates should then be dried as rapidly as
possible in air which is dust-free, and used almost immediately.
Hypersensitized plates can be kept for several weeks in a refrigerator,
but hypersensitization increases the likelihood of fogging and can be
TABLE 7.3
APPROXIMATE RANGES OF SENSITIVITY OF EASTMAN SPECTROSCOPIC
PLATES
Plate type
Short X*
0
J
H
G
T
D
B
C
E
F
U
N
K
R
L
P
M
Z
2,000 A
4,600
4,600
4,600
5,500
4,600
4,600
4,600
5,600
4,500
6,600
6,600
7,300
7,400
2,000
8,200
8,600
10,000
Long X*
.5,200 A
5,300
5,000
6,000
6,200
6,200
6,600
6,800
6,900
6,900
7,600
8,600
7,900
8,500
8,600
9,200
10,000
12,000
* All emulsions show some sensitivity in the sensitivity range 2000-50QO A.
carried further if done just before the plates are used.' It is very
important that the temperature of the sensitizing bath be controlled
accurately.
It is convenient to have available a light-tight box in which sensitized plates can be dried with a blast of air blown from a fan.
The Eastman Kodak Coinpany recommends that the developer
whose components are listed in Table 7.4 be used with Eastman spectroscopic plates and states that it is of particular value with infrared
sensitive plates. This formula is available in prepared form, and
can be used without dilution.
§7.16] .
PHOTOGRAPHY OF THE SPECTRUM
165
Quantitative control and use of the photographic emulsion are discussed in Chapter 13.
TABLE 7.4
KODAK DEVELOPER D - 1 9
(For high contrast on spectroscopic plates)
Stock Solution*
Water, about 125° F (50° C)
Elon
Sodium sulfite, desiccated
Hydroquinone
Sodium carbonate, desiccated
Potassium bromide
Cold water to make
500 cc
2.2 grams
96.0 grams
8.8 grams
56.0 grams
5.0 grams
1 liter
* Dissolve the chemicals in the order given. Use without
dilution. Average time of development is about 3 min at
68° F (20° C).
GENERAL
7.1.
7.2.
7.3.
7.4.
7.5.
REFERENCES
C. B . N e b l e t t e , Photography:
Principles and Practice, 4 t h ed. N e w
Y o r k : D . Van N o s t r a n d C o m p a n y , Inc., 1942.
C. E . K. Mees, The Theory of the Photographic Process. N e w Y o r k :
T h e Macmillan C o m p a n y , 1942.
K . H e n n e y and B . Dudley (Eds.), Handbook of Photography. N e w Y o r k :
McGraw-Hill Book C o m p a n y , I n c . , 1939.
A. C. H a r d y and F . Perrin, The Principles of Optics. N e w Y o r k :
McGraw-Hill Book C o m p a n y , I n c . , 1932.
Photographic Plates for Use in Spectroscopy and Astronomy,
various
editions. Rochester, N . Y . : E a s t m a n K o d a k Co., 1933-1948.
CHAPTER
8
Light Sources for Spectroscopy
L I G H T SOURCES, OR SOURCES O F RADIANT E N E R G Y , m a y
be
classified
according t o (a) the m e t h o d used for exciting radiation, (b) t h e t y p e
of spectrum emitted, or (c) the spectral region to which t h e source is
best adapted (infrared, visible, ultraviolet, or extreme ultraviolet).
I n terms of method of excitation, there are four principal categories
of sources: (1) thermal radiators, (2) arc sources, (3) discharge tubes,
and (4) spark sources (see General Reference 8.3). T h e r m a l r a d i a t o r s
emit radiation as a result of heating of the radiating surface, as when a
current of electricity h e a t s a m e t a l filament t o incandescence. Arc
sources emit radiation as a result of the maintenance of a comparatively low-voltage ionic electrical discharge between suitable
electrodes, under conditions in which the material of t h e electrodes is
evaporated into the arc stream a n d provides a large proportion of t h e
conducting a n d radiation-emitting ions. Discharge-tube sources also
emit radiation as the result of t h e maintenance of an ionic discharge,
b u t the source of t h e ions is a gas a t low pressure in an enclosed
container, and little, if any, electrode material passes into t h e ion
stream. Spark sources emit radiation .as-a result of comparatively
high-voltage disruptive discharges between suitable electrodes.
T h e distinction between arc, spark, and discharge excitatien is
often only approximate. T h u s , m a n y carbon arcs depend for their
emission primarily on t h e heating of t h e tips of t h e carbons t o incandescence as a result of b o m b a r d m e n t by the ion stream. Again, when;
spark sources are operated a t progressively higher current densities/
t h e y begin t o p a r t a k e more and more of the characteristics of arcs.
I n addition to the four principal categories mentioned, t h e r e are
certain other methods of excitation, such as b o m b a r d m e n t b y
cathode rays or excitation of fluorescence or resonance radiation,
t h a t are sometimes useful in spectroscopy.
W i t h respect t o t h e t y p e of spectrum emitted, sources are con166
§8.1]
LIGHT SOURCES FOR SPECTROSCOPY
167
veniently classified into three categories according to whether they
yield continuous, line, or band spectra (Fig. 1.5). This distinction is,
again, somewhat arbitrary. Continuous spectra often have lines
superimposed upon them, and line or band spectra often exhibit an
appreciable continuous background. This result is to be expected,
sinf e, in general, continuous spectra arise from thermal radiation or
nonquantized atomic or molecular-energy transitions (dissociation
spectra, for example), line spectra arise from atomic-energy transitions, whereas band spectra arise from molecular-energy transitions;
and all these emission mechanisms coexist to some extent in almost
every source (see Chapters 10 and 11).
Classification of sources according to the spectral region to which
they are best adapted is likewise arbitrary to some extent, but is
useful in the choice of sources for special applications. Sources for
the infrared are discussed in Chapter 17 and for the vacuum ultraviolet in Chapter 19.
GENERAL CHARACTERISTICS OF SOURCES
8.1. Spectral Energy Distribution. In choosing a source for a
particular application, it is desirable to select one that emits radiant
energy predominently in the spectral region to be explored. This
selection is frequently difficult to achieve in practice For example,
incandescent lamps are excellent sources for many applications in the
visible region, yet they radiate more total energy in the infrared than
in the visible. The greater the atomic or molecular energy transitions involved in the excitation, the shorter the wavelength region in
which the radiant energy may be expected to predominate (see
Chapter 10). Thus thermal-emission sources are mainly useful for
the infrared and visible regions, arcs for the visible and near ultraviolet, and discharge tubes and sparks for the visible, ultraviolet, and
extreme ultraviolet regions.
Sources emitting continuous spectra are particularly useful in
making absorption measurements (Chapters 14 and 17). Those
emitting line spectra are useful in studying atomic structure and in
qualitative and quantitative emission spectrum analysis (Chapters 15
and 16), and in applications in which it is desired to isolate approximately monochromatic radiation.
The general spectral-emission characteristics of various sources are
summarized in Table 8.1.
LIGHT SOURCES FOR SPECTROSCOPY
168
a
.2
w^
Qj
^
X
.2"3Ti
-0
1
Isl
*S
=« J
ra
.*
Q,r3
„
8 N^g •&.
„ 2 !>
^
§
^'
»
M "^3
IS
.2
1%
t
t> JJ -S
•s 3 6
!3
C3
B
P
O
"a
.n.
c
£
in
«3
o
C
w
P.
-Si
> 5 «*«
<;
0
0
0
4J
3
CO
4>
0
^
*-3
-^
• S a g
&
o
C=35
s
s .-^i 0 0 g T
§ ^ 5^ -^ ^ c
H <;
3
<
CO ^
.M
; -
0
ti
CO S
CO
-0
« -S K S
«3 w 0 re
O'
2
-2 C JJ 3 3
3 ~ 3 g «j
n 0 u, p 2 u
B
3
u
S
°
f
p
S
jH S
. iCO
d
° j<«
M - •^«
0
5^ "
S
;0
3
CO
-M
.a
S
^
-0
^
to
a
£3
o.
M
. 2 E3 .2
tc T j
<
*^
Js
^ 1
^1
*-f->
o
4--
go...
en
3
+j
Jj2
&
o
§•
0
1
0
o
r3
in
-a
3
0
to
3
o
3
s
o
O
3
3
O
3
.3
.3
3
3
0
o
O
0
Si
0)
0
to '-'
3
0 r3
.3
tCO 0
3
0 -0
0
fc«
tio
.3
.•S 3 .-a 3 c ^
^ .3 & .9 « 3
C3
. 3
c; 0
3
u
Si
0
S
. 3
3J 0
3
0
3
. a
4J 1 *
3 T3
0
3
•^3
P
1
bo
c
03
(U
S
0)
C
c;
3
3
.^
eg
3
«
a
g
«
• >
w
e3
"i
.&
D
^
3
«
t»
3
w
t j '?
"%
-a
'> -d
'^.-^
i > ri
OJ nj
b3
1~, U.
P!
t'.
&
3
4J
fw
"0
3
4->
o
m
a
o
-Q
.i<
5
-a
e
«
-M
OJ
...>
^
'O
3
*.S'
•>
'>
c3
'>rt
•*
tH
4-)
t-M
">
t4
.
T3 4J
«J D
•0
> ^
3
.
oj
rt 3 3
-.J
3
3
OJ
0"
3
3
'co
>
hS
"«
^
1
a;
S
.3
^1
'«
•?
•? 9
I
CQ
•kJ
ID
-Q
X
0)
«;
s
73
0)
T3
•1SI
w
,
^
"3
03
_g
§s1
-
.2 " .a 3
§
ca
•g
f
,
1« s -
^
13
rt
S
0
c3
•
9i
• ^
t,
0
C3
S
Q
CO
a
§8.2]
LIGHT SOURCES FOR SPECTROSCOPY
169
8.2. The Power Output of Sources. The total radiant or luminous power output is sometimes of importance in choosing a source
for spectroscopic applications. The steradiancy (radiant power output per unit projected area of source per unit solid angle) or brightness
(luminous power output per unit projected area of source per unit
solid angle) is, however, usually more important, since this is the
limiting factor that Ordinarily determines the amount of radiant
power transmitted through the spectroscopic system onto the radiation receiver (photographic plate, photocell, eye, and so on), as discussed in Chapter 6. For example, an ordinary fluorescent lamp is
an excellent source for general illumination but a very poor source for
visual absorption spectrophotometry, for precisely the same reason
in both instances: while its total luminous power output is comparatively high, its brightness is comparatively low.
Radiant power is measured in watts, milliwatts (mw), or microwatts iixw). Luminous power is measured in lumens. One lumen
is the luminous flux in unit solid angle from a uniform source of
1 candle power. Since the luminous effect of a given amount of
radiant power depends on the wavelength of the radiation and the
spectral sensitivity of the observer's eye, the ratios lumens/watt and
watts/lumen can be expressed only for each particular wavelength and
in terms of a particular spectral visibility curve (Fig. 1.1). Data
representing the lumens/watt at various wavelengths for a "standard
observer" have been adopted by the Illumination Engineering
Society on the basis of the average spectral visibility curves of a large
number of normal persons. From such data and from data on the
spectral energy distribution of a source, the luminous power of the
source may be determined by step-by-step or graphical methods.^
Steradiancy is measured in watts (or mw or tivf) /cra^/steradian, where
the area in cm^ refers to the projected area of the source on a plane
perpendicular to the direction in which the radiance is taken. Brightness is measured in lumens/cm^/steradian, or in candles/cm'', the area
in cm^ having the same significance as in the case of steradiancy.
The radiant or luminous power output deteriorates with time in
some sources, such as incandescent lamps and mercury arcs, but not
in others, such as open carbon arcs and sparks wherein provisions
are made to keep the interelectrode gap constant as the electrodes
wear away. Such long-time changes are usually not important, but
' A. C. Hardy and F. H. Perrin, The Principles of Optics.
Book Company, Inc.. 1932.
New York: McGraw-Hill
170
L I G H T SOURCES FOR
SPECTROSCOPY
C§8.3
if emission must be kept at or near the max^imum, it is desirable to
replace deteriorating sources routinely after a fixed, limited period of
operation. Short-time fluctuations of radiance or brightness are
often significant in spectroscopic measurements, particularly if the
radiation-measuring system is not an integrating device like the
photographic plate, which averages out random fluctuations.
Radiance or brightness is more uniform over extended areas of
some sources, such as ribbon-filament lamps, than others, such as coilfilament lamps. A uniform source is to be preferred in instances in
which different portions of a final image field, formed by radiation
emitted from different portions of the source, are to be compared.
8.3. Practical Considerations. Practical factors in the choice of
sources include simplicity of construction and operation, ruggedness,
useful life, availability, and cost. Special considerations regarding
the choice of sources for absorption spectrophotometry, qualitative
and quantitative emission spectrum analysis, infrared spectroscopy,
spectroscopy of the extreme ultraviolet, and Raman spectroscopy are
discussed in the chapters dealing with
these subjects.
THERMAL EMISSION SOURCES
8.4. Spectral Characteristics of
Blackbody Radiation. A blackbody is
a body that absorbs all radiant energy
incident upon it, neither transmitting
0
lOpOO eopoo 30,000 40p00 50POO
nor reflecting any of this radiant
Wavelength in A
energy.* No substance behaves as a
Fig. 8.1. Intensity distribuperfect blackbody, but very close aption as a function of wavelength
for the radiation from a blackproximations to blackbodies can be
body at various temperatures.
constructed (see § 8.5).
The spectrum of a blackbody radiator is continuous. From the
short-wave end, the spectral intensity curve rises rather sharply to a
maximum and then tapers off more gradually toward still longer'
wavelengths (Fig. 8.1). The position of the maximum depends on
the temperature of the radiator, in accordance with Wien's displacement law:
'
,
XmT = b
(8;.l)
* It should be recalled that this discussion is confined to the optical region of the
spectrum (infrared, visible, and ultraviolet), hence we are here concerned with blackbody characteristics within the optical region.
§8.4]
LIGHT SOURCES FOR SPECTROSCOPY
171
where X™ is the wavelength; in angstroms, of maximum intensity;
T is the temperature, in degrees Kelyin; and & is a constant = 2.884
X 10'. The total radiation depends on the temperature of the
radiator in accordance with the Stefan-Boltzmann law:
W = aT'
(8.2)
where W is the total radiant power, in watts/cm- area of source, and
(T is a constant = 5.735 X 10"^^. The distribution of intensity as a
function of wavelength is given closely by Planck's radiation law
(shown here as applied to spectral energy measurements of band
width d\):
JxrfX = 4p^ d\
(F - 1
(8.3)
where J\ is the spectral radiant intensity (watts per steradian per cm
wavelength), d\ the spectral band width (in cm), A the area of the
source (in cm^), Ci the first radiation constant (1.177 X 10"^^ watts
cm^), C2 the second radiation constant (1.4320 cm deg), and e the base
of natural logarithms (2.718+). The curves of Fig. 8.1 were computed from Eq. (8.3).
The Planck radiation formula is somewhat cumbersome to handle,
and for many practical purposes the Wien formula, which is an
approximation of simpler form, may be used without causing appreciable error. This is of the form
JxdX = ^CiX"'e"^dX
(8.4)
the constants having the same values as in Eq. (8.3). The errors
introduced by using the Wien equation are smallest for small values
of XT. When XT is less than 0.5 cm deg, these errors are less than
probable errors of measurement. Extensive tables giving the radiant
power per unit wavelength interval at various wavelengths will be
found in the International Critical Tables.^ Methods of applying
the radiation laws to the sensitivity standardization of photographic
plates and photocells are discussed in § 13.5.
Those thermal radiators which are not blackbodies do not radiate
precisely in accordance with the above laws, but the correspondence is
2 International Critical Tables. New York: McGraw-Hill Book Company, Inc.,
1929.
172
LIGHT SOURCES FOR SPECTROSCOPY
[§8.5
sufficiently close to permit computation of the approximate c h a r a c teristics of the radiation to be expected from such sources when their
operating temperatures are known.
I n presenting d a t a regarding nonblackbody t h e r m a l r a d i a t o r s ,
reference is frequently m a d e to t h e brightness t e m p e r a t u r e , radiation
t e m p e r a t u r e , or color t e m p e r a t u r e . T h e brightness temperature or
radiation, temperature is t h e t e m p e r a t u r e a t which a blackbody would
have the same brightness or radiate the same power in a particular
wave band as the given nonblackbody radiator. Since blackbodies
are more efBcient radiators t h a n nonblackbodies, brightness or
radiation temperatures are less t h a n the true t e m p e r a t u r e s of nonblackbody radiators. T h e color temperature is the t e m p e r a t u r e a t
which a blackbody would have to be maintained to m a t c h the visible
color of a particular nonblackbody radiator, for example a m e t a l .
T h e color temperature m a y be m u c h higher t h a n t h e actual t e m p e r a t u r e of the nonblackbody radiator, since there is a tendency for t h e
spectra of such nonblackbody radiators as metals to be shifted t o w a r d
t h e blue as compared with those of blackbody radiators. Optical
pyrometers of the vanishing-filament t y p e measure brightness
temperatures.
8.5. Blackbody Radiators. Blackbody radiators are useful as
standards in the visible a n d infrared regions because their radiation
Fig. 8.2. Cross section of a simple blackbody, consisting of an electrically heated
metallic tube with small hole, a, through which radiation is observed.
distribution is completely determined by their t e m p e r a t u r e , as indicated b y E q s . (8.1) t h r o u g h (8.4). A close approximation t o
blackbody radiation is obtained b y viewing a uniformly heated c a v i t y '
of opaque material through a hole t h a t is small in proportion t o t h e
size of t h e cavity.' Various designs of practical cavity-type blackbodies have been described (General Reference 8.1). One of these
is illustrated in Fig. 8.2.
' P. K. Kichtmyer and E. H. Kennard, Introduction to Modern Physics. New York:
McGraw-Hill Bool!; Company, Inc., 1942.
§8.6]
LIGHT SOURCES FOR SPECTROSCOPY
173
8.6. Incandescent Electric Lamps. These are useful sources for
visible, near infrared, and near ultraviolet radiation, because they
have highly uniform and predictable spectral intensity characteristics.
Thus according to Forsythe (General Reference 8.1), a new gas-filled,
single-coil, 115-volt, 100-watt lamp, operated at a color temperature of 2870°K with a rated' life of 750 hr and a luminous efficiency
of 15.4 lumens per watt, has a spectral intensity distribution in the
visible region that can be matched by that of a blackbody operated
at a temperature somewhere in the range from about 2865 to 2875°K.
Since the radiation characteristics of incandescent lamps are well
known and are reliably constant when the lamps are new and are
operated at rated voltage, such sources serve as good secondary
standards, when calibrated in comparison with blackbodies, for determining the sensitivities of thermopiles and other radiation-sensitive
devices.
Data on the brightness and radiance of tungsten filaments are
shown in Tables 8.2 and 8.3. Table 8.4 shows the color temperatures
and maximum brightnesses of various lamps.
TABLE 8.2
DATA FOB BRIGHTNESS AND RADIANCE OF TUNGSTEN*
Temperature,
°K
-
2500
2600
2700
2800
2900
3000 •
3100
3200
3300
3400
3500
3655t
Color
temperature.
Candles
per cm^ *
Total lumens
per watt
input
Radiant
• flux, watts
per cm^
2557
2663
2770
2878
2986
3094
3202
3311
3422
3533
3646
3817
237.5
347.0
498.0
694.0
949.0
1257.0
1647.0
2110.0
2685.0
3370.0
4220.0
5740.0
11.67
14.28
17.26
20.43
23.80
27.10
31.0
34.6
38.5
42.6
45.9
53.1
67.2
80.6
95.6
112.5
132.5
154.5
177.5
203.0
232.0
264.0
300.0
360.0
* W. E. Forsythe and E. M. Watson, Jour. Opt. Soc. Am., 24, 114-118 (1934).
* Equivalent to lumens per cm^ per steradian.
t Melting point of tungsten.
C§8.6
LIGHT SOURCES FOR SPECTROSCOPY
174
TABLE
DATA
8.3
FOR S T E R A D I A N C Y O F T U N G S T E N * AT 2800
°K
Watts/cm^/steradian
in 100 A b a u d
Xin A
3,000
4,000
5,000
6,000
7,000
8,000
9,000
10,000
8.8
1.46
5.98
1.29
1.95
2.44
2.06
2.70
TABLE
X
X
X
X
X
X
X
X
10-*
10-2
10-2
10-'
lO-i
10-1
10-1
10-1
8.4
COLOR T E M P E R A T U R E AND M A X I M U M
INCANDESCENT
BRIGHTXESS OF
VARIOUS
LAMPS
Lamp
Color t e m p . , ° K
Regular 50-watt
75-watt
100-watt
200-watt
1000-watt stereopticon
Photoflood
6.4-volt a u t o
R i b b o n filament
Projection
2670
2705
2740
2810
3175
3360
2905
2800
3311
M a x . brightne.ss,
candles/cm^
469
563
605
781
2065
3100*
965*
694*
2110*
' Computed.
For spectroscopic applications, concentrated, uniform sources of
high brightness (or radiance) are desirable. These requirements are^
most closely met by projection lamps, automobile headlight bulbs,
and ribbon-filament lamps. The latter two types may be used
advantageously in applications in which the emission is required to be
free from short-time fluctuations, since they are low-voltage devices
that may be operated from storage batteries.
The radiant power output of incandescent lamps decreases with
time, as a result of evaporation of metal from the filament onto the
inside of the bulb. The life L is markedly decreased by increase in
§8.7]
LIGHT SOURCES FOR SPECTROSCOPY
175
the operating voltage V, as shown by the following expression
(General Reference 8.1):
8.7. Enclosed Metallic Arcs with Incandescent Electrodes. In
this category are several sources in which the bombardment of electrodes by ions in an arc stream raises the temperature of one or more
electrodes to incandescence, causing them to emit radiation. These
might be classed as arc sources, but since the radiation they emit is
principally from the incandescent electrodes, they are included here
with thermal radiators. All these sources must be operated in series
with ballast resistors or reactors, since they have the negative potential-current characteristics of arcs (see
§8.10).
The Pointolite is an enclosed tungsten-electrode arc in an argon atmosphere. For DC operation, it is
constructed with a tungsten ball as
anode and a tungsten rod-and-coil
filament in series as cathode (Fig. 8.3).
The AC Pointolite is similar in principle but has two tungsten balls that Fig. 8.3. DC Pointolite. b. Bimetallic strip; K, key for-startoperate alternately as anodes on suc- ing; Ri and R2, ballast resistors.
cessive half cycles. Usual sizes for
the DC Pointolites are 30, 100, 500, and 1000 candle power, and for
the AC, 150 candle power. Further data are given in Table 8.5.
The General Electric photomicrographic lamp is a variant of the
S-1 sun lamp, described in § 8.14. It contains a small cup-shaped
electrode, about 0.25 cm in diameter, placed slightly behind a timgsten
ring of somewhat larger diameter. These electrodes, between which
the arc is maintained, are connected by a V-shaped coiled-tungsten
filament. The atmosphere in the bulb is argon together with mercury vapor from an excess of metallic mercury. The bulb must be
burned base up, so that the mercury pool remains near the tip. The
lamp is operated on alternating current from an autotransformer.
The cup-shaped tungsten electrode becomes very hot during operation, emitting visible radiation of intense brightness (see Table 8.5).
Considerable radiation also arises from the arc stream, the mercury
lines being clearly visible in the spectrum.
176
LIGHT SOURCES FOR SPECTROSCOPY
C;
o o o
o
o o
o •??o
<
(p
j-H
<!>H ^
-
V5 O O O
t- O O O
W t - CO O
r-H
J^ »<
:t
>
J3
(W
CO i > i >
O
o ^
o o o
o
o
r H r»o
H CO
O
o
G^ CO J >
1>
2*0 o o - o
g o o o o
^ CO © * 1—1 0 5
Si I C S ^ C« CO
sg "'a
H
u o
0 ~S-
:SJ
bCTJ
't^
o o o
o o o
rOS
H GO
rH 00
^
fl
"^
g
s
? o o o o
.5=
O
£ O
»0 O
g C3 » 0 - ^
O
G^
iC
^
^
D
OJ
(H
OJ
*o
CO
S
§ i S
W i W5 « 5
oi ©i o^
, O
O
GO
O
" ^ t-; t o
O O i-H
TS
a
a
<
W5
eo q q
o »o ©* o<
i-H
©^
O
CO
O
>0
•*
O
>0
rH
rt
J > I-H O
GO Q* G^
O
GO
GO
©* o »o o
O
1-H CD
w
O
>5
>
"*
>0
rH
J2
'3D
tn
o
B
i j
c3
^
O
O
CO O
^
i-H ©^
O
t-(
°o
Ooo
a<<
OU U U
QOQP
(J
V
u
u
M
O
u
V
3
o
CJ
'a
o
'o
"o "§,
.*_) 4 J •**
•Os . Oa wJ
y
t^
cj
U
tH
y
tH
O
tH
cd cj c€
ppp-p
^ ^ ^ ^
[§8.7
S8.8]
LIGHT SOURCES FOR SPECTROSCOPY
177
The Western Union concentrated ar&- * operates in argon at atmospheric pressure by ionic bombardment of a metallic film of
metallic zirconium or zirconium oxide. It is a DC lamp, operated
from a suitable rectifier that provides high voltage for starting and
low voltage for operation. Alternatively, it may be operated in
series with a suitable ballast resistor and r-f choke directly from
110-volt DC mains. Means such as a Tesla-coil vacuum leak tester
must then be provided for starting. The cathode consists of a cup
of tungsten, molybdenum, or tantalum packed with zirconium oxide;
the anode is a sheet or plate of similar metal. The lamp is operated
under suitable conditions until a thin metallic film of molten zirconium forms on the surface of the oxide in the cathode cup, after
which it is ready for service. The spectrum exhibits a continuum
from the incandescent zirconium metal (temperature about 3000°K)»
superposed upon which are argon and zirconium lines from the arc
stream. The brightness, which is very high initially (Table 8.5),
increases with age, whereas the spot size and total candle power
decrease. The current density is ,about 900 amp/cm^ for the 2-watt
size, whereas the diameter of the luminous spot is only a few thousandths of an inch.
Descriptions of several special incandescent-electrode metallic arcs
in addition to those described herein, are given in General Reference 8.1.
8.8. Low-Temperature Thermal Radiators. Several thermal radiators that operate at comparatively moderate or low temperatures
(about 2000°K or less) are useful as emitters of infrared radiation
(see Chapter 17).
The Nernst glower (§ 17.1) is a high-resistance element made principally from zirconium, yttrium, and thorium oxides, which is maintained at .an appropriate radiation temperature by passage of an
electric current after preheating to make it conducting. The preheating may be accomplished automatically by an electric heater, or
a Bunsen flame may be used. To ensure constancy of output,
photocell-electronic regulating circuits may be employed. I t is
claimed that one such device maintains the radiant emission to
within 0.1 per cent of the average value. A typical glower for
110-volt AC or DC operation is in the form of a rod 1 mm in diameter
and 10 mm long, and consumes 1 amp at 95 volts. This source has
' W. O. Buckingham and C. R. Deitert, Jour. Opt. Soc. Am., 36, 245 (1946).
' The Concentrated-Arc Lamp: A New Type of Light Source. Water Mill, N. Y.:
The Western Union Telegraph Co.
178
LIGHT SOURCES FOR SPECTROSCOPY
[§8.9
a negative temperature coefficient and must be operated in series with
a ballast resistor. The radiant emission is high in the near infrared
but low in the far infrared. The visible emission is sufficiently high
to permit use in this region when a uniform source of moderate brilliance in the form of a rod is desired.
The Globar rod (§ 17.1) is a resistance heating element commonly
used in electric furnaces and heaters. I t has a high emissivity in the
far infrared and is advantageous for measurements beyond 10 /i. If
operated at above-normal voltages, it may have a radiance as high as
60 watts/cm^, at a brightness temperature of about 1800°K.
The Welsbach mantle has high emissivity in the infrared beyond 8 jj.
as well as in the visible, whereas its near infrared emissivity is relatively low. A piece of Welsbach mantle heated by a gaseous discharge serves as a satisfactory source for far infrared work.
Hot glass, which may be heated by embedded wires carrying an
electric current, is a good emitter in the far infrared.' Heated films
of crystalline powders^ have been used as selective radiators in the
infrared.
8.9. Other Thermal Radiators. W. M. Cohn" has developed a
thorium lamp that depends for its emission on bombardment of a
thorium target with an electron stream at 25 kv, 1 ma. This source
is nearly free from infrared and red radiation and has a continuous
spectrum extending well into the ultraviolet.
Photoflash lamps'" have maximum intensities of about 360,000
candles and flash durations of 0.03 to 0.06 sec. They emit radiation
through the electrical ignition of a thin sheet of crumpled aluminum
in an atmosphere rich in oxygen. The maximum temperature reached
' is about 9500°K.
Luminous flames, such as those of the kerosene lamp, yield continuous spectra as a result of. the heating of carbon particles in theflame. They are of insufficient brightness or steadiness to. be of
interest in modern spectroscopy. Nonluminous -flames, such as that
of the Bunsen burner, are often used to excite characteristic emission- ^
spectrum lines of Ba, Ca, Na, Sr, and other elements by .the insertion /
of salts of these substances in a hot portion of the flame>- This method'
is mainly useful for demonstrations or as a source of alkali-metal lines
' C. H. Cartwright, Phys. Rev., 35, 415 (1930).
8 A. H. Pfund, Jour. Opt. Soc. Am., 23, 270 (1933).
3 W. M. Cohn, Physik. Zeitschr., 32, 559 (1931).
" W. E. Forsythe and M. A. Easley, Jour. Opt. Soc. Am., 21, 685 (1931).
<4
§8.10]
LIGHT SOURCES FOR SPECTROSCOPY
179
when a suitable discharge lamp is not available. Many methods of
introducing the materials into the flame have been suggested; one
of the simplest and most effective is to soak a strip of filter paper in
a solution of the salt &nd wrap the paper around the Bunsen burner
so as to form a tube extending about half an inch above the top of
the burner. By the use of oxyacetylene or oxypropane flames, sufficient temperatures may be obtained to excite the characteristic line
spectra of more than a dozen elements for quantitative emission
analysis, and such sources are used in flame photometry.
A graphite-tube furnace has been developed by King (General
Reference 8.1) for studying the high-temperature emission spectra of
various substances.
Exploded wires afford a means of obtaining radiation of high
intensity, particularly in the ultraviolet. The. method consists in
discharging a high v6ltage (about 50,000) from a condenser (of say
0.2 to 0.5 /xf) through a thin wire^" or through an asbestos fiber
saturated with a solution of metallic salt,ii' '^ with a minimum of
inductance in the circuit. The wire or fiber may be mounted in a
groove in a block of insulating material. The discharge is extremely
noisy. The spectrum is continuous at ordinary atmospheric pressures except for absorption lines arising from the vapor of the metal
exploded.
OPEN ARCS
8.10. Electrical Characteristics. Arcs between electrodes of materials having high thermal conductivity, such as metals, tend to extinguish more readily than those between materials of low conductivity,
such as graphite. This characteristic accounts for the fact that it is
very difficult to maintain an arc between metallic electrodes on
alternating current, whereas with carbon electrodes this is not true.
The equation relating voltage and current in an open arc is -P
V = A + f^
(8.6)
where A and B are constants, / is the current, and x depends on the
anode material and is equal to 1 for carbon, 1.38 for tungsten, and
0.67 for copper. It will be noted that this equation indicates a
" J . A. Anderson, Astrophys. Jour., 51, 37 (1920).
12 R. A. Sawyer and A. L. Becker, Astrophys. Jour., 57, 98 (1923).
" W. B. Nottingham, Am. Inst. Elect. Eng., 42,12 (1923); Phys. Rev., 28, 764 (1926).
180
LIGHT SOURCES FOR SPECTROSCOPY
[§8.11
negative potential-current characteristic; as the voltage increases,
the current decreases, and vice versa. I t is generally necessary,
therefore, to use a suitable ballast resistor (or a reactor for AC operation) in series with the arc to achieve stability of operation., When a
DC arc is operated from a suitable rectifier, the ballast resistor may
be dispensed with by designing the transformer supplying the rectifier
tubes to have effective regulation in controlling the DC output. It
is often useful to include a suitable reactor in series with a DC arc to
damp out incipient oscillations and to build up a voltage if the arc
starts to die out.
Direct-current carbon arcs operate at voltage drops of from 30 to
60 volts across the arc terminals, AC carbon arcs at 80 volts or more,
and DC iron and copper arcs at 20 to 90 volts.
8.11. Carbon Arcs. There are two principal types of carbon arcs:
incandescent arcs, in which the incandescent ends of the electrodes are
the main source of radiation, and flame arcs, in which the radiation
comes primarily from the arc space.
Incandescent arcs may use solid or cored carbons. In DC arcs, the
maximum brightness is in the positive crater. The total candle
power is directly proportional to the current. The brightness is
proportional to the current density. The carbon diameter must be
increased as the current is increased to achieve effective stable operation, but higher current densities are attainable, in general, with the
larger carbon sizes. The plain carbon arc operates at about 55 volts,
4 to 8 amp for small laboratory arcs and 20 to 40 amp for projector
arcs. The spectrum of the incandescent carbon arc consists of a
continuum arising from the approximately blackbody emission of the
hot electrodes (principally the anode'in DC arcs), upon which is
superposed a series of lines characteristic of the materials vaporized
into the arc stream.
Flame carbon' arcs use hollow carbons packed with core materials
of various substances; salts of strontium, calcium, cobalt, Qr sodiumi
are frequently employed. The usual burning position is vertical.'
The highly luminous area is an extended arc stream, roughly ellipsoidal
in shape, the projected area of which is approximately equal to that
of a circle twice the diameter of the carbons used. For the same
current consumption, the carbon size used is generally somewliat
larger than that for incandescent arcs (see Table 8.6).' The spectrum
consists principally of closely spaced lines emitted by excited atoms
in the arc stream. By varying the core material, it is possible to
§8.12]
L I G H T SOURCES
FOR
SPECTROSCOPY
181
modify the spectral emission in particular regions substantially.
Appropriate cores give high emissivity in the ultraviolet.
TABLE 8.6
CHARACTEBISTICS OF CARBON ARCS AS LUMINOUS SOURCES
(Based on data in General Reference 8.1)
Type
Plain carbon
(low intensity)
Suprex carbons
High intensity
>
Flame arcs, white flame
" Volts
55
55
55
32
34
35
90
55
Amp
5 •
30
40
40 .
50
65
195
60
Carbon
diam., cm
power*f
0.8
1.2
1.3
0.6
0.7
0.8
1.6
1.5
1,350
8,500
11,900
8,150
11,600
17,100
88,400
15,900
•
* Mean spherical candle power for flame arc; candle at 20° for high intensity arc
and at 30° for Suprex 0.7 and 0.8 cm carbons; horizontal candle power in all other
cases.
t Emission from crater of positive carbon only in the case of plain and high-intensity
arcs; from the entire area of the positive carbon in the case of Suprex arcs.
I n continuously operated arcs, t h e maintenance of constant interelectrode distance as the electrodes wear away m a y be accomplished
b y hand-feed devices or b y a u t o m a t i c clockwork or motor-drive
mechanisms. Very reliable feeding mechanisms, which should find
application in t h e spectrographic laboratory, have been developed for
motion-picture projector arcs.
F o r AC operation, high voltages (about 2500 t o 5000) are sometimes used in spectrographic work t o trigger t h e discharge on each
half cycle, t h e supply circuit being designed so t h a t the voltage drops
almost instantaneously t o normal arc-operation values as soon as t h e
arc is established.'*
8.12. Metallic Arcs. Arcs of m a n y metals, such as copper,
nickel, cadmium, thallium, and tungsten, are used as spectroscopic
sources in t h e study of emission spectra (Chapters 9 and 15). As an
alternative to placement of the sample in a hollow carbon electrode.
" O. S. Duffendack and K. B. Thomson, Proc. Am. Soc. Testing Materials, 36, II,
301 (1936); commercial arcs of this type are available for use in spectrochemical
analysis.
182
LIGHT SOURCES FOR SPECTROSCOPY
C§8.12
arc electrodes may be formed directly from metallic samples for
qualitative and quantitative emission-spectrum analysis (Chapters 15
and 16). Horizontal rotating electrodes have been used in iron arcs
to permit long-continued operation. King'" and Gerdien and Lotz'*
Fig. 8.4. Pfund arc.
developed special arcs for studying the spectrum of iron and other
metals, in which water-cooled electrode holders permit the use, of
extremely high current densities.
Pfund ^' developed a very steady type of metallic arc, shown in
Fig. 8.4, which though ordinarily used with iron electrodes can also
be used with electrodes of copper and some other metals. Owing to
i^A. S. King, Astrophys. Jour., 62, 238 (1925).
^^ H. Gerdien and A. Lotz, Zeitschr. tech. Phy.iik, 4, 157 (192S); 5, 515 (1924).
" A. H. Pfund, Astrophys. Jour., 27, 298 (1908).
r
/
§8.12]
LIGHT SOURCES FOR SPECTROSCOPY
183
the reproducibility of the lines produced by the iron arc of the Pfund
type, it has been adopted as a source for secondary standard lines
(§ 9.8). The source is specified by the International Astronomical
Union as "the Pfund arc
operated between 110 and
250 volts, with 5 amps
or less, at a length of
12-15 mm used over a
central zone at right angles
to the axis of the arc, not
to exceed 1-1.5 mm in
width, and with an iron rod
6-7 mm diameter as the
upper pole and a bead of
oxide of iron as the lower
pole."
With protection
from air currents the Pfund
arc can be drawn out to
lengtlis pf 20 mm or more.
The lower electrode is usually made half an inch in
diameter, tapered conically
to a cup that holds the
bead of iron oxide. To
avoid deforming the cup
when the arc is first struck,
a small bead of iron may
be deposited in the cup,
the arc being struck with
a nail held in an insulating
Fig. 8.5. Arc or spark electrode holder.
handle. The upper elec- (Courtesy Baird Associates, Cambridge, Ma.ss.)
trode should be made
negative, and a suitable ballast resistor and reactor should be used
to operate the arc from 220 or more volts direct current, to ensure
stability.
Figure 8.5 shows a convenient type of arc holder for use with rod
electrodes. It is characterized by having adjustments that give
vertical control of the arc height, and horizontal and arc-length
adjustments, so that the operator can keep an arc of proper length
centered on the spectroscope slit.
184
LIGHT SOURCES FOR SPECTROSCOPY
[§8.13
ENCLOSED ARCS
8.13. Electrical Characteristics. Enclosed arcs, like open arcs,
have negative potential-current characteristics and must be operated with series resistors and reactors, or from special transformer or
rectifier circuits having appropriate regulation characteristics.
The potential drop across a particular arc and the current through it
depend on the pressure of the conducting vapor between the electrodes. After the arc is started, a rise in temperature ensues, resulting in an increase in the pressure of the vapor, a decrease in the current, and a rise in the potential drop between the electrodes. The
final operating condition depends on the equilibrium temperature
between the arc and its surroundings. This temperature in turn
'depends on whether the arc is ventilated or is cooled by an air blast
or by running water. The equilibrium condition is apt to be rather
unstable unless the operating temperature is controlled automatically
within narrow limits by such means as an intermittent or variable
air blast.
The usual operating voltages and currents for low-pressure metallicvapor enclosed arcs are respectively about 30 to 200 volts and 3 to
8 amp. For high-pressure metallic-vapor arcs, the operating voltage
may be 800 or more. The excitation in the arc stream increases as
the pressure is lowered, so that vacuum arcs may show higher
excitation than sparks.
8.14. Low-Pressure Mercury Arcs. These arcs usually operate at
pressures of about 1 atmosphere or less. They are good sources for
isolation of monochromatic radiation corresponding to the principal
mercury lines, for example in the excitation of the Raman effect or
in monochromatic irradiation experiments. Water-cooled arcs, such
as the Kromayer and Burdick therapeutic lamps, operate at lo.wer
temperatures and pressures than air-cooled arcs; and yield spectra
comparatively free from continuous background.^
^,
i
There are three principal types of low-pressure mercury arcs:'
(1) those with two liquid mercury electrodes, (2) those with a cathodei
of liquid mercury and an anode consisting of a spiral of tungsten wire
(DC Uviarcs), and (3) those with tungsten or oxide-coated metal
electrodes and an atmosphere that contains, in addition to mercury,'a
small amount of argon or neon for starting the arc by ionization.
The first two types of lamps are-normally started by tilting the arc
mechanically until contact is established between the electrodes, as a
§8.14]
LIGHT SOURCES FOR SPECTROSCOPY
185
result of the flow of metallic mercury through the arc tube, and then
restoring the tube to its initial position so that the contact is broken.
The tungsten coil must be made positive in the Uviarc; otherwise the
arc will go out or the coil will quickly be burned out by bombardment
with mercury ions. A reactor should be used in the circuit to provide
an induced high voltage t a help maintain the discharge when contact
is broken. Many ingenious electrical and electromechanical devices
have been designed for starting arcs of the first two types automatically. However, when automatic starting is desired, it is now
more customary to employ arcs of the third type. These depend on
thermionic emission from the electrodes (or from an auxiliary filam.ent) to excite ions in a gas such as argon, and on the use of sufficiently high interelectrode potentials so that the accelerated argon
ions excite rnercury ions by collision. Such arcs, if properly designed,
will operate on 60-cycle alternating current with extinction of the arc
at each half cycle.
Arcs of the third type are obtainable in envelopes of quartz or of
ultraviolet-transmitting glasses, such as Corning Corex. The latter
are frequently used for "sun lamps" in order to restrict the transmitted
radiation approximately to the spectral range of sunlight, the shortwave limit of which is about 2900 A.
Data regarding the radiant emission of low-pressure mercury arcs
at wavelengths corresponding to the principal mercury lines have been
published by McAli^ter^* and others (see General Reference 8.2).
Alternating-current Uviarcs^' in quartz arfe convenient line sources
of ultraviolet radiation down to about 1850 A. The electrodes are
helices of wire, the interstices of which are filled with rare-earth
oxides. The electrode material contributes little to the nature of the
arc stream, so these lamps are sometimes classed as discharge tubes
rather than arcs, but their operating pressure ( | to 1 atmosphere) is
higher and their potential gradient (about 12 volts/cm) lower than
those characteristic of discharge tubes. A small amount of argon,
added to the completely vaporized mercury in the tube, serves to
initiate the discharge. The quantity of mercury to be used is determined by the fact that the voltage gradient varies directly as the
xV power of the mass of mercury per unit tube length and inversely
as the f power of the inside tube diameter.^" Alternating-current
i«E. D. McAlister, Smithsonian Misc. Collect, 87, No. 17, 1 (1933).
1' L. B. Johnson and S. B. Webster, Rev. Sci. Inst, 9, 325 (1938).
2" W. Elenbass, Physics, 4, 747 (1937).
186
LIGHT SOURCES FOR SPECTROSCOPY
C§8.1S
Uviarcs are operated from high-reactance transformers, which provide the necessary ballast for stable operation. Maximum output is.
attained in about 4 min after the arc starts. The lamps may be
burned in any position. In all cases it should be remembered thatthe radiation from a mercury arc changes greatly as it warms up.
The General Electric S-1 sun lamp employs tungsten electrodes,
connected by a tungsten filament, in an atmosphere of mercury and
argon. Its spectrum shows the mercury emission lines superposed
upon a continuum from the tungsten filament and electrodes. The
lines of shorter wavelength than 2664 A are absorbed almost completely if the bulb is Corex D glass; this transmission limit is shifted
to the 2894 A line if Corning glass No. 690 is used.
8.15. High-Pressure Mercury Arcs. Several high-pressure mercury arcs have been developed for use primarily as luminous
sources.^1"^* These operate at comparatively high current densities
and at pressures ranging from 1 to 80 atmospheres (Table 8.7). Their
spectra exhibit, in general, much stronger continua and a greater
TABLE 8.7
DATA ON HIGH-PKESSUBE MEKCUKT Ancs^''
Type
H-3
Watts (lamp)
Lumens (100 hr)
Pressure (atm)
Maximum lumens/cm^/steradian
Source length (cm)
Source diameter (cm)'
85
3000
30
900
1.8
0.4
H-4
100
3500
8
400
2.4
0.75
H-5
250
10,000
4
300
4.3
1.3
H-6
1,000
65,000
80
30,000*
2.5
0.2
*The average value appears from, computatidn to be about 6500.
proportion of energy in the visible region than is characteristic of low- pressure mercury arcs. Like AC Uviarcs (§ 8.14), they employ
oxide-coated electrodes, together With an atmosphere of argon for
starting the arc, and are operated on alternating current from trans- '
formers with sufficient reactance to serve as ballast.. After starting, /
these sources require several minutes of operation to reach maximum'
total radiant output.
'''—'
21 B. T. Barnes, W. E. Forsythe, and W. J. Karash, General Electric Rev., 42, 540
(1939).
.
I
22 E. B. Noel, Jonr. Appl. Physics, 11, 425 (1940).
23 B. T. Barnes-and W. E. Forsythe, Jovr. Opt. Soc. Am., 27', 83 (1937).
2< L. J. Buttolph, Jour. Opt. Soc. Am., 29, 124 (1939).
§8.15]
L I G H T SOURCES F O R
SPECTROSCOPY
187
The arc envelopes kre of quartz. These are surrounded by jackets
o? glass except i n thA case of the H - 6 water-cooled lamp (Fig. 8.6),
for which outer jacktts of either quartz or glass may be obtained.
Some of the lamps with glass outer jackets cannot be operated with
the outer jacket removed without damage to the arc envelope seal
and hence are useful only for visible radiation and ultraviolet to
about 3000 A. With a quartz outer jacket, the H-6 lamp gives a
Fig. 8.6. High-pressure, water-cooled, 1000-watt quartz mercury arc. (a)
Quartz-arc capillary tube containing mercury vapor. (6) Housing, consisting of
quartz or glass tube with metal end fittings, to permit cooling the arc tube with
flowing water.
strong continuum in the ultraviolet to about 2270 A, with reversals
at 2350 and 2537 A. The lines are superposed upon the continuum
but are not prominent. The H - 3 and H-4 lamps with outer jackets
removed (or vfith, holes punctured in the jacket for transmission of
radiation) yield the mercury line spectrum in the ultraviolet superposed upon strong continua. The short-wave limit of emission is
about 2120 A, and the 2537 A line is reversed.
Because of fts high radiant emission from a small emitting area
(Table 8.7), the H-6 arc is an excellent source for applications in
which high radiance is required in the range from about 2700 A
through the visible and near infrared regions, and in which a strong
continuous spectrum is desirable. For shorter wavelengths or for the
isolation of monochromatic radiation, other sources may be preferable
even.when the highest radiance is required. On alternating current,
the H-6 arc must be operated in the horizontal position; on direct
current, it may be operated vertically, with the cathode as the upper
electrode, using a supply capable of deUvering 1220 volts at 1.2 amp
188
LIGHT SOURCES FOR SPECTROSCOPY
[§8.16
and a ballast resistor which, at 1.2 amp, will reduce the potential
across the arc to about 840 volts.
8.16. Other Enclosed Metallic-Vapor Arcs. The 200-watt G.E.
sodium-vapor arc has an output of 10,000 lumens and a brightness of
about 6 candles/cm^, the radiation being concentrated primarily in
the sodium lines near 5890 A.^^ Commercial enclosed metallic-vapor
arcs yielding spectra of cadmium and zinc, as well as sodium are
available.^* Enclosed metallic-vapor arcs have been designed for
emission of characteristic lines of antimony, bismuth, iron, lead,
potassium, selenium, tellurium, and tin (General References 8.1, 8.2).
8.17. Enclosed Carbon Arcs. With suitably designed enclosures,
carbon arcs may be operated at increased pressures in various inert
atmospheres^^ (including nitrogen, argon, helium, and hydrogen) or at
reduced pressures as compared with the normal operating pressures
of open arcs. The effect of increasing the pressure is to increase the
operating temperature, accentuate the continuous background, and
broaden the characteristic spectrum lines. The effect of reducing the
pressure is to narrow the emission lines and to yield lines characteristic
of higher excitation energies.
DISCHARGE TUBES
8.18. General Characteristics. A distinction has been made between arcs and discharge tubes on the basis that arcs are characterized
by appreciable contribution of the electrode material to the ionic
discharge stream and to the emission of radiation. This distinction
is not always clear-cut. There are, however, other distinguishing
characteristics that mark discharge tubes as different from arcs:
(a) they operate at lower pressures (usually less than OX)l-^atmosphere),
lower current densities, and-lower temperatures than arcs; (b) a.
higher potential gradient (up to several hundred volts per cm) may
be required to maintain the discharge; and (c) the spectra emitted
show lines of higher excitation energy and other differences.
Discharge tubes are usually operated from spark coils or trans-/
formers supplying high voltage (2000 to 20,000 volts) and low current'
(4 to 60 ma). Spark coils have the advantage for some applications
of supplying undirectional pulses, the induced voltage developed
^ Electric Discharge and Other Lamps. London: Adam.Hilger, Ltd., 1940.
25 F. Paschen, Ann. d. Physik, 12, 509 (1932).
§8.19]
LIGHT SOURCES FOR SPECTROSCOPY
189
when the primary interrupter makes contact being much less than
that when it breaks contact. Transformers for operating discharge
tubes should have sufficient reactance to limit the secondary current
to a safe operating maximum. Sign-lighting transformers are designed with such characteristics and are available in a sufficiently
wide variety of specifications to meet most needs for AC dischargetube operation. High-voltage rectifier circuits may be used in special
instances in which direct current is required. Finally, electrodeless
discharge tubes may be excited by placing them in a high-frequency
field.
As a general rule, the brightness or radiance of discharge tubes is
comparatively low. With certain exceptions, therefore, they are not
so well adapted to spectroscopic applications requiring highly concentrated light sources of great intrinsic brightness or radiance as to
applications in which extended sources of high total radiant output
are desired.
8.19. Glow-Discharge Tubes. A convenient form of Geissler
tube for spectroscopic use has two enlarged portions, containing the
electrodes, connected by a constricted tube (Fig. 8.7). Such tubes
°€
So
Fig. 8.7. Geissler tube.
are made either of glass or quartz, and may be obtained unfilled
(with stopcocks for filling) or filled with various gases such as argon,
helium, hydrogen, neon, nitrogen, or mercury vapor. The electrodes
may be of a plain metal, such as tungsten, or may be oxide-coated.
Excitation is often supplied by a spark coil, but a small sign-lighting
transformer (about 3000 volts, 6 ma) serves equally well. Geissler
tubes are chiefly useful for demonstration purposes and to obtain
narrow lines for reference standards or for interferometry (see
Chapter 20).
Another type of discharge tube that is commercially available and
has somewhat greater br ghtness, or radiance, than the usual Geissler
tube has been described by Ryde." It contains two closely spaced
electrodes in a compact envelope, the discharge being viewed through
" J. W. Ryde, Nature, 112, 944 (1923).
190
LIGHT SOURCES FOR SPECTROSCOPY
[§8.19
a window in one electrode. Glass or quartz envelopes may be obtained, with atmospheres of argon, CO2, helium, neon, nitrogen, or
oxygen. The spectrum of the neon tube is rich in sharp lines that
are useful as secondary wavelength standards. These lamps are
intended primarily for DC operation at 300 to 450 volts with oxygen,
nitrogen, and CO2, and at 200 to 250 volts with helium, argon, and
neon. They may, however, be operated on alternating current.
The current consumption is 5 to 15 ma.
Hollow-cathode tubes,'^^- ^^ with an atmosphere of inert gas, have been
designed in which radiation is emitted almost exclusively from the
cathode glow inside a hollow electrode closed at one end (Fig. 8.8).
ri
^
Fig. 8.8.
Hollow-cathode discharge tube.
A, Anode; C, cathode.
Spark lines of the metal comprising the cathode occur in the spectTum. These tubes are especially valuable for producing sharp lines
for interferometry (§ 20.2), or for spectroscopic analysis of small
quantities of material (Chapter 15).
Mercury-Vapor Discharge Tubes. When mercury vapor at low
pressures is admixed with a small amount of neon or argon in a discharge tube with oxide-coated electrodes, an easily started source is
obtained in which the ultraviolet radiation is largely concentrated
in the 2537 A mercury resonance line. This fact has been made use
of in the design of a large variety of discharge tubes for supplying
(together with appropriate filters), approximately monochromatic
radiation at 2537 A (see General Reference 8.2), to be used, for
example, in the excitation of Raman spectra or in ultraviolet photomicrography. With appropriate cooling it is possible to operate
^^F. Paschen, Ann. d. Physil.; SO, 901 (1916).
» H. Schuler, Phy.nk. Zeiuchr., 22, 2G1. (1921).
§8.19]
L I G H T SOURCES FOR SPECTROSCOPY
191
such tubes at moderately high-current densities that, together with
the concentration of emission in the resonance hne, give radiance at
2537 A as great as 10 times that achievable with usual mercury arcs.^"
Hydrogen Discharge Tubes. One of the most convenient and practical sources for providing a continuous spectrum throughout the
visible and ultraviolet regions (especially useful in absorption spectrophotometry) is the hydrogen discharge tube.''- ^- Such tubes are
Fig. 8.9.
Hydrogen discharge tube for absorption spectroscopy.
Adam Hilger, Ltd., London.)
(Courtesy
operated at hydrogen pressures ranging from 1 to 10 mm of mercury,
sX. applied voltages from 3000 to 5000 and at currents from a fraction
of an ampere to several amperes. At the higher current densities the
tubes must be jacketed and cooled by running water. The discharge
is usually viewed end on through a quartz window to increase the
effective brightness.'' Several commercial hydrogen discharge tubes
are available for spectroscopic use (Fig. 8.9).
'1 G. Kornfeld and F. Muller-Skjold, Zeitschr. pliysik Chem., B31, 223 (1936).
» E. O. Lawrence and N. E. Edlefson, Rev. Sci. Inst., 1, 45 (1930).
=2 G. B. Kistiakow.sky, Rer. Sci. In.if., 2, 549 (1931).
^ R. W. Wood, Physical Optics. New Vork: The Macmillan Company, 1934.
192
LIGHT SOURCES FOR SPECTROSCOPY
[§8.20
Electrodeless Discharge Tubes. If an electrodeless tube containing
an appropriate gas at low pressure is placed in a high-frequency electromagnetic field, such as that of a Tesla coil or radar transmitter, a
glow discharge in the gas will be excited under proper conditions.
Various methods of construction and operation have been described.'^"
One method of excitation is to surround the tube with a close-fitting
coil of wire carrying the high-frequency electrical current. Sources
of this type have been used in therapeutic applications and as probes
to determine the extent of high-frequency fields. They have found
use in spectroscopy for the production of spectra of multiply ionized
atoms or gases or metallic vapors and are particularly useful in the
vacuum ultraviolet.
ELECTRIC SPARKS
8.20. General Characteristics. The electric spark is an electrical
discharge across a gap separating two electrodes between which a
high potential difference exists. The potential gradient necessary to
initiate such a discharge depends on the gas pressure in the gap, the
ionization potential of the gas, the shape of the electrodes, and other
factors. For sharply pointed electrodes in air at atmospheric pressure the required gradient is about 12,000 volts/cm.
Cold emission of electrons from the cathode as a result of the high
potential gradient plays an important part in starting the discharge.
In this respect, sparks differ from arcs, in which thermionic emission
accounts primarily for the contribution of electrons to the discharge
stream. After breakdown occurs, an oscillatory discharge takes
place, the frequency and duration of -jvhich depends upon the constants of the electrical circuit. Once the train of succeeding oscillations -and sparks has died out, the gap remains quiescent until the
potential gradient has been built up again to the point at which a
disruptive discharge occurs. During the oscillatory discharge, electrode material enters the discharge stream as a result of ionic bom-/
bardment of the cathode. This effect, again, distinguishes sparks
from arcs (see General Reference 8.3), in which vaporization by heat
is largely responsible for the entry of electrode material into the arc
stream.'^
'
« J. G. Winans, Rev. Sci. Inst., 9, 203 (1938); see also General Reference 8.2.
^ H. Kaiser and A. Wallraff, Ann. d. Physik, 34, 297 (1939).
§8.21] .
LIGHT SOURCES FOR SPECTROSCOPY
193
As sparks are operated at higher current densities and higher electrode temperatures, they begin to behave more and more hke arcs.
Indeed, under suitable circumstances the transition to an arc discharge may be complete.
Spark spectra show the emission lines of singly and multiply
ionized atoms in addition to those of neutral atoms which are characteristic of arc spectra (see Chapter 10). The emission lines of atoms
of the electrode material normally predominate, in terms of total
radiant emission, over those of any gases present in the gap, and the
latter may be suppressed almost entirely by use of a series inductance
(Chapter 15).
8.21. The Spark in Air and Other Gases. For spectroscopic use,
it is convenient to employ electrodes about 3 to 4 mm in diameter
with wedge-shaped opposing ends (Fig. 8.10).
The electrodes are mounted with the formed
edges parallel to each other and to the optical
axis of the spectrograph, so that wandering of the
spark along the edges does not displace it laterally
with respect to the axis. The gap between the
electrodes may be from 2 to 8 mm.
A spark of these specifications may be oper- shaped^ spark Plicated from an induction coil ("spark" coil) but trodes.
is much more convenient to use with a highvoltage transformer, the primary of which is supplied with power
from a 110- or 220-volt AC line. The transformer should be rated
at 0.25 to 1.0 kva and should develop a secondary voltage of at least
10,000 and preferably 15,000 or 20,000 volts.
If a spark is operated directly from a spark coil or transformer, the
capacitance of the circuit is insufficient to permit appreciable storage
of electrical energy at the discharge potential. Under this circumstance, discharges occur quite frequently, the spark is "thin" and
comparatively nonluminous, and the radiance is low, being primarily
from emission by atoms of the gas in the gap rather than from those of
the electrode-material. If, however, a capacitor of appropriate value
is connected in parallel with the spark gap (Fig. 8.11), the energy
dissipated during each oscillatory discharge is greater, and appreciable
quantities of electrode material appear in the gap and contribute to
the radiant emission. Although the discharges occur less often (and
with lower oscillation frequency), they are of so much greater radiance
that the integrated radiation during a given period of time is con-
194
L I G H T SOURCES FOR
SPECTROSCOPY
[§8.21
siderably larger. It is customary, therefore, to use a capacitor in
the circuit so as to obtain a "hot," bright spark in which the spark
Unes of the electrode material predominate.
Increasing the value of the capacitor augments the brightness of
each disrujjtive discharge. Obviously, however, the capacitance
cannot be increased indefinitely. With any transformer of given
power rating and a circuit of given resistance, ultimately a capacitance
will be reached which is so great that the transformer cannot charge
it to a potential sufficient to cause breakdown of the spark gap
within the time of a half cycle. Then the condenser will simply be
charged with opposite polarities during succeeding half cycles, without
any disruptive sparks taking place. Before this condition is reached,
the spark will become irregular. The appropriate circuit constants
Fig. 8.11. Electrical circuit for operating a spark. T, High-voitage step-up
transformer; P, primary; S, secondary; C, condenser; L, self-inductance (used
if it is desired to suppress air lines); 0, spark gap.
may be computed from principles set forth in standard electrical
engineering texts. For example, if P is the power in watts required
to charge a condenser of capacity C in farads to a maximum voltage Fo
at every half cycle from an AC circuit of frequency F cycles per
second, then P = [CVo'F. If C is 0.02 ^i, Fo is 15,000 volts, and
F is 60 cycles per second, P — 270 watts, or approximately 0.25 kva. Actually, the power rating of the transformer used should be considerably greater than this for satisfactory operation. The practical
approach to the problem of optimum capacitance' is to use a capacitor
of multiple sections and, with a particular transformer and gap, t o '
determine by trial the capacitance which gives a bright but regular
spark. For use with 0.25 to 1.0 kva, 15,000- to 20,000-volt transformers and gaps of 4 to 5 mm, the optimum capacitance usually lies .
between 0.003 and 0.03 ^f.'"
T
= J. A. Anderson, Astrophys: Jour., 59, 76 (l9-ii).
§8.23]
LIGHT SOURCES FOR SPECTROSCOPY
195
T h e introduction of an inductance coil in the oscillatory circuit
(Fig. 8.11), which reduces the frequency of oscillation (since F
= --—7Y7^>
t e n d s t o reduce t h e intensity of emission lines arising
from t h e atmosphere in which t h e spark operates, and gives rise t o a
hotter spark. With large values of inductance and capacitance,
essentially t h e entire arc spectrum of t h e electrode material appears
in addition t o spark lines. Usual values of the self-inductance L of
the coil range from 15 microhenries to 1 millihenry; the value essential for effective suppression of t h e air lines in any particular case
m a y be determined by trial. A suitable coil m a y be made by winding
40 t u r n s of N o . 18 copper wire in a single layer on a 4-in.-diameter
insulating t u b e and providing a t a p every five t u r n s .
One difHculty with sparks is their tendency t o be irregular. M e t h ods of overcoming this difficulty include t h e use of a rotating synchronous spark gap in series with t h e gap used as a source" and t h e
use of a low-power, high-voltage i n t e r r u p t e d spark t o ignite a n d
control a high-power, low-voltage spark connected in parallel with it.^*
8.22. T h e H o t Spark in Vacuum. T h e average excitation energies
in t h e spark discharge increase as t h e voltage and capacitance are
increased. B y operating a spark in a v a c u u m , under which conditions
high breakdown potentials are required, a n d by using large capacitors
a n d transformers of high power a n d voltage rating, it is possible t o
obtain bright sparks -high in emissivity m t h e far ultra violet. ^^~^'
Millikan a n d Sawyer'^ and Edlen^" used gap lengths from 0.2 t o
2 m m , voltages of 50,000 or more, a n d capacitances of 0.01 to 0.5 yuf.
A fixed or rotating external gap is used in series with t h e v a c u u m gap
t o obtain uniform discharges (see C h a p t e r 19).
If wires are exploded (§ 8.9) in v a c u u m instead of air, t h e spark
spectrum ofl^he-.yire material is obtained instead of a continuum.
8.23. T h e Underwater Spark. A spark between metallic electrodes under water yields a continuous spectrum extending to a b o u t
" O. Faussner, Archiv f. Eisenhutienwesen, 6, .551 (1932).
38 M. F. Hasler and H. W. Dietert, Jour. Oft. Soc. Am., 33, 218 (1943).
39 R. A. Millikan and R. A. Sawyer, Phys. Rev., 12, 107 (1918).
1" B. Edlen, Zeitschr. f. Phydk, 100, 621 (1936).
" R. A. Millikan, Astrophys. Jour., 52, 47 (1920).
*2 R. A. Sawyer, Astrophys. Jour., 52, 286 (1920).
« E . Carter, Astrophys. Jour., 55, 162 (1922).
196
LIGHT SOURCES FOR SPECTROSCOPY
[§8.24
2000 A in the ultraviolet.""*' The source is of high intrinsic brightness and is of particular value in absorption spectrophotometry of the
ultraviolet (Chapter 14). Electrodes of various metals may be used,
tungsten steel being particularly satisfactory. The spark gap (3 or4 mm in length) is housed in a watertight container, with a quartz
window. Distilled water is used in the container. A high-frequency,
high-voltage electrical supply from a Tesla coil is used to energize the
spark. The primary of the Tesla coil is connected through a spark
gap to a circuit consisting of a capacitor in parallel with the secondary
of a high-voltage transformer (about 20,000 volts, 1 kva), with a
primary for operation from 110 or 220 volts alternating current.) If
an open spark gap is used in the oscillatory exciting circuit for the
Tesla coil, operation is extremely noisy; quieter operation may be
achieved by the use of a quenched gap.
8.24. The Spark as a Source in Qualitative and Quantitative
Analysis. If it is desired to observe the spark spectra of solid conducting materials for purposes of identification or quantitative analysis, these may be made the electrodes of a spark gap. The spectra
of nonconductors or of liquids may be observed by introducing them
into suitable spark gaps having electrodes of appropriate materials;
alternatively, conducting liquids may be made to serve as the electrodes. Various methods of applying spark spectra to qualitative
and quantitative'analysis are discussed in Chapters 15 and 16.,
MISCELLANEOUS SOURCES
8.25. Cathodoluminescence Devices. Radiation may be excited
by bombarding a gas or vapor with accelerated electrons. The principal problem in so doing is to provide an evacuated space for,the
acceleration of the electrons, and a gas space in which impacts may
occur, without the use of a barrier between them, ^his result has
been accomplished in various cathodoluminescence devices*'' *^ by
bombarding a metal with electrons in an emjlosure that may be con« H. J. McNicholas, Nat. Bur. Standards Jour. Res., ,1, 939 (1928).
« I. Wyneken, Ann. d. Physik, 86, 1071 (1928).
• « B. Wrede, Ann. d. Physik, 3, 823 (1929).
" H. Hertz, Wied. Ann., 19, 809 (1883^.
*» .4. S. Kingand E. Carter, Astrophys. Jour., 44, 303 (1916); E. Carter and A. S.
King, Astrophys. Jour., 49, 224 (1919).
§8.28]
LIGHT SOURCES FOR SPECTROSCOPY
197
tinuously evacuated. Small amounts of the metal are vaporized;
under appropriate conditions the vapor can be confined largely to
the immediate vicinity of the metal target. The atomic beam for
producing very narrow lines by this method is discussed in § 20.2.
8.26. Fluorescence, Phosphorescence, Resonance Radiation, and
Chemiluminescence. Sources involving these mechanisms for the
emission of radiation are of low intrinsic brightness and are of interest
in spectroscopy primarily from the standpoint of studying the spectra
characterizing the phenomenon concerned. Fluorescence is readily
excited in many substances by irradiating them with ultraviolet
radiation. The 3650 A and 2537 A mercury arc or discharge-tube
lines are particularly convenient for this purpose.
8.27. Pulsed Discharge Tubes. • By discharging a condenser
through a suitable tube with a minimum of inductance and resistance
in the circuit, it is possible to obtain brilliant flashes of light of
extremely short duration. Anderson^^ used this method to excite
hydrogen discharge tubes, employing a 2-juf condenser charged to
35,000 volts. Current densities of the order of 25,000 amp/cm^ were
obtained. The brightness was extremely high, approximating that
of a blackbody at 40,000°K. Edgerton^" has carried out extensive
investigations of pulsed discharge tubes for application to highspeed photography, for aerial photography at night, and for other
uses. In applications in which a maximum of radiance is required
and a minimum duration of radiation is unobjectionable, such sources
are extremely 'useful.
8.28. The Sun as a Source of Radiation. The sun is a source of
high intrinsic brightness, yielding radiation extending from the far
infrared through the visible and ultraviolet regions. Its spectrum is
continuous but contains thousands of absorption lines, the Fraun' hofer lines. As the sun's radiation reaches the earth, it is modified
furtlier by absorption by CO2, O2, water vapor, and ozone in the
. various layers of the atmosphere. Typical values for the radiant
power of sunlight at the earth's surface in various spectral regions at
noon on a clear day are given by Luckiesh.'^
" J . A. Anderson, Astrophys. Jour., 75, 394 (1932).
™H. E. Edgertou, Elec. Eng., 50, 327 (1931); II. E. Edgerton and K. J. Germeshausen. Rev. Sci. Inst, 3, 535 (1932).
*i Matthew Luckiesh, Oermiddal, Erythemal, and Infra-Red Energy. New York:
D. Van Nostrand Company, Inc., 1946.
198
LIGHT SOURCES FOR SPECTROSCOPY
GENERAL REFERENCES
8.1.
8.2.
8.3.
W. E. Forsythe (Ed.), Measurement of Radiant Energy. New York:
McGraw-Hill Book Company, Inc., 1937.
F. F. Heyroth, The Chemical Action of Ultraviolet Rays. New York:'
Reinhold Publishing Corporation, 1941.
Leonard B. Loeb, Fundamental Processes of Electrical Discharge in Gases.
New York: John Wiley & Sons, Inc., 1939.
CHAPTER
9
Identification of Spectrum Lines
T H E ATOMS OF T H E CHEMICAL ELEMENTS t h a t have been s t u d i e d
spectroscopically are found t o emit, in their various stages of ionization, milhons of spectrum hnes of different wavelengths. Some of
these lines are much stronger t h a n others, so t h a t most of t h e light
emitted b y atoms appears in a smaller number of lines, of which some
350,000 h a v e been measured a n d listed as t o parent atom. Less t h a n
half of these have been classified as t o exact mode of origin in t h e
atom.
E a c h of t h e hundreds of thousands of molecules produced b y combinations of t h e elementary atoms emits m a n y characteristic bands.
T h e study a n d elucidation of b a n d spectra is a very important aspect
of spectroscopy t h a t is still in its infancy. Only in t h e case of diatomic molecules have m a n y bands been classified.
E v e r y spectrum line has a definite wavelength, characteristic of t h e
a t o m or molecule t h a t emits it a n d dependent to only slight degree on
the electrical a n d magnetic surroundings of t h a t a t o m or molecule.
T h e most precise means of identifying a spectrum line is by its wavelength, which in much of t h e spectrum can be determined t o seven
significant figures, in some cases t o eight. Other less positive means
of identification are b y observation of t h e intensity of t h e line relative
to other lines in t h e same spectrum, of t h e patterns formed in t h e
spectrum b y groups of related lines arising from t h e same atom, a n d
of t h e behavior of t h e line when t h e source of light from which it is
emitted is subjected t o various external influences, such as variations
in t e m p e r a t u r e , pressure, excitation, or electric or magnetic field.
I n general, elements on t h e left-hand side of t h e periodic table emit
comparatively simple spectra. T h e complexity of t h e spectra
emitted increases for t h e elements in t h e middle of t h e table and
diminishes again slightly for those on t h e right. T h e b r e a d t h of
spacing of lines in t h e patterns formed b y groups of spectrum lines,
199
'200
IDENTIFICATION OF SPECTRUM LWES
F'""
IS
' si
S a>
o Sb
^
S
R 0 "
11
;^•3
0)
a
0.
OJ
0
B
U
4)
M
a>
O
«r^a
3 "o g
J^ •s
ta
•*->
o
4)
U3
O
f »Z
J
"13 1
<y •>
a
*->
3
tH
i
?s ?!
5-6
.;^
g? 1
C3
^ .a
. --^
^
-1 -0
OS
bi S
ti
be
?'
S | i>
^-J
rt
pQ
O
'73
QJ
>*-^
b
§9.2]
IDENTIFICATION OF SPECTRUM LINES
201
called multiplets, emitted by an element increases from top to bottom
of the periodic table. This widening causes overlapping of patterns
and an apparent increase in complexity for the elements lying low
in the table. Thus the simplest spectra are emitted by the elements
in the upper left-hand corner of the table, and the most complex by
those in the middle bottom region. The two extremes are well
represented by hydrogen and uranium.
9.1. Identification of Lines and Bands by Appearance. Experience soon brings the practicing spectroscopist considerable familiarity
with the appearance of characteristic spectra, so that he can readily
identify groups of lines at a glance. The color of the visible lines as
seen in a spectroscope gives a first clue; thus the two yellow lines of
sodium,»at 5896 and 5890 A, known as the D lines, are familiar to
almost every scientist. On a spectrogram the color is lost, but a
much more precise means of identification is substituted—the position
of the line on the plate relative to other lines. In Fig. 9.1, several
characteristic spectra are shown. The spectra of zinc and cadmium
show triple groups of lines, of intensity diminishing toward shorter
iiiiiliiimnii;
Fig. 9.2;
Comparison spectrum for identiflcation purposes. The upper spectrum
is that of iron, and the lower that of copper.
wavelengths, the triplets of cadmium being somewhat more widely
separated than those of zinc. The spectra of copper, potassium, and
rubidium contain obvious doublets, and the spectra of titanium and
vanadium show more complex regularities. Iron shows few regularities that are immediately obvious, yet they exist in profusion.
The cyanogen bands are emitted strongly by CN molecules formed
in the carbon arc burning in air and are easy to identify by their
appearance.
9.2. Identification by Comparison Spectra. Lines in an unfamiliar
spectrum can conveniently be identified by using the method of comparison spectra. The spectrum of the unknown material is photographed on the same plate as that of some riiaterial whose spectrum
is well known, as in Fig. 9.2. Known lines of the familiar spectrum
can then be identified at intervals across the spectrogram, and by
202
IDENTIFICATION OF SPECTRUM LINES
[§9.2
approximate interpolation the wavelengths of the unknown lines can
be determined. The unknown may then be identified by the use of
wavelength tables. In preparing such spectrograms, it is important
to avoid lateral displacement of the plate between the recording of
the known and the unknown spectra. If the plate is racked up or
down, one cannot be sure that this condition is fulfilled, and so use
is made of diaphragms or occulters to cover different parts of the slit
or the plate during each of the exposures. The Hartmann diaphragm,^ shown in Fig. 5.4, can be used on the slit of a stigmatic
spectrograph and produces a spectrogram of the type showm in
Fig. 9.3. It is important to remember that the image of the slit
;!m;!!':^ii|iii-;i:i;
'';::jni'ii.:''iMiiii
Fig. Q.3. Comparison spectrogram taken with a Hartmann diaphragm. Top
and bottom spectra are of the iron arc. The center spectrum is lead.
on the plate is inverted; hence the upper portion of the slit corresponds to the lower spectrum. Care should be taken not to jar the
spectrograph when adjusting the diaphragm.
Even with astigmatic instruments, sufficient separation between
two spectra can be obtained with a Hartmann diaphragm at the slit
to make identification possible if the astigmatism is small. With
spectrographs having great astigmatism,, it is useful to provide an
occulter directly in front of the plate and as close to it as possible.
This diaphragm can be moved up and down to uncover various portions of the line length on the plate for the production of comparison
spectra.
For quick identification by comparison spectra, a low- or mediumpressure quartz mercury arc such as the Lab-arc or IJviarc is a useful
source, because it .can be kept readily available and furnishes a
limited number of intense lines well distributed throughout the
visible and ultraviolet regions. The beginning spectroscopist should;
' J . Hartmann, Zeits. f. Instrumentenlcunde, 20, 57 (1900').
§9.3]
I D E N T I F I C A T I O N OF S P E C T R U M
LINES
203
acquaint himself with the approximate wavelengths of the principal
groups of mercury lines, as given in Table 9.1.
To provide comparison spectra having greater numbers of lines, a
copper arc or spark or an iron arc will be found useful, the latter
having the more complex spectrum. The wavelengths of the principal iron lines have been very carefully measured, and the spectrum
is so rich that a known line can be found every few angstroms.
In many cases it is useful to expose a spectrogram to light from both
a quartz mercury arc and an iron arc, the spectrum of the former
being used for preliminary orientation and that of the latter for
precise determination of the wavelengths of unknown lines.
TABLE 9.1
GROUPS OF MERCUBY LINES USEFUL FOR WAVELENGTH IDENTIFICATION
Group
Color
1
Red
2
Yellow
3
4
Green
Blue
5
Violet
6
7,
Ultraviolet
Ultraviolet
Ultraviolet
Ultraviolet
Ultraviolet
Ultraviolet
Ultraviolet
Ultraviolet
Ultraviolet
Ultraviolet
Ultraviolet
Ultraviolet
s'
9
10
11
12
•13
14
15
16
17
Wavelengths
Approximate
intensity
6234 A
6152
5790
5770
5461
4359-4348
4078
4047
3663-3654-3650
3342
3132-3126
3026-3022
2967
2893
2804
2753
2652
2537
2482
2399-2378
10
20
50
50
100
20
8
10
70
10
40
30
10
10
20
10
8
30
5
3
9.3. Spectrum Chartsior Comparison. Each spectrograph has its
own characteristic dispersion curve, relating the wavelengths of spec-trum lines to their positions on a spectrogram. Users of spectrographs find it convenient to accumulate a set of standard spectrograms
on which wavelengths are marked for quick identification of lines on
future plates. Spectrograms of unknown materials produced with
[§9.3
IDENTIFICATION OF SPECTRUM LINES
the same instrument can then be fitted over these standard plates,
corresponding known lines being superimposed, and other lines identified by inspection. Lines of striking intensity or pattern can be
204
ADAM HILGER. LTD. LONDON
Fig. 9.4. The Judd-Lewis spectrum comparator.
quickly identified by this means, and key lines caii be marked on
the new spectrogram.
The spectra on such standard plates should not be crowded together, but a sufiicient width of clear plate should be lefl l)etween to
permit recording wavelengths of important lines in India ink. When
§9.3]
IDENTIFICATION OF SPECTRUM LINES
205
a m a r k is t o be permanent it should be placed on the emulsion side of
the plate, b u t if temporary only, on t h e glass side. T h e dry emulsion
will t a k e ink better if roughened with a rubber eraser.
Comparison of plates t a k e n on t h e same spectrograph can be accomplished b y superposing plates on a viewing box, b u t various devices
are manufactured commercially for doing this without introducing
actual contact between the plates. Of these, the Judd-Lewis Spect r u m Comparator, Fig. 9.4, is typical. A suitable optical system
enables t h e observer to see the image of one spectrogram superposed
on t h a t of the other, a n d by adjusting t h e position of one plate with
a horizontal a n d a vertical screw or rack a n d pinion he can bring a n y
spectrum line on this plate into coincidence with the corresponding
line on t h e other. A library of s t a n d a r d films with wavelengths of
i m p o r t a n t lines marked on t h e m is provided with the spectrum comp a r a t o r manufactured by the Applied Research Laboratories, for use
with their grating spectrograph. These films can be projected
directly on t h e screen of the spectrum comparator for rapid identification of lines.
Although a set of marked spectrograms built u p about a given
i n s t r u m e n t forms the most useful t y p e of reference library for identification of lines by comparison, charts a n d atlases of spectra are
obtainable which can be of considerable assistance. These are of
greatest use for visual comparison of line p a t t e r n s . T h e value of t h e
published charts is somewhat limited b y variation in t h e dispersion
characteristics of different spectroscopic instruments. Grating spectrographs, however, give dispersion t h a t is nearly uniform with wavelength; a n d b y varying the magnification in a projection device, one
can provide any dispersion desired. Hence grating spectrograms can
usually be projected directly on commercial charts made with grating
instruments and can be compared with t h e m with relatively little
error as a result of differences in dispersion.
T h e principal atlases of spectra t h a t have been printed or are
available as photographic reproductions are listed in Table 9.2. P r o b ably the most generally useful of the charts are those of the spectrum
of iron, since iron lines are so frequently used for purposes of wavelength identification.
M a n y of these iron charts have marked on
t h e m the positions of the principal lines of other elements, so t h a t
these can be identified by projecting unidentified lines directly on
t h e chart.
"r
206
IDENTIFICATION OF SPECTRUM LINES
[§9.3
TABLE 9.2
LIST OF CHARTS AND ATLASES OF SPECTRA*
1. J. M. Eder and E. Valenta, Atlas typlscher Spektren. Vienna: A. Holder,
1911. Contains reproductions of more than 600 photographs of flame, arc,
and spark spectra, taken with glass and quartz prism spectrographs and with
low-dispersion gratings. Note Rowland scale of wavelengths.
2. J. Hardei, Atlas de Spectre d'Arc. Paris: G. Doin, 1926. Contains 54
charts of sensitive lines of the elements, with the iron spectrum, taken with a
prism, from 3500 to 2500 A.
3. F. Lowe, Atlas der Letzten Linien der Wichstigsten Elemente. Dresden:
Steinkopf, 1928. Contains charts of sensitive lines for 43 elements taken
with a small quartz spectrograph in the region 4700 to 2200 A.
4. G. Scheibe and C. F. Lindstrom, Tabellen des Funlcen- und BogenspeJdrum des Eiiens, etc. Berlin-Steglitz: R. Fuess, 1933. Chartso fare and
spark spectra, in the range 3700 to 2300 A.
5. W. J. Crook, Metallurgical Spectrum Analysis. Stanford University
Press, 1935. Gives 20 charts of sensitive lines of the elements in the range
5670 to 2796 A, and the iron spectrum in the ranges 5671 to 5058 A and
3433 to 2794 A, using grating.
6. A. Gatterer and J. Junkes, Atlas der Restlinien. Castel Gandolfo, Italy:
Specola Vaticana, 1937. Contains 28 photographs of spectra in the range
8000 to 2200 A, in arc and spark for 50 elements.
7. W. R. Brode, Chemical Spectroscopy. New York: John Wiley & Sons,
1939. Contains 35 charts for the range 5090 to 2310 A, in the arc, showing the
spectral lines of iron, with indicated positions of lines of other elements.
8. A. Gatterer and J. Junkes, Spektren der seltenen Erden. Castel Gandolfo, Italy: Specola Vaticana, 1945. Contains 45 pages of spectrum charts
of the rare earths taken with prism spectrograph. Arc spectra, 7600 to
2265 A; spark spectra, 4350 to 2265 A.
9. A. Hilger, Ltd., London. Charts of spectra taken with a quartz prism
spectrograph of R.U. powder (mixture of 50 elements) and of iron, copper,'
neon, and helium.
,
10. Charts of the Iron Spectrum Photographed at the Massachusetts Institute
of Techriology. Cambridge, Mass.: Jarrell-Ash Co. Contains 10 marked and
mounted spectrum charts, each 20 in. long, <jf the iron arc taken with a 35-ft
concave grating.
A d a m Hilger, Ltd., has issued a set of spectrum charts showing a t
least seven i m p o r t a n t lines of each included element. This firm also
furnishes a so-called R . U. {raies ultimes) powder, which contains a
mixture of chemical elements in such proportions-lhat.seven or more
lines of the most important elements will appear when t h e m a t e r i a l '
is b u r n e d in an arc. This powder is convenient for use in qualitative
* Based on an article by G. R. Harrison, Jour. App. Phys., 10, 760 (1939), by permission of that Journal.
§9.4]
IDENTIFICATION OF SPECTRUM LINES
207
spectrographic analysis, to determine the presence or absence of any
given chemical element.
9.4. Identification of Lines by Wavelength Determination. Spectrum lines can be identified most accurately by measurement of their
wavelengths. In a simple spectrum, a line can usually be identified
if its wavelength is determined with a precision of a few tenths of an
angstrom, or even more. Several other strong lines of the element
can be looked for to establish the presence or absence of the element
in the source. In more complex spectra it often becomes important
to know wavelengths to within ±0.01 A, whereas for term analysis
of a really complex spectrum, wavelength precision to ±0.001 A is
desirable. Fortunately, the broadest lines, whose wavelengths are
most difficult to determine precisely, usually arise from the simplest
spect'ra. Lines emitted by such atoms as the rare earths or uranium
are usually sharp and well defined. Sharp lines can be measured
with large diffraction gratings to within a few thousandths of an
angstrom, and when improved standard lines become available, precision to ±0.001 A should not be unusual. With interferometers,
wavelengths of sharp lines can be measured to ±0.0001 A, or 1 part
in 50 million or better (Chapter 20).
Wavelength measurement with an ordinary spectrograph involves
determining the location of a line on the plate relative to known lines
and also determining the positions of the known lines so the dispersion
of the plate can be computed. The dispersion of a prism spectrograph varies so rapidly with wavelength that it is necessary to use a
carefully plotted dispersion curve, or if very precise determinations
are needed, to make interpolation calculations of the type discussed,
in § 9.6. Known comparison lines are then needed on the plate at
very close intervals. For a small prism spectrograph, where precision
to only 0.1 A is required, a dispersion curve similar to that shown in
Fig. 9.5 can be plotted on cross-section paper. The scale of the plot
should be fairly large, 0.1 mm on the position scale corresponding to
0.1 A if the fifth figure of the wavelength is to be considered significant. Thus a sheet of paper a meter long would be required to cover
3000 angstroms, and to reach greater accuracy would require use of
an inconveniently large sheet.
Since a long narrow sheet of paper is more readily obtainable and
handled than a large square sheet, it is useful to break up the interpolation process into two parts, a linear portion, marked A in Fig. 9.6,
and a residual part. Part A need not be plotted, since linear compu-
208
IDENTIFICATION OF SPECTRUM LINES
[§9.5
tations are readily carried out, and the residual can be plotted as a
correction curve in a form similar to Fig. 9.7.
Wavelength, measurements should always be supplemented by
7000
\
6000
\
\
s
5000
X in
•
\
V
4000
3000
2000
100
2D0
300
Plate Distance in mms
AOO
*
Fig. 9.5. Dispersion curve for a prism spectrograph.
intensity estimates, since the intensity of a line often gives secondary
information that is revealing and confirmatory (§9.9).
9.5. Measurement of Spectrograms. Distances along the plate
can be measured with a celluloid rule calibrated in centimeters, if
7000 I . - '
•'
r
100
200
Plate Distance in mms
'^
300
*"
400
Fig. 9.6. Approximation of a dispersion curve C by a straight line, A.
§9.6]
IDENTIFICATION OF SPECTRUM LINES
209
estimates to 0.1 mm will suffice for the desired precision. Lines produced by small prism spectrographs can be measured in this way with
an uncertainty of 1 to 10 angstroms, depending on the spectral region
involved, and those produced by large grating spectrographs to
within 0.05 A. Some spectrographs are provided with wavelength
or frequency scales. These are inferior in 'accuracy to separate
millimeter scales but reduce the necessary computation.
Improved precision can be obtained by using a spectrum magnifier.
A typical instrument of this sort, with 20 mm scale calibrated to
0.1 mm, is manufactured by Bausch & Lomb. It is arranged to be
placed directly against the emulsion side of the plate, thus avoiding
errors due to parallax. This scale can be read to dbO.Ol mm. The
eyepiece is adjustable and should be carefully focused on the scale
by each observer.
+ 1600
+ 1200
+ 800
+ 400
Correction
flA.inA
°
-400
-800
-1200
i, 0
100
200
Plate Distance in mms
300
40O
*~
Fig. 9.7. Correction curve, B, to be used with the linear approximation of
Fig. 9.6.
For measurement of Raman spectra, in which weak and diffuse
lines are common and extreme precision is unnecessary, positive
prints enlarged approximately 10 times are useful. Such lines are
difficult to measure on the negative under magnification, especially on
the fast and grainy plates used in photographing Raman spectra.
9.6. Use of the Comparator. Precise wavelength measurements
are carried out on a spectrum-measuring engine or wavelength comparator. A typical comparator, manufactured by Adam Hilger,
Ltd., is illustrated in Pig. 9.8. Comparators are of two basic types,
those in which the microscope (and the observer's eye) is moved, and
those in which the plate is moved. The former is the simpler method,
since the microscope can be carried on a comparatively short carriage
210
IDENTIFICATION OF SPECTRUM
Fig. 9.8.
LINES
[§9.6
Hilger wavelength comparator, Model L76.
Fig. 9.9. Optical diagram of Jarrell-Ash projection comparator JA-200. 1,
Lamp; 2, 11, condensing lenses; 3, slit and filter; 4. 14, 13, first-surface mirrors;
5, 7, 13, projection lenses; 6, plate being measured; 12, comparison plate; 8,
screen occulter; 9, adjustable slit; 10, photocell (barrier-layer type); 16, screen:
17, observer's eye. (Courtesy Jarrell-Ash Company, Boston, Massachusetts.)
§9.6]
IDENTIFICATION OF SPECTRUM LINES
211
which shdes on ways t h a t need be only as long as the length of this
carriage plus the a m o u n t of motion desired. When the entire plate
is moved, the length of the ways m u s t be a t least equal to the length
of the plate carriage plus the a m o u n t of motion required. T h u s , t o
read from end t o end of a 10-in. plate without resetting would require
a comparator more t h a n 20 in. long. However, this arrangement
results in higher over-all precision and greater convenience.
Fig. 9.10.
Jarrell-Ash projection comparator, Model JA200.
T h e plate is usually observed t h r o u g h a microscope provided with
an eyepiece in which cross hairs are mounted to serve as fiducial
m a r k s . T h i s microscope should magnify t h e spectrum lines by n o t
more t h a n 15 diameters, and 10 X will be found satisfactory for all
spectra b u t those containing the sharpest lines. I t is undesirable t o
magnify spectrum lines by more t h a n this amount, since the line
consists merely of an elongated array of silver grains, and the eye
m u s t be able to estimate the center of gravity of this array. Con-
212
IDENTIFICATION OF SPECTRUM LINES
[§9.6
t r a s t , as well as density and s y m m e t r y , enters into this j u d g m e n t .
Observation of a very diffuse line is sometimes facilitated b y use of a
diminishing lens. Usually, however, a change in magnification during
a series of measurements along a plate is not practicable.
Some modern comparators use t h e projection system. A beam
from a low-power l a m p is sent t h r o u g h the plate as in a projection
lantern, and a n image of t h e plate is thrown on a screen in front of
t h e operator. A typical optical system of this sort is shown in
Fig. 9.9, a n d a commercial instrument using this principle in Fig. 9.10.
Advantages of this m e t h o d are ease of superposing comparison spect r a ; improved comfort of t h e operator, who need not refocus his eyes
between setting on the line and reading the comparator scales; a n d
the possibility of using a fiducial m a r k
of a n y desired shape, this being changeI
able a t will by drawing an India-ink line
on t h e ground-glass projection screen.
,g,
/u%
, ,
..
Favorite forms of fiducial m a r k are
shown in Fig. 9.11. T h e chief disadFig. 9.11.
Convenient
vantages of projection comparators are
forms of fiducial marks for
, ,
„
,
., .,.
use in projection comparalack of compactness and susceptibility
tors.
to errors introduced by t h e h e a t from
the projection l a m p .
Before measuring a plate, it is desirable t o place dots a t t h e ends
of a few identified lines, t o serve as reference marks. T h e plate t o be
measured (or film fastened on a glass plate) is clamped, emulsion
side u p , on the comparator carriage. This carriage is moved on
ways by means of a screw, usually of 1 or 2 m m pitch. T h e screw
is t u r n e d by a handle m o u n t e d on a d r u m t h a t is usually calibrated
with divisions m a r k i n g 0.01 m m plate travel. Verniers m a k e estim a t e s possible t o 0.001 m m . D u r i n g measurement of a spectrogram it is i m p o r t a n t t h a t t h e
plate be m o u n t e d so t h a t the spectrum being measured is closely ,
parallel t o the ways of the c o m p a r a t o r and so t h a t the distance from /
t h e emulsion surface t o t h e microscope oc projection lens r e m a i n s /
suflSciently constant t o ensure' t h a t t h e plate will not go out of focus'
while being traversed^from one end t o the other. T h e plate should,
be m o u n t e d wijh the emulsion side up, since measuring the lines
through the glassimay introduce errors due to variations in refraction.
I n some comparators a means is provided of freeing the carriage from
the screw t h a t drives it, so t h a t t h e plate can be slid rapidly from
§9.6]
IDENTIFICATION OF SPECTRUM LINES
213
one end to the other to make sure that the lines remain in focus and
in the field..
The eyepiece of the comparator should first be focused on the cross
hairs, and then the microscope should be carefully focused on the lines
by the parallax method, in which one moves the eye slightly from side
to side to make sure that the image of a cross hair moves with that
of the spectrum line. One cross hair should be adjusted to be closely
parallel to the lines (though it is not necessary that these be at right
angles to the screw), to ensure that the same part of each line will be
measured from one end of the spectrum to the other. Change of
inclination of the line of traverse across the plate will obviously change
the observed dispersion.
With measurements being started at one end of the plate, a line is
brought into view and the comparator is adjusted until an accurate
setting of the vertical cross hair on the cente*" of the line is achieved.
The observer then makes entries in his notebook, recording the screw
reading in millimeters and the drum reading in thousandths of a
millimeter followed by an estimate of the intensity of the line. Intensity estimates may be made on a scale of 0 to 10, 0 to 100, 0 to 1000,
or even greater. At first, intensity estimates will be haphazard, but
after some practice the observer should be able to make fairly selfcortsistent estimates. After deciding on the line intensity, he may
put down in a Remarks column some comment such as d, R, s, h,
6T other notation as given in Table 9.3, to describe the character of
the line.
TABLE 9.3
DESCRIPTIVE NOTATION FOR SPECTRUM LINES*
bh
c
d
h
a;
a.
r
R
s
w
TV
Band head
Complex line
Double line
Hazy, diffuse, nebulous
Asymmetrical, heavy toward long wavelengths
Asymmetrical, heavy toward short wavelengths
Slightly self-reversed
Heavily self-reversed
Sharp
Wide
Very wide
* Adapted from Massachusetts Institute of Technology Wavelength Tables, General Reference 9.3.
214
IDENTIFICATION OF SPECTRUM LINES
[§9.7
Comparators usually have a certain amount of backlash between
the screw and the nut, and settings should be made by approaching
all lines from the same side. If the operator decides that he has
overshot the center of gravity of a line, he should reverse the screw
by at least half a turn and approach the line again.
To reduce backlash, comparators are sometimes provided with
counterweights which keep the carriage pushed against the screw
which drives the nut. Others are provided with split nuts with
built-in springs that automatically keep the tension uniform and
reduce the backlash. Even when one of these provisions is made,
however, it is desirable to approach all spectrum lines from the same
side in any series of mea'surements.
With a carefully made comparator, it should be possible to repeat
single readings to within 0.002 mm or even better, but this limit will
depend on many factors besides the judgment of the operator, such
>as backlash, the thickness and viscosity of the oil film separating the
nut from the screw, friction in the bearings, elasticity in the metal,
and so on. Comparators are available for astronomical work in
which motions in two directions can be measured. Additional data
on comparator construction are given in § 9.11.
9.7. Calculation, of Wavelengths. When wavelengths are to be
determined to a precision greater than d=0.1 A, they must be calculated unless an automatic comparator is used (§ 9.11). If standard
reference lines are close together, linear interpolation can be used,
especially with grating spectrograms for which the dispersion curve is
almost linear when wavelength is plotted against position on the plate.
With prism spectrograms, the wavelength-dispersion curve is far from
linear, though a somewhat flatter curve is obtained when wave
numbers or frequency units are plotted instead of wavelengths.
The plate factor of a spectrogram should always be measured in
angstroms per millimeter. The tendency of some beginning spectrographers to calculate "dispersion" in angstroms per inch is to be discouraged because of the convenience of using a common system"
among all spectroscopists.
As an example of linear interpolation, take the case of two lines A
and B, shown in Fig. 9.12, lying between two known lines whose,
wavelengths are as given in the figure.. By means of a comparator,
the distance from standard 1 to line A is measured as 4.88 mm, to
line B as 7.96 mm and to standard 2 as 11.21 mm. Since the distance
between the two standards is 11.21 mm and the difference in their
§9:7]
IDENTIFICATION OF SPECTRUM LINES
215
wavelengths is 13.56 A, the plate factor in this region is 1.209 A per
millimeter. The distance from standard 1 to line A can then be
multiplied by this plate factor to give a wavelength difference of
5.90 A, which, when added to the wavelength of standard 1, gives
4696.21 A as the wavelength of line A. This process may be repeated
to determine the wavelengths of unknown lines by extrapolation for
a short distance beyond the second standard, the limitation being
Standard spectrum
Unknown spectrum
A B
Fig. Q.12. Wavelength measurement by linear interpolation.
that the departure from linearity must not be greater than the
maximum error that can be tolerated in the wavelength determinations.
Where greater precision is desired, nonlinear interpolation formulas
can be used with either grating or prism spectrograms. Hartmann's^
interpolation formula is the best known of these. It may be written
simply as
C
X = Xo +
d — da
where X is the wavelength of the unknown line, Xo is a wavelength
which is constant for a given plate, C is a constant for the plate, and
d is the distance measured along the plate from do, which is some
definite point on a linear scale. The three constants Xo, C, and do
can be calculated by substituting the wavelengths and comparator
readings of three known lines in the above equation and then solving
the three equations simultaneously. On substituting in the equation
the values of the constants so determined, the wavelength of any
unknown line can be calculated from its position.
Hartmann's formula is used much less frequently nowadays than
formerly, since the wavelengths of so many lines are known with
high precision that linear interpolation between known lines suffices
for most wavelength determinations. Furthermore, prism spectrographs are seldom used for precise wavelength determinations, and
linear interpolation serves with diffraction gratings.
The method of coincidences is sometimes used in measuring wave2 J. Hartmann, Astrophys. Jour., 8, 218 (1898).
216
IDENTIFICATION OF SPECTRUM LINES
[§9.8
lengths with diffraction gratings, though it has been restricted in the
past principally to setting up wavelength scales, and the modern
interferometer has made this procedure unnecessary. The method
of coincidences makes use of the fact that in diffraction-grating spectra
a first-order line of wavelength X should theoretically occur in the
same position as a second-order line of wavelength X/2, a third-order
line of wavelength X/3, and so on. Since diffraction gratings are
imperfect optical instruments, this method gives only approximate
results, and gross errors may result if it is relied on entirely. Rowland's original wavelength scale,' built up by this system, ultimately
was found to have errors in some regions of the spectrum of several
tenths of an angstrom. Errors produced by using the method of
coincidences vary from grating to grating but are usually of the order
of a few thousandths or hundredths of an angstrom. With a given
grating, such errors depend on the density of exposure. This effect
is closely related to target pattern and to the variation of line shape
with order and with density of exposure.
i
Overlapping orders are sometimes of value in identifying spectrum
lines obtained with diffraction gratings, since it Js often possible to
identify lines of different orders by their appearance. Thus a secondorder line at 2300 A can readily be distinguished from a first-order line
at 4600 A, because the contrast of the photographic emulsion is less
for the shorter wavelength than for the longer; one line will have a
grayish tone and the other will be a dense black. The appearance of
the Rowland ghosts (§5.3) and the lengths of lines can also be used on
occasion to identify lines from different orders.
For very accurate measurements of wavelengths, interferometers
should be used, as discussed in Chapter, 20.
9.8. Standards of Wavelength. The lengths of light waves can be
measured directly by means of interferometers (§ 20.2). Since the determination of a wavelength to 1 part in 5,000,000 or better is a very
delicate and lengthy process, the procedure has been adopted of
determining the wavelengths of a few lines very precisely and then >
making measurements relative to these standard lines with ordinary /
spectrographs. In practice, the red cadmium line at 6438.4696 A is'
taken as the primary standard of wavelength, and several hundred
other lines have been measured relative to this with interferometers',
and defined as secondary and tertiary standards.
' H. A. Rowland, Collected Physical Papers.
§9.8]
IDENTIFICATION OF SPECTRUM LINES
217
If new measurements of the wavelength of the primary standard
were always made in terms of the standard meter bar, the accepted
wavelengths of all spectrum lines would have to be changed every
time improved measurements of either were made. To avoid the
necessity of making such corrections, the International Union for
Cooperation in Solar Research^ in 1907 adopted the following resolution : "The wavelength of the red ray of light from cadmium produced
by a tube with electrodes is 6438.4696 A, in dry air, at 15°C on
the hydrogen thermometer, at a pressure of 760 mm of mercury, the
value of g being 980.67. This number will be the definition of the
unit of wavelength." Hence the International Primary Standard of
Wavelength Cd 6438.4696 A is exactly correct by definition, even
though its value is known in terms of the standard meter bar to only
about 1 part in 10,000,000. An angstrom, written A, is thus defined
to be
of the wavelength of the cadmium red line, and not, as
.D4OO.4D9D
previously, 10~^ cm (written A).
The International Astronomical Union^ has set up a number of
international secondary standards of wavelength, using only lines that
have been measured concordantly and independently in at least three
laboratories, usually with Fabry-Perot etalons (§ 20.6). Many of
these secondary standards are lines of neon and krypton, and are
known relative to each other and to the primary standard to within
0.0001 A, or about 1 part in 50,000,000. A still larger number of
iron lines have been measured with the etalon interferometer, and
some of these have been adopted as secondary standards. These
lines are broader than those of the rare gases "and presumably are
correct only to within 0.001 A. Table 9.4, page 218, contains a list of
publications giving the adopted secondary standards.
The standard Pfund arc (§ 8.12) is used for producing the iron
secondary standards, light being taken only from regions not closer
than 7 mm to an electrode (to avoid wavelength shifts due to strong
electric fields near the electrodes, known as pole effect). This precaution can be observed only at wavelengths below 4500 A, however,
since at longer wavelengths it is found necessary to use a shorter arc
to bring out the desired lines; but pole effects are less likely to occur
to lines in this spectrum region;
The secondary standards leave little to be desired with respect to
* Trans. Int. Union Solar Res., 2, 109 (1907).
5 Trans. Int. Astron. Union, 6, 79 (1938).
218
IDENTIFICATION OF SPECTRUM LINES
[§9.8
TABLE 9.4
TABLES OF STANDABD WAVELENGTHS*
1. Commission 14, Wavelength Standards, Trans. Int. Astron. Union, I,
35 (1922), contains 402 secondary and tertiary standards in the range 7032
to 3370 A, for iron and neon, given to 0.001 A, with intensity range 1 to 10.
Trans. Int. Astron. Union, 2, 40 (1925), contains 4 neon Hnes in range 75376929 A, given to 0.001 A, adopted as secondary standards. Other lines Hsted
but not adopted. Trans. Int. Astron. Union, 3, 77 (1928), contains 384 lines
in the range 7535 to 3370 A, in arc and discharge for iron and neon, given to
0.001 A, with intensity range 1 to 10. Revision of standards, including provisional standards as well as adopted standards. Trans. Int. Astron. Union, 4,
58 (1932), contains 312 lines in range 8062 to 341 A, in standard arc, vacuum
arc, and discharge, for six elements, given to 0.001 and 0.0001 A, 3 iron and
10 krypton standards adopted. Corrections to 1922 table given.
Trans.
Int. Astron. Union, 5, 81 (1935), contains 201 lines in the range 10,216 to
580 A, in standard arc, vacuum arc, and discharge, for seven elements, given
to 0.001 and 0.0001 A, with intensity range 1 to 1500. Krypton and neon
standards adopted. Trans. Int. Astron. Union, 6, 79 (1938), contains 271
lines in the range 3845 to 2100 A, in arc and discharge, for iron and krypton,
given to 0.001 and 0.0001 A, with those adopted as standards.2. W. F. Meggers, Nat. Bur. Standards Jour. Res., 14, 33 (1935), contains
91 lines in the range 10,216 to 7164 A, in the short iron arc,(given to 0.001 A,
with intensity range 1 to 1500. Nat. Bur. Standards Jo)^r. Res., 18, 543
(1937), contains 242 lines in the range 3497 to 2100 A, in Wie standard iron
arc, given to 0.001 and 0.0001 A.
3. W. F. Meggers, Proceedings Sixth Conference on Spectroscopy (Wiley,
New York, 1939), page 116, contains 346 lines in the range 7032 to 2447 A, in
arc and discharge, for iron, neon, and krypton, given to 0.001 and 0.0001 A.
Collection of all adopted secondary standards.
4. F. Twyman and D. M. Smith, Wavelength Tables for Spectrum Analysis,
2d ed. (Hilger, London, 1931), page 12, contains 505 lines in the range 6750 to
2327 A, for standard arc and discharge, for iron, helium, and neon, given to
0.001 A, with intensity range 1 to 10. Collection of adopted standards and
other accurate measurements.
5. W. R. Erode, Chemical Spectroscopy (Wiley, New York, 1939), page 387.
In list of principal iron lines, gives adopted secondary standards.
precision b u t are somewhat lacking in number, distribution, suitability, /
for obtaining various forms of excitation, a n d desirable variety a n d
uniformity of physical characteristics. A discussion of the 1938 s t a t u s
of wavelength standards has been given b y W. F . Meggers,^ president
of t h a t commission of t h e I.A.U. charged with t h e responsibility foi;
these standards.
,
* Based on a list given by G. R. Harrison, Jour. App. Phys., 10, 760 (1939), by
permission of that Journal.
§9.9]
IDENTIFICATION OF SPECTRUM LINES
219
Spectral regions shorter than 2447 A and longer than 6678 A are
thus far unprovided with official wavelength standards, but for most
routine purposes wavelengths of known spectrum lines suitable for
use as standards in these regions can be obtained from wavelength
tables.
Wavelengths measured before 1910 or thereabouts are on a different
scale from that of the international angstrom and should be corrected
before use. Most such measurements are on Rowland's scale.
Kayser (General Reference 9.1, Vol. VI, page 891 (1912)), gives
corrections that should be applied to change original angstroms (A)
to international angstroms (A).
In addition to the International Secondary Standards, a group of
tertiary standards has been set up, but these are relatively unimportant, and many wavelengths given in spectroscopic tables are of
comparable accuracy. The tertiary standards have been measured
with large diffraction-grating spectrographs. Such wavelength determinations are limjted in precision by several factors, most of which
are related to the fact that many diffraction gratings produce unsymmetrical spectrum lines, which vary in shape with density of exposure.
The center of gravity of the line may shift with density owing to the
nonlinearity of the characteristic curve of the emulsion and the
complex forms of the lines produced by the grating. It is usually
necessary to photograph at one time many lines of widely different
intensities. When a broad region of the spectrum is covered in a
single exposure, as is becoming increasingly common, it is difficult to
have all standard lines of the proper density.
Standards measured by interferometers can be effectively supplemented by those obtained by computation, if we make use of the
Ritz combination principle, that each spectral line corresponds to the
difference in energy between two levels (§ 10.1). In a complex atom
each energy level may give rise to many spectrum lines, and when the
wavelengths of a number of lines have been measured with sufficient
accuracy to determine a large number of the energy levels precisely,
the wavelengths of other lines arising from these levels can be computed with a high degree of accuracy. By this means, wavelength
scales can be smoothed out and made self-consistent. This procedure
is particularly valuable for the infrared region.
9.9. Intensity Estimates. An important part of the description of
a spectrum line is an estimate of its intensity relative to other lines
in the same spectrum. Ability to estimate intensities on a uniform
220
IDENTIFICATION OF SPECTRUM LINES
[§9.9
scale can be acquired only by experience. A good intensity estimate
is often more satisfactory than a very precise intensity determination,
because the actual intensity of a spectrum line depends on many
factors, including the type of spectrograph and photographic emulsion
used, the method of source excitation, and the characteristics of the
atom emitting the line.
It is important to differentiate clearly between two needs for intensity estimates. The fundamental need is that of determining the
probability that an atom will emit a certain spectrum line under certain conditions, relative to its probability of emitting another line.
The second need is that of the spectroscopist who is interested merely
in the actual density of a line as it can be expected to appear on his
spectrum plate. For this purpose, no method of determining spectral
intensities has been developed which is more satisfactory than a good
visual estimate.
One of the most uniform intensity scales is that of A. S. King, who,
in the course of a long lifetime spent at Mount Wilson Observatory
in measuring spectrograms and estimating intensities, developed a
remarkably uniform and self-consistent scale. Russell * has shown
that King's intensity estimates vary approximately as the sjjaare root
of the true intensity of the line.
y^
The older intensity scales ordinarily ran from \ to 10, with additions
at both ends. When a spectroscopist found lines fainter than those
he had been calling 1 he called them 0 or 00, and often progressed
as far as 0000. If the line were stronger than 10, he might call it
12 or 15. Modern workers have found that an expanded scale is
useful, particularly since the tendency in making an eye estimate is
to compress the true intensity scale, strong lines appearing relatively
weaker than is justified. Meggers and other workers at the Bureau
of Standards use a scale that-goes up to 10,000, and the intensity scale in the Massachusetts Institute of Technology Wavelength Tablet
runs from 2 to 10,000, lines fainter than 2 having been arbitrarily
excluded. Even on these expanded scales all numbers are not used,- ,
the numbers most commonly used by spectroscopists being 0, 1, 2, /
3, 5, 8, 10, 12, 15, 20, 25, 35, 50, 60, 70, 80, 100, 120, 150, 200, 300,'
400, 500, 700, 1000, 1200, 1500, 2000, 3000, 5000, and 10,000, with ,
occasional interpolations between these.
When one estimates the intensity of a line, its width and shape
6H. N. Russell, Proc. Nat. Acad. Sci., 11, 314 and 322 (1925).
§9:10]
IDENTIFICATION OF SPECTRUM LINES
221
probably contribute as much to t h e estimate as its maximum density.
A skillful estimator takes into account t h e general integrated intensity
of t h e line, including something for self-absorption (§ 10.9), as shown
b y the changed shape of the line.
T h e precise determination of intensity is a complex and difficult
process, since so m a n y variables m u s t be t a k e n into account.'' This
precision is n o t needed and, in fact, is not w h a t is wanted for spectral
intensities t o be used in connection with wavelength tables.
9.10. Catalogues of Wavelengths. Wavelengths are conveniently
listed in t w o types of catalogues, t h e first giving in order of wavelength t h e lines known t o be emitted b y a particular type of a t o m or
molecule, a n d the second listing wavelengths in order, followed b y
t h e a t o m or m^olecule of origin.
Spectrochemical qualitative analysis requires use of only a few of
t h e most sensitive lines of each element, a n d these are most conveniently obtained from brief tables listing from 500 to 5000 lines.
Such tables are given in Appendices I a n d I I of this book.
T h e best-known inclusive tables of spectrum wavelengths for atomic
lines are listed in Table 9.5. O u t s t a n d i n g is Kayser's m o n u m e n t a l
Handbuch der Spedroscopie,
General Reference 9.1, with d a t a on
180,000 lines. M a n y wavelength d a t a included in this are now outd a t e d , however. Although p a r t s of Volumes V I I and V I I I contain
d a t a obtained for certain elements as recently as 1934, no d a t a obtained later t h a n 1911 are included for some 40 elements.
TABLE 9.5
INCLUSIVE TABLES OF SPECTRUM LINES*
1. H. Kayser, Handbuch der Spedroscopie (Hirzel, Leipzig), Vol. V (1910),
contains about 20,000 lines in the range 39,100 to 1030 A, in arc, spark, discharge and flame, for 45 elements, A to N, given mostly to 0.01 A, with intensity range 1 to 500. Contains table of air lines and also some band heads.
Note Rowland scale. Volume VI (1912) contains about 50,000 lines in the
range 8000 to 2000 A, in arc, spark, discharge, and flame, for 41 elements,
Na to Zr, given mostly to 0.01 to 0.001 A, with intensity range 1 to 10. Also
contains tables of iron lines, wavelength range 9000 to 2200 A, principal lines,
wavelength range 91,000 to 1850 A, and band heads. Note mostly Rowland
scale.
'See G. R. Harrison, Jour. Opt. Soc. Am., 17, 389 (1928); G. R. Harrison and H.
Engwieht, ibid., 18, 287 (1929); R. S. Seward, Phys. Rev., 37, 344 (1931).
* Based on an article by G. R. Harrison, Jour. App. Phys., 10, 760 (1939) by permission of that Journal.
222
IDENTIFICATION OF SPECTRUM LINES
[§9.10
2. H. Kayser and H. Konen, Handbuch der Spectroscopie (Hirzel, Leipzig),
Vol. VII (Parts 1, 2, 3, 1923-1934), contains about 54,000 lines in the range
75,000 to 70 A, in arc, spark, discharge, and flame for 49 elements, given to
0.01 or 0.001 A, with intensity range 1 to 10. Also table of air lines. Volume
VIII (1932) contains about 22,000 lines in the range 29,300 to 250 A, in arc,
spark, and discharge, for 19 elements Ag to Cu, given to 0.01 or 0.001 A, with
intensity range 1 to 10.
3. Landolt-Bornstein, Physikalisch-chemische Tahellen, Erganzungsband I,
.page 336, II, page 529, III, page 703 (Springer, Berlin, 1931-1935), contains
about 10,000 linesin the range 74,360 to 291 A, in arc, spark, and discharge,
for 81 elements, given mostly to 0.01 A. Contains also some band heads and
tables of intense and ultimate lines.
4. G. R. Harrison, Massachusetts Institute of Technology Wavelength Tables
(Wiley, New York, 1939), contains 109,275 lines, listed according to wavelength, in the range 10,000 to 2000 A, in the arc, spark, and discharge, for
87 elements, given mostly to 0.001 A, with intensity range from 1 to 9000.
Air lines and some band heads included.
5. F. Exner and E. Haschek, Die Spelctren der Elemente bei normalem Druck
(Deuticke, Leipzig and Vienna, 1911), Vol. II, contains about 60,000 lines in
the range 6800 to 2200 A, in the arc, for 67 elements, given to 0.01 A, with
intensity range 1 to 1000;. 214 band heads also listed. Volume I I I contains
about 56,000 lines in the range 6800 to 2200 A, in the spark, for 78 elements,
given to 0.01 A, with intensity range 1 to 1000; 107 bands also listed. Note
Rowland scale.
6. P. Auger and others, Donnees numeriques de spectroscopie (GauthierVillars, Paris, 1910-1936), contains about 150,000 lines (including duplicates)
in the range 10,000 to 2000 A, in the arc, spark, flame, and discharge, for
76 elements, given to 0.1 and 0.01 A, with intensity range 1 to 1000. Some
bands also included.
7. W. Jevons, Report on Band Spectra of Diatomic Molecules (The Physical Society, London, 1932), contains data on bands of 142 molecules.
8. W. Weizel, Handbuch der ^ ExperimentalphysiJc, Erganzungsband .1
(Akademische Verlagsgesellschaft, Leipzig. 1931), contains data on bands of
about 150 molecules.
A catalogue containing more t h a n 300,000 entries, comprising all
wavelength measurements on atomic lines given in the literature u p
to 1939, has been compiled a t t h e Massachusetts I n s t i t u t e of T e c h nology by W P A workers, b u t this has not appeared in printed form.
Kayser's Tabelle der Hawptlinien der Linienspektren aller. Elemente,
appearing in its latest edition in 1939 under the editorship of K a y s e r
and Ritschl (General Reference 9.2), lists 27,000 lines in t h e order of
wavelengths in the range 90,850 to 33 A, in arc, spark, f ,nd discharge
tube for 88 elements, with wavelengths given to one or two figui;es
after the decimal.
§9.11]
IDENTIFICATION OF SPECTRUM LINES
223
In 1939, the Massachusetts Institute of Technology Wavelength
Tables (General Reference 9.3), giving the 109,275 strongest lines
lying between 10,000 and 2000 A from neutral atoms and those in
their first stage of ionization, were published. In these tables, the
lines are listed in order of wavelength, followed by the parent element.
Abbreviated tables of wavelengths useful in spectrochemical analysis
are listed in Table 15.2.
Which lines of a given element will appear strongest when vanishingly small concentrations of the element are caused to emit light
depends to some extent on the type of spectroscopic equipment used,
on the methods of recording and observing the spectrum, and on the
type of excitation used. Tables of the most sensitive spectrum lines
will be found t c differ somewhat, depending on whether they are
based principally on lists compiled by the earlier workers (see § 15.2),
whether they depend on theory,* or whether they involve observations
made with modern equipment and the new ultraviolet and infrared
photographic emulsions ot increased sensitivity.
The published descriptions of spectra are far from complete, even
for comparatively strong lines. Most complex elements have not
been thoroughly studied at wavelengths longer than 6000 A, in the
range 2000-2500 A, and in the vacuum ultraviolet. Such elements
as ruthenium, rhodium, thorium, and uranium have been fairly
thoroughly studied in recent years, but much work remains to be done
on many similar elements.
9.11. The Harrison Automatic Comparator. The measurement
and'reduction of spectrograms can be greatly facilitated by the use of
an automatic comparator. Such a measuring engine,^ in use at the
Massachusetts Institute of Technology since 1938, is capable of
measuring in 120 sec an entire 20-in. plate containing perhaps 2000
spectrum lines, recording on a motion-picture film to seven-figure
precision the wavelengths in angstroms of all lines, and providing on
the same film a density trace of the lines. Such a machine is of great
value when large numbers of measurements are to be made from a
number of plates taken with a given spectrograph. For the measurement of occasional single plates taken on different spectrographs, it is
of value in giving results in terms of distance along the plate, the
8 W. F. Meggers, Jour. Opt. Soc. Am., 31, 39 (1941).
" G . R. Harrison, Jour. Opt. Soc. Am., 25, 169 (1935); G. R. Harrison and J. P.
Molnar, ibid., 30, 343 (1940).
224
IDENTIFICATION OF SPECTRUM LINES
[§9.11
reduction from linear measurement to wavelength being carried out
with a computing machine.
A spectrum plate is clamped to the carriage shown in the rear
center of Fig. 9.13 and is moved by the comparator screw at constant
speed across a beam of light, which throws on the circular white screen
an image of that portion of the plate passing across it at any moment.
The operator can move the plate in either direction by turning the
handle shown in the lower right-hand corner of Fig. 9.13 or by operating an electric drive in the forward or reverse directions.
Fig. 9.13. The Harrison automatic comparator.
An automobile headlamp operated by a storage battery is mounted
below the comparator case. Light from this lamp is projected by
a condenser lens through the plate being measured, on to a microTessar lens, which produces an enlarged image of the plate on the
screen some 6 ft distant.
Figure 9.14 shows the appearance of a portion of a wavelength
record obtained with the automatic engine. The wavelength of a
§9.11]
IDENTIFICATION OF SPECTRUM LINES
225
line which is passing across a scanning slit in the screen is recorded
with a single flash of a stroboscope lamp as the density peak passes.
Condensers discharge through this lamp when a signal is received
from an electrical circuit, indicating that the maximum of the spectrum line is passing across the slit.
Though the relation between distance along the plate and wavelengths of the lines recorded thereon is not linear, the instrument is
arranged to record wavelengths directly in angstroms. By means of
decimal gears, the first two figures of the dispersion of the plate can
be set into the machine. The next four figures are controlled by a
variable-speed unit driven by a cam specially cut for the spectral
region being covered, each cam having been laid out initially in terms
of standard iron spectra obtained with the spectrograph used. The
seventh figure is controlled by a photoelectric device that automatically follows the outer edge of an inked curve. Plate after plate taken
«nii ' l l f liill I
Fig. 9.14. Wavelength record obtained with the automatic comparator.
in the same region of the same spectrograph can be measured with a
single cam and curve. When plates are changed, small variations
may be found, but these can be reduced to any desired value. Variations as great as 0.01 A may be tolerated if necessary, since standard
lines are recorded on all plates and these are measured with the
unknowns. Wavelength values read from the film can then be
changed in the seventh figure in accordance with the error found in
the seventh figure of the standard line readings.
An electron multiplier tube is used behind the slit of the projection
system to measure the light passing through it. The output of the
multiplier is fed to an electrical network which causes the stroboscopic
lamp to flash at the instant when the peak of a spectrum line is symmetrically disposed across the slit.
The maximum picker, as the electrical line-measuring device is
called, does not, as might be supposed, operate on the slope of the
density curve of the line. The graininess of the plate makes such
226
IDENTIFICATION OF SPECTRUM LINES
[§9.12
operation undesirable, and the saturation of the photographic emulsion at high densities flattens the tops of dense lines. Instead, the
line peak is taken as being halfway between two areas of equal density
measured at the points of approximately maximum contrast for the
narrowest line that can be resolved by the spectrograph on which the
plate was exposed.
This automatic comparator has been found to possess several advantages beyond the more than hundredfold gain in speed of measurement and computation which it provides. The comparator screw
does not change temperature during the few seconds required to
measure a plate, and its nut is pushed with uniform speed, so that
an oil film of constant thickness is kept in the screw-nut contact.
Wavelength values arc available instantaneously so that any line
can be identified by inspection if the machine is stopped or is operated
by hand. If the operator prefers, he can run the comparator by
hand, using hand settings, and pressing a button whenever he has set
the center of a line on the fiducial mark.
I t is found that the reproducibility of setting on narrow lines by
the automatic method is from three to five times more precise than
that of eye setting. In a test spectrum, lines of various breadths
were picked by the machine 30 times in each direction with an average
deviatioji of ±0.0004 A, whereas the average internal deviation of
hand-and-eye setting by an experienced operator was ±0.0020 A on
the same machine and plate.
9.12. Limitations of Wavelength Measurement with Diffraction
Gratings. The principal sources of errors in wavelength determinations by means of gratings are as follows (see also § 20.4):
1. The inadequacy of wavelength standards in certain spectral
regions, particularly of standard lines of suitable intensity.
2. The displacement on spectrograms, relative to standard lines,
of lines to be measured.
_ ;
3. Displacements caused by strong neighboring lines or by blends
with impurity lines due to bands.
4. The natural breadths of some lines and the complex structures
of others (§ 20.1).
5. Actual variations of wavelengths, in standards or unknowns,
with excitation conditions.
6_ Incorrect identification of lines.
7. Uncertainties of setting on line maxima.
§9.12]
8.
9.
10.
air of
IDENTIFICATION OF SPECTRUM LINES
227
C o m p a r a t o r errors.
Errors in computation.
Corrections required to express all measured wavelengths in
s t a n d a r d density.
Great gaps still exist in our knowledge of spectrum lines. Accurate
wavelengths of approximately a million atomic a n d ionic lines will
probably be needed in order t o give astronomers, physicists, a n d
chemists all' the material of this sort t h a t they can effectively use.
Fewer t h a n one-third of^these spectrum lines have been measured
and assigned t o p a r e n t atoms.
GENERAL REPEEENCBS
9.1.
9.2.
9.3.
9.4.
9.5.
H. Kayser, Handbuch der Spectroscopie, Vols. 1-6, to 1912; H. Kayser
and H. Konen, ibid., Vols. 7-8, to 1932. Leipzig: S. HirzeL
H. Kayser and R. Ritschl, Tabelle der Hauptlinien der Linienspektren
aller Elemente. Berlin: Springer, 1939.
G. R. Harrison, M.I.T. Wavelength Tables. Cambridge, Mass: Technology Press. New York: John Wiley & Sons, Inc., 1939.
G. R. Harrison, "Compilations of Spectroscopic Data". Jour. App.
Phys.. 10, 760 (1939).
G. R.'Harrison, "The M.I.T. Wavelength Project," Reports on Progress
in Physics, Phys. Soc. Lord., 8, 202 (1941).
CHA^PTER
10
The Origins of Atomic Spectra
ANY THEORY OF THE ORIGIN.OF SPECTRAL LINES must explain
observed spectra quantitatively in terms of models of the atoms
from which the spectra arise. I t is necessary not only to explain the
intensity of emitted or absorbed radiation as a function of wavelength
but also to give a quantitative basis for the understanding of such,
details as the effects on the spectrum of temperature and of electric/
and magnetic fields. Spectroscopic theory has contributed much to
our ideas about the structure and mechanics of atoms. Some
familiarity with the elementary aspects of the theory is therefore
helpful in visualizing the processes involved in the emission and
absorption of radiation.
SPECTRAL SERIES AND ATOMIC ENERGY STATES
The most striking regularity in the spectra of many atoms is the
classification of the spectral lines into series. The frequencies of the
several members of a given series can be represented numerically by- a
simple formula such as
,
"i ="- - jTc^r
^^^-^^ -
Here vi is the frequency of a line in the series, a so-called series
member; voo is the series limit, or series member of highest frequency;;
R and Ci are constants for the entire series; and i is an integer that/
runs from 1 to infinity. As i gets larger and,larger, the spacing of
the lines in the series gets smaller and smaller, until the series finally
converges on the series limit. Such series, of which an example ,is
shown in Fig. 10.1, are frequently found and are called Rydberg series
after the proposer of. Eq. (10.1). R is called the Rydberg constant.
The relationship between diilerent series in the same spectrum is
228
§10.1]
THE ORIGINS OF ATOMIC SPECTRA
229
significant inasmuch as the frequency voo for one series is equal
to the expression R/{i + Ci)- for another series, i being some small
integer. Thus a more general form of Eq. (10.1) is
by which all the series for certain atoms may be represented. Here
the value of i is fixed at some small integer for a particular series, and
j takes on integral values up to very large numbers to give the frequencies for the various members of that series.
3600
4000
I
1
4400' 4800
1
Hs Hy
Series (Violet)(Blue)
limit
I
5200
5600
6000
6400
I
1
I
I
H|a
(Blue-green)
6800A
I
Hex
(Red)
Fig. 10.1. The Balmer series in the spectrum of the hydrogen atom.
10.1. The Hydrogen Atom. The empirical formula for the
various series in hydrogen has a particularly simple form:
-R-(l~l)
(10-3)
The constant Rg is the Rydberg constant for hydrogen. If we set
i = 1 and j = 2, 3, 4 , 5 , . . . , we get the members of the Lyman far
ultraviolet series in hydrogen, beginning with a strong line at 1216 A.
Similarly, i = 2,,and j = 3, 4, 5, 6, . . . , gives the Balmer series
(visible and near ultraviolet) and i = 3, j = 4, 5, 6, 7, . . . , the
Paschen series (near infrared). Thus the frequencies of the members
of all the known series in hydrogen can be expressed as the difference
between two terms, the frequency value of each term being expressed
as
v. = ^
(10.4)
in which n is"an integer.
As we shall se.e in § 10.5, terms such as those given by Eq. (10.4)
can be identified with stationary energy states or levels in atoms, and
spectral lines associated with the transitions between these states.
[§10.1
THE ORIGINS OF ATOMIC SPECTRA
230
The hydrogen terms are plotted on an energy-level diagram in Fig.
10.2, and the transitions between them, which give rise to the various
hydrogen series, are indicated by arrows connecting the initial and
final levels for each transition. Thus each arrow indicates the transiQuanlum
Number.n
n=<o
n=|
n=5
n=4
Lyman
Series
Poschen
Series
Balmer
Series
:
' " ', ^
^
.
n=3
ID
Ertergy
Wn
o
•a-
ro oo
•3-
ID
r
1
g
S5J5
m
O 54 2
RHH
4
n=2
•
<
01
<<<
•s>
g C\J
..-Wn=-R„h
n=i
Fig. 10.2. The energy levels of the hydrogen atom.
tion giving rise to a particular spectral line. Expression (10.4) can
be rewritten in energy units as
- ,'
•fF„-
-Riih
(10.5);
where 1F„ is the energy of a stationary state, the factor —h (Planck's
constant) being necessary to convert frequency units into ergs and to
conform to the useful convention of having the highest energy that
which corresponds to n = » and the lowest that for n = 1 (compare
Fig. 10.2).
§10.1]
,
THE ORIGINS OF ATOMIC SPECTRA
231
The concept of stationary energy states in atoms was first put
forth by Bohr.'^ The mechanics used by Bohr has since been superseded by quantum mechanics, but the expression he obtained for the
energies of the stationary states in the hydrogen atom is also given
by quantum mechanics. It is
In this expression, n is the reduced mass of the central nucleus and
electron [in the hydrogen atom n = nieVip/ime + flip), m^ = electron
mass, mp = proton mass], e is the electronic charge, and Z the atomic
number of the central nucleus (which is 1 for the proton). The
constant RH can thus be expressed* in terms of fundamental constants :
i?H = ^ '
(10.7)
It is interesting to consider the^ physical significance of the energy
levels shown in Fig. 10.2. The lowest level {n = 1) is the so-called
ground state, and in a collection of hydrogen atoms in a gas, it will be
the stationary state in which most of the atoms exist for most of the
time. To raise a hydrogen atom to a higher level, energy has to be
supplied to the atom, the amount of energy required for elevation to
a particular level Wn being known as the excitation potential of that
level. This excitation energy can be supplied by radiation, in which
case the atom absorbs radiation of frequency
' = —I—
It can also be supplied by collision with atoms, ions, or electrons. If
it is supplied by collision with electrons in an electrical discharge,
the electrons must necessarily have a kinetic energy at least as large
as Wn — W\. Since the kinetic energy is imparted to the electrons
by their falling through a potential drop of a certain number of volts,
the kinetic energy of the electrons can be expressed in terms of volts.
1 N. Bohr, mi. Mag., 26, 1 (1913).
* Equation 10.7 gives RH in frequency units {v). It is often expressed in wavenumber units (<r). <7 = v/c. In wave-numbers the value of Rg is 109,677.76 cm""',
t Also sometimes called critical •potential.
232
THE ORIGINS OF ATOMIC SPECTRA
[§10.2
In turn, an excitation potential can be expressed in volts, since this
potential corresponds to the minimum kinetic energy that an electron
must have to cause the transition to the upper state. Electron volts
are frequently used as units for energy levels, because the numerical
values of the levels then lie in the range from 1 to 100 volts. One
volt corresponds to 8066 cm~\ so the wavelength region from 12,500 A
(8000 cm-i) to 125 A (800,000 cm-i) corresponds to the range 1 to
100 volts.
The excitation potential of the level for n = co has a special
significance. It corresponds to the energy required to remove the
electron in the hydrogen atom to an indefinitely large distance from
the proton, in other words, to ionize the atom. This energy, called
the ionization potential, is Wo:. — Wi, or Rah, and has the value
13.59 electron volts. Ionization potentials for other atoms are listed
in Table 10.1.
Several other atomic systems are known whose energy-level scheme
is that of Fig. 10.2 except for a change of scale. These are all ions
consisting of one electron and a heavy atomic nucleus. As one can^
see from Eq. (10.6), the energy levels differ from those of hydrogen
only in the values of Z^ and ix. If the atomic nucleus is that of
helium, n is only slightly different (about 0.05 per cent) from n in
the hydrogen atom, but Z^ has the value 4. If the nucleus is lithium,
Z^ = 9, and so on. Hence the scale in Fig. 10.2 has to be changed
by about a factor of 4 for the He+ ion, 9 for the Li++ ion, and so on.
The various series for these ions lie in the far ultraviolet, as can
readily be verified by calculation from Eq. (10.3). The ionization
potentials are also increased by the approximate factors 4 for the
ionization of He+ to He++, 9 for Li++ to Li+++, and so on (see Table
10.1). The series for ionized helium were known before Bohr's
work on the hydrogen spectrum, but to what atom or ion they were
due was not clear. Bohr's highly exact calculation of the frequencies
of the lines in the He+ series was simultaneously a satisfying solution
to a puzzling problem in the origin of spectra and a highly convincing
corollary to his theory of the hydrogen spectrum.
10.2. Quantum Numbers in Atomic Spectra; The number n in
Eq. (10.6) is a quantum number. It is not the only quantum number,
however, associated with the various energy states; but because it is
the most important insofar as the energy values in the hydrogen atom
are concerned, it is called the principal quantum number. There are
three other quantum numbers associated with each electron in an
§10.3]
THE ORIGINS OF ATOMIC SPECTRA
233
atom, usually designated I, m, and s. Each of these is concerned
with one of two types of angular momentum which every electron in
an atom possesses. The quantum numbers I and m are related to
the orbital angular momentum of the electron in its motion about the
atomic nucleus, whereas the quantum number s is associated with the
spin of the electron. If the atom be compared to a miniature solar
system with the sun and earth in the roles of nucleus and electron,
respectively, the orbital angular momentum of the earth is that associated with its annual trip around the sun, whereas its spin angular
momentum is due to its daily rotation on its own axis.
The quantum numbers I and m take on integral values, just as
does n, but these values are not completely independent of n. The
numerical values that I can have range between n — 1 and 0.
Similarly, m can never be larger than I, but in contrast to I, it can
assume negative values down to — I. The physical interpretation
of I and m pictures Z as a numerical factor expressing the orbital
angular momentum in terms of the quantum mechanical unit of
angular momentum, h/^ir*; m, on the other hand, is the number
related to the orientation of the orbit in space. A positive and
negative pair of values of m correspond to the same orbital orientation, but one (say the positive m) corresponds to a clockwise
motion of the electron in its orbit, and the other to a counterclockwise
motion.
Before we consider the quantum number s, we must discuss the
spectra pf some relatively simple atoms, in which the energy levels
depend not only on n but also on I.
10.3. Series in Atoms with Many Electrons. An atomic spectrum
increases in complexity as the number of electrons directly involved
in its production increases. However, the number of electrons
involved need not be the total number of electrons in the atom. I t
frequently happens that the energy-level scheme for an atom depends
primarily on one or two electrons only, the other electrons maintaining
a constant set of quantum numbers (and therefore energies), regard-
* The reader can readily verify that the units of h (erg-seconds) have the dimensions
of angular momentum, which is linear momentum (dimensions mass X velocity)
multiplied by a moment arm (dimension length). The numerical relationship between
I and orbital angular momentum is
Orbital angular momentum = v/(/ -|- 1) I — 1
\2x/
TABLE
IONIZATION POTENTIALS
Ionization p o t e n t i a l 3
Atomic
number
Symbol
II
I
H
1
2
3
4
5
6
He
Li
Be
B
C
III
—
13.59
24.46
5.36
8.26
11.22
1 54.38
75.26
18.12
25.00
24.27
121.8
153.1
37.75
47.65
9.28 1
7
8
9
10
11
12
N
0
F
Ne
Na
Mg
14.48'
13.55
17.34
21.45
5.12
7.61
29.47
34.93
34.81
40.9
47.06
14.96
47.40
54.87
62.35
63.2
70.72
79.72
13
l-t
15
16
17
IS
Al
Si
P
S
CI
A
5.96
8.12
10.9
10.30
12.95
15.68
18.74*.
16.27
19.56
23.3
23.67
27.76
28.31
33.35
30.0
34.9
39.69
40.75
19
20
21
22
23
24
K
Ca
Sc
Ti
V
Cr
4.32
.6.09
(6.7)
6.81
6.71
6.74
31.66
11.82
(12.8)
(.13.6)
14.1 •
(16.6)
25
26
27
28
29
30
Mn
Fe
Co
Ni
Cu
Zn
31
32
33
3-1.
35
36
Ga
Ge
As
Se
Br
Kr
5.97
8.09
10.5
• 9.70
11.80
13.93
37
38
39
40
41
42
Rb
Sr
4.16
5.67
6.5
6.92
MO
7.35
43
44
45
46
47
48
Tc
Ru
Rh
Pd
Ag
Cd
(4.8)
7.5
7.7
8.3
7.54
8.96
In
Sn
5.76
7.30
49
50
Y
Zr
Cb
•
.
7.41 •
7.83
7.81
7.61
7.68
9.36
,
?
.
234
46.5
50.96
24.61
(27.6)
(29.6) •
(32.1)
15.70
16.16
17.3
18.2
20.2
17.89
(34.0)
(31.7)
(33.8)
(35.9)
(29.5)
- 40.0
20.43
15.86
20.1
21.3
19.1
24.4
(30.6)
(32.0)
28.2
33.9
25.7
31.4
27.36 .
10.98
12.3 /
13.97
(13.5)
(15.2)
(47:0)
(43.0)
20.4
24.00
24.2
(27.0)
(14.6)
(10.4)
(18.0)
ia.8
21.4
17.30
(29.3)
(28.6)
(31.0)
(33.4)
35.9
38.0
18.79
14.53
27.9
30.5
•
10.1
OF THE
ATOMS IN VOLTS*
Ionization potential 5
Atomic
number
, Symbol
51
52
Sb
Te
I
Xe
Cs
Ba
La
Ce
Pr
Nd
Sd
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72 .
73
74 '
75
76
77
78
79
80
81
82
83
, 84
85
86
87
88
89
90
91
92
93
94
95
96
—
-
Sm
Eu
Gd
Tb
Dy
Ho
Er
Tm
' Yb
Lu
Hf
Ta
W
Re
Os
Ir
Pt
Au
Hg
Tl
Pb
Bi
Po
At
Rn
Fr
Ra
Ac
Th
Pa
U
Np
Pu
Am
Cm
. I
8.35
8.96
10.6
12.08
3.87
5.19
'5.6
6.54
5.76
6.31
?
6.55
5.64
6.65
6.74
• 6.82
?
? ,
?
7.06
?
?
?
?
(7.8)
(8.7)
?
8.88
9.18
10.38
6.07
7.38
7.25
(8.21)
(9.59)
10.70
(4.0)
5.25
?
?
?
?
?
?
5
?
II
III
18.8
(19.3)
19.4
(21.1)
23.4
9.95
11.4
14.8
?
?
?
11.4
11.4
?
?
?
?
?
?
?
24.7
30.5
(31.4)
(32)
(35)
?
(20.4)
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
24.1
(20.0)
(24.8)
(26.7)
(28.0)
(30.46)
34.3
29,7
(31.9)
25.42
(27.8)
(30.3)
(30.0)
(32.3)
(34.3)
?
29.4
?
?
?
?
?
?
"•?
*
?
?
(14.0)
(13.2)
(14.6)
(16.0)
(17.4)
19.95
18.65
20.32
14.90
16.6
(18.6)
(18.1)
(20.0)
(21.5)
10.10
?
?
?
?
?
?
?
?
* The values in Table 10.1 have not been critically reviewed; uncertain or estimated
values are given in parentheses. Higher ionization potentials may be found in
General Reference 10.3, p. 200, and in E. Lisitzin, Soc. Scient. Fennica, Comm. Phys.Math., X, No. 1 (1938). The latter reference contains an extensive bibliography of
papers dealing with ionization potentials.
235
236
THE ORIGINS OF ATOMIC SPECTRA
[§10.3
less of changes in the quantum numbers of the one or two electrons.
The reason is that electrons in an atom tend to arrange themselves
in concentric shells about the atomic nucleus according to their •
quantum numbers, all the electrons in a given shell having the same .
value of n. Because each shell has a limited quota of electrons (as
we shall see in § 10.9), electrons in excess of this quota are forced into
shells of larger radius (higher n). A shell whose quota is filled has on
the average a symmetrical arrangement of electronic orbits that corresponds to a spherical distribution of electric charge. The term
shell originates from this fact.
A spectroscopically important situation occurs when the total .
number of electrons in a neutral or in an ionized atom exceeds by 1
the number just needed to fill the quota of a shell. In this case, the
excess electron moves approximately as does the electron in the
hydrogen atom, that is, in the field of a charged central nucleus,
because the spherical charge distribution of the other electrons acts'
approximately as though the whole charge were concentrated at the"
nucleus. Such an electronic arrangement is characteristic of the
atoms of the alkali metals lithium, sodium, potassium, rubidium, and
cesium. The series found in the'spectra of such atoms are similar to
the hydrogen series but follow Eq. (10.2) instead of Eq. (10.3). The
number of series is larger, however, because a distinct series is obtained for each value of the constants Ci and C2 in Eq. (10.2). Moreover, two series can converge on the same wavelength if they share
in common the constant Ci but have different d values.
The various energy levels of the alkali atoms that give rise to these
series correspond to various values of the quantum numbers of the
single outermost electron, the quantum numbers of all the other
electrons in the atom (those in the filled shells) staying fixed. The
numbers i and j in Eq. (10.2) are the values of the principal quantum"
number n for this electron, in strict analogy with the case of the
hydrogen atom. The constants Ci and C2, however, intrude because ,
the energy levels depend on the quantum number I of the outer/
electron as well as on n, in contradistinction to the energy levels in'
hydrogen. The reason can be seen qualitatively with the help of the
orbital picture, if we recall that the quantum number I is- a measure
of the orbital angular momentum of the electron, larger I corresponding to larger angular momentum. A value of I small icompared with
the value of n therefore means that the angular momentum is small
and indicates that the electron's orbit is not circular but a highly
§10.3]
THE ORIGINS OF ATOMIC SPECTRA
237
eccentric ellipse. I n a circular orbit, (maximum I) the outermost
electron moves a t a uniform distance from the inner, filled shells
(Fig. 10.3a), and the situation resembles hydrogen. I n the eccentric
orbits, however, the electron orbit " p e n e t r a t e s " the shells t o a greater
or lesser extent (Fig. 10.3b) a n d during t h e penetration is a t t r a c t e d
much more strongly by the positively charged nucleus. T h e energy
of t h e electronic orbit is thereby lowered, t h e reduction depending on
t h e extent of the penetration, t h a t is, on t h e eccentricity, which in
t u r n is determined by the q u a n t u m n u m b e r I for a given value of n.
(a)
(b)
Fig. 10.3. The penetration of an inner shell by an electron moving in an
eccentric orbit
lit. (a) Circular orbit {I equal to n — 1). (b) Elliptical
orbi (Z
ptical orbit
smaller than n
1).
These effects are illustrated in Fig. 10.4, where the energy levels
of t h e sodium a t o m are plotted t o scale with those of hydrogen. T h e
filled shells in sodium contain 10 electrons, the eleventh or outermost
being forced into orbits with n equal t o 3 or larger. T h e energy
levels* depend on both n and I. If n = 3, 4, 5, . . . , and I = 0, t h e
energies are
W
(n - 1.35)2
(10.8)
* The units of the energies in Eqs. (10.8)-(10.11) will be ergs per atom if RNB is
in reciprocal centimeters. KNB has the value 109,735 cm"'.
238
[§10.3
THE ORIGINS OF ATOMIC SPECTRA
When 1=1,
the expression is
W =
(w -
(10.9)
0.87)2
When / = 2, the expression is
TF =
Rva/lC
(10.10)
{n ~ 0.01)2
When / = 3 or larger (n > 3, since I is always less t h a n n), t h e
expression is
W =
HYDROGEN
~ RysJlC
(10.11)
SODIUM
s-levels
p-levels d-levels
f-levels
!^t in cm
he
-12.186
-27,420
4.-41,510
•109,678
Fig. 10.4. Comparison of tlie energy levels of the hydrogen and sodium atoms.
This last expression agrees exactly with E q . (10.5), which means t h a t
the energies of orbits not eccentric enough t o penetrate the filled
shells ( t h a t is, orbits with large Z) do not depend appreciably on I
b u t only on n.
§10.3]
T H E O R I G I N S OF A T O M I C
SPECTRA
239
The various series arising from these levels in the sodium atom are
summarized in Table 10.2, in which the numerical values of n and I
for the upper and lower levels of each transition are indicated. It
will be notieedthat the lower level for each series has a value of 3 for n.
Accordingly, each series is analogous to the Paschen series in hydrogen, for which the lower level has n = 3, but the different series in
sodium arise from the differences in energy associated with different /
values.* The short wavelength limit of the Paschen series occurs
in the near infrared at 8206 A, which is also approximately the limit
of the fundamental series in sodium, but the other series all have
limits in the violet or ultraviolet. The shift is illustrated in Fig. 10.4,
which shows the drop in energy of the levels of small I value in
comparison with the corresponding levels in hydrogen.
TABLE 10.2
SERIES IN SODIUM
i
Value of /
.Series
Sharp
Principal
Diffuse
Fundamental
Upper
level
Lower
level
0
1
2
3
1
0
1
2
Value of n
A;,
difference
-1
+1
+1
+1
Upper
4.
a,
3,
4,
5,
4,
4,
5,
6,.
,5,
5,.
6,.
Lower
.
..
. .
. .
3
3
3
3
Constants (Eq. 10.2)
C2
(upper)
(lower)
-1.35
-0.87
-0.01
0.00
-0.87
-1.35
-0.87
-0.81
Because all the lines in the sharp series of the alkali metals originate
from upper levels with 1 = 0, spectroscopists have fallen into the
practice of calling all electrons with I = 0, s electrons {s for sharp);
similarly, electrons with I = 1, p electrons (p for principal); with
I = ^, d electrons; with i = 3, / electrons. For higher values of I,
the notation proceeds alphabetically (g for Z = 4, ft for Z = 5, and
so on). I t will be noticed from Table 10.2 that the several series
result from transitions during which I changes by one unit, either
plus or minus, never by zero, two, or three. This result may be
* There are two spectral lines for which this analogy does not hold, however: the
first line in the so-called principal series (transition from n = 3, Z = 1 to re = 3, / = 0)
and the first line in the diffuse series (transition from n = 3, ( = 2 to n = 3, i = 1).
The upper and lower levels for each of these two transitions have the same value of n
and accordingly in the hydrogen atom would have identical energies.
240
THE ORIGINS OF ATOMIC SPECTRA
[§10.4
summarized by giving the selection rule (§ 10.6) for I: In any oneelectron spectrum, the change in I, At, during a transition is always
plus or minus 1.
AZ = d=l
Similarly, one may say
An = 0, 1, 2, 3, . . . . CO
(10.12)
(10.13)
that is, there is no restriction on the amount by which n may change
during a transition. In particular, these selection rules hold for the
other alkali metals, all of which have series similar to those of sodiura
but with different values of n for the ground state.
10.4. Multiplicity in Atomic Spectra. One important feature of
the sodium spectrum has not been mentioned—the double character
of the individual lines. For instance, the first "line" in the principal
series—the famous D lines of sodium—is well known to be a doublet
whose individual wavelengths are 5890 and 5896 A. A composite
"line" such as this one, consisting of several components of related
origin that are usually closely spaced, is known as a multiplet. Multiplicity is the rule rather than the exception inatomic spectra, although
in very complicated spectra it may be difficult to discern in the ric)i
assortment of lines those which are associated as the components of
a particular multiplet.
Multiplicity arises from electron spin. The spin of the electron
produces a magnetic field, and between this ^eld and the electron
moving in its orbit there is electromagnetic interaction that affects
the energy of the electron. This effect depends quantitatively on the
size and direction of the spin magnetic field, both of which properties
are severely restricted. The spin field always has the same size (that
is, the electron spin has a constant angular velocity), and its direction
is limited to one of two specific directions, namely, parallel or antiparallel to some other field that serves as a reference direction. The
other field may be the magnetic field of the electron moving in its.
orbit, the field of another spinning electron, or an external field due
to an electromagnet. Associated with these two directions are the
two possible values of the spin quantum number s, namely, s = + J
for parallel direction and s = — | for antiparallel direction.* '
* The reason.for s = ± j , rather than s = ± 1 , is that the spin angular momentum
. 1
A
• h
IS - X — , not
2
27r
27r
§10.4]
THE ORIGINS OF ATOMIC SPECTRA
241
In the sodium and other alkali atoms the orbital magnetic field of
the single outermost electron provides the reference direction. If the
spin field is parallel to the orbital field, s — -\-\. Thus each energy
level in Fig. 10.4 becomes two levels,* the higher one with s = + | ,
the other with s = — | . The difference between the two levels
n = 3, ^ = 1, s = + i and n = 3, Z = 1, and s = —| in sodium is
about 17 cm'i (or 0.002 volt), and this is the order of magnitude of
multiplet separation in light atoms. In heavy atoms such as mercury
the separations may be a hundred times larger.
Whereas a single electron produces a doubling of the energy levels,
two or more electrons have a more complicated effect. Fortunately,
the fact that the spin quantum number s for each electron can have
only two values simplifies matters. The multiplicity or splitting up
of each level on account of spin depends on the numerical values of the
sum, S, of the spin quantum numbers. The multiplicity is
M = 2S + 1
(10.14)
The sum S, which is always taken to have a positive sign, can have
various values, of course, and the multiplicities M will vary accordingly. For two electrons, S = 0 or 1, depending on whether the two
spins are antiparallel or parallel. Hence atomic spectra that arise
from two electrons (helium and the alkaline-earth metals Be, Mg,
Ca, Sr, and Ba) can have multiplicities of 1 (that is, no splitting of
levels because of spin) or 3 (splitting of each level into three). Spectra depending on three electrons can have S = | or f (hence M = 2
or 4). I t can readily be seen that the values of S will range in steps
of one from 0 to A'^/2 for N electrons if 'N is even, and from \ to
A^/2 if A^ is odd. Hence an atom or ion containing an even number
of electrons will have only odd multiplicities, and one containing an
odd number will have only even multiplicities.
The summing process by which the individual spin quantum
numbers s add up algebraically to give a total spin S for the atom
also is applicable to the individual orbital quantum numbers I, which
add up to a total orbital quantum number L. The addition process,
however, depends on the quantum numbers m, which, as we have
mentioned, indicate the orientation of the orbit in space, and can
have negative values {counterclockwise orbits) as well as positive ones
{clockwise orbits). The sign of L is taken as positive, but its numer. * This doubling does not occur in the set of levels for which I = 0, since there is
no orbital magnetic jield when 1 = 0, and hence no effect of spin on the energy.
242
THE ORIGINS OF ATOMIC SPECTRA
[§10.4
ical values cannot be so succinctly given as can those for tS because
of the greater variety of possibilities of the individual m values. The
maximum possible value of L, however, is the arithmetical sum of the
individual Us. By way of example, we can consider three electrons
with the respective sets of quantum numbers as follows:
rei = 2, Zi = 1, mi = + 1 , 5 i = + 1
na = 3, Z2 = 2, m^ = + 2 , st = + 5
ns = 4, Z3 = 3, -Ms = + 3 , Si = + i
Here
and
S = si + S2 + ss = f
Z = mi + TO2 + "13 = 6
The total energy of a given level depends on S and on L, and also
on the way in which the total spin interacts or "couples" with the
total orbital angular momentum. S and L couple tO give a total
angular momentum for the atom. This angular momentum is called
J, is taken as positive, and lies between a minimum value of the
numerical difference between L and S and a maximum value of the
sum L -\- S.
It is customary for spectroscopists to symbolize a given atomicenergy level or term by a term symbol. The term symbol represents
the numerical values of the different quantities L, S, and J on which
the energy depends, and for which the slection rules are stated. The
term symbol is
in which M, the multiplicity, is written as a number calculated
from Eq. (10.14); that is, M = 2S + 1. J is also written as a
number and varies with the coupling between L and S, but has a
definite value for a particular energy level. L is not written as a
number but as a capital letter. The letter symbol is S if i .= 0,
P if L = 1, i) if i = 2, and so on,' in strict analogy with the previously discussed convention of writing s for I ==' 0, and so on. The ',
term symbol is sometimes written with a principal quantum number n,'
as a prefix when there is no ambiguity as to which electron or elec-^
trons possess that value of n. For example, the two upper levels of
the D lines in sodium (Fig. 1,0.4), which arise from one electron for'
which n = 3,1 = 1, s = i, so that i = 1, S = |, i ¥ = 2, J = 1 - J
and 1 4" |> will have the symbols
S^Pi
and
3 2P,
§10.4]
THE ORIGINS OF ATOMIC SPECTRA
243
T h e lower level of t h e D lines arises from an electron with n = 3,
/ = 0, s = §, so t h a t i = 0, S = i , I f = 2, J = | . Hence this
level will have the term symbol 3 ^S>. T h e two Z)-line transitions
can t h e n be represented briefly by t h e symbols
3 25} - 3 2P}
and
3 ^Sj - 3 ^Pj
I t is customary to write the lower level of the transition first.
Three spectroscopically useful rules about the energy relationships
among multiplet levels can be stated with the help of the L, S, J
q u a n t u m numbers.
>
1. Hund's rule: Of the energy levels associated with a given
electronic configuration (that is, a set of electrons with their n a n d
/ q u a n t u m numbers fixed b u t with m a n d s not fixed), the levels of
highest multiplicity (highest value of S) will have the lowest energy.
Of t h e levels with highest multiplicity, t h a t one with m a x i m u m L
will h a v e t h e lowest energy.
2. Landes interval rule: T h e energy difference between two adjacent levels in a multiplet (same L a n d S, various J values) is proportional t o t h e J value of the higher of t h e two levels.
3. Inversion rule: I n electronic configurations in which a shell of
electrons is less t h a n half full, t h e lowest J value in a multiplet has
t h e lowest energy; when a shell of electrons is more t h a n half full, t h e
highest J value in a multiplet has t h e lowest energy.
These rules are exceedingly useful in understanding t h e spectra
of polyelectronic atoms and ions, b u t their application requires some
information about the q u a n t u m numbers n and I of the various electrons in t h e atom. Fortunately, this information is usually available
with t h e help of the basic rule known as the Pauli principle, which is
discussed in § 10.9.
LINE INTENSITIES IN ATOMIC
SPECTRA
. Although line intensities cannot be determined with the precision
of wavelength measurements, t h e y are an important property of the
spectrum, and the theory must deal with t h e m quantitatively. I t is
beyond t h e scope of such a brief discussion as this to include a n y of
the details of the q u a n t u m mechanics on which the theory is based.
Our discussion will therefore merely summarize some results of
244
THE ORIGINS OF ATOMIC SPECTRA
[§10.5
quantum mechanics with the help of which a simplified treatment of
intensities can be given.
10.5. Some Basic Results of Quantum Mechanics. When the
equations of quantum mechanics are set up and solved for a simple
atom or molecule, they lead to the result that the total energy of the
atom or molecule cannot have every value between minus infinity and
plus infinity but is restricted to a relatively small set of special values.
These are the energy levels, or stationary states, of the atom or
molecule. Ordinarily a stationary state is described in terms of a set
of quantum numbers upon which the energy, W, of the stationary
state depends in some definite algebraic way. However, in those
instances in which the energy expression does not contain all the
quantum numbers, more than one set of quantum numbers will correspond to the same energy levels. For example, if the states of an
atom are described in terms of four quantum numbers n, I, m, s, and
the energy expression has the simple form
W - constant -i- v?-
(10.15)
then all states with the same value of n have the same energy,
regardless of the values of the other quantum numbers.
I t is usually customary to lump together all the various states in
an atom or molecule having the same energy. The number of states
of equal energy so grouped together is termed the degeneracy of the
resultant state. Synonymous terms for degeneracy are statistical
weight and a priori probability. I t is usually denoted by gr,-, where i
is an index number referring to the particular group of states lumped
together.
I t is helpful in considering the energy levels of an atom or molecule,
whether these be known from experiment or from theory, to make a
diagram of them. Such diagrams, of which Figs. 10.2 and 10.4i are
simple examples, are widely used in the systematic, understanding of
spectra of all kinds. If the energy levels depend on a quantum
number n in the way expressed by Eq. (10.5), in which the constant
is given a negative value, then the energy for n ~ '^ will be zero and
that for n = 1 will be lowest (that is, largest negative number). As
was mentioned in § 10.1, the horizontal lines in the figure represent,
the energies of the various stationary states, and the energy values
other than those represented by horizontal lines .are energies that the
atom under consideration cannot possess.
As long as an atom or molecule remains in a given stationary state,
§10.6]
THE ORIGINS OF ATOMIC SPECTRA
245
its energy is fixed and it neither loses nor gains energy from its surroundings. Therefore, if it is to emit or absorb radiation, it must
change to another energy level. Because the various energy levels
differ from one another by fixed amounts, the amounts of radiant
energy which the atom or molecule can gain or lose are fixed in size.
It is a result of quantum mechanics that there is a definite relationship
between the frequency of the emitted or absorbed radiation and the
change in energy of the atom or molecule. As was mentioned in
§ 10.1, this relation is
J'1,2 =
T
(10.16)
where vi^i is the frequency of the radiation, h is Planck's constant
(6.6 X 10~^^ erg-seconds), and Wx and TF2 are, respectively, the
energy levels of the atom before and after the atom has emitted (or
absorbed) light. If TFi is smaller than W^, the atom has gained
energy, that is, has absorbed radiation. If Wi is larger than W2,
the atom has lost energy in the process, that is, has emitted radiation.
10.6. Selection Rules and Intensities of Spectral Lines. The
change from one energy level td another is termed a transition. It is
clear that the jrequeficy of radiation emitted or absorbed during a
transition depends not on the properties of the system before or after
the transition but only on the energy difference between the two
stationary states. The intensity with which radiation is emitted or
absorbed, on the other hand, is very much dependent on the nature
of the initial and final states. Suppose we have a container full of
similar atoms all in a stationary state Wi. These atoms can conceivably radiate energy of a particular frequency j'1,2 if there is a
lower energy level W2 such that
Vl,i
h
(10.17)
The rate at which radiation of frequency j'1,2 is emitted depends on
the number of atoms making the transition from Wi to W2 per second.
In fact, the intensity of radiation of frequency j'1,2 will be
/ = constant X Ni^ihvi,^
(10.18)
where iVi_>2 is the number of atoms going from state Wi to state Wi
per second, and hvi,2 is the energy given out by each transition. To
understand what Ni-^2 depends on, let us consider the behavior of
246
THE ORIGINS OF ATOMIC SPECTRA
[§10.6
the atoms in our container in state TFi. They can make transitions
to W2, of course, but it is also likely that there are a good many other
stationary states W3, Wi, Wi, and so on, to which transitions are
possible. Since these states have different energies, transitions to
them will not result in radiation of frequency J'1,2; moreover, if an
atom makes a transition from Wi to W3, for example, it will no longer
be able to make the transition W\ to W2. Hence the number A^i_>2
will depend on the relative chance that an atom in state Wi will go
to state Wi in preference to states Wz, Wi, Wi. . . . This relative
chance is termed the transition probability for the transition Wi-^Wi
and is found to depend on the characteristics of the two stationary
states.
The calculation of transition probabilities by quantum-mechanical
methods leads to the interesting general result that for a given atomic
system there are a great many transitions for which the transition
probability is zero; that is, the two states concerned do not combine
to emit or absorb radiation. Sjpce the characteristics of the two
states are determined by their quantum numbers, it is possible to
express transition probabilities in terms of quantum numbers, and in
particular, to state what the relationships are between the quantum
numbers of two states that do not combine. A generalization of
transition probabilities expressed in terms of the quantum numbers
is termed a selection rule, because the rule_ enables one to make a
selection of the pairs of states which combine (the so-called allowed
transitions), and of the pairs which do not combine {forbidden tr&nsitions) from the various pairs. Selection rules usually take the form
of a statement of the changes in quantum numbers associated with
allowed transitions. In terms of the 1 quantum numbers L, S, J
(§ 10.4), the selection rules* governing allowed transitions are
Ai
AS
AJ
A/
=
=
=
=
0, ± 1
0
=t 1 (all J values)
0 (all J values except J = 0)
(10.19)
* The rules are valid only for atoms in which the L,S coupling (Russell-Saunders
coupling) holds. L,S coupling is a good approximation for atoms of low atomic number •
but is not so good for heavy atoms. For heavy atoms, in which the so-called j,j
coupling or intermediate types of coupling obtain, the rules hold only roughly. Especially the selection rule AS = 0 is no longer strictly valid, and many "intercombination" lines, corresponding to, transitions in \vhich AS = 1 or more, are observed. The
well-known "resonance line" in mercury at 2537 A is an example. Here AS = 1.
§10.6]
THE ORIGINS OF- ATOMIC SPECTRA
247
The symbol A i stands for the difference between the value of L in
the initial state and that in the final state. AS and AJ have analogous
significance.
i
An interesting situation arises when a given energy state in an atom
or molecule may not combine with the lowest energy state (the socalled ground state) in the atom, or with aiiy of the intermediate
energy states between it and the ground state. If, in such a case, the
atom by some means gets into the given energy state, it will be unable
to drop to a lower state by the emission of radiation and hence will
remain in the upper state for an indefinite length of time. It may
ultimately change from the state by the absorption of radiation, or
by a collision with another atom, but transitions by these mechanisms
are usually slow compared with those which emit radiation. When
an atom remains in an excited stationary state for a long time, that
state is said to have a long "lifetime" and is called a metastahle state.
The lifetime of an atomic state generally is of the order of magnitude
of 10"^ sec, but for metastable states it may be many orders of
magnitude larger.
Evidently, from the foregoing remarks, it is desirable to have some
knowledge of what energy states are occupied by the atoms or
molecules in a system whose spectra we wish to understand. Atoms
and molecules may be raised to upper states {excited states) from the
ground state by various means, including electrical discharge and
heating. If the excitation occurs by heating and if thermal equilibrium among the atoms or molecules is approximately established, the
number of atoms or molecules in a stationary state of energy W^ compared with the number in the ground state W\ is given by the expression
in which Ni and A^2 are the numbers of atoms in states 1 and 2,
gi and g2 the respective degeneracies of states 1 and 2, k is Boltzmann's constant (1.371 X 10~^* erg per degree per molecule), and
T is the absolute temperature. The right-hand side of Eq. (10.20)
is sometimes referred to as the Boltzmann factor for state 2, when
state 1 is the ground state.
The same quantum-mechanical calculations that lead to the selection rules also provide certain generalizations about intensity relationships within multiplets. These intensity rules are as follows:
248
THE ORIGINS OF ATOMIC SPECTRA
[§10.6
A. The sum rules:
1. The sum of the intensities of all lines of a multiplet that begin
their transitions from the same energy level is proportional to the
degeneracy gi of that level. In terms of J, Qi = 2J + 1.
2. The sum of the intensities of all lines of a multiplet that end
their transitions on the same energy level is proportional to the
degeneracy gi of that level ((/,• = 2J + 1).
B. The rules for relative intensities within a multiplet (all L, S, J
values are those of the final state):
For AL = + 1 , AJ = + 1 :
J
{L + J + S + 1)(L + J + S){L + J - S){L + J - S - I)
Irel -
J
For AL = + 1 , AJ = 0:
-{L + J+S+l){L+J-S){L-J+S){L-J-S-l){ZJ
+ l)
J{J + 1)
ForAL = + 1 , AJ = - 1 :
_ (L-J+S-l)(X-J+S)(X-J-g-l)(£-J-S-2)
^'"^' ~
(J + 1)
ForAi = 0, AJ = + 1 :
^
- (L+J+S+l)(i+J-S)(L'-J + S+l)(L-J-S)
T-rel —
irel -
For AL = 0, AJ
J
[LjL +
'"' ~
For AL = 0, AJ
_ - (X + J
"' ^
J
= 0:
1) + J(J + 1) - SjS + 1)]^ (2J + 1)
J {J + 1)
= - 1 :
>
+S +2)(L - J +S){L + J - g + ! ) ( £ - J -S
(J + 1)
-1)
For AX = - 1 , AJ = + 1 , 0, - 1 : The formulas for AL = -1
can be obtained from those for AL = + 1 by. reversing the sign of A J
and using the L, J, S values of the initial rather than the final state.
A factor involving the frequencies of the multiplet lines and the temperature of the source has been omitted from the sum rules and
relative-intensity formulas. Usually the relative values of .this factor
for the different lines are not far from unity.
,
As an example of the application of these rules, there may be cited
the studies of Harrison and collaborators.^
2 G. R. Harrison and H. Engwicht, Jour. Opt. Soc. Am., 18, 287 (1929).
§10.7]
THE ORIGINS OF ATOMIC SPECTRA
249
10.7. The Effect of External Influences on Atomic Spectra. The
energy-level scheme of an atom determines the radiation frequencies
which the atom can emit, and the selection rules control the transitions
between these levels. Before one can say, however, what the spectrum of an assemblage of atoms will look like, the number of atoms
populating each energy level must be known. The populations of
the energy levels are quite sensitive to the environment of the assemblage, and therefore this environment is important for the appearance
of the spectrum.
Among the most important factors in the environment are temperature and pressure. When thermal equilibrium prevails in an assemblage of atoms, the populations of the various energy levels decrease
exponentially with the height of the levels above the ground level and
increase exponentially with the temperature [see Eq. (10.20)]. This
statement means that as temperature goes higher, the populations of
other energy levels than the ground level increase, and therefore the
number of atoms that can make transitions to lower levels increases.
Consequently, the intensities of the spectral lines emitted by an
assemblage of atoms such as those in a star or in a vapor heated by
a furnace are strongly dependent on temperature and in turn are a
clue to the temperature of the emitter.
When the excitation to upper levels is accomplished by thermal
means, the populations of the higher levels are in general much
smaller than those of the lower ones. Therefore the intensities of
spectral lines of short wavelength, which must come from high levels,
are also small. This statement does not mean, however, that all lines
of long wavelength are more intensely emitted than lines of short
wavelength, because a line of long wavelength can result from a
transition between two closely spaced high levels. This is an example
of the general rule that the temperature dependence of the intensity
of a spectral line is given by the temperature dependence of the
population of the energy level from which the radiating transition
starts. I t should be noted that while the intensities of lines that arise
from high starting levels are small, the temperature variation of this
intensity, percentagewise, is very large [see Eq. (10.20)].
An interesting situation arises when radiation is emitted by an
assemblage of atoms that is divided into two regions, one at a higher
and the other at a lower temperature. Such a situation may exist
under proper circumstances in a gas discharge tube, in which the
atoms at the center of the tube are at high temperatures while the
250
THE ORIGINS OF ATOMIC SPECTRA
[§10.7
ones on the periphery (near the walls) are at lower temperatures.
Those spectral lines which originate at the higher energy levels will
be emitted by the high-temperature part of the assemblage with
relatively higher intensities than by the low-temperature part; but
before this radiation can leave the tube, it must pass through the
region of the low-temperature atoms. Here the lower energy levels,
which are the final levels of the transitions giving rise to radiatton
from the high-temperature atoms, act as starting levels in the lawtemperature atoms, and the radiation is absorbed. The ground
level is particularly important in this process, since its population
is always large.
The net result of the absorption of some of the radiation before it
leaves the discharge tube is that certain lines, especially those involving the ground level, are markedly diminished in intensity. The
diminution may be so great that the lines actually are weaker than
the continuous background present in every high-temperature spectrum and hence appear as dark lines against a Jjright background.
Such lines are said to be reversed.* The outstanding examples are
the famous Fraunhofer lines in the sun's spectrum, which arise from
the absorption by atoms in the relatively low-temperature outer
envelope of the sun's atmosphere. An example frequently met in
the laboratory is the reversal of the 2537 A line in the high-pressure
mercury arc. Sometimes a bright line is seen with a dark line at its
center; this is called self-reversal. If the line is merely changed in
shape on passing through the assemblage, it is said to be self-absorbed.
Another, but minor, influence of temperature on line spectra is
the broadening of lines because of the Doppler effect. An atom
moving away from the slit of a spectrograph along its optic axis emits
radiation of frequency v — Av, whereas an atom moving toward the i
slit with like velocity emits radiation of frequency v -f- Av. These
frequencies are measured with respect to the frequency v emitted by
a stationary atom, and the value of Av is
Av = V -.'
where v is the velocity of the moving atom and c that of light. Since
the average velocity of the atoms will increase with temperature, the
* This kind of reversal of spectral lines is to be distinguished from the "reversal" of
very intense lines ion the photographic plate. The latter, a purely photographic
phenomenon s^ssociated with overexposure, is independent of the source of radiation
(Compare § 7.2, page 145).
§10.8]
THE ORIGINS OF ATOMIC SPECTRA
251
average Av will also increase; that is, the observed spectral line will
broaden. The magnitude of the effect is indicated by the increase in
the half width* of one of the sodium D lines at 5890 A that results
from a temperature change from 300 to 3000°K. The half width
changes from 6.5 X 108-secT\ or 0.0075 A, to 20.5 X 10^ sec~S or
0.024 A.
The influence of pressure on spectral lines arises from the increase
in the number of atoms per unit volume which occurs when pressure
is increased. The intensity of both line and continuous spectra may
increase with elevation of pressure simply from the increase in the
number of radiating atoms. If self-absorption occurs, the elevation
of pressure may reduce the intensity of certain lines. The chief
effect of more atoms per unit volume, however, is a higher number of
collisions per second between atoms. The exact frequency of a
spectral line depends on the energy difference between the initial and
final levels. Since these levels are shifted slightly during a collision,
the frequency of the line shifts. The effect of increased collisions is
twofold—a widening of the spectral line, called pressure broadening,
and a shift to lower frequencies. Pressure broadening is of the
greater practical importance, because the line width limits the accuracy with which its wavelength can be measured, as well as the
resolution of closely spaced lines. The order of magnitude of pressure
broadening in atomic spectra is about one-third of a wave number per
atmosphere.
10.8. The Stark and Zeeman Effects. The most fruitful kind of
external influence that can be exerted on an atomic spectrum is that
of an electric or magnetic field. An electric field affects the spectrum
because it adds an additional electric force to that already existing
between the atomic nucleus and the electrons. The result is a
"splitting" of those atomic energy levels which are degenerate, and a
consequent splitting of spectral lines. This splitting, called the
Starh effect, was not discovered until 1913 because of the tremendously
strong electric fields required to produce it to an observable extent.
Field strengths in excess of 100,000 volts per centimeter are required.
Field strengths of this order, however, are quite weak compared to
the fields existing in atoms and molecules. For example, the field of
* The term half width means the width of a spectral line measured from a point
on one side of the line where the intensity is half that of the peak to the corresponding
point on the other side of the line (compare Fig. 6.1).
§10.9]
THE ORIGINS OF ATOMIC SPECTRA
253
mystifying features such as the 14 rare earths, and has enabled the
chemist to understand the electronic basis of chemical valence. The
keystone to this explanation is the Pauli exclusion principle, which
will be discussed briefly in terms of its application to the periodic
table.
10.9. The Pauli Exclusion Principle and the Periodic Table. The
Pauli principle was foreshadowed by Bohr's building-up principle^ in
1921. It was announced by Pauli^ on empirical grounds in 1925 and
Fig. 10.5b. The Zeeman efiect in rhodium, after Molnar and Hitchcock."
T h e top spectrum shows near ultraviolet lines in the normal spectrum of rhodium
in the absence of a magnetic field. The middle gives the perpendicularly polarized and the bottom the parallel-polarized Zeeman effect produced by a magnetic
field of 90,500 oersteds.
has since been demonstrated theoretically to be a consequence of the
fundamentals of quantum mechanics. It states that no two electrons
in the same atom can have the same four quantum numbers. Since
the quantum numbers themselves are subject to severe restrictions
{n = integer, I = integer smaller than n, m = positive or negative in* N-. Bohr, Zeitschr.f. Physih, 9, 1 (1922). "
*
6 W. Pauli, Zeitschr.f. Physik, 31, 765 (1925).
8 J. P. Molnar and W. J. Hitchcock, Jour. Opt. Soc. Am. 30, 523 (1940).
254
THE ORIGINS OF ATOMIC SPECTRA
[§10.9
teger equal to or smaller than I, and s = ± | ) , the Pauli principle and ^
the energy rules pretty definitely dictate the electronic structure of
an atom or ion.
The electrons, in arranging themselves around the atomic nucleus,
will try to occupy those energy levels which are lowest; that is, all
electrons seek to have a principal quantum number n = \. The
Pauli principle, however, excludes all but two electrons from the
level n = \, since of necessity Z = 0 and m = 0 when n = 1. Thus
the electrons with n = 1 can differ only in their spins, one having •
!v s = + I and the other s = — | . The other electrons in the atom
must have n = 2 or larger A collection of electrons with the same
principal quantum number, such as the pair of electrons with n = 1,
is called a shell.* A shell is said to he filled or comfleted if the number
of electrons therein is the maximum permitted by the Pauli principle.
When n — \, this maximum is 2, but for larger n values the maxima
are larger. When n = 2, Z = 0 or 1, and therefore m = 0 or — 1, 0, + 1 .
Consequently, the shell for n = 2 can have eight electrons, one pair
each for Z = 0, m = 0; Z = 1, TO = — 1; Z = 1, m = 0; Z = 1,
TO = + 1 .
In shells where several values of Z are possible, the question of
which Z value has lowest energy arises. The answer, as we have seen
in the case of sodium atom (I?ig. 10.4), is that lower I values (eccentric
orbits) correspond to lower energies. Therefore in an unfilled shell,
the electrons will prefer the levels of lowest Z.
Another important energy question arises when n is larger than 3.
As one can see by referring to Fig. 10.3, the space between levels of
different n is very small when n is more than 3. For these levels, the
value of Z is of comparable importance to that of n in determining the
energy, and one cannot say offhand whether the level n = 4, Z = 0,'
will have lower or higher energy than the level n = 3, Z = 2. For
the sodium atom, the energy-level.diagram indicates clearly that the
level n = 4, Z = 0 is the lower. As a result, electrons will enter this
level in preference to the level n = 3, Z = 2; in'other words, electrons
will enter the shell M = 4 before the shell n = 3 has been completed.'
' Shells are frequently denoted by capital letters:
n
1
2
3
4
5
6
,.
Symbol
K
L
M
N
0
P
Q
§10.9]
THE ORIGINS OF ATOMIC SPECTRA
255
This preference is illustrated in the periodic table by potassium, the
atom with 19 electrons. Of these 19, two are in the shell n = 1,
eight in shell n = 2, and eight in the shell n = 3 (2 with n = 3,
1 = 0, and 6 with n = S, I = 1). There is still room for 10 electrons
in the shell n = 3 (n = 3, Z =" 2), but because the energy of the level
n = 4, Z = 0 is lower, the nineteenth electron enters this level. As
a result, potassium has an electronic structure similar to sodium,
with a single outermost s electron. Likewise, in the atom with
20 electrons (calcium), the nineteenth and twentieth electrons both
enter the level n = 4, Z = 0. However, in the atom with 21 electrons
(scandium), the twenty-first electron must choose between n = 4,
1=1 and n = 3, Z = 2, because the level n = 4, Z = 0 will hold only
two electrons. As one can see from Fig. 10.4, the level n = 4, Z = 1
is higher than n = 3, Z = 2. Hence the twenty-first (and the next
nine electrons thereafter) enters the levels n = 3, Z = 2.
Thus from a combination of the Pauli principle with the energy
characteristics of electrons as determined by their quantum numbers,
the ^ound-state (lowest-energy) configuration of an atom or ion of
any number of electrons can be understood. For chemists this is of
great importance because of the relationship between the electronic
structure of atoms and their chemical properties. The periodic table
of the elements, for example, which was first determined from chemical
properties, is completely explained on the basis of the periodicity in
electronic configurations of the atoms. In Table 10.3 are summarized
the electronic structures of the chemical elements. For detailed discussion of these structures, one should refer to any standard text on
atomic spectra, of which several are listed in General References 10.1
to 10.6.
For the spectroscopist, who is interested in oZZ the electronic energy
levels, an understanding of the structure of polyelectronic atoms is
likewise of paramount importance. Spectroscopic principles that
follow from the previous reasoning are given herewith:
Similarity of spectra of isoelectronic atoms and ions. The operation
of the Pauli principle is not dependent on the size of the positive
charge on the nucleus of an atom. The primary effect of a change
in this charge is a change in the scale of the atomic-energy levels.
We indicated this in our discussion (§10.1) of the spectrum of the
helium ion He+, the energy levels of which were those of the hydrogen
atom, except for a scale factor of 4 associated with the double charge
on the helium nucleus. As a result of the Pauli principle, one can
256
T H E O R I G I N S OF A T O M I C
SPECTRA
[§10.9
TABLE lO.S
ELECTRONIC STRUCTUKES OP THE ATOMS*
Element
N u m b e r of
electrons
H
1 He
1
2
Li
Be
B
C
iN
0
P
Nc
3
4
5
0
7
8
9
10
Na
Mg
Al
Si
P
S
CI
A
11
12
13
14
15
IG
17
13
K
Ca
Sc
Ti
V
Cr
Mn
19
20
21
22
23
24
25
Inner
electrons
Outermost
Is
(ls)=
T!
1•S
3
w
=5
•3
§
^
•
I
-1
a
1
<
electrons
Ground term
symbol
'S,/2
'So
2s
(2«)2
(2s)22p
(2s)H2p)^
(2s)H2p)=
(2s)2(2p)«
(2s)2(2p)=
(2s)K2p)«
^S,/2
'So
^Pl/2
'Po
^83/2
V2
3s
(3s)2
(3s)23p
^Sl/2
'So
^Pl/2
'Po
^S3/2
V2
'Pzl2
'So
(ssnspr
(3s)2(3p)'
(3s)'(3p)^
(3s)H3p)=
(3s)H3j))''
4s
(4s)2
3<Z(4s)2
(3d)2(4s)^
(3d)=(4s)2
{My is
(Sd^iisY
^^3/2
'So
^Sl/2
'So
=D3/2
'f2
^i^3/2
'S3
"85/2 '
* Column 1 gives the chemical symbol. Column 2 gives the number of electrons
surrounding the nucleus (that is, the atomic number) of each chemical element. Cok.
umn 3 shows the number and distribution of inner electrons, which are those electroifs
unchanged during chemical reaction. For brevity, these, are expressed in terms of the
electronic configuration of the rare-gas atoms. The inner electrons in Li, for example,
are the two Is electrons, which have the same arrangement that one finds in the;JHe
atom. In Na, the inner electrons are the two Is, the two 2s and the six 2j) electrons,
which is just the arrangement in Ne, and so on. Co'uuin 4 lists the remainder of the
electrons. The "exponent" of the bracketed symbojs indicates the number of/electrons of the type within the brackets. For example, (Is)^ means "two Is electrons."
The ground term symbols in column 5 are explained in the text (§ 10.4). Although
these symbols can be worked out from the electronic configuration in column 4, it
must be remembered that the reverse procedure is the one actually followed. The
ground term symbol is determined by an analysis of the spectrum, and from it the
electronic configuration is inferred. When the spectroscopic evidence for the ground
term symbol is not clear-cut, the symbol is marked " ? " .
§10.9]
THE ORIGINS O F A T O M I C
TABLE
257
SPECTRA
lO.S—Continved
ELECTEONTC STBUCTUBES OF THE ATOMS
Element
Number of
electrons •
Fe
Co
Ni
Cu
Zn
Ga
Ge
As
Se
Br
Kr
26
27
28
29
30
31
32
33
34
35
30
Bb
St
Y
Zr
Cb
Mo
Tc
Ru
Rh
Pd
Ag
Cd
In
Sn
Sb
Te
I
Xe
37
38
39
40
.41
42
43
44
45
46
47
48
49
50
51
52
53
54
Cs
Ba
La
55
50
57
Ce
Pr
Nd
58
59
60
61
62
03
64
65
C6
67
Sm
Eu
Gd
Tb
Dy
Ho
Inner
electrons
Outermost electrons
'{3d)\-isy
(3d)'(4«)2
w
3J
n
1
<
•
m
1>
CI
o
^
M
«
(34«(4«)2
(SdyHs
(Mynsy
isdyiis^ip
Ground term
symbol
'Di
'F,/2
Wt
'S,/2
'So
^Pl/2
(3d)i«(4s)2(4j')'
=Po
(3dyo(,isy{ipy
{3dyo(,isy(4py
{sd)io(4sy(ipy
^S3/2
Isdyiisyiipy
'So
5s
{5sy
4rf(5s)2
(4rf)2(5s)2
(idySs
=S,/2
(uyss
'S3
^P2
'Pzl-i
'So
^/)3/2
'F2
'D,/2
{idyss
(idyss
{4d)'>5s
(idyo
(idySs
«Z)<,/2
(4rf)i»(5«)2
'So
(4(i)i»(5s)25p
(4d)«(5.?)2(5p)2
^Pl/2
(idyo{5syl5py
^S3/2
(4d)i»(5.?)»(5p)4
'P2
^P3/2
H<{)">(5sy(5py
(•idy''{5sy{5py
?
'F,
'F,l2
'So
^Sl/2
=Po
'So
§^
ds
^S,/2
(6«)2
X «
5d{6sy
'So
'D,i,
(4/)6(0^)2
'Hi
'hit
'h
'Ihl2
'F„
O
(4/)'(6s)2
»S7/2
a;'
(4/)'5d(6*)2
«A
(4/)9(6«)2
(4/)i«(6s)2
«i/l6/2
?
^/s
?
(4/)"(6«)'
*/i./2
?
(4/)2(6*)2
en
"%
(4/)H0«)2'
(ify{6sy
(4f)H6s)2
1
?
?
258
[§10.9
THE ORIGINS OP ATOMIC SPECTRA
TABLE 10.3—Continued
EuscTBONic STR'JCTURES OF THE ATOMS
Element
Er
Tin-
Yb
Lu
Hf
Ta
W
Re
Os
Ir
Pt
Au
Hg
Tl
Pb
Bi
Po
At
Rn
Fr
Ra
Ac
Th
Pa
U
Np
Pu
Am
Cm
N u m b e r of
electrons
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
80
87
88
89
90
91
92
93 ^
94 '
95
96
Inner
electrons
O u t e r m o s t electrons
Ground term .
symbol
d
{ifYKdsy
'Hs
0-2
(4/)"(6«)2
^/^7/2
{if)'Kesy
'So
--D3/2
^ ^
^ ^
W3
"^
»2
C3
O
at
(4/)"5rf(6«)2
{4f)"(Sd}KSs)'
(4/)"(5d)'(6s)'
(4/)«(5rf)^(6s)2
(4/)"(5rf)5(6s)2
(4/)"(5rf)''(6«)«
(4f)"(5rf)'(6«)2
(4/)"(5rf)96«
(4/)"(5d)'»0s
(4/)"(5rf)'»(6.)2
(4/)»(5(Z)"'(6«)^6p
(4/)"(5rf)"'(0«)2(6p)2
(4/)"(5rf)"'(6«)KM'
(4/)"(5rf)"'(6«)2(6p)<
(4/)"(5d)i"(6«)K6p)5
(4/)W(5(;)"'(6«)2(6p)«
m
7s
{7sy
6d{7sy
{6dy{7sy
VI
{5fy(id{7sy
C3
o
(5/)56d(7s)
(o/)*6d(7«)2
(5/)^6(Z(7s)2
(5j-)«6d(7^)2
(5/)'ed(7^)2
03
?
^/2
'F^„
'D,
'S,i,
'D,
'F,i,
'D,
^Sl/2
'So
^Pl/2
^Po
.
's,h
'Pi
^Pzl2
'So
^S,/2
'So
'-D3/2
'F2
^A'il/2
'ie
"iVls/.;.
'A'4
»//3/2
"Dj
?
?
?
?
?
make the general statement that the energy levels-of^ two atoms or ions
with the same number of electrons (which are called isoelectronic with,
one another) are closely similar except for a'scale factor. The closer
the two nuclear charges, of course, the closer the similarity and the
smaller the scale factor.
y'
This similarity of isoelectronic spectra was first recognized empirically and was enunciated as the displacement law by Kossel, and
Sommerfeld': The spectrum of an ionized atom of net positive charge
' W. Kossel and A. Sommerfeld, Verh. der deutsch. phys. Ges., 21, 240 (1919).
§10.9]
259
THE ORIGINS OF ATOMIC SPECTRA
C4
-
•
o<
o :
^^ ;
rH
GO
H
to
=+= +
o»
r^
E
o+„t
^
CO
G*
+ +
fa ° *-
rH
»^
^
CO
O
1^
i>
o
o
&*
^ fa u
^
:0
GO
l5
,^10 +
O M ,«
H
IZ
H
rH
+
^ + +
.-+ t
GO
o<
rH
^
GO
^ ^ H
-H
CO
+ +
el
-3
a;
N
'3 3
is ^ 3
3 M 3
<U 3 O
'-2
o
iO
•*
(5<
t*
»10
260
THE ORIGINS OF ATOMIC SPECTRA
[§10.9
c has a structure closely analogous t o t h a t of the neutral a t o m c
places ahead of t h e ionized a t o m in t h e periodic table.
The alternation of multiplicities.
According to t h e displacement
law, a neutral a t o m and t h e singly ionized a t o m just following it in
t h e periodic table should show t h e same kind of spectrum, including
the same multiplicities.
As was mentioned in § 10.4, atoms and ions
with even numbers of electrons show odd multiplicities, a n d vice
versa, a n d so isoelectroiiic atoms a n d ions should be expected t o show
i h e same multiplicities. Two neutral atoms adjacent in the periodic
"^ table differ b y one electron, a n d accordingly show respectively even and
odd, or odd and even, multiplicities.
This p a t t e r n is illustrated in
T a b l e 10.4, which combines t h e displacement law a n d t h e law of
alternation of multiplicities for t h e first long row in t h e periodic t a b l e .
T h e maximum multiplicity of 8 in M n , Fe"*", a n d Co++ is calculable
from t h e Pauli principle: as t h e shell with n = 3 fills u p beyond t h e
halfway point, more a n d more electrons have to pair u p their spins,
with consequent decrease in t h e possibilities for multiplicity.
GENEKAL REFERENCES
10.1.
10.2.
10.3.
10.4.
10.5.
10.0.
R. F. Backer and S. Goudsmit, Atomic Energy States. New York:
McGraw-Hill Book Company, Inc., 1932.
E. U. Condon and G. H.'Shortley, Theory of Atomic Spectra. New
York: The Macmillan Company, 1935.
G. Herzberg, Atomic Spectra and Atomic Structure. New York:
Dover Publications, Inc., 1944.
F. K. Richtmyer and E. H. Kennard, Introduction to Modern Physics.
New York: McGraw-Hill Book Company, Inc., 1947. .
Y.HojsinskY, Introductory Quantum Mechanics. New York: PrenticeHall, Inc., 1938.
H. White, Introduction to Atomic Spectra. New York: McGraw-Hill,
Book Company, Inc., 1934.
C H A P T E R 11
Molecular Spectra and Molecular Structure
ATOMIC SPECTRA ARISE FROM ELECTRONS IN MOTION about a single
heavy positively charged nucleus. The motion of the atom as a
whole, which is essentially unaccelerated straight-line motion,
influences atomic spectra only in a relatively minor way through the
Doppler effect. In a molecule, on the other hand, the electrons move
about two or more heavy positively charged nuclei, and the nuclei
themselves not only move together in a straight-line translation but
also rotate and vibrate periodically about their center of gravity.
These latter motions, being periodic, exhibit spectroscopic activity
on their own account, and in addition influence the electronic spectra
of the molecule. This influence, which is missing from atomic spectra, results in a highly complicated fine structure. The details of the
electronic spectra of molecules, as well as those spectra concerned only
with molecular vibration and rotation, will be considered briefly.
ENERGY LEVELS IN MOLECULES
A fundamental statement about energy levels in molecules is that
the molecular energy, Wjaoh is the sum of its translational, rotational,
vibrational, and electronic energies:
••
•
IF^oI =
IFt,ans +
IF,ot +
IFvlb +
IFe,
(11.1)
The translational energy, ^Ftrans> has no significant effect on molecular
spectra and will henceforth be disregarded.
11.1. Rotational Energy Levels in Molecules. The rotational
energy, Wnti of a molecule is directly related to the angular velocity
with which the molecule rotates. It is a result of quantum mechanics
that a molecule may not rotate with any arbitrary velocity but only
with certain restricted velocities, and therefore that only certain restricted values of W^ot are possible. The set of possible IFrot values
261
262
MOLECULAR S P E C T R A A N D
STRUCTURE
[§11-1
may be schematically represented by an energy-level diagram
analogous to the diagrams previously given for atomic energy levels.
The appearance of the rotational energy-level diagram is strongly
dependent on the geometrical form of the molecule, because the rotational energy depends on the moments of inertia of the molecule,
which in turn depend on the molecular geometry. Every molecule
except those whose atoms all lie on a straight line has three intersecting, mutually perpendicular axes, passing through the molecular
center of gravity, about which the molecule can rotate and with
respect to which the three moments of inertia are calculated (see
-c-axis
Cannot rotate
obout c-axis
Fig. U . l . How molecules rotate,
Spherical top (uranium hexafluoride).
A symmetrical top (water)^
(a) ,Linear molecule (acetylene),
(c) Symmetrical top (ammonia),
(b)
(d)
Fig. 11.1). Eachijiroment is associated with one of the three axes
and is determined numerically by multiplying the jnass of each atom
by the square of its distance from the axis, and adding up these
products for all the atoms in the molecule. Symbolically, ,
Ja =
^miTu
(11.2)
Here la is the moment of inertia about axis a, and m^ and r,a are,
respectively, the mass and the distance from a of atom i. The sum
§11.1]
MOLECULAR SPECTRA AND STRUCTURE
263
includes all A^ atoms in the molecule. Equations identical with
Eq. (11.2) except for the different values of Vi give the moments /s and
/c about the two other axes h and c. Ordinarily, the three moments
of inertia have different numerical values, but sometimes the molecule
has a symfnetrical geometry that leads to two or even three equal
values. In the former instance, the molecule is known as an asymmetrical-top rotator* and in the latter two as a symmetrical-top rotator
and a spherical-top rotator, respectively.
When all the atoms in a molecule lie on a straight line, the moment
of inertia of the molecule about that straight line as an axis is zero,
because all the r^ values are zero. In such a case, the molecule cannot
rotate about the straight line determined by the atoms, but only
about intersecting axes perpendicular to it. The moments of inertia
about these intersecting axes are equal. Such a molecule is a linear
rotator.
The rotational energy, W^at,, is entirely kinetic. It depends on the
rotational quantum numbers of the molecule in different ways for the
different kinds of rotators. For the linear and spherical-top rotators,
only one quantum number, J, is involved,! and the equation is
P^rot = Ji-f + V)Bhc
where B = „ „ r-
(11.3)
The values of J are 0, 1, 2, 3, . . . , oo.
In the symmetrical top, the equation is
W,ot = J {J + l)Bhc + K^(C - B)hc
(11.4)
h
where C = „ „ ^ • The quantum number K can have both positive
VlTT^clc
and negative values but can never exceed J numerically.
There is no general expression for the energy levels of the asymmetrical top, which depend in detail on the relative values of la, lb.
* The word top is used here in the sense of the child's toy, which, incidentally, is
usually a symmetrical-top rotator.
tThis quantum number is not to be contused with the atomic quantum number J
of § 10.4. In the atomic case, J depicts the total electronic angular momentum,
whereas J in the molecule usually is the index to the total angular momentum of
rotation of the nuclei about the molecular axes. In both cases, however, the relation
between J and the angular momentum is the same, namely.
Angular momentum = Vj(J
-j- 1) —
2x
264
MOLECULAR SPECTRA AND STRUCTURE
[§11.1
and Ic- However, the energy-level scheme can be inferred qualitatively from the fact that it approaches that of the symmetrical top
when two of the values of the moments of inertia approach one
another.
It will be noted that the rotational energies go up as the square
of / ; that is, as / gets larger, the spacing between levels gets larger.
This result is in contrast to the electronic levels in atoms, which get
closer together as the principal quantum number gets larger. Moreover, the energies of rotation are inversely proportional to the moment
1-
-L
Wrot
"1
"\
J= 0
K=0
K=l
K=2
. K=3
K=4
K = 5.6-
Fig. 11.2. Rotational energy levels for molecules.
of inertia, so t h a t for heavier molecules, both the energy levels a n d
their spacings are smaller t h a n for, lighter molecules. Figure 1112
shows t h e lowest energy levels of a symmetrical t o p in which /<, == I t
= \Ic- The energy levels in the column K = 0 are also those of a
linear or spherical-top rotator whose moment of inertia is Jj, since
Eq. (11.4) reduces to Eq. (11.3) when K = 0. The rotational
energies of various kinds of molecules are summarized with examples
in Table l l . l .
The preceding discussion is valid only for molecules that can
§11.1]
MOLECULAR SPECTRA AND STRUCTURE
265
r o t a t e freely; therefore t h e rotational energy-level diagrams are valid
only for molecules in t h e vapor s t a t e . I n liquids and solids, rotation
is usually hindered or stopped entirely b y intermolecular forces a n d ,
with rare exceptions, t h e rotational energy levels are "smeared o u t "
TABLE 11,1
ROTATIONAL ENERGIES IN MOLECULES
Type of
rotator
Linear
Moments of
inertia
Rotational
energies, Wroi
Ia = h;Ic = o J(J + \)Bhc
Geometrical
forms
Examples
Straight
line
All
diatomic
molecules;
carbon dioxide; acetylene
Tetrahedron
Methane;
phosphorus
{PA)
Spherical top
Symmetrical
top
Ia= h - h
Ia = h ^ Ic
J{J +
\)Bhc
Octahedron
Uranium hexafluoride
Cube
None known
Any symmetry
with only one
threefold
or
higher symmeJ{J + \)Bhc
+ K\C - B)hc try axis, such
as:
Trigonal
mid
pyra-
Trigonal prism
Plane
Asymmetrical
top
la^
h9^
h
No general
pression
ex-
Ammonia
Cyclopropane
hexagon Benzene
Any symmetry
lacking a three- M o s t p o l y fold or higher atomic molesymmetry axis cules
*
into a c o n t i n u u m of levels. Accordingly, t h e study of r o t a t i o n a l
spectra is almost entirely confined t o those molecules which can be
obtained in satisfactory concentrations in t h e vapor state.
266
MOLECULAR SPECTRA AND STRUCTURE
[§11.2
11.2. VibrationaJ Energy Levels. A molecule containing iV atoms
has SN kinds of motion. Three of these are simple translation, and
three, as we have just seen, are rotations about the three axes of
inertia (two if the molecule is linear). There are therefore SN — 6
additional kinds of motion (SN — 5 if the molecule is linear). These
are all vibrational, and associated with each kind is a vibrational
frequency. As long as the vibrations are not violent, that is, are
restricted to amplitudes less than about one-tenth the average distance between atoms, the vibrations are essentially harmonic. The
actual form of the vibrational pattern may be quite complicated, but
it is always possible to regard any harmonic vibration whatever as
the superposition of two or more simple vibrations called the normal
vibrations of the molecule.
In a normal vibration, the atoms all move with the same frequency,
which is independent of the amplitude of vibration, and their displacements from the positions they occupy in the nonvibrating molecule vary with time in pure sine-wave fashion. There are 3iV —_ 6
normal vibrations, and the 3vV — 6 frequencies associated with them
are called the fundamental frequencies of the molecule. These frequencies may all be different, or there may be several pairs or triples
of vibrations with tnfessame frequency. Because the forms of the
normal vibrations are strongly dependent on the symmetry of the
molecule, symmetry is of great help in the determination of the vibrational amplitudes and frequencies. For example, it can be shown
that any molecule containing a threefold or higher axis of symmetry
will have a certain number of its frequencies occurring in pairs, and
a molecule with two or more threefold or higher axes will have a
certain number of frequencies occurring in triples. Symmetry is also
definitive for selection rules in vibrational spectra. The normal
vibrations of several simple molecules are shown qualitatively in
Fig. 11.3, in which the displacements of the atoms are depicted by
arrows and by -j- and — symbols. The latter indicate respectively
displacements above and below the plane of the page.
/
The vibrational energy levels of a molecule form a simple pattern
as long as the vibrations are harmonic, since'in this case each vibration
contributes a set of energy levels to the pattern that is independent
of the energy levels of the other vibrations. The relation between
the energies W^n, of one vibration whose frequency is vi and the
vibrational quantum number 'i)i for that frequency is
TFvib = (DI + i)/ivi
' (11.5)
§11.2]
MOLECULAR SPECTI\A AND STRUCTURE
267
where Vi = 0, 1, 2, 3, . . . . There is a separate q u a n t u m number, v,
for each vibration, and the energy-level scheme is obtained by adding
u p the equations like E q . (11.5) for all the SN — 6 vibrations:
37V-6
(11.6)
T h e following characteristics of the energy levels of a single frequency
m a y be n o t e d :
1. T h e energy levels are equally spaced, the spacing being hv,
2. T h e lowest energy level {v = 0) is n o t a t zero energy b u t a t
^hv. This minimum energy, which remains with the vibrator even
at t h e absolute zero of temperature, is called the zero-point energy.
(Q) Diatomic Molecule AB:
(I)
(I')
(c) Non-linearTnafomic
Molecule ABC:
A
B
C
(bj Linear Triatomic Molecule ABAA
B
A
(I) — •
•
(2)—•
••
'-•
(3, y
4
1
(4)
•^
(I)
f-
^
(2)
(3)
*^
(d) Planar Symnnetncal Molecule BA
(2)
j^
(5)
[3)
(6)
-^
^ r-N
Fig. U.3. The normal vibrations of some small molecules.
P'igure 11.4 shows a few of the energy levels of a single vibrational degree of freedom. Such levels would constitute the entire vibrational
energy p a t t e r n for a given electronic state in a diatomic molecule, or a
268
[§11.2
MOLECULAR SPECTRA AND STRUCTURE
small part of the pattern for a polyatomic molecule. The lower levels
have been drawn with approximately equal spacing, in accordance
with Eq. (11.5), but the higher ones are crowded together somewhat
to illustrate the effect of anharmonidty.
This effect is always present
and is a consequence of the fact that the forces between atoms cannot
V-R
5000
u=4
4000 ••
3000 -.
8
—
7
—
.
V--^
5 —8 - —
3
in cm-'
0
7 - —
^
2000
4 _
3 -_
8 --7 -—
V=l
6
1000
'-
5-_.
4
—
3 —-
500
Zero-point-energy
Fig. 11.4. The vibrational energy levels of" a diatomic molecule.
be described accurately by simple force constants but vary with
interatomic separation in a more complicated fashion.
In order to indicate the relatiye magnitudes of vibrational and rotational energy levels, a few rotational levels are superimposed on the
§11.2]
269
MOLECULAR S P E C T R A A N D S T R U C T U R E
lower vibrational levels. The levels are those for a diatomic molecule
whose vibrational frequency, is 1000 cm"' and whose rotational
TABLE 11.2
t
CHARACTERISTIC VIBRATION FREQUENCIES OF GROUPS OF ATOMS
Group
Single
bonds:
C—C
C—N
C—0
C—F
Frequency,, Wavelength,
cm~'
micron.s
C—CI,
H
Cac
H—Cfst
H
Car
H-C,i
H—Na„
H—Ni„,
H—0
H—F
H—Si
H—P
1000
1100
1150
1200
650
750
3350
3020
3080
2890
3400
3375
3600
3959
2260
2350
10.
9.
8.5
8.3
15.4
13.3
3,0
3.30
3.25
3.46
2.9
2.95
2.78
2.59
4.42
4.26
H—S
H—CI
2585
2886
3.87
3.46
H—Se
H—Br
H—I
2230
2559
2231
4.48
3.91
4.48
c—s
Group
Double
bonds:
C^Cet
^^^*^cni
vv=Car
C=N
C^Oanh
C=v)est
v^^^Uaci
C=Oliet
C=S
Triple b o n d s :
CsC
C= N
^^^iso
c=o
Some larger
groups:
MethyKCHs)
Ether(C - O
-C)
Terminal
C =CH2
Trifluoro
(CFa)
NitroC-NOz)
Frequency,
cm"'
Wavelength,
microns
1640
1590
1595
1640
fl775
[1825
1745
1710
1695
1530
6.10
6.30
6.27
6.10
5.63
6.48
5.73
5.85
5.90
6.54
2220
2250
2150
2165
4.50
4.45
4.65
4.62
1385
565
7.22
17.70
1 985
jl410
1380
10.15
7.09
7.25
1565
6.39
The above frequencies in cm~' and wavelengths in microns are approximate because
group frequencies vary somewhat from molecule to molecule. In general, a frequency
value ending in two zeros means the group frequency varies as much as ± 100 cm~'
from the value given. A single final zero means ± 5 0 cm""', and a final 5 means ± 1 0 —
20 cm~'. H—O and H—N frequencies are reduced from the listed values by 200 cm""'
and are broadened to a width of 100 cm"' or more by hydrogen bonding such as exists
in liquid H2O and NH3.
Abbreviations: ac, acetylenic; an, acid; al, aliphatic; am, amine and amide;
anh, acid anhydride; ar, aromatic; cnj, conjugated; est, ester; et, ethylenic; ini,
imide; iso, isocyanide; ket, ketone and aldehyde.
constant B [Eq. (11.3)] is 15 cm~'. Each of these levels corresponds to a particular value of v, the vibrational quantum number.
270
MOLECULAR SPECTRA AND STRUCTURE
[§11.3
and to a particular value of / , the rotational quantum number.
An interesting result of the study of the vibrational spectra of
thousands of molecules is that many of the vibrational frequencies
of a molecule are essentially those of very small groups of atoms
within the molecule, and that these frequencies are characteristic of
the groups of atoms regardless of the composition of the rest of the
molecule. This fact is of great usefulness in the applications of
spectroscopy to the study of molecules containing more than three
or four atoms. I t serves as a basis for qualitative analysis of molecules and for the elucidation of molecular structure. Not all molecular
frequencies, however, are group frequencies, since each molecule also
contains vibrations which are characteristic of the molecule as a whole
and which are strongly dependent on its over-all structure and compositioh. A list of characteristic group frequencies is given in
Table 11.2.
«
11.3. Electronic Energy levels. lElectrons in molecules have four
quantum numbers, as do electrons in atoms, and moreoever are
subject to the restriction of the Pa'uli principle. They may be
classified into three kinds: those which belong to a single atom,
those which are shared by two adjacent atoms, and those which are
shared by more than two atoms.
The first kind of molecular electron consists of those electrons in
the inner shells of the atoms in the molecule. " They differ little from
the corresponding electrons in the atom when it is not part of a molecule, and they do not contribute to the binding forces that hold the
molecule together. Their contribution to the molecular electronic
levels may be disregarded.
The second kind occurs in all molecules, and necessarily all electrons in diatomic molecules are either of this or of the preceding kind.
An electron that is shared by two atoms has a set of quantum numbers which is similar to the set of atomic quantum numbers. The
difference lies in the replacement of the quantum number m by a
new number (usually called X), which is simply the index to the
orientation of the electronic orbit with respect to the interatomic
axis. The restrictions on the four quantum numbers, n, I, X, and s
are the same as those in atoms. Specifically, X can be either positive
or negative but never larger numerically than I. In diatomic molecules, the X's and spins couple in a manner quite analogous to the
coupling of Z's and spins in atoms. The algebraic total of the X's
(called A) represents the projection of the total orbital angular momen-
§11.3]
MOLECULAR SPECTRA AND STRUCTURE
271
turn'of the electrons on the interatomic axis. This total is denoted
by capital Greek letters S,n,'A, $, T, . . . , where A = 0 , 1 , 2, 3, 4 . . . .
Term symbols, by analogy to atomic symbols, are written
MA
For example, the lowest electronic state in H2, N2, and HCl is ^S;
in O2, ^S; and in NO (nitric oxide), ^11. Most chemically stable
diatomic molecules have ^S ground states. There is usually no subscript analogous to the atomic quantity J, because the total angular
momentum of the molecule changes from one rotational state to
another, and therefore the total angular momentum is not determined by the electrons alone, as it is in atoms. However, other
symbols are often added to ^A to give additional information about
the electrical structure of the molecule. These are discussed in detail
by Herzberg in General Reference 11.2.
The close analogy between electronic quantum numbers in atoms
and those" in diatomic molecules enables one to trace the ancestry ot
diatomic quantum numbers to the atomic quantum numbers of the
outer electrons in the two atoms of which the molecule is composed.
There are well-defined rules for this correlation, which lead not only
to an understanding of the electronic quantum numbers in the molecule but also to at least a semiquantitative estimate of the locations
of the. various electronic energy levels.
An important feature of the electronic energy levels in molecules
is the dependence of the electronic energy on the distance between
atoms. Figure 11.5 shows this dependence for two different electronic states in a diatomic molecule. The curve for each state,
called a 'potential curve or Morse curve, shows the interatomic distance
at which the electronic energy is a minimum (that is, the equilibrium
distance r^) as well as the energy difference between this minimum
and the limiting value reached by the electronic energy as the interatomic distance gets very large. This energy difference is the dissociation energy De of the molecule. In addition, vibrational energy
levels of the two electronic states are drawn roughly to scale on the
curves. The two intersections of each vibrational level with the
potential curve give the maximum and minimum values of r during
the course of the vibration corresponding to that level.
The two electronic energy levels shown in Fig. 11.5 are low-lying
levels that would correspond to small values of n. The levels for
higher values of n converge as n gets larger, just as do the levels in
MOLECULAR SPECTRA AND STRUCTURE
272
[§11.3
atoms. This convergence appears experimentally in t h e form of
R y d b e r g series in the v a c u u m ultraviolet absorption spectra of m a n y
molecules. T h e limit t o these series is an ionization potential of t h e
molecule, just as it is in atoms.
80,000
'
2
r"
^VO^
Interatomic disfanee.r,,
in Angstrom units
Fig. U.S. Variation of electronic energy with interatomic distance in a diatomic
^,
molecule.
'O
T h e third kind of electrons in molecules consists of those which
cannot be localized on a single a t o m or a pair of atoms. I t is difficult
§11.3]
MOLECULAR SPECTRA. AND STRUCTURE
273
to assign quantum numbers (except for spin) to this kind of electron,
since the meaning of the quantum numbers varies from one molecule
to another as the geometrical form and number of atoms in the molecule change. A common practice of molecular spectroscopists'^' ^ is
therefore to assign a term symbol to each energy level arising from a
collection of such electrons in a molecule without attempting to
evaluate the quantum numbers for the individual electrons. These
term symbols are useful in the evaluation of selection rules for transitions between the various levels and serve as a description of the distribution of electrons of this Isind in the molecule on the basis of the
molecular geometry. Unfortunately, it is not feasible to interpret
these term symbols by means of the conventional structural formulas
of the organic chemist, or by means of Lewis' "dot structures," and
we must refer the interested reader to the articles just cited for further
information. I t can be stated in general, however, that electrons of
this kind have much smaller energies than the other two kinds and
therefore give rise to spectra in the near ultraviolet, the visible and
even the near infrared.
The best examples of the third type of electrons are found in
molecules possessing what the organic chemist terms conjugated
double bonds. Each atom in the conjugated chain of atoms contributes one such electron to the molecule's collection, but these electrons
may be regarded as belonging collectively to all the atoms in the chain
rather than to a particular atom. The energy levels associated with
such electrons are characterized by relatively low energy and by
small and rather regular separations from each other. It has long
been recognized that these levels are of great importance with respect
to both the chemical properties and the visible and ultraviolet absorption spectra of the molecule. They are highly characteristic of the
geometrical configuration of the molecule, and of the number and
kind of double bonds in the conjugated chain.
An important empirical statement concerning electronic energy
levels in molecules is that the energies of electrons in small groups
of atoms are often influenced very little by the attachment of neighboring groups of atoms of varying kinds. This concept was first
advanced by dye chemists, who noted that certain groups of atoms
i R . S. MulUken, Jour. Chem. Phys., 7, 121 (1939).
^ See General, Reference 11.6, page 7 7 ^ .
^ See, for example, L. Pauling, Nature of the Chemical Bond, page 5 of 2d ed., Cornell
Press, Ithaca, 1944.
274
MOLECULAR SPECTRA AND STRUCTURE
[§11.3
in an organic compound cause characteristic absorption of visible
light irrespective of the nature of the rest of the compound. Such
groups are called chromophores, that is, color carriers, but they are
not by any means limited to groups absorbing in the visible region
of the spectrum.
As an example of a chromophore, one may consider the benzene
ring. The first excited electronic level in benzene lies some 38,000
cm~^ above the ground state, giving rise to absorption at about
2650 A in the ultraviolet. Many compounds containing the benzene
ring, such as alkyl benzenes and the halobenzenes, show absorption
very close to this wavelength. Certain other derivatives show the
absorption but' at somewhat altered wavelengths (about 2950 A in
the case of aniUne). When a substituent group, such as the amino
group in aniline, has a noteworthy effect on the wavelength at which
absorption occurs, it is called an auxochrome. In the special case
that the change in wavelength is to longer wavelengths, that is, to
lower frequencies, the group is called hathochromic or coJor-Jowering,
the lowering referring of course to the frequency and not to the intensity of the color. The distinction between a chromophore and an
auxochrome begins to lose significance when the effect of the auxochrome is so large that the characteristics of the chromophore are
altered greatly, as when an auxochromic group can conjugate with
a chromophore to lengthen a conjugated chain, In such a case the
lengthened conjugated system is more properly regarded as a new
chromophore in its own right.
A chromophore can absorb light by virtue of transitions between
electronic levels, either of the two-atom kind or the many-atom kind.
An example of the two-atom kind is, the carbonyl group as found in
aliphatic ketones and aldehydes. This chromophore has a characteristic absorption at approximately 2800 A which is associated with
a transition between two electronic levels associated with the carbonyl group alone, as is shown by comparison of the spectra of
formaldehyde, higher aldehydes, and aliphatic'k<jtqnes. The benzerie
nucleus cited above is an example of a many-atom chromophore. / A
more striking example is the porphyrin ring system found, with
assorted auxochromic substituents, in many natural pigments such
as hemin and chlorophyll.
' v.
In Table 11.3 are listed some chromophoric groups whose absorption lies between 1700 and 6000 A. The colurnns marked X^ax and
log € give, respectively, the longest wavelength at which an absorption
§11.3]
275
MOLECULAR SPECTRA AND STRUCTURE
maximum occurs in the spectrum of the chromophore and the common logarithm of the molar extinction coefficient e at that maximum
(e is defined in § 14.2). The values of X^ax and log e are approximate.
They vary somewhat from compound to compound having the same
chromophore and change also with state of aggregation, temperature,
solvent, molecular geometry, and other factors.
TABLE 11.3
CHHOMOPHOHIC
Chromophore
"max
log 6
C=C
(C=C)j
(C=C),
(C=C)4
CsC
—COOH
—CONH2
1900
2200
2600
2900
1800
2100
2100
3.5
4.2
4.6
4.8
<2
1.6
2.2
'C=N
C= N
C=0
C=S
N=N
N=0
—NO2
1900
1700
2800
3300
3700
6600
2700
3.7
<2
1.3
1.0
1.2
1.3
1.2
'
GROUPS
Chromophore
—0N=0
—ONO2
Rings:
PhenylPyridylNaphthylCyclopentadiene
Pyrrole
Pyrimidine
QuinoHne
Anthracene
^max
log e
3700
2700
1.7
1.2
2700
2600
3100
2.4
3.2
2.4
2440
2400
2450
3100
3800
3.4
2.4
3.5
3.8
2.8
For each chromophore only those atoms are listed which share
the electrons responsible for the absorption. Other atoms or groups
such as hydrogen atoms and methyl groups are omitted because of
their small effect on Xmax and log e. These omitted groups do not
themselves absorb at wavelengths above 2000 A. Certain generalizations of approximate validity can be used to extend the usefulness
of Table 11.3:
1. When two chromophores are joined by an aliphatic link (for
example, —CH2—) their absorptions are additive; for example, log e
for two identical chromophores is 0.3 larger than log e for one.
When two chromophores are attached directly, they form a new
chromophore whose absorption does not usually resemble that of
either.
2. The addition of a C ^ C link to a conjugated chain increases
Xmax by roughly 300 A and log e by about 0.3; the replacement of
276
MOLECULAR SPECTRA AND STRUCTURE
[§11.4
C = C in a chromophore by C = C does not greatly affect either
Xmax or log 6.
The reader will find extensive tables of chromophores and auxochromes, together with a comprehensive bibliography, in General
Reference 11.1. More detailed discussion of electronic energy levels
in diatomic molecules is given by General Reference 11.2, and in
polyatomic molecules by General Reference 11.6.
MOLECULAR SELECTION RULES AND THE
APPEARANCE OF MOLECULAR SPECTRA
In even the simplest molecules, the rich collection of energy levels
associated with the electronic, vibrational and rotational degrees of
freedom leads to a highly complicated spectrum. Fortunately, the
levels can be disentangled to a considerable extent, and the interpretation of molecular spectra is facilitated by the consequent possibility
of considering electronic, vibrational, and rotational transitions more
or less separately. This division does not mean that the three types
of transitions do not occur simultaneously but only that transitions
of one kind have relatively small effects on the energy levels of the
other kinds. The spectra associated with the three types are considered briefly.
11.4. Pure Rotational Spectra. Equation (11.3) shows that the
rotational energy of linear (including diatomic) and spherical-top
molecules depends only on a single quantum number J, which can
have any integral value. The selection rules for this quantum number in pure rotational spectra, that is, spectra in which no vibrational '
or electronic energy changes occur, are as follows:
AJ = 0 for molecules that do not have a permanent electric dipble
moment.*
•
.
A J = d=l for molecules that have a permanent electric dipole
moment.
These rules mean that molecules which have no permanent dipole
moment cannot have a pure rotational spectrum. Examples of such
molecules are all symmetrical linear molecules (H2, O2, N2, CQjj.
C2H2, and so on) and all spherical-top molecules (CH4, UFe, and t h e \ ^
* A-molecule has a permanent electric dipole moment if on the average its center of
negative charge does not coincide with its center of positive charge.
§11.4]
MOLECULAR SPECTRA AND STRUCTURE
277
like). Unsymmetrical linear molecules (CO, HCl, HCN, N2O), on
the other hand, show a pure rotational spectrum, the appearance of
which is determined by the energy levels (Pig. 11.2), the selection
rule AJ = ± 1 , and the number of molecules populating each level.
The last factor depends on the temperature of the molecules; the
higher the temperature, the greater the relative populations of the
higher levels [compare Eq. (10.20)]. The combination of Eq. (11.3)
with the selection rule AJ = ± 1 leads to the relationship
0- = 2 / S
(11.7)
where a is the wave number in cm"' of a line in the pure rotational
spectrum, for which the upper energy level has the quantum number / .
Since this J can have only integral values from 1 to «=, the rotational
lines occur at equal frequency intervals. The appearance of the
Intensity
of
absorption
20
30
40
50
Fig. 11.6. The pure rotational absorption spectrum of a linear molecule.
resulting spectrum is shown schematically in Fig. 11.6, in which the
height of the lines represents their relative intensity. The maximum
in intensity occurs at the J value that corresponds to the most highly
populated initial level and thus depends on external conditions (the
temperature) as well as on the rotational energy levels of the molecule
itself.
The rotational line spectrum shown in Fig. 11.6 can in principle
be either an emission or an absorption spectrum. However, such
spectra are usually observed in absorption because of the wavelength
region in which they occur. * Since the constant B is inversely proportional to the moment of inertia, the larger the molecule, the lower
the wave number at which the absorption takes place. In practice,
the absorption maximum seldom occurs above 100 cm~\ which
* The pure rotational Raman effect of a few molecules has been studied. The
selection rules are different in this case; in particular, a permanent electric dipole
moment is not necessary for the effect.
278
MOLECULAR SPECTRA AND STRUCTURE
[§11.5
means that the wavelength region is usually longer than 100 ix. Pure
rotational spectra thus lie very far in the infrared—so far, in fact,
that few have been observed. Recently, however, the pure rotational
absorption of numerous molecules has been measured in the "microwave" region of the spectrum, for which X is approximately 10,000 /x,
or 1 cm, and tr is 1 cm~^ The technique of microwave spectroscopy
is outside the scope of this book, but certainly the spectroscopist can
expect much from this field of research. Because of its development,
the importance of pure rotational spectra has been greatly increased.
The energy-level scheme for a symmetrical-top rotator is more
complicated (Fig. 11.2) than that of a linear rotator, but the nature
of the selection rules simplifies the actual spectrum considerably.
The rules for AJ are the same as before, and in addition AK = 0.
Hence symmetrical top molecules that do not have a permanent
electric dipole moment do not exhibit pure rotational absorption,
whereas those with a dipole moment have a spectrum as given by
Eq. (11.7), similar to that shown in Fig. 11.6. Since neither the
energy-level pattern nor selection rules for asymmetrical top rotators
can be expressed in general terms, little can be said about the appeajrance of the spectrum of such molecules without explicit knowledge
of the moments of inertia and the electric dipole moment. ' T h e
spectrum is ordinarily quite irregular and complex. For details the
reader may consult Chapter I of General Reference 11.3.
Finally, it should be recalled that the foregoing discussion is valid
only for molecules in the vapor state, and that ordinarily molecules
in solids and liquids cannot rotate freely; that is, they have no simple
set of rotational energy levels such as those summarized in Table 11.1.
11.5. Vibrational Spectra. The,wave-number range covered by
molecular vibrations is approximately 100 to 5000 cm~^ which means
that the wavelength range is 2 to 100 /x. The experimental study
of vibrational spectra is carried out by means of infrared absorption
spectra in this spectral region, and in the visible region of the spectrum
by means of the Raman effect (Chapter 18). Most infrared studies
are made with prism spectrometers, on which type of instrument the
practical upper wavelength limit is 25 n, pr low wave-number limit
of 400 cm~i. By means of the Raman effect, however, much lower
wave numbers—50 cm~^ or even less—are observable.
Although the vibrational energy-level p'attern given by Eq. (ll.6)
is very complex, the equal spacings of the levels associated with each
frequency, and the selection rules, combine to simplify the vibrational
§11.5]
MOLECULAR SPECTRA AND STRUCTURE
279
spectrum. The selection rules restrict the changes in the 3N — 6
quantum numbers Vf as follows:
1. Only one of the quantum numbers Vi can change during a transition between vibrational levels caused by the emission or absorption
of radiation.
2. This change, Av, is + 1 or — 1 .
3. For certain vibrations, Av must always be zero.
These rules lead to the result that each frequency observed in the
spectrum is identical with the frequency of one of the molecular
vibrations. However, the complete set of molecular frequencies will
not necessarily appear in the spectrum, because those vibrations to
which the selection rule A?; = 0 always applies will not be observed.
The geometrical form of the molecule determines the number of
vibrations of this kind. In the infrared spectrum, Av is always zero
for vibrations during which the electric dipole moment remains
unchanged. For example, in b and d of Fig. 11.3 the vibrations
marked (1) have A« = 0 and therefore do not appear in infrared
absorption. In the Ilaman effect, the vibrations that do not change
the molecular refractivity have Av = 0. Examples in Fig. 11.3 are
(b) 2, 3, and 4, and (d), 4.
Since the geometrical form—-often called the symmetry—of a
molecule is definitive for the vibrational selection rules, the classification of the vibrations of a molecule according to their symmetry is a
necessary preliminary to a determination of their occurrence in their
spectrum, or spectroscopic activity. Once this classification has been
made, it is possible solely on the basis of symmetry to say how many
of the SN — 6 vibrations will appear in both the infrared and the
Raman spectra, how many will appear in one and not the other, and
how many will be forbidden to appear in either. Conversely, if the
infrared and Ilaman spectra are known and the molecular symmetry
is not, the latter may be inferred from the spectra with the help of
the selection rules.
Tables by means of which the vibrations of molecules may be
classified and their spectroscopic activity determined have been
worked out for all the various likely symmetries (see, for example.
General Reference 11.3, Tables 35, 36, and 55). Two examples of the
sort of conclusions which can be drawn from such Tables are as follows:
1. A molecule that contains a center of symmetry (for example, the
linear molecule ABA in Fig. 11.3b) will never have the same vibration
280
MOLECULAR SPECTRA AND STRUCTURE
[§11.6
active in both infrared and Raman spectra. A corollary of this rule
says that symmetrical diatomic molecules (H2, O2, and the like)
cannot absorb infrared radiation by vibrational transitions-; that is,
they will have no vibrational spectrum, just as they have no pure
rotational spectrum.
2. A fundamental frequency in the Raman effect which is found
to be polarized arises from a vibration which is totally symmetrical,
that is, during the course of which the symmetry of the molecule does
not change (examples: all the vibrations (1) in Fig. 11.3).
The vibrational selection rules on \ ^ c h the above discussion is
based are ample to explain the chief feamres of vibrational spectra,
but they must be modified if the spectra are to be interpreted in detail.
The rules are derived on the assumption that the vibrations are
harmonic and that the molecules are in the vapor state, that is, free
from the disturbing effects of close neighboring molecules. If these
assumptions are not strictly justified, the selection rules must be
altered. The chief effects are to allow more than one quantum
number v to change during a transition between vibrational levels,
and to permit Av to be 2 or even more. These alterations permit the
appearance of so-called overtones and combination tones in the spectrum. Such overtones in general are of feeble intensity in comparison
with the fundamentals, the frequencies permitted by the simple
selection rules. The Raman effect and infrared absorption differ
markedly in this respect, however, the Raman effect showing stricter
adherence to the simple rules. Overtones in the infrared spectra of
gases in turn are weaker than those in the spectra of liquids. Fortunately, the third selection rule, which excludes certain vibrations '
from infrared absorption or the Raman effect or both, is less affected
by these disturbing influences. Again the Raman effect adheres
closely to the rules, as does infrared absorption in the vapor. Only
in the infrared spectrum of liquids does one find significant deviation. '
11.6. Rotational Fine Structure in Vibrational Spectra. We have
already noted that vibrational frequencies are from hundreds' t o .
many thousands of times larger than those of rotation. One result
is that the frequency of a pure vibrational transition, in which only a
vibrational quantum number changes, differs Very little percentagewise from that of the corresponding transition in which both vibrational and rotational quantuni numbers change. In consequence,
every pure vibrational transition is observed spectroscopically in the
§11.6]
. MOLECULAR SPECTRA AND STRUCTURE
281
immediate neighborhood of a collection of vibrational-rotational
transitions. The various members of the collection differ only in
the rotational quantum numbers involved, and the collection is called
a vibration-rotation band.
The positions of the rotational lines in such a band for a diatomic
molecule are given to a fair approximation by an expression similar
to Eq. (11.7):
"• = <''vlb ± 2JB ,
(11.8)
in which o-yib is the wave number in cm~' corresponding to the pure
vibrational transition, and / takes the values 1, 2, 3, . . . . The near
infrared vibration-rotation absorption band of hydrogen chloride gas.
Fig. 11.7a, illustrates the structure of such a band. The left-hand
side of this band, called the P-branch, is given by Eq. (11.8) with
the minus sign, corresponding to the selection rule AJ = — 1 ; and
the right-hand side, the ij-branch. by the plus sign (A/ = -f-1). Each
numerical value of J in Eq. (11.8) refers to the higher of the two
rotational quantum numbers involvec^ in the transition, as it does
in Eq. (11.7). This statement means, however, that the J value is
that of the initial rotational energy level for the P branch, and that
of the final rotational energy level for the R branch.
I t will be noted that Eq. (11.8), with J restricted to integers higher
than zero, gives no wave number corresponding to the pure (j^u,. This
limitation, a consequence of the selection rule AJ = ± 1 , results in
a gap in the center of the vibration-rotation band. The center of
this gap is the wave number crvib [compare (a) in Fig. 11.7]. Such a
gap always appears in the infrared vibration-rotation absorption
bands of diatomic molecules and in certain of the bands of linear
polyatomic molecules. It always appears when the dipole moment
of the molecule vibrates parallel to the molecular line, which is the
only way the moment can vibrate in a diatomic molecule. Bands
with the central gap are called parallel bands.
In certain of the vibrations of linear polyatomic molecules, such as
No. 3^ in Fig. 11.3b, the dipole moment vibrates in a direction perpendicular to the molecular line. The rotational selection rule is
then A J = 0, ± 1 . As a.result, the P and R branches are joined by
a third, the Q branch, for which AJ = 0. Since there is no change
in rotational energy when J does not change, the various members
of the Q branch all have the same frequency, which is that of the pure
vibration. Hence the Q branch occurs in the center of the band.
282
MOLECULAR SPECTRA AND STRUCTURE
[§11.6
Such a band, with P , Q, and R branches, is termed a perpendicular
band and is illustrated in Fig. 11.7b.
Although few molecules have such simple vibration-rotation band
structures as those shown in Figs. 11.7a and 11.7b, these can serve
(0)
34,
76.^„:L^I
1
Abs.
R
2700
2800
2900
3000
3100
3000
cm"'
"
HCI vapor, Zju parallel band
cm
C^H, vapor, 3.3/x perpendicular
band
(d)
(b)
Abs.
Abs.
3000
COj vapor, I5/.1 perpendicular band
3100
C^H, liquid, 3.3/< perpendicular
band
(e)
Abs.
1500
1550
1600
cm''—
1650
1700
H j O v a p o r , &.7jj. band
- /
,'
Fig. U.7. Typical vibration-rotation bands in infrared absorption;
as a basis for a consideration of what happens t o t h e appearance of
the band structures when various complications are introduced.
Some of the complications are
(a) Complex eaergy-level systems of symmetrical a n d a s y m m e t rical tops.
§11.7]
. MOLECULAiR SPECTRA AND STRUCTURE
283
(b) Large moments of inertia that lead to unresolvable continuous
bands.
(c) Smearing out of rotational levels in liquid and solid phases.
(d) Changes in the energy-level system produced by the mechanical interactions between rotation and vibration-and by centrifugal
distortion of the molecule.
In Fig. 11.7 the gross effects of these various complications are
illustrated. Figure 11.7e shows a complex band in the spectrum of
water vapor, whose molecule is an asymmetric top. I t will be noted
that the regularities present in the bands of the linear molecules have
disappeared. Figure 11.7c shows a band for a molecule (benzene)
whose moments of inertia are all so large that resolution of the lines
associated with individual rotational transitions is impossible. This
limitation results, as we have seen in Eq. (11.3), from the fact that
the spacing of the rotational levels is inversely proportional to the
moments of inertia. The rotational band in such a case is simply the •
"envelope" of unresolved lines. Even this envelope is smeared out,
however, if the molecular rotation is hampered or eliminated by
condensation of the vapor to a liquid or solid. The same vibrational
band as observed in the liquid state is shown in Fig. 11.7d. The
complications introduced into the rotational energy-level scheme by
centrifugal distortion of the molecule or by vibration-rotation interaction are important for the precise interpretation of the spectra but
are beyond the scope of this brief account.* ^
Finally, it should be mentioned that the Raman effect could, in
principle, be used as well as infrared absorption for the study of the
rotational structure of vibration-rotation bands. In practice, it is
almost never so used because of the difficulty, discussed m Chapter 18,
of obtaining the Raman spectra of gases. The rotational structure of
Raman lines in the liquid state is suppressed to about the same
extent as that of the infrared absorption band for liquid benzene
shown in Fig. 11.7d.
11.7. Electronic Spectra of Diatomic Molecules. These spectra
arise from transitions during which quantum numbers associated with
electronic, vibrational, and rotational energy levels all change. The
region where such spectra are found, which may be anywhere from
the vacuum ultraviolet to the near infrared, is determined by the
electronic levels. The changes in vibrational and rotational quantum
' See, for example. Chapter IV of General Reference 11.3.
284
MOLECULAR SPECTRA AND STRUCTURE
[§11-8
numbers introduce a fine structure (analogous t o the rotational fine
structure of vibrational bands) from which the n a m e band spectra
is derived.
T h e electronic q u a n t u m n u m b e r s described in § 11.3 are subject
to selection rules similar to atomic selection rules [Eqs. (10.19)]:
t h e total electron spin S does not change during a transition (AS = 0),
and the angular m o m e n t u m projection A m a y change by only one u n i t
or not a t all (AA = 0, ± 1 ) . Complications are introduced, however,
because the angular m o m e n t u m of molecular rotation can couple with
the electronic spin a n d orbital angular m o m e n t a (compare RussellSaunders coupling in atoms—§ 10.6, footnote). This coupling can
t a k e place in various ways which are beyond the scope of our discussion b u t which give rise to various additional selection rules. R e g a r d less of these additional rules, however, the general s t a t e m e n t can be
made t h a t the t o t a l molecular angular m o m e n t u m — s p i n , orbit a n d
r o t a t i o n — m u s t change b y one unit or not a t all.
11.8. Vibrational Structure of Diatomic Spectra. T h e vibrational
selection rules are much more diverse t h a n those for vibration-rotation
spectra (§ 11.6) because of the entirely different basis on which t h e y
are determined. This diversity stems from the relative sluggishness
with which atomic nuclei move in a molecular vibration compared t o
the speed of electrons in their orbits. T h e latter speed is of the order
of 10* cm/sec, whereas t h e velocity t h a t vibrating nuclei reach is
a t most one-hundredth of this value a n d usually is much smaller still.
I n making a transition from one orbit t o another, an electron travels
a distance of approximately 10"* cm and therefore will require only
a b o u t 10~i* sec for t h e switch. D u r i n g so short an interval, t h e
vibrating nuclei will travel less t h a n 10"^" cm, t h a t is, less t h a n oneh u n d r e d t h of the inter nuclear distance.
i
Since the internuclear distance changes by less t h a n 1 pe» cent,
for all practical purposes it remains fixed dyring the electronic transition. Hence the molecule will go from one, electronic level to another
without variation of r in Fig. 11.5. If the value^of r initially is r/',
t h e molecule being in the ground electronic state and not vibrating
(v" = 0), absorption of radiation will carry the molecule t o an upper
level (point A on the upper curve in Fig. 11.5) without change, in r.-*
I n t h e upper state, however, the internuclear separation r / ' no longer
is a n equilibriuni separation. T o restore equilibrium, the molecule
* It is standard usage to designate upper state quantities with a single prime and
lower state quantities with a double prime.
§11.8]
MOLECULAR SPECTRA. AND STRUCTURE
285
in the upper level moves from r / ' (point A) to r / (point B), picking
up linear momentum a^it does so. This momentum, however, carries
it past rj to a point of potential energy equal to that at point A, that
is, to point C. Point C is not a position of equilibrium, so that the
molecule moves toward r/ again, and again picks up momentum,
which carries it back to r^" (point A). The whole cycle is then repeated; that is, the molecule vibrates between points A and C.
This amplitude of vibration corresponds to v' ~ 2. Hence the
change, Av, on going from the lower curve {v" = 0) to the upper
{v' = 2) is + 2 .
If the vertical ordinate through Te" had intersected the upper curve
at some other level, say v" = 5, Av would have been + 5 . In actual
molecules, the value of AD is not restricted sharply to any one value
but can have a range of values, a few of which are much more likely
than others. The most probable values correspond to the most
u I.I Miit'iiiBiinBUii •Oil
Mtcir.
i i — •
aivi
'••
Fig. 11.8. Band progression in the electronic spectrum of the
nitrogen molecule Naintense bands in the spectrum. This method of determining At; in
diatomic electronic spectra is called the Franck-Condon frinciple.*' ^
The collection of vibrational bands associated with a given electronic transition is called a band system. In the analysis of such a
collection with respect to the values of v' and v" for each band, it is
frequently possible to assign a regular series of bands to a single value
of v' and to successive values of v", or vice versa. Such a regular
series, called a j)rogression, is illustrated in Fig. 11.8 for the molecule
Na. Another kind of regularity occurs when two Morse curves
(Fig. 11.5) are related in such a way that Av is a constant regardless
of the values of »' and v". Groups of bands with the same value, of
Av then occur together and are called sequences.
With the help of Fig. 11.5 it can be seen that the value of Av may
be quite different for transitions beginning on the upper curve from
its value for those beginning on the lower. If we assume a non4 J. Franck, Trans. Farad. Soc, 21, 536 (1926).
^ E . U. Condon, Phys. Rev., 28, 1182 (1926); 32, 8S8 (1928).
286
MOLECULAR SPECTRA AND STRUCTURE
[§11.9
vibrating molecule in the upper state {v' = 0), the internuclear separation before transition will be r/, and therefore the transition will
take the molecule from point B to the intersection of the ordinate r/
with the lower curve, or to point X. This transition has A«; = — 3
(or —4) rather than + 2 as before.* Hence the energy change associated with the transition is quite different, and the emitted band
will occur in the spectrum at some considerable distance from the
absorption band Av — -|-2. This is simply another way of saying
that the emission-band spectrum of a molecule and its absorptionband spectrum can be, and experimentally often are, quite different
in appearance. Part of the difference, it should be added, comes
from the diversity of vibrational excitation imparted by the electrical discharge or high-temperature flame required to excite the
emission spectrum.
A valuable by-product of the vibrational analysis of diatomic
spectra is the possibility of tracing the vibrational levels of the ground
state to such large quantum numbers that the amplitude of vibration
corresponds almost to dissociation of the molecule. When .this
tracing can be done, as is often the case, the energy of dissociation
(De" in Fig. 1L5) can be measured directly and accurately. Indeed,
the most accurately known heats of dissociation, such as H2 and the
halogens, have been determined in this way. However, when the
dissociation energy is very large, as in N2 and CO, it may prove
difficult to follow the usual procedures because of the tremendous
spectral range involved. This difficulty has given rise to considerable
controversy over the interpretation of the spectra of N2 and CO.
11.9. Rotational Fine Structure of an Electronic-Vibrational Band.
Each electronic-vibrational transition of the sort just described gives
rise to a single band of which numerous examples can be seen' in
Fig. 11.8. Accompanying the transition is a set of rotational energy
changes that give the band its fine structure just as in the case of the
vibrational-rotational bands depicted in Fig. 11.7. The rotational
selection rules in the two cases are quite similar: A J = 0, ± 1 except
when the electronic transition takes place between two ^S levels, for
which case AJ = ± 1 . These two rules result in perpendicular and
parallel bands, respectively.
The basic difference between the rotational structure of electronic
bands and that-6f vibrational bands arises from the fact that the
' By definition, Ac = JJ'
§11.9]
MOLECULAR SPECTRA AND STRUCTURE
287
moment of inertia of a diatomic molecule changes little in passing
from one vibrational level to another in the same electronic state but
changes considerably (sometimes as much as a factor of 2) when the
electronic state changes. Therefore the spacing of the rotational
levels differs markedly in the two electronic states. I t is usually,
though not always, smaller in the upper electronic level, corresponding
to a larger moment of inertia, because a higher electronic state usually
has weaker binding between the two atoms, and hence a larger internuclear distance.
An expression for the P, Q, and R branches of an electronicvibrational band analogous to Eq. (11.8) can be derived from the
above selection rules and Eq. (11.3) applied to the two electronicvibrational levels:
cr = a e,.vib± {B' ± B")J + {B' - B")J'
(11.9)
This expression gives the wave numbers in cm~' of the rotational lines
in the P branch if the first minus sign and the second plus sign are
taken. The Q-branch formula results if the first plus sign and the
second minus are used, whereas the R branch is given by the first and
second plus signs. J takes the values 1, 2, 3, . . . , but as before
(§ 11.6) the significance of J differs for each branch. In the P branch
the J value is that in the lower electronic level; in the R branch it is
that in the upper electronic level; and in the Q branch the two are
of course the same, since AJ = 0. The reason for starting J at one
instead of zero for the Q branch lies in the coupling between rotation
and electronic angular momentum which is always present in molecules giving rise to perpendicular bands I t will be noted that
Eq. (11.9) reduces to the form of Eq. (11.8) when B' = B", that is,
when the moments of inertia are the same in both upper and lower
states.
The presence of the J^ term in Eq. (11.9) has the result that, for
high enough J values, all three branches lie on the same side of
o'ei-vib- Since B' is usually smaller than B" (that is, lb > h"), this
side is the low-frequency side. The lines of highest J are always very
weak, which means that the three branches ordinarily fade out
towards the red end of the spectrum from o-ei.vib- Such bands are
said to degrade toward the red. On the violet side of (Xei-vib there is
a frequency of maximum value that occurs in the R branch shortly
before the J^ term overtakes the J term of opposite sign. This
maximum ^equency, which depends on the magnitude of the differ-
288
MOLECULAR SPECTRA AND STRUCTURE
[§11.9
ence between B' and B", is called the band head and is clearly to be
distinguished from o-ei-vib. which is called the band origin. Experimentally, the band head is more prominent and can be measured
directly. The band origin, on the other hand, has to be determined
by analysis of the rotational branches.
The foregoing discussion is illustrated in Fig. 11.9, which shows
the structure of two bands degraded toward the red. When B' is
(a) PARALLEL BAND
R-branch
(NO Q-BRANCH)
78910 ll 12 13 14
R-branch 6|543 2
Band-Head
15
i
16^7
18-^ R-branch
1 2
3
4
5
6-P-branch
Band Ofigin ( f f ^ i . ^ i i , )
A(b) PERPENDICULAR BAND (THREE BRANCHES)
P-branch L
Low J's
Q-branch I Low Ss\
I
R-branch
l— Low J's - ^
Head
origin
J ^ H i g h J's _ _ ^
High J's
I
High J's
\-
Fig. 11.9. The rotational fine structure of diatomic bands.
larger than B", the above discussion needs to be changed by replacing
"low-frequency" by "high-frequency," "red" by "violet" and vice
versa, "maximum" by "minimum," and "R branch" by " P branch.'
The several members of the P, Q, and R branches vary in intensity
in much the same way as they do in vibration-rotation bands (Fig.
11.7), the most intense lines corresponding to those originating from
the most highly populated rotational levels. Since the population
factor is given by an expression of the form of Eq. (10.20), in which
all quantities except the temperature T are either known or obtainablefrom the spectr,um, a measurement of intensity distribution among
the various branches can be used to determine T. In this way the
effective temperatures of molecules in flames and electrical discharges.
§11.10]
MOLECULAR SPECTRA AND STRUCTURE
289
which might be difficult to measure by other means, can be determined with reasonable accuracy.
The rotational fine structure can be made to yield information
about the moments of inertia of the molecule in the upper and lower
electronic states, and hence, if the identity of the molecule is known,
about the internuclear distances in the two states. There are two
ways in which this information can be obtained, to both of which an
assignment of the various rotational lines to their initial and final
levels is prerequisite. Once this assignment has been accomplished,
the first method proceeds by working out an empirical quadratic
formula relating the observed frequencies of various lines to their
J numbers. The coefiicient of the J term is numerically equal to
B' + B", and the coefficient of the J^.term is equal to B' - B", in
accordance with Eq. (11.9) for the P and R branches. Hence B'
and B" can be evaluated from these coefficients.
The other and more commonly followed procedure consists of
finding the energy difference between the individual rotational levels
by taking the difference between appropriate lines in the band. For
example, the difference between the eighth and ninth rotational levels
in the ground state is the difference between the Q-branch transition
J ' = 9 ^ J " = 9 and the jR-branch transition J' = 9 ^ J" = 8.
From such energy differences the constants B' and B" can be obtained, along with additional information such as the effect of
centrifugal force on the intermiclear distance.
The foregoing discussion applies to diatomic molecules in the
vapor phase. As soon as most molecules are condensed to liquids
or solids, the previously mentioned cornplications ensue. The rotational energy levels are "smeared out" by the collisions and by the
strong electric fields of closely packed molecules. Accordingly, diatomic spectra are studied almost exclusively in the vapor state. For
a comprehensive discussion of the manifold details of diatomic
spectra, the reader is referred to General Reference 11.2.
11.10. Electronic Spectra of Polyatomic Molecules. It can
readily be seen from the foregoing simplified account of the main
features of the electronic spectra of diatomic molecules that even
the simplest polyatomic molecule should have a still richer electronic
spectrum. The electronic energy levels can be much more complicated (especially in molecules with conjugated double bonds),
there are 3N — 6 vibrational degrees of freedom, and there are three
moments of inertia, usually all different. It is not surprising, then.
290
MOLECULAR SPECTRA AND STRUCTURE
[§11.11
that few polyatomic spectra have been thoroughly analyzed. There
is much active work in this field, however, and extensive experimental
as well as theoretical progress is being made.
There is, to be sure, a vast literature (see Chapter 14) of visible
and ultraviolet absorption spectra of polyatomic molecules of all
kinds. Most of this material, however, was obtained on molecules
dissolved in some relatively transparent solvent, and by means of
spectrographs of low resolving power. In consequence, even the
coarsest features of the spectra are usually "washed out," and only
rather broad generalizations, of the sort incorporated in Table 11.3,
have been possible from the data. It seems probable that this state
of affairs will improve markedly by progress along these lines:
1. Investigation of more and more molecules in the vapor state,
at low pressures but with long optical paths, with spectrographs of
high dispersion.
2. Study of spectra over wide temperature ranges, and especially
at very low temperatures.
3. Extension of the study of spectra to shorter wavelengths.
4. Improvement of means of excitation of polyatomic spectra, so
that emission spectra as well as absorption and fluorescence spectra
can be used in the study of the upper electronic levels.
It was indicated in § 11.3 that it is difficult to assign quantum
numbers to individual electrons in most polyatomic molecules, and
that instead the over-all distribution of the electrons with respect
to the various atomic nuclei is described by a term symbol. These
term symbols carry the multiplicity (2S + 1) as a superscript, just
as in atomic and diatomic term symbols, but the remainder of the
symbol indicates simply the geometry of the over-all electron distribution in the molecule. The usefulness of such symbols lies in
the fact that selection rules can be expressed in terms of them, and,
in turn, observed spectra can be interpreted with the help of the
selection rules to give the symbols for tlie' various observed states.
A complete discussion of these symbols and the associated selection
rules can be found in General Reference 11.6. The only one of these
rules we will mention is the one already given for atoms and diatomic
molecules: AS = 0.
11.11. Vibrational Structure of Electronic Spectra in Polyatomic
Molecules. The vibrational selection rules in polyatomic eliectronic
spectra are based on an extension of the Franck-Condon principle.
§11.12] '
MOLECULAR SPECTRA AND STRUCTURE
291
To a first approximation the result of this extension is the rule that
the only vibrations for which Av differs from zero are the totally symmetrical vibrations [compare §§ 11.2 and 11.5 and vibrations marked (1)
in Fig. 11.3]. For these vibrations the most likely values of Av are
determined, as for diatomic molecules, by the relative interatomic
distances in the upper and lower electronic .states. This rule is of
first importance for the interpretation of the spectra of highly symmetrical molecules, in which the number of totally symmetrical
vibrations is small, because it sweepingly simplifies the vibrational
pattern in the spectrum.
In the benzene molecule, for example, there are 30 vibrational degrees of freedom, with 20 distinct frequencies. Only two of these are
totally symmetrical, one of which is a pulsation or "breathing' of the
ring and the other a vibration of the hydrogens. In the first excited
electronic state of benzene, the ring size is slightly larger than in the
ground state, but the carbon-hydrogen distance is practically the
same. As a result, the most probable A«)'s for the totally symmetrical
ring vibration are 2 and 3, whereas AJ> = 0, 1 and 4, 5, 6, . . . are less
probable but still permitted. The Av for the hydrogen vibration, on
the other hand, is zero. In consequence, the most prominent feature
of the absorption spectrum of benzene is a simple progression of evenly
spaced bands that resembles roughly a progression of diatomic bands
because only one vibration is involved. The ultraviolet absorption
spectrum of benzene in solution is given in Fig. 11.10a, along with the
more complicated vapor spectrum of Fig. 11.10b. The absorption
spectrum of permanganate ion in aqueous solution is also shown in
Fig. 11.10c. Presumably, this ion has only one totally symmetrical
vibration, and therefore its electronic absorption shows a simplified
vibrational structure because of the action of the Franck-Condon
principle.
It is apparent from the foregoing that the more complete our
knowledge of the vibrational energy levels of a polyatomic molecule,
the better our chance of making a satisfactory interpretation of the
vibrational fine structure of its electronic spectrum. For this reason
the results of infrared and Raman spectra are frequently an indispensable adjunct to the understanding of the data obtained in the visible
and ultraviolet regions.
11.12. Rotational Fine Structure in Electronic Spectra. The rotational fine structure of polyatomic electronic bands provides less
information, generally speaking, than does that of diatomic bands.
292
MOLECULAR SPECTRA AND STRUCTURE
[§11.12
because of the much larger moments of inertia, the usually smaller
fractional difference between the moments in one electronic state and
another, and the complex patterns associated with the band struc(a)
Absorption
Intensity •
(b)
2400
2500
A in A
2400
2500
2600
»-
Absorption
Intensity
2600
X in A
»-
Absorption
Intensity
4500
5000
\
• 5500
6000
in A
Fig. 11.10. Vibrational fine structure of electronic spectra in polyatomic
molecules, (a) The absorption spectrum of benzene in cyclohexane solution in
the 2600 A region, (b) The absorption spectrum of benzene vapor in the 2600 A
region, (c) The absorption spectrum of permanganate ion in the green region of
the visible spectrum.
tures for symmetrical and asymmetrical tops. These characteristics
render the resolution of the rotational fine structure exceedingly
difficult for any but the smallest and lightest molecules. Naturally,
§11.13]
MOLECULAR SPECTRA AND STRUCTURE
293
even for such molecules it is necessary to study the spectra in the
vapor state and to keep the pressures as low as possible. Some
progress has been made in the analysis of rotational fine structure in
the ultraviolet spectra of carbon disulfide" (linear rotator), formaldehyde (quasi-symmetrical top), and nitrogen dioxide and sulfur
dioxide (asymmetrical tops).
The selection rules for the rotational fine structure are similar to
those for the infrared, but several new features enter that make the
transition possibilities more diverse. As for diatomic molecules, the
moments of inertia may differ in the upper and lower electronic states,
leading to fine-structure expressions like Eq. (11.9). In addition,
however, all polyatomic rotators except asymmetrical tops have
degenerate vibrations, such as (3) and (4) of Fig. 11.3b, which can
give rise to vibrational angular momentum. This angular momentum
can couple with the angular momentum of rotation in various ways,
and it increases the complexity both of the energy level scheme and
of the transitions between levels. The possibility also exists that
the actual geometrical structure of the molecule may be different in
two electronic states; for instance, it may be a symmetrical top in
the ground state and an asymmetrical top in an upper state.
From the preceding statements it can be seen that for both theoretical and experimental reasons the rotational structure will be
difficult to observe and to analyze. For further details, the reader
is referred to General Reference 11.6.
11.13. The Effects of External Influences on Molecular Spectra.
The external variables that affect molecular spectra are much the
same as those which affect atomic spectra (§ 10.7), but their relative
importance is quite different. One reason is the difference in methods
of excitation of atomic and of molecular spectra. The latter are
studied more frequently in absorption than in emission, and are not
often excited by arc and spark discharges. The high-voltage, highfrequency discharge, with or without electrodes in direct contact with
the vapor being studied, is used for excitation of emission spectra,
especially for small molecules.
Fluorescence is of importance in the analysis of many molecular
spectra. I t is studied with a technique not greatly different from
that used for the Raman effect (see Chapter 18). The chief requirement is an intense source of sharply monochromatic radiation of a
wavelength that corresponds to the energy difference between two
electronic levels whose band systems are under investigation. The
294
MOLECULAR SPECTRA. AND" STRUCTURE
[§11.13
correspondence need not be exact so long as the minimum energy
required is available, and it is usually better to have the exciting
radiation correspond to a combined electronic-vibrational transition
for which the transition probability is high. The spectrum obtained
in fluorescence depends not only on the exciting wavelength but
also on temperature and pressure, as indicated below. One limitation of fluorescence spectra (shared also by absorption spectra) exists
because of the selection rule AS = 0. Since most molecules are in
the ground state before excitation of fluorescence and since this state
is usually a singlet, in general only singlet levels in the upper state can
be studied by this means.
Temperature. The effect of temperature on molecular spectra is
of great importance. As we have already seen, the distribution of
intensity among the rotational lines is strongly temperature-dependent. The same statement can be made for vibrational bands in
electronic spectra. In ultraviolet absorption spectra, a temperature
effect is noticed for both totally symmetrical and nontotally symmetrical vibrations. The intensities of the former change as v"
changes (see Fig. 11.5), because, for example, a totally symmetrical
band for which D" = 1 —>«' = 5 will grow in intensity if the number of
molecules in the level v" = 1 increases. Nontotally symmetrical
bands, for which Av = 0, will be affected similarly. The change in
the appearance of the spectrum is even more striking here because
of the fact that in general v" and v' for a given vibration are different, and therefore the bands due to transitions «;" = 0 —> j;' = 0,
v" = I ^y v' = 1, v" = 2 —^ v' = 2, and so on, occur at different
places in the spectrum. The fine structure in between the tall peaks
of the benzene vapor spectrum (Fig. 11.10b) is largely associated withi
different v" values for nontotally symmetrical vibrations and shows a
strong temperature variation in intensity.
Conversely, the quantitative variation of intensity with temperature can frequently be the means for confirmation or rejection of the
analysis of a band system. If an interpretation of the spectrum
asserts that a given band or progression of bands arises from a
vibration in the ground state with v" = 2, for a frequency of v"', the
assertion can be checked by comparison of the observed effect of
temperature on band intensity with that calculated from Eq. (10.20).
• Moreover, it can be seen that the absorption spectrum will be considerably simplified if it is obtained from molecules with T near 0°K,
because then all vibrations in the ground state have v" = 0, and
§11.13]
MOLECULAR SPECTRA AND STRUCTURE
295
only totally symmetrical vibrations can give rise to transitions to
upper vibrational levels with v' different from zero.*
Pressure. The effects of pressure on molecular spectra are also
somewhat more complex than those on atomic spectra (see § 10.7).
The broadening of atomic lines by collisions, which of course increase
in number with increase in pressure, has a molecular analogue.
Atoms, 'however, can only exchange electronic and translational
energy upon collision, whereas molecules in addition trade vibrational
and rotational energies. It has been found, moreover, that it takes
very many collisions (of the order of 10^ or more for each molecule)
to establish a new balance among the various degrees of freedom once
the old balance has been upset in some way. Hence the spectrum
emitted by an assembly of molecules whose equilibrium has been
disturbed during excitation will depend on the extent to which
equilibrium has been restored prior to the emission of radiation.
In the fluorescence of vapors, for example, the equilibrium among
molecules may be upset by the exciting radiation, because the molecules are transferred to an upper state in which the molecules have
higher vibrational energy than their temperature warrants. If this
extra vibrational energy can be given up to other degrees of freedom
(vibrational as well as rotational and translational) before the molecule radiates (a time lag of about 10~^ sec exists between excitation
and reradiation), the emitted spectrum will correspond roughly to one
in which all the v' values are zero or at most very small. This situation obtains when the molecules are under relatively high pressure
(say 1 to 10 atmospheres at room temperature), so that the requisite
number of collisions can take place in 10~* sec. If the'pressure is low
(say 10""^ atmosphere at room temperature), an insufficient number of
collisions occurs before reradiation,t and the resulting spectrum
contains bands for which the v' values are high (values given by the
Franck-Condon principle for transitions from the vibrationless ground
state) as well as low (in those molecules which have made sufficient
collisions). The general result is that "high-pressure" fluorescence
spectra are somewhat simpler, involving primarily the upper vibrational state in which all v' values are equal to zero, and show vibrational spacings corresponding to different v" values. Low-pressure
* To a more refined approximation, vibrations that are not totally symmetrical can
have A» = 2 as well as A» = 0, but these transitions are extremely weak.
t If the lifetime of the excited state is longer than 10~* sec for some reason, the
pressure values must be revised downward.
296
MOLECULAR SPECTRA AND STRUCTURE
[§11.13
fluorescence (sometimes called resonance fluorescence) may be more
complex, since it involves .both upper state and ground state vibrations.
Because the probabilities for allowed transitions between various
levels in the same molecule may vary over a wide range, it is always
valuable in the study of absorption spectra, both infrared and ultraviolet, to vary the number of molecules in the absorbing path as
much as possible. There is no difficulty in making this number
indefinitely small, but frequently vapor pressure furnishes an upper
limit. Even when no limit is set by vapor pressure, increase in the
number of molecules per unit volume by increasing pressure is not
always desirable because of collision broadening. A solution to this
problem can sometimes be made by simple increase in the path
length of the radiation through the vapor. On occasion, to be sure,
the broadening produced by pressure may be useful, as for example
when one wishes to measure directly the mtegrated intensity of an
entire band,* or even band system, without resorting to the difficult
and usually inaccurate procedure of measuring the intensity of each
resolvable line and summing.
Electric and Magnetic Fields. The effects of electric and magnetic
fields are relatively unimportant in molecular spectra. The Stark
effect is extremely small and cannot be observed in molecules with the
usual spectrographic techniques. In the microwave region, however,
the Stark effect- on the pure rotational spectrum (§ 11.4) is readily
observable and is frequently of use in the detection of weak lines, in
the assignment of rotational quantum numbers, and in the measurement of perm£trient electric dipole moments. The Zeeman effect isusually small or missing altogether. Singlet electronic levels predominate in molecules, and there is no need for the Zeeman effect as
a means of series identification. Magnetic rotation spectra are sometimes useful in the analysis of diatomic spectra.'' It also appears
likely that the Zeeman effect on nuclear spins will be of value in the
determination of such spins by means of molecular microwave spectra.
Condensed Phases. One influence of first importance is that of the
state of aggregation. It has been indicated se.veral times in the foregoing discussion that change from the vapor to a condensed state—
liquid, solid,, or solution—results in considerable change both in the
^ E. B. Wilson, Jr., and A. J. Wells, Jour. Chem. Phys., 14, 578 (1946).
' See General Reference 11.8, page 729.
§11.14]
MOLECULAR SPECTRA AND STRUCTURE
297
energy-level scheme and in the selection rules. The rotational levels
are most drastically affected, since the close approach of neighboring
molecules in condensed phases hinders or halts entirely molecular
rotation. The vibrational levels are usually affected only slightly,
but in certain vibrations—mainly those involving the binding of
hydrogen to strongly negative atoms like fluorine, oxygen and
nitrogen—the difference in frequency between gas and liquid phases
may be as much as 10 per cent. The effect on vibrational selection
rules is more pronounced. Since the rules are based on molecular
symmetry and since the random intermolecular forces can never
increase the symmetry and usually destroy it, it may be expected
that the rules are invalidated to a greater or lesser extent. This
expectation is reahzed for infrared vibrational spectra (§ 11.5) but
very slightly for Raman spectra. It is difficult to generalize concerning the effect on vibrational selection rules in electronic spectra.
Certainly in many instances the rules do not seem to be drastically
altered.
The question of the effect of the condensed phases on electronic
levels is a serious one because of the immense amount of ultraviolet
and visible spectroscopic data obtained on molecules in solution. The
extent of the effect on the levels themselves of collisions and electric
fields due to near neighbors must be clearly differentiated from solvent
effects that actually alter the structure or composition of the molecule and are not properly considered as "condensation" effects at all.
True condensation effects seem to vary considerably with the nature
of the molecule and its solvent but not so drastically as to prevent
identification of levels in the vapor phase with those in the liquid.
Hydrocarbon molecules dissolved in hydrocarbon solvents show the
least effect on energy levels, selection rules, and transition probabilities, whereas polar molecules dissolved in water or other highly
polar solvents show the largest. On the other hand, a familiar but
striking example of structural alteration of molecules in solution is
provided by chemical indicators, whose color (related to the electronic
energy scheme) changes drastically with change in concentration of
hydrogen ion or other chemical substance. The subject of solvent
effects on absorption spectra does not lend itself to sweeping generalization, however, and it is impossible to summarize briefly and justly
the results of the large amount of investigation in this field.
11.14. Summary of Molecular Spectra. In order to summarize
the results of the foregoing sections, the energy levels and transitions
MOLECULAR SPECTRA AND STRUCTURE
298
SAOAAOJOILU .JO
[§11.14
pajCXJ/Uf
uoijc/~K>sqo
J
(-•<SV)
U0141SUOJ4 sa-ifoi^,
+
\
\
^
A6-ISU3 y<3M
3
u
"S
B
s6^
a
O
^•^XDuBnbt
T
UOIjajiO^B -
v_
M
~7
S
BDU30^3J0njJ.
s/SAB-j /Duoi-fOJqiy\
a404^-punQj^
-I
II
^§
N 6
^Ojauj
joinoBt&^-ij
II
MOLECULAR SPECTRA AND STRUCTURE
299
responsible for t h e various kinds of molecular spectra are presented
in one comparative diagram in Fig. 11.11.
GENERAL REFERENCES
11.1.
11.2.
11.3.
11.4.
11.5.
11.6.
11.7.
11.8.
E. A. Braude, "Ultraviolet Light Absorption and the Structure of
Organic Compounds," Ann. Reports of Chem. Society (London),
XLII, 105 (1945).
G. Herzberg, Molecular Spectra and Molecular Structure: I: Diatomic
Molecules. New York: Prentice-Hall, Inc., 1939.
G. Herzberg, Infrared and Raman Spectra. New York: D. Van Nostrand Company, Inc., 1945.
R. deL. Kronig, The Optical Basis of the Theory of Valency. London:
Cambridge University Press, 1935.
L. Pauling and E. B. Wilson, Jr., Introduction to Quantum Mechanics.
New York: McGraw-Hill Book Company, Inc., 1935.
H. Sponer and E. Teller, "Electronic Spectra of Polyatomic Molecules," Rev. Mod. Phys., 13, 75 (1941).
G. B. B. M. Sutherland, Infrared and Raman Spectra. London:
Methuen & Co., Ltd., 1935.
H.Vi.Vfood, Physical Optics. New York: The Macmillan Company,
1934.
Additional references on ultraviolet, infrared, and Raman spectra will be
found in the General References for Chapters 14, 17, and 18, respectively.
CHAPTER
12
The Measurement of Spectral Intensities
FOUR PRINCIPAL MEAVS ARE AVAILABLE
for the determination of
intensities in spectra:
1. The radiation may be absorbed on the blackened surface of a
radiometric device, which uses the heating effect of the radiant
energy to obtain an electrical or other readily measured signal.
2. The radiation may actuate a photoelectric device of some sort.
In such a device the electrical signal is produced by the direct conversion of the energy in the radiati(xa-*ather than through its heating
effect.
3. The radiation may be permitted to fall on a photographic emulsion for a controlled period of time. The density of the silver deposit
produced in the emulsion on development can be made a measure
of the total radiation.
4. The radiation can be determined by the human eye. So severe
are the limitations of the eye as a light-measuring device, however,
that it is useful only for special types of photometry, some of which
will be discussed in Chapter 14.
These four methods are compared in Table 12.1. The significance
of the columns in the table is as follows: Wavelength range means the
spectral region over which the method can be used. Sensitivity
(sometimes called responsivity) is the slope of the curve relating the
response of the receiving device in each method to the radiant energy
required to produce that response, whereas linearity refers to the
closeness with which such a curve approaches a straight line. For
many receivers this curve for radiation of one .wavelength will differ
from that for another wavelength. If the differences are negligibly
small, the receiver, has high neutrality; that is, it responds as well to
one wavelength as to another. The cumulative property and the
300
c^l^\^
T H E M E A S U R E M E N T OF S P E C T R A L I N T E N S I T I E S
301
panoramic 'property are possessed to a significant extent only by the
photographic emulsion. The former is the ability of the emulsion to
respond to light of exceedingly low intensity by prolongation of the
time of exposure. The 'panoramic property means that the photographic emulsion can simultaneously register different beams of radiation on different parts of the plate or filin.
TABLE 12.1
PROPERTIES OF VARIOUS METHODS
FOR THE MEASUREMENT OF SPECTRAL INTENSITIES
Method
Wavelength
range
Sensitivity
Linearity
Radiometric
Photoelectric
Photographic
Visual
All wavelengths
10-30,000 A
10-12,000 A
4000-7500 A •
Low
High
High
High
Excellent
Good
Poor
Very poor
Method
Neutrality
Cumulative
property
Panoramic
property
Radiometric
Photoelectric
Photographic
Visual
Excellent
Poor
Poor
Poor
None
Fair
Good
None
None
None
Excellent
Limited
It is apparent from a glance at Table 12.1 that no single method of
radiation measurement is superior to the others. The most suitable
one for a given purpose depends on the details of the spectroscopic
procedure, the speed with which results must be obtained, and other
factors. In this and the following chapter, the various methods are
described, and the circumstances to which each is best suited are
discussed.
RADIOMETRY
A radiometer is any device for the detection and measurement of
radiant energy by means of its heating effect. Since the heating
effect is strictly proportional to the amount of the radiant energy
which does the heating, accurate measurement of the heating gives
302
THE MEASUREMENT OF SPECTRAL INTENSITIES [§12.1
an accurate indication of the incident radiant energy if none is
allowed to go astray by reflection or otherwise. Therefore the
radiometric method is usable for the measurement of radiation of any
wavelength that can be effectively absorbed. Absorbing surfaces can
be made that are more or less uniformly black to radiation from the
vacuum ultraviolet through the far infrared, and in consequence the
radiometric method can be used throughout the entire optical range.
At wavelengths below 1 n, however, the sensitivity of radiometric
devices is markedly inferior to that of photoelectric and photographic
detectors. Radiometric methods are therefore seldom used in this
part of the spectrum except for calibration purposes or other circumstances under which linearity and spectral neutrality of response are
essential.
On the other hand, only radiomet;ric devices can be used at wavelengths longer than about 3 ii because no other kind of detector is sensitive in this region. This statement means that all infrared spectrometers except those operating in the photoelectric infrared use thermal
detectors. Because of the importance of thermal detectors for this
purpose, they will be considered in some detail, along with the
auxiliary equipment needed for amplification and recording of the
detected radiation.
Of the many physical properties of substances that are dependent
on temperature, the change of electrical resistance with temperature
and the thermoelectric effect lend themselves most readily to the
detection of minute temperature changes. Devices which use
these effects are known, respectively, as bolometers and thermocouples.
12.1. Bolometers. The bolometer is a device, usually in the form
of a short, narrow strip, for the detection of radiation by the change of
electrical resistance that accompanies the temperature rise produced
in the device by radiation. Since the temperature rise produced by
a given amount of radiant energy will be greater as the heat loss and
specific heat of the bolometer are smaller, it is desirable to keep the
mass of the bolometer to a minimum. In a small bolometer the rate
of the temperature rise will also be faster than in a large one, which
may be useful if a greater speed of response is desired. For a given
temperature rise, the change in electriqal resistance will depend on the
temperature coefficient of resistance, which suggests the use of materials with high temperature coefficients.
The bolometer is ordinarily used in some modification of the
§12.2] THE MEASUREMENT OF SPECTRAL INTENSITIES
303
Wheatstone bridge circuit, in which one of t h e other arms of the bridge
is another bolometer strip. Radiation to be measured does not strike
this latter strip, which otherwise is subject t o t h e same environmental
influences, including bridge current, as those operating on the active
strip. T h e constant bridge current flowing through the two strips is
called t h e heating current, because it results in a bolometer t e m p e r a t u r e
t h a t m a y be 50°C above the ambient t e m p e r a t u r e . T h e unbalance
in the bridge caused b y radiation produces a voltage linearly proportional t o t h e radiation and also proportional t o the heating current.
T h e o p t i m u m measurement of this voltage demands a careful matching of t h e electrical characteristics of t h e bridge circuit to those of t h e
amplifying a n d recording system, which adds another factor t o be
considered in t h e choice of bolometer material a n d design.
Because of t h e number and variety of t h e factors involved, n o
single bolometer design has a clear-cut superiority over all others.
This situation is reflected in the n u m b e r of different materials and
designs t h a t h a v e been used successfully and in the lively controversies
over their respective merits.
12.2. M e t a l Bolometers. M e t a l strips can b e produced and handled
which are as t h i n as 0.1 ft. I t is hardly convenient t o make t h e m
smaller t h a n a few millimeters in length, however, and since in t h e
measurement of spectral intensity their length and width are related
t o t h e size of t h e exit slit of the spectrometer, metal bolometers usually
h a v e dimensions of t h e order of 0.5 cm X 0.5 m m X 1 M. I n this
size their electrical resistance will be a few ohms for such metals as
nickel a n d platinum.
I n operation t h e strip m a y be suspended from wire leads or supported on some kind of nonconducting backing. I n t h e latter case
t h e thermal contact between the strip a n d its support will be extensive
a n d t h e t e m p e r a t u r e rise of the strip when exposed to radiation will
be smaller because of heat transfer t o t h e mounting. This h e a t
transfer will increase the minimum a m o u n t of detectable radiation
b u t will speed u p the rate at which equilibrium temperatures are
reached, a result t h a t m a y be desirable. If t h e strip is mounted in a
gas-filled container, it will also lose h e a t b y gaseous conduction.
Evacuation of t h e container increases t h e t e m p e r a t u r e rise produced
by a given a m o u n t of radiation, b u t t h e time, required to reach
temperature equilibrium will be simultaneously increased.
Whether t h e advantage of increased sensitivity obtained by v a c u u m
304
THE MEASUREMENT OF SPECTRAL INTENSITIES
[§12.3
operation is offset by the increased time of response depends on the
amplifying and recording system. In general, evacuation of the
bolometer housing is desirable. However, the strip can be operated
with higher bridge currents when gas conduction is available to remove the electrical heating. For this reason metal bolometers have
been operated on occasion under a few millimeters pressure of
hydrogen gas.
Most metals are good reflectors in the infrared, and therefore metal
bolometers must be blackened by evaporated metallic blacks to
absorb the radiation."^ The amount of blackening is rather critical,
since too much of it will both impede the flow of heat from the black
to the strip and increase the heat capacity of the bolometer.
At room temperature, the temperature coefficients of resistance of
metals used as bolometers are about 0.3 to 0.5 per cent per degree
centigrade. By way of illustration of the orders of magnitude involved, the following figures are given for a nickel bolometer of about
20 ohms resistance. In a certain electrical setup, a minimum change
in resistance of about 10""^ ohm can be detected, corresponding to a
temperature rise of the order of 10^^°C. The amount of radiant
power required to produce this temperature -rise depends on the
structure of the bolometer and, for a given bolometer, on the way in'
which the radiant power is supplied. If no heat were lost by any
mechanism, however, such a temperature rise would be produced by
an amount of radiant energy equal to the specific heat of the bolometer material (nickel) times its mass. For a bolometer of lO"" gram,
the product will be 4 ergs per degree or 4 X 10"* erg for 10~*°C. If
radiant power is supplied to the bolometer at the rate of 1 microwatt,
none of which is lost by reradiation and other processes, 4 X 10""^ sec
would be a sufficient time to raise the temperature 10~5°C; 0.04 sec
would be required if the power is 10""* /xw. This latter figure is
somewhat smaller than the minimum detectable power realized in
practice with metal bolometers used in infrared spectrometers.
Details of the construction and use of metal bolometers will be
found in many articles and books (see General Reference 12.5).
12.3. Semiconductor Bolometers. The large (negative) temperature coefficient of semiconductors (for example, -^ 15 per cent per °C
for cuprous oxide, as compared with +0.3 per cent for nickel) makes
1 A. H. Pfund, Rev. Set. Inst., 1, 397 (1930); Jour. Opt. Soc. Am., 23, 270, 375 (1933).
§12.5] THE MEASUREMENT OF SPECTRAL INTENSITIES
. 305
them potentially valuable as bolometer materials, but until recently
practical difficulties sucb as their very high resistance prevented
their use. Becker^' ' and coworkers have described briefly a
"thermistor" type of semiconductor suitable for use as a bolometer
material in infrared spectrometers. The temperature coefficient of
the thermistor material is about —5 per cent per °C, and the resistance of a strip 3 X 0.2 X 0.01 mm is some 4 megohms. According
to the references cited, such a bolometer, mounted in thermal contact
with a quartz backing, will respond with a temperature rise of
2 X 10"" °C when irradiated at the rate of 2 X 10'* watt for 3
milliseconds. This rise corresponds to a resistance change of about
0.3 ohm and, under operation in a particular Wheatstone bridge
circuit that includes a 400-volt drop across the bolometer, results
in a bridge output of 3 X 10""" volt.
One shortcoming of the thermistor bolometer is its appreciable
transmission of radiation in the neighborhood of 6 /i. This is not
too serious a matter, however, and can be minimized by coating the
bolometer with blackening or with some other material that absorbs
uniformly in the 6 ju region.
12.4. A Superconductor Bolometer. The tremendous resistance
change associated with the transition from the normal to the superconducting state of certain metals and semiconductors at very low
temperatures suggests the possibility of a superconductor bolometer.
This possibility has been realized by Andrews and coworkers,^ who
utilized the semiconductor columbium nitride, found by Horn^ to
become superconducting at .about — 257°C. The temperature coefficient of this substance in the transition range is as much as
5000 per cent per °C. It is evident that the difficulties of bolometer
operation at these very low temperatures are considerable, but certainly the utility of the superconductor bolometer for spectrometric
purposes deserves study.
12.5. Thermocouples and Thermopiles. The thermoelectric effect, in which two similar bimetallic junctions kept at two different
temperatures generate an electromotive force, may be used to detect
2 J. A. Becker and H. K. Moore, Jour. Opt. Soc. Am., 36, 354 (1946).
' W. A. Brattain and J. A. Becker, Jour. Opt. Soc. Am., 36, 354 (1946).
* D. H. Andrews, R. M. Milton, and W. DeSorbo, Jour. Opt. Soc. Am., 36, 518 (1946).
= See F. H. Horn, W. F. Brucksch, Jr., W. T. Ziegler, and D. H. Andrews,
Rev., 61, 738 (1942).
306 .
THE MEASUREMENT OF SPECTRAL INTENSITIES [§12.5
radiant energy by the temperature rise the radiation produces in one
junction. If the two junctions are constructed as nearly alike as
possible and are subjected in use to the same conditions, the temperature difiFerence between the two junctions can be restricted essentially
to that produced by the irradiation of one junction. The junction
receiving the radiation is known as the active junction and the other
as the compensating junction, and the thermocouple is said to be
compensated.
The voltage developed by a small temperature difference between
a pair of junctions depends linearly on this difference and on the
thermoelectric powers of the two metals. A. given quantity of radiation will produce a larger temperature rise in a system of lower heat
capacity; and therefore the heat capacity, and hence the mass, of the
junction should be kept as small as possible. The selection of the
metals for the junction, however, cannot be made simply on the basis
of thermoelectric powers alone. The thermal and electrical conductivities are also involved, as is shown in detail in the theory as
developed by Cartwright.* From a consideration of the various
factors involved, Cartwright has reached the conclusion that a junction of pure bismuth with an alloy of 5 per cent tin and 95 per cent
bismuth repreSgnJA a satisfactory compromise among the several
conflicting factors involved. However, various other materials have
been used with equally good or better results' (see also General Reference 12.5).
The construction of a compensated vacuum thermocouple using the
above metal-alloy junction is described in great detail by Strong and
Cartwright.* To keep the heat capacity of the couple small, wires
of pure bismuth and bismuth-tin alloy are made about 3 mm long
and 0.625 mm in diameter. The soldered junction of these fine wires
is scarcely larger than their diameters and is thus much too small for.
use with a spectrometer, the exit slit of which is many times greater.
The transfer of radiant energy to the junction is made with the help
of a thin metal strip called a receiver, which' is about the same shape
and size as a conveniently formed image of the widened exit slit.
The receiver is cemented or soldered to the junction to give good
" C. H. Cartwright, Zeitschr.f. Physik, 92, 1S,S (1934). .
' D . F. Hornig and B. J. O'Keefe, Rev. Sci. Inst, 18, 479 (1947).
' J. Strong and C. H. Cartwright in Chapter VIII, General Reference 12.3.
§12.5] THE MEASUREMENT OF SPECTRAL INTENSITIES
307
thermal contact and is blackened to increase its absorption. In
physical characteristics the receiver differs little from a bolometer
strip, but its function is simply to collect radiation and transfer it
to the thermocouple. In order to simulate the active junction as
closely as possible, the compensation junction is also equipped with
a receiver, which, however, is not subjected to the radiation to be
measured.
The thermocouple is mounted in a case which is evacuated to
10""^ mm or better. The use of a vacuum reduces the loss of heat by
gas conduction from the couple and may increase the sensitivity of
the couple to radiation as much as twentyfold. However, there is an
attendant decrease in the speed of response, which may have certain
disadvantages discussed below in connection with amplifiers.
One way of increasing the electromotive force generated by thermoelectric means is to use several thermocouples in series, an arrangement called a thermopile. The many factors that must be considered*
make it impossible to say dogmatically that the improved performance
of a thermopile over a single couple will warrant the trouble of making
the extra junctions. Actual practice indicates a widespread conviction to the contrary on the part of infrared-research workers who
make their own detectors as well as by commercial concerns producing
infrared spectrometers. There is a technique for the production of
thermopiles, however, by which multiple junctions are just as readily
produced as single junctions. The procedure^" consists of making
the metallic junctions by successive deposition of the two metals in
a vacuum evaporator. A separate pattern or mask is used during the
evaporation of each metal to form the strips of that metal, and the
junctions occur at areas where the two masks have open areas in
common. The metals, usually bismuth and antimony, are deposited
on a thin supporting film of plastic such as cellulose acetate or
Formvar. The evaporation technique has also been used for the
fabrication of fast thermocouples" and bolometers.'^
' See Cartwright, footnote 6, page 306.
1" L. Harris and coworkers. Rev. Sci. Inst., 4, 454 (1933); 5, 153 (1934); Jour. Opt.
Soc. Am., 30, 519 (1940).
" See, for example, L. C. Roess and E. N. Dacus, Rev. Sci. Inst., 16, 164 (1945).
"2 See, for example, B. H. Billings et al.. Jour. Opt. Soc. Am., 36, 354 (1946).
308
THE MEASUREMENT OF SPECTRAL INTENSITIES [§12.6
Other thermopile design and construction details will be found in
articles'ihy Hornig and O'Keefe^, Pfund^^ and Coblentz (General
Reference 12.1, page 191).
12.6. Other Thermal Detectors. Various detecting devices in
addition to bolometers, thermocouples, and thermopiles have been
used extensively at one time or another for the measurement of
infrared radiation. For sundry reasons they find little application
today. The vane radiometer, most familiar in the form of the simple
Crookes radiometer, has been refined by Nichols and others'^ to a
point where it is at least as sensitive, if not more so, than the detectors
considered above. However, the construction of these instruments is
so difficult, their maintenance and operation demand such elaborate
precautions, and they require so much time for a single reading that
they are rarely used today for spectroscopic purposes.
Another detecting device that has interesting possibilities is the
pneumatic radiometer known as the Golay cell. This device, a
predecessor of which is the Hayes cell,^^ measures radiation by the
pressure increase in a gas chamber accompanying the temperature
rise caused by absorption of the radiation." The radiation is not
absorbed by the gas itself but by a thin metal film in contact with the
shown in Fig. 12.1. The temperature of the gas is raise'd then
by gaseous conduction of heat away from the film. The small
pressure increase is observed by the deflection of one of the walls of
the gas chamber, which is made very thin and flexible. Measurement'
of the deflection can be made optically or electrically. In the Golay
cell, shown in Fig. 12.1, the deflectable wall is used as a mirror, and
the amount of light reflected from it through a matched gridwork is
measured photoelectrically. It has a time constant of about 3 X 10^"*
sec and in comparison with other types of infrared receivers is reported to show a sensitivity" several times better than those of
conventional bolometers and thermocouples. The Golay cell is made
commercially by the Eppley Laboratories of Newport, R. I.
" A. H. Pfund, Rev. Sci. Inst., 8, 417 (1937).
» E. F. Nichols, Phys. Rev., 4, 297 (1897).
15 H. V. Hayes, BCT. Sci. 7nsi., 7, 202 (1936).' '
i«H. A. Zahl and Marcel Golay, Rev. Sci. Inst, 17, 511 (191,6). See also R. A.
Weiss, Jour. Opt. Soc: Am., 36, 350 (1910).
" See, for example, H. H. Nielsen et al.. Jour. Opt. Soc. Am., 36, 338 (1946).
§12.7] THE MEASUREMENT OF SPECTRAL INTENSITIES
309
12.7. Amplification and Recording Methods in Radiometry. The
electrical output of bolometers and thermocouples under steadyradiation is a small DC voltage that may be measured directly with
a sensitive galvanometer of appropriate characteristics. When the
galvanometer is pushed to the limit of its sensitivity, however, in the
measurement of voltages near 10""^, serious difficulties are encountered.
A sensitive galvanometer is sensitive to other things besides the
e.m.f. the experimenter wants to measure, particularly to mechanical
vibrations and stray electrical interference. These can be reduced
respectively by supports such as the Julius suspensions^ and by
Image of upper half of line grid when flexible mirror is flat
Fig. 12.1. The pneumatic radiometer or Golay cell.
(Courtesy the Eppley Laboratory.)
careful shielding, but often cannot be eliminated. A galvanometer
suspension also indulges in erratic torsional fluctuations associated
with Brownian motion, which ultimately set a limit to the voltage
which can be measured. Moreover, the relatively long time required by a sensitive galvanometer to come to full deflection is a
decided inconvenience when thousands of readings have to be
taken.
In addition to these troublesome features, a more serious difficulty
arises from the imperfect compensation of thermocouples and bolometers. As a result, the "zero reading" of the detector output in the
absence of radiation does not stay at zero but slowly changes in one
direction or the other. This phenomenon, called drift, is troublesome
'* Vibrationless mountings tor galvanometers are discussed in Chapter XIV of
General Reference 12.3.
310
THE MEASUREMENT OF SPECTRAL INTENSITIES [§12.8
to correct and adds much labor in the form of extra zero readings to
the process of obtaining an infrared absorption spectrum with a
galvanometer and scale. Drift can also introduce inaccuracy in
galvanometer readings when the drift during one reading is an
appreciable fraction of the deflection.
It has proved possible to eliminate drift by virtue of the fact that
its rate of change with time is slow. The method (see below, §§ 12.8
and 12.9) is to "chop" the radiation from the infrared source with a
shutter several times a second and then to amplify the output of the
thermal detector with an AC amplifier tuned to the chopping frequency. To eliminate drift by this kind of procedure, as well as to
minimize the other difficulties mentioned above and to reduce the
labor of obtaining an infrared spectrum, various kinds of amplifiers
and automatic recorders have been introduced.
12.8. Photorelays. The photorelay is a device for amplifying the
deflections of a primary galvanometer by means of a photocell and a
secontl_^galvanometer. The first photorelay, that of Moll and
Burger," actually used a thermopile rather than a photocell, but the
principle of later devices is the same. Light reflected from the mirror
of the first galvanometer falls on the surface of a sensitive photocell
of some kind. The photocurrent thereby generated is sent to a
second galvanometer but is balanced potentiometrically so that at
the zero reading of the first galvanometer the second galvanometer
also reads zero. When the mirror of the first galvanometer suffers a
slight deflection, the amount of light falling on the photocell changes
and a current flows through the second galvanometer. The optical
system can be arranged to give a linear relationship between the two
galvanometer deflections. Since the magnitude of the photocurrent
can be made quite large by use of an appropriate optical arrangement
and intense light source, tremendous amplification of the primary
galvanometer deflection is possible. I t is relatively'easy by means
of the photorelay to amplify the primary galvanometer deflections to
such an extent that the Brownian fluctuations therein are readily
observable.
Of the several modifications of the photorelay,^" that made by
" W. J. H. Moll and H. C. Burger, Phil. Mag. (6), 50, 621 (1925).
2»A. H. Pfund, Science, 69, 71 (1929); R. B. Barnes and R. Matossi, Zdtschr. f.
Physik, 70, %i (1932); C. II. Cartwright, Rev. Sci. Inst., 3, 221 (1932).
§12.9] THE MEASUREMENT OF SPECTRAL INTENSITIES
311
Pfund and called by him the "resonance radiometer" is t h e most
significant. T h e photorelay described in the preceding p a r a g r a p h
unfortunately does n o t eliminate zero drift. Pfund pointed out t h a t
by tuning the primary and secondary galvanometers t o the same
period, one can make the photorelay particularly sensitive t o electrical
impulses of t h a t period and much less sensitive to other impulses,
especially those of drastically differing period,* such as a slow zero
drift. T h e response of the thermocouple or bolometer to radiation
is made t o v a r y with the period t o which the galvanometers are t u n e d
by " c h o p p i n g " the radiation a t t h a t period. A rotating sector or
pendulum-controlled shutter m a y be used for this purpose.
v!
I n actual practice, the resonance radiometer is somewhat complicated t o use and has a relatively slow response time, b u t t h e
fundamental idea of eliminating drift by periodic interruption of t h e
radiation a n d sharp tuning of t h e photorelay t o t h a t period is very
sound. T h e replacement of the photorelay by a t u n e d A.C amplifier
results in la simpler over-all system, in widespread use.
12.9. Alternating-Current Amplifiers. I n recent years great advances have been made in the design of vacuum-tube amplifiers,
particularly in the development of low-frequency, sharply tuned a m plifiers. One of the most significant of these advances, due t o
Scott,^' is t h e so-called " T w i n - T " negative feedback circuit. I n this
circuit, sharp tuning of the amplifier t o a narrow band of frequencies
is achieved by negative feedback from the late stages of the amplifier
t o t h e first stage. T h e negative feedback cancels t h e gain of t h e
amplifier, b u t b y the insertion of sharply t u n e d filters in the feedback
line, a narrow range of frequencies can be eliminated from the feedback. F o r this frequency range, the gain of the amplifier is not
nullified. W i t h such an arrangement it is possible t o construct
amplifiers of low frequency (10 cycles per second or less) and very
narrow pass b a n d (1 to 2 per cent of the t u n e d frequency is a t t a i n able). Several such amplifiers for use with bolometers or thermocouples in infrared spectrometers have been described.'' ^^' ^'
* The resonance radiometer does not eliminate the effect of Brownian motion, since
the random fluctuations of the primary galvanometer are superimposed on its harmonic oscillations.
» H. H. Scott, Proc. Inst. Radio Eng., 26, 226 (1938).
™L. C. Roess, Rev. Sci. Inst., 16, 172 (1945).
^ N. Wright and L. W. Herscher, Jour. Opt. Soc. Am., 37, 211 (1947).
312
THE MEASUREMENT OF SPECTRAL INTENSITIES [§12.9
The specifications of the AC amplifier of course depend on the
radiation detector and the measuring and recording system with
which it is used. Clearly, the frequency at which the radiation
beam is interrupted must not be so high that the detector is unable
to respond to the interruptions. It was remarked above in the discussion of bolometers and thermocouples that those features of a
detector which make it very sensitive, such as vacuum housing, also
decrease the speed with which maximum response is attained; and
in general it may be said that high speed of response is gained only
at a price of decreased sensitivity, and vice versa. The chopping
frequency must therefore be a compromise that is not too high for
the speed of the detector and not too low for good amplifier design
and recording speed. Chopping speeds as low as 1 cycle per second ^^
have been used with vacuum thermocouples, * and as high as 40 cycles
per second with bolometers.' A compromise of 15 cycles per second
has been reported for a thermistor bolometer mounted on a glass
backing.'
In addition to the great advantage of the tuned AC amplifier for
the elimination of drift, good amplifier design will also eliminate
sensitivity to mechanical vibration and to stray electrical interference. The electrical analogue of the Brownian motion of the
galvanometer suspension, however, is still present in the form of
random voltage fluctuations (the so-called Johnson noise^^) in the
thermocouple or bolometer. The effect of Johnson noise may be
reduced by sharpening the tuning of the amplifier, since the noise is
proportional to the square root of the pass-band width. This kind
of reduction is limited, however, by considerations of speed. The
response time of the amplifier is essentially the reciprocal of the band
width; and halving the band width, which will reduce the noise by
only I / V 2 , doubles the response time. Thus the optimum band
width is a compromise between low noise and speed. There is obviously no point to making the amplifier Response time materially
faster than that of the indicating device to which the amplifier output
is fed. Since this latter is usually not much faster than 1 sec, band
widths are rarely narrower than 1 cycle per'second. Many amplifierrecording systems permit control of both noiSe level and recorder
* Vacuum thermocouples need not be this slow.
a factor of 10 or more.
^* J. B. Johnson, Phys. Rev., 32, 97 (1938). .
They have been made faster by
§12.11] THE MEASUREMENT OF SPECTRAL INTENSITIES
313
speed so that the optimum combination of the two factors may be
selected.
12.10. A Direct-Current Amplifier. Despite the advantage of the
AC amplifier in the elimination of zero drift, there is a DC amplifier* that has been widely adopted for use in conjunction with
a vacuum thermocouple. This amplifier^^ is a DC amplifier in
the sense that a small DC voltage fed into the input terminals
appears as a highly amplified DC voltage at the output terminals.
As far as the circuit is concerned, it is an AC amplifier not drastically
different from those described in the references cited in the previous
section. The conversion of the DC input to low-frequency alternating current (75 cycles per second) is accomplished by a motor-driven
commutator. The same motor simultaneously drives a second commutator that rectifies the amplified 75-cycle output to direct current.
Used in conjunction with a commercial pen recorder, the amplifier
permits measurement of voltages as low as 10^^, which is approaching
the Brownian motion limit of a sensitive galvanometer.
12.11. Recorders. The wearisome task of taking thousands of
galvanometer scale readings in the course of charting a single spectrum makes highly desirable some method for automatic registration
of these readings. For rapid industrial work this is imperative.
Automatic recording was, in fact, the first of the automatic techniques
to be used and was introduced as long ago as 1895 by Langley and
Angstrom. The recording method consisted, in principle, of replacing
the galvanometer scale by photographic paper wrapped on a cylinder,
the axis of the cylinder being parallel to the long dimension of the
scale. Rotation of the cylinder at a fixed rate with respect to the
traversal of the spectrum causes the galvanometer spot to trace out
a permanent record of its displacements and establishes a definite
relationship between angular position of the cylinder and the spectrometer setting.
The great advantage of recording on photographic paper is its
ready addition to the standard galvanometer setup, and the principal
disadvantage is the delay involved in the photographic processing.
It has been used widely, and numerous special photographic arrange-
* This amplifier may be purchased from the Perkin-Elmer Corporation, Glenbrook,
Conn.
^ M. D. Liston, C. E. Quinn, W. E. Sargeant, and G. G. Scott, Rev. Sci. Inst.,
17, 194. (1946).
314
THE MEASUREMENT OF SPECTRAL INTENSITIES [§12.11
ments have been described. Reproductions of photographic records
have often appeared in the literature."' '"^' "• ^^
A simple device which makes possible the substitution of pen-andink for photographic recording has been described by Pompeo a n d
Penther.^' T h e light beam from the final galvanometer falls on a
split-cathode photocell m o u n t e d on a carriage holding a pen. T h e
photocell o u t p u t controls a motor t h a t drives the carriage in such
fashion t h a t the galvanometer beam is always centered on t h e photocell. T h e carriage track is m o u n t e d parallel to the axis of a revolving
cylinder, and t h e moving pen records on paper wrapped a b o u t t h e
cylinder.
T h e development of high-gain amplifiers like those described in t h e
preceding section occurred simultaneously with the commercial
availability of fast, low-voltage pen-and-ink recorders suitable for use
with t h e m . M a n y of these recorders are well suited for use with
radiometric devices, and accordingly most of the infrared spectrometers now on t h e m a r k e t are equipped with standard commercial
recorders. T h e Electronik high-speed strip recorder of the Brown
I n s t r u m e n t Company, shown in Fig. 12.2, is a representative example.
PHOTOELECTRIC
MEASUREMENT
OF
INTENSITY
M a n y of the electrical properties of m a t t e r are affected by light,
so there are various kinds of "photoelectric effects." Three kinds of
photoelectric phenomena have been a d a p t e d to t h e measurement of
spectral intensity: the photoemissive effect, which is t h e ejection of
electric charge from m a t t e r through the agency of radiation; t h e
photocondudive effect, which is the change of electrical conductivity
produced by radiation; and the photovoltaic effect, which is the generation of a potential difference between two electrodes as a result of t h e
irradiation of one of t h e m .
Since a b o u t 1930 the characteristics of photocells of these t h r e e
types have been improved so much t h a t photoelectric m e t h o d s are
now used t o a n extent comparable with photography in t h e detection
a n d measurement of spectra. Especially striking advances have been
m a d e in (a) t h e reduction of t h e threshold of radiant power t h a t is
^ F, A. Firestone, Rev. Sci. Inst., 3, 163 (1933).
" N. Wright and H. M. Randall, Phys. Rev., 44, 39 (1933).
28 N. Wright, Ind. Eng. Chem., Anal, ed., 13, 1 (1941).
29 D. J. Pompeo and C. J. Panther, Rev. Sci. Inst., 13, 218 (1942).
§12.11] THE MEASUREMENT OF SPECTRAL INTENSITIES
315
needed to produce a recognizable response from the photocell;
(b) development of techniques for the manufacture of stable, reproducible photosensitive surfaces; (c) extension of the spectral range
of sensitivity well into the infrared and ultraviolet regions; and
(d) development of new means for amplification of the photocurrent.
These improvements are still in progress, and the procedures described below for the photoelectric measurement of spectra will
unquestionably undergo important changes in the future.
In the evaluation of a photocell as a measuring device for spectral
radiation, the properties of primary importance are its sensitivity.
Fig. 12.2. The "Electronik" high-speed strip recorder.
(Courtesy Brown Instrument Company.)
linearity, wavelength range, neutrality, and threshold of detection.
These properties have been defined, with the exception of threshold,
which may be taken as the radiant power needed by the cell to produce
an electrical signal that is at least as large as the electrical noise generated in the cell. Unfortunately, it is difficult to determine an
absolute value of the noise, because a measured value depends not
only on the cell itself but also on the characteristics of the amplifying
and indicating systems. For this reason comparisons of the thresholds of detection in photocells of different types must be made with
care and with due allowance for any differences in circuits.
316
THE MEASUREMENT OF SPECTRAL INTENSITIES [§12.12
12.12. Photoemissive Cells and Electron-Multiplier Tubes. For
the measurement of spectral intensity, the photoemissive cell is probably the best all-round photocell with respect to all the above characteristics except wavelength range. It is linear over widely different
intensities of illumination. Its threshold of detection is low at room
temperature and can be made still lower by refrigeration, if necessary.
The sensitivity is high and the working stability of the cell is satisfactory. A typical cell is shown in
Fig. 12.3, and a convenient circuit for
the measurement of low light levels in
Fig. 12.4.
In the form of the electron multiplier,
the photoemissive cell promises to be
more widely used in spectrophotometers
than any other kind of photocell. This
is a combined photoemissive cell and
electronic DC amplifier in a single
envelope. Its operation is illustrated
in Fig. 12.5. The basic principle is the
phenomenon of secondary electron
emission from a treated metal surface
under bombardment by primary electrons. The importance of the pheK
l^^^HI
nomenon lies in the fact that a single
•f
l^^^B*
M primary electron may give rise to four
HL .
^ ^
M or five secondary electrons. In the
'
•*• electron multiplier the first primary
electrons are those ejected by radiaFig. 12.3. A typical
tion falling on the photocathode C.
photoemissive cell.
(Courtesy RCA Laboratories.)
These electrons are accelerated to
the first anode (electrode No. 1) by
an electrical potential of, say, +100 volts applied to it. The
photoelectrons impinge on the first anode with sufficient velocity to
eject four or five secondary electrons each. The newly released
secondaries are then accelerated to electrode No. 2 by a higher
potential (for example, +200 volts above the photocathode), where
each in turn gives rise to several secondary electrons, The process
continues in this fashion through a number of stages (six in Fig. 12.5),
the number of electrons being multiplied at each stage by a constant
factor of approximately 4.5. The electrons from the final electrode
1
§ 1 ^ 1 3 ] THE MEASUREMENT OF SPECTRAL INTENSITIES
317
are collected a t the anode A, whence they are conducted away a n d
nieasured as photocurrent.
T h e photoelectric properties of the electron multiplier are determined by the photocathode. T h e over-all amplification m a y range
FP-54
Fig. 12.4. A simple photocell circuit for the measurement of low light levels
(order of microlumens). Representative values of the electrical components
would be; 5i = 2 volts; B2 = B3 = B4 = B5 = i volts; RL = 1000 megohms;
R^ is a balancing resistor; P = photocell {e.g., RCA-929); C = photocathode;
G = galvanometer.
from one t h o u s a n d for a six-stage t u b e to as much as a million for a n
11-stage t u b e . For precise spectrophotometry, the potentials supplied t o the several stages must be carefully controlled (to ± 0 . 1 per
cent approximately), because a voltage fluctuation of x per cent in a n
ri-stage multiplier changes the amplification factor by about nx per
cent. Suitably stable power supplies have been designed,'"' '^ or
C
2
Path of electrons:
Photocathode
Electrode 1
2
3
4
5
6
A
0 volts
+100
+200
„
+ 300
„
+ 400
..
+ 500
..
+ 600
„
+ 700
..
Fig. 12.5. Diagram of a six-stage electron multiplier tube.
B batteries m a y be used. T h e spectral sensitivity curves of t w o
types of photocathodes are shown in Fig. 12.6.
12.13. Photoconductive cells. When photoelectric detection of
spectra beyond 1 ^ is desired, it is necessary to use photoconductive
3» F. V. Hunt and R. W. Hickman, Rer. Sci. Inst., 10, 6 (1939); \V. R. Hill, Proo.
Inst. Radio Eng., 33, 38 (1945).
» G. H. Dieke and H. M. Crosswhite, Jour. Opt. Soc. Am., 35, -171 (191.5).
318
THE MEASUREMENT OF SPECTRAL INTENSITIES [§12.14
cells; these can be constructed with workable sensitivity to wavelengths as great as S ;i or even longer. In the past the most practical
photoconductive cells used a thin film of elemental selenium as the
material whose electrical conductivity increased markedly upon
illumination, although many other substances with this property
were known. Cells of superior qualities are now available whose
photosensitive element consists chiefly of the sulfide of lead or of
thallium.'^
The thallous oxysulfide or "thalofide" cell shows optimum response
to radiation of about 0.95 yu> and has a useful sensitivity to about
1.4 /i. I t is easy to use and has quite a low threshold of detection.
Its chief shortcoming is a nonlinear sensitivity curve. The lead
sulfide cell is usable out to 3 n or even beyond. Despite the wide
1.0
- . >v
4) •*-
-'.>
OB
\ /\\///A\
W
y^
\\ \x\Ar
j K.
1I \\y/
\ri /
0.6
0) c 0 4
O
a: C
a>
0.2
I\
^ '"^
\
^^
^^^-^
\
no
D\
—
0
1
I-
2
3
Fig. 12.6. Spectral sensitivities of photocells. Each curve is relative to its
own maximum sensitivity. ( 4 ) Cesium-antimony photoemissive cell; (B)
cesium oxide-silver photoemissive cell; (C) thalofide photoconductive cell;
(/)) lead sulfide photoconductive cell.
range of sensitivity, it is relatively neutral over most of this range, in
contrast to the thalofide cell, which has a very steep sensitivity-vs.wavelength curve in the range 0.95 to 1.5 /t. The threshold of detection of the lead sulfide cell is higher than that of the thalofide cell, but,
refrigeration of the lead sulfide surface with "dry ice" both lowers
the threshold and extends the usable range to longer wavelengths.
Like the thalofide cell, the lead sulfide cell'is nonlinear.
'
The spectral sensitivity curves of thalofide and lead sulfide' cells
are shown in Fig. 12.6. Other kinds of photoconductive materials,
for example lead selenide, show promise of workable sensitivity at
even longer wavelengths.
12.14. Photovoltaic Cells. In the form of the "barrier-layer" cell,
the photovoltaic cell is probably the most convenient kind of photocell
52 R. J. Cashman, Jour. Opt. Six:. Am., 36, 356(A), (1946).
§12.15] THE MEASUREMENT OF SPECTRAL INTENSITIES
319
obtainable. It requires no auxiliary voltage supply, since it utilizes
the incident radiation to generate rather than to modulate a voltage.
Under proper circumstances its linearity is good, and for any photometric purpose where plenty of light is available, the photovoltaic cell
is simple and easy to use. Unfortunately, most spectrophotometers
run short of light at one end or the other of their operating spectral
range, and the low sensitivity of the photovoltaic cell in these regions
is a great handicap under such circumstances. In addition, the
amplification of the response presents special problems due to the low
impedance of the cell. For these reasons photovoltaic cells are little
used in spectrophotometers. The spectral sensitivity curve of a
typical photovoltaic cell does not differ much from curve A of Fig.
12.6. The peak of sensitivity is at slightly longer wavelength, say
0.56 M.
12.15. The Incorporation of the Photocell in the Spectrograph.
Because the photocell does not discriminate between two different
wavelengths, it is necessary to isolate each wavelength at which
intensity measurement is to be made. The spectrograph in which
photoelectric detection is used must therefore function as a monochromator. A monochromator is provided with an exit slit, and the
first requirement to be met in photoelectric detection is the arrangement of the photocell to utilize the maximum amount of monochromatic radiation emergent from the exit slit. This arrangement is
ordinarily a simple one. The size of the exit slit and the angle at
which the monochromatic beam emerges are often such that the
photocathode area in the photocell can intercept the entire beam if
the photocell is placed at the proper orientation and within the proper
distance from the slit.
If it is desired to place other devices in the region between the exit
slit and the photocell (for example, an absorption cell for absorption
spectrophotometry as in Chapter 14), an optical system is needed to
carry the beam through this region and then to project it upon the
photocathode. The nature of this system depends on the function
it is required to perform. Usually it consists of a collimating lens
to render parallel the beam from the slit, and a focusing lens to project
the beam on the photocathode. A sketch of this optical arrangement
is shown in Fig. 12.7.
A second requirement to be met concerns the method of scanning
the spectrum, that is, the method of bringing the various wavelengths
successively to the exit slit. In most monochromators the spectrum
320
THE MEASUREMENT OF SPECTRAL INTENSITIES [§12.16
is scanned by rotation of the dispersing element, whether it be prismatic or diffractive. Under this arrangement the exit slit is usually
fixed in position, and therefore the entire optical system from the
exit slit to the photocell may also be fixed. No addition is required
to the optical system mentioned in the previous paragraph, apart
from some method of making changes in lens positions and other corrections required by the change in wavelength of the emergent beam.
Sometimes the optical or mechanical arrangement of the monochromator precludes scanning of the spectrum in this way. An
alternative procedure is to keep the dispersing element in a fixed
position and to move the exit slit along the surface in space on which
the spectrum is focused. Necessarily, the emergent beam will move
with the slit, and this motion must be compensated in some way to
keep the beam properly directed onto the photocathode. Such compensation can sometimes be effected by optical means, but it is
E?*'*
silt
First
lens
Absorption
cell
Second
lens
Photocell
Fig. 12.7. Simple optical system for photocell.
usually found preferable to keep the photocell in fixed relationship
to the exit slit. This requires moving the entire slit-photocell system
during the scanning of the spectrum, which is somewhat awkward.
However, with grating spectrographs whose mountings -prevent
scanning of the spectrum by rotation of the grating, such as the
Paschen or Wadsworth mountings, this procedure is always followed.
The construction and-use of various types of photoelectric spectrophotometers are considered in Chapter 14.
12.16. Amplification and Recording of Photocurfents. Because of
the high impedance of both photoemissive and phofoconductive, cells,
electronic ainplification of their photocurrents is readily achieved.
The problem of amplification has two aspects, however, one concerned with the sensitivity of the photocell. and the other with its
threshold of detection.
As defined at the start of the chapter, sensitivity is the slope of
the curve relating the photocurrent to the radiant power w;hich produces it. Amplification of the photocurrent by a given factor multi-
§12.16] THE MEASUREMENT OF SPECTRAL INTENSITIES
321
plies the slope of this curve by this same factor and therefore increases
the sensitivity by this factor. Inasmuch as the currents produced in
a photocell by the light intensities available in spectrometers are
usually so weak as to require a sensitive galvanometer for measurement, amplification of the photocurrent is essential if it is to be
recorded, for example, on a commercial recording milliammeter.
One purpose of amplification, then, is to elevate the photocurrent to
a usefully high level.
The threshold of detection of a photocell, on the other hand, is a
property that cannot be altered (except for the worse) by mere
amplification of the photocurrent. I t was defined previously as the
radiant power required to produce an electrical signal equal to the
electrical noise present for various reasons in the current output of
the photocell. An important point to be remembered is that in
amplification of all components of the photocurrent the noise is
amplified along with the signal, so that such amplification does not
alter the signal-to~noise ratio (except for the worse, in case the
amplifier introduces appreciable noise). Therefore mere amplification
of all photocurrent components does not lower the threshold of
detection.
The electron-multiplier tube furnishes a simple example of these
two aspects of amplification. A certain amount of radiant power
incident upon the photocathode (C in Fig. 12.5) will produce a photocurrent between C and anode 1. The sensitivity of the first stage of
the photocell is the ratio of this photocurrent to the radiant power
which produces it. Now the photoelectrons incident upon anode 1
liberate a larger number of electrons, which in turn impinge upon
anode 2. This larger number constitutes an amplified photocurrent,
and a correspondingly increased sensitivity. The sensitivity increases with each stage in this way, the over-all amplification amounting to several powers of ten. This increase in sensitivity, achieved
so neatly within the confines of a single vacuum tube, is the great
virtue of the electron multiplier.
On the other hand, the threshold of detection of a multiplier tube
is no lower than that of the first stage, because the photocathode is
constantly emitting electrons by virtue of the thermionic effect. The
thermionic current forms the chief part of what is usually called the
dark current because it is best measured in the absence of light, which
would of course give rise to a photocurrent. The dark current depends on the temperature of the cathode, on the potential of the
322
THE MEASUREMENT OF SPECTRAL INTENSITIES [§12.16
first anode with respect to the cathode, and on other factors, but
on the average is constant when all these factors are constant. If the
dark current were absolutely constant, it could be subtracted out of
the total current and would not limit the least detectable signal.
However, the real dark current fluctuates slightly in random fashion
about its average value. These current fluctuations fix a threshold
for the least detectable signal that is about equal to the mean value
of the fluctuations themselves; that is, the ratio of least detectable
signal to the noise is about unity. Any smaller signal tends to get
buried in the noise.
The amplification stages of the multiplier amplify the dark current
with its fluctuations indiscriminately along with the photocurrent.
The ratio of photocurrent to noise is therefore .not altered* during
the amplification, and the size of the least detectable signal is not'
lowered.
In most photoelectric spectrometers the optical system feeds the
photocell sufficient light (except perhaps at the long and short wavelength limits of the instrument), so that the prime objective of the
amplifying system is the elevation of the photocurrent to levels
suitable for recording. Discussion of the many types of amplifying
systems is outside the scope of this book, and the reader is referred to
treatises and journal articles on electronic amplification for information on this subject (General Reference 12.4). However, it is worth
while to consider briefly the possible steps by which the threshold of
detection might be lowered with the help of selective amplification.
We have seen in § 12.9 that amplification by a tuned AC amplifier
of the alternating output of a bolometer or thermocouple can elimi-:
nate long-period drifts in thermoelectric recording. There the objective is the amplification of a signal alternating with much higher
frequency than that of the drift. The fluctuations in dark current
from a photocathode, on the other hand, are not confined to a par^
ticular period but are spread out over a "noise spectrum." This
means that a tuned AC amplifier will amplify some of the noise but
only that part of it which falls in the frequency band to which the
amplifier is tuned. The narrower this band, the smaller the total
fraction of the noise amplified. Hence the threshold of detection can
be lowered if the light to be measured is interrupted by a shutter or
* The amplification stages may introduce noise of their own, thereby raising the
minimum detectable signal. This effect is usually small compared to the one under
discussion.
§12.16] THE MEASUREMENT OF SPECTRAL INTENSITIES
323
"chopper" so that the resultant photocurrent alternates at the
frequency to which the amplifier is tuned.
A practical advantage for the tuned amplification of alternating
photocurrents stems from the fact that photocells have in general a
high speed of response. Because of this high speed, the chopping rate
can be fast; that is, the alternating current from the photocell has a
high frequency. It is easier to construct stable high-gain AC amplifiers for which the frequency is above, say, 100 cycles per second
than for lower frequencies. Hence the difficulty encountered m the
AC operation of thermocouples and bolometers, where chopping
frequencies well below 100 cycles per second must be used, is not met
witji in the amplification of photocurrents.
It might be supposed that the threshold of detection can be reduced
to indefinitely low values by reduction in the band width of the tuned
amplifier. Such a reduction would reduce the fraction of the noise
amplified, to be sure, but would be attended by a complication mentioned in § 12.9: an AC amplifier with a band width of | cycle has
I / V 2 as much noise as a similar one with a band width of 1 cycle,
but it exhibits twice as large a time lag (roughly two seconds compared with one). For this reason, the threshold of detection of
radiant power can be reduced by selective amplification only at the
expense of a much larger increase in the time required to detect a
change in the radiant power.
Since the thermionic current is a function of the temperature of the
photocathode, another possibility for reducing the dark current and
the random fluctuations thereof lies in refrigeration of the photocathode. The anticipated reduction is very great, because of the
exponential factor in the thermionic current-temperature relationship.
In actual practice,'* the reduction in thermionic current that accompanies a temperature drop from room temperature (300°K) to that
of liquid air (90°K) may be as much as ten-thousandfold, which
corresponds to a reduction in thermionic fluctuations of a factor of 100.
The radiant power level which will produce an electrical signal equal
to the noise at 90°K is phenomenally low—of the order of 10~^' lumens
under certain circumstances. Usually it is neither necessary nor
expedient to operate at so low a threshold, but for certain applications, notably the study of the Raman effect (Chapter 18), light
levels of this low value are typical.
' R. W. Engstrom, Jour. Opt. Soc. Am., 37. 420 (^1947), and references there cited.
324
THE MEASUREMENT OF SPECTRAL INTENSITIES [§12.17
The output of an AC amplifier is an alternating current. This
current has to be rectified, either electronically or by some other
means, to be recorded or otherwise registered. Discussion of rectification methods is outside the scope of this book. It is of interest to
note, however, that when the rectified output is fed to a galvanometer
or other output meter, such as a recording potentiometer, the time lag
inherent in such a meter in effect imposes a narrow band width even
on a broad-band amplifier: fluctuations of any kind, whether noise or
signal, which have a period of, say, 0.001 sec, will not be registered by
a galvanometer of a period of a second or larger. In consequence, if a
slow galvanometer is used with a DC or broad-band AC amplifier, the
sluggishness of the galvanometer "filters out" all noise with appreciably shorter period than that of the galvanometer itself. This kind
of filter will not eliminate drift from thermocouples, because the
period of thermocouple drift is so long, but it can be highly useful in
improving the ratio of DC signal to noise from a photocell.
12.17. Photoelectric Spectrometers. Spectrometers using photoelectric detection are manufactured by various firms and are in
widespread use. Several of these, with attachments making them
useful for special purposes, are described in Chapters 14 and 16.
Among the first descriptions of the adaptation of a large-grating
spectrograph to photomultiplier detection and measurement of emission spectra was that of Dieke and Crosswhite.'^ One condition to be
fulfilled in their adaptation was that of easy change of the instrument,
in which the grating was in a Wadsworth mounting, from photoelectric to photographic use. For this reason, a scanning method had
to be devised which interfered with the photographic procedure as
little as possible. The photographic plateholder was therefore
altered so that a small carriage could be mounted on it, and could be
moved smoothly along the focal plane in the direction perpendicular
to that of the exit slit. Upon the carriage were mounted the exit slit
(in such a way that the slit always 'moved in the focal plane) and,'the
electron-multiplier tube. The divergent beam from the slit, fell
directly on the multiplier's photocathode, which was at a distance
no larger than 3 or 4 cm.
The spectrum was scanned by motion of the. carriage along a track
in the plateholder. Since the effective length of the plateholdier was
only 50 cm, the coverage of a wider spectral range involved a resetting
of the entire plateholder in the usual fashion for the Wadsworth
mounting. The fact that the plateholder could be used inter-
§12.17] THE MEASUREMENT OF SPECTRAL INTENSITIES
325
changeably for photographic and photoelectric detection enabled t h e
o p t i m u m focus for a new plateholder setting to be determined photographically. T h e spectral region covered b y t h e new setting could
t h e n be scanned photoelectrically with t h e assurance t h a t n o
further focusing was needed.
T h e o u t p u t of t h e photomultiplier was fed t o a Leeds a n d N o r t h r u p
recording microammeter, usually without intervening amplification.
Figure 1.4 shows photographic a n d photoelectric records of t h e same
emission spectrum t a k e n with this instrument. A sketch of t h e
optical arrangement of the spectrograph is given in Fig. 12.8.
Fig. 12.8. Photoelectric detection of spectra with a Wadsworth mounting.
Si, entrance slit; M, coUimating mirror; G, concave grating; Sa, exit sUt; P,
photocell mounted on carriage; F, focal curve along which exit slit and photocell
move.
GENERAL REFERENCES
12.1.
W. E. Forsythe (Ed.), Measurement of Radiant Energy. New York:
McGraw-Hill Book Company, Inc., 1937.
12.2. H. J. Reich, Theory aM Application of Electron Tubes. New York:
McGraw-Hill Book Company, l a c , 1939.
12.3. J. Strong, Procedures in Experimental Physics. New York: PrenticeHall, Inc., 1938.
12.4. P, E. Terman, Radio Engineering. New York: McGraw-Hill Book
Company, Inc., 1937.
12.5. V. Z. Williams, "Infrared Instrumentation and Techniques," Rev. Sci.
Inst., 19, 135 (1948).
12.6. V. K. Zworykin and E. Wil son, Photocells and Their Applications,
2ded. New York: John Wiley & Sons, Inc., 1934.
CHAPTER
13
Photographic Photometry
OUTSTANDING ADVANTAGES, OF PHOTOGRAPHIC PHOTOMETRY
for
spectroscopic purposes are its high sensitivity, its feature of integrating the light which falls on the emulsion over the entire period
of an exposure, the possibility it gives of recording the intensities of
a large number of beams of light (spectrum lines) simultaneously,
and the fact that it gives a permanent record in simple form. Its
disadvantages are extreme variation of sensitivity with wavelength,
non-linearity of response, its requirement of careful control of a number
of variable factors whose importance is easily overlooked, and the
delay introduced by photographic processing. The advantages outweigh these disadvantages, however, and photographic photometry
is used widely for determining the relative intensities of spectrum lines
in the visible and ultraviolet regions, and hence for quantitative
spectrochemical analysis.
The properties of the principal photometric methods were compared in Table 12.1. Comparison of beams of different wavelengths,
called heterochromatic photometry, requires a "neutral" or "nonselective" photometric method, in which sensitivity does not vary
with wavelength. Unfortunately, only the radiometric methods are
neutral, and these are often insufficiently sensitive to be used directly.
They can, however, be used to standardize more sensitive nonneutral methods, which can then be made to serve for heterochromatic
photometry. The photographic emulsion is also nonlinear in response
and therefore must be calibrated at each wavelength in terms of
beams of light of known relative intensity.
Many problems, such as those involving measurements of the
attenuation of a beam of light by absorption, scattering, reflection, or
interference, require that comparisons be made only between an original beam and a weakened beam of the same wavelength; these
involve homochromatic photometry. Plate standardization is then not
326
§13.1]
PHOTOGRAPHIC PHOTOMETRY
327
required, but calibration or its equivalent is still needed. Special
niethods of homochromatic photometry have been developed which
use the photographic emulsion merely as a null indicator or which
introduce other short cuts that avoid the necessity of direct calibration; these come under the heading "spectrophotometry" and are
discussed in detail in Chapters 14 and 16.
Although photographic photometry and the sensitometry of photographic emulsions (discussed in Chapter 7) are superficially similar, a
fundamental "difference exists between them. As most plates and
films are used for camera photography, commercial sensitometry is
usually only semiquantitative and is mainly concerned with the
qualitative response of an emulsion to fairly intense white or colored
light. Photographic photometry is usually applied to the precise
quantitative comparison of faint monochromatic beams of light,
which may be visible or invisible. The proper use of a photographic
emulsion for intensity measurements requires control of so many
variable factors that for years the method was believed incapable of
yielding precise results. However, it is now possible to obtain
without difficulty photographic results self-consistent to within
± 2 per cent.
13.1. Photometric Characteristics of the Emulsion. The general
characteristics of the photographic emulsion were discussed in
Chapter 7. There it was pointed out that the response of an emulsion
to a beam of light depends on at least seven factors: the intensity i
of the light, its wavelength X, the time of exposure t, the nature of
the emulsion, and the time, type, and temperature of the development it undergoes. In addition, various minor factors must be
controlled, such as the effect on any developable patch of the conditions in the surrounding emulsion. Because of the difficulty of
controlling all these factors accurately, photographic photometry
cannot be used for absolute photometric measurements. It is best
used as a null method, but if a plate or film is carefully calibrated
when this method is used, it can be made to function as the equivalent
of a direct-reading instrument.
Variable factors can best be controlled by keeping constant those
which need not vary. In each measurement using homochromatic
photometry, only the intensity i of the light beam being measured
and the density d, which is a measure of this intensity, are essentially
involved, and all other factors can be kept constant. In heterochromatic photometry the wavelength X of the light also is varied.
328
PHOTOGRAPHIC PHOTOMETRY
[§13.1
Tliree methods of determining plate response have come into
common use: (1) measurement of density with a densitometer, which
is the most direct and precise;, (2) estimation of the least visible
density that can be seen by eye and correlation of this density with
intensity—a quick and convenient method but only semiquantitative;
and (3) visual search for adjacent areas of equal density. This last
is a quick method but usually can be applied only with the special null
methods of absorption spectrophotometry discussed in Chapter 14.
The fundamental law of monochromatic photographic photometry
was first enunciated by Hartmann in 1899 approximately as follows:
If two light beams of the same wavelength produce equal densities on a
given plate in the same time of exposure, they are equal in intensity.
That is, if all the auxiliary variables are kept constant and di = ^2,
then ii = i^. When two beams of unequal intensity are to be
compared, one need only determine the ratio by which the stronger
beam must be reduced in intensity to make its density equal to that
produced by the weaker beam. With this equality of intensity, the
emulsion is being used as a null indicator.
In practice, it is not necessary to make d^ exactly equal to d\,
though the highest precision is attained when this is done, and when
the two exposed areas are close together and similarly isolated on the
photographic plate. I t is possible
under controlled conditions to
determine how d vaxies with i.
Then an unknown intensity can
Density
be interpolated between two known
values of i, by interpolating the
'density that this intensity pror
duces on the curve expressing the
LoQiQ intensity
^
d- log i relation as in Fig. "13.1,
Fig. 13.1. Calibration curve re- where a typical calibration curve is
lating photographic density to the
shown, similar to the characteristic
logarithm of light intensity.
curve of § 7.2.
It is customary to plot density against the common logarithm of
the intensity producing it, on account of the relatively simple shape
of the curve which results and for convenience in covering a wide
range of intensities. Other functions of the blackening of the plate
and the light intensity could be plotted against one another and could
be used satisfactorily for interpolation purposes if a smooth and
reproducible curve resulted. Sometimes d is plotted directly against i
§13.1]
PHOTOGRAPHIC PHOTOMETRY
329
when work is being done in the underexposure region or with X rays.
A few workers prefer to use blackening (§ 13.15) instead of density.
Since the variable that is usually read directly is the galvanometer or
microammeter deflection of a densitometer, which is inversely proportional to the opacity of the emulsion being measured, it is convenient to plot this deflection directly on double logarithmic paper
so that a curve of d against logio i results. This inverted curve, of
the form shown in Fig. 13.2, is more convenient than Fig. 13.1 for
calibration purposes (see § 13.2).
If the calibration curve is determined over an intensity range
suflaciently wide that all desired i values can be interpolated, more
accurate results can be obtained
than when extrapolation is required.
Each individual plate or film used
LoQio
must be calibrated, preferably at Deflection
wavelengths within 25 A of all
wavelengths reduced. It is not
satisfactory to expose two plates or
LoQio Intensity
^
films simultaneously, develop and
Fig. 13.2. Calibration curve refix them together, and then measlating densitometer deflection to
ure one in terms of a calibration light intensity.
curve determined on the other.
Because every emulsion varies somewhat in sensitivity over its
surface, it is wise to keep calibration and unknown exposures as close
together as possible. It is also desirable, because of the Eberhard
effect (§ 7.13), to surround known and unknown spectral regions to
be measured by areas of similar density. Thus it is unwise to calibrate
a plate by means of a continuous spectrum if individual spectrum lines
are to be measured on it; artificial spectrum lines should be produced
by using a diaphragm to cut off parts of the continuous spectrum, care
being taken to avoid errors due to diffraction.
Since film is coated in large areas that are cut up subsequently,
films are likely to be more uniform than plates, which are coated on
the concave side of sheets of glass of only moderate size and later are
cut into several pieces over which the thickness of the emulsion may
vary somewhat.
To fulfill the condition that the wavelengths of the unknown and
calibration exposures should be the same, a plate should be calibrated
not merely with another spectrum line of the same wavelength but
preferably with one of the same shape, size, and other characteristics.
330
PHOTOGRAPHIC PHOTOMETRY
[§13.1
to reduce errors due to the Eberhard effect. This recommendation
suggests the desirabihty of actually using the line under measurement
to produce its own calibration curve, as is done in the single-exposure
automatic calibration methods discussed below. Keeping the time,
temperature, and character of the development the same for both
the known and unknown exposures presents no problem if we put
both on the same piece of plate or film and arrange that the developer
be evenly distributed by one of the special methods discussed later.
Much work has been done on the exact form of the relationship
between the different variables mentioned above. This information
is largely irrelevant to the purposes of photometry, since in practical
work we can achieve the desired results without it. However, knowledge of the qualitative relationships between the variables is convenient as a guide in selecting the best working conditions for photographic photometry. These are discussed in Chapter 7.
The photographic plate as ordinarily used commercially is exposed
to white light for a small fraction of a second. There is a large
literature dealing with the characteristics of emulsions under these
conditions. Much, less is known regarding their behavior when
exposed to monochromatic light for minutes and hours. Plates and
films are used on three illumination levels, which have been called,
respectively, the photographic level, the spectrographic level, and the
astronomical level. These may be taken as involving light-intensity
ratios of approximately 1,000,000 to 1000 to 1, corresponding to times
of exposure of 0.02 sec, 20 sec, and 5 hr. Qualitatively, the responses of an emulsion are similar under these three conditions, but
they differ greatly quantitatively.
HOMOCHROMATIC
PHOTOMETRY
Homochromatic photometry can be used whenever a beam of light
is to be compared with itself after being absorbed, scattered, or
attenuated by other means, or when two or more spectrum dines
having wavelengths not more than 25 A apart are to be compared.
It is also the first step in heterochromatic photometry, which can be
thought of as a number of separate problems in homochromatic
photometry that must be correlated. Homochromatic photometry
requires calibration of the response of the emulsion to light of varying
intensity.
§13.2]
PHOTOGRAPHIC PHOTOMETRY
331
13.2. Calibrating the Emulsion. An obvious method of calibrating the emulsion would be to expose various areas on it to a spectrum
of constant intensity, for equal times of exposure, through a series
of neutral filters that transmitted equal fractions of the light of
each wavelength to the plate. The undiminished light intensity
at each wavelength could then be called 100, and that passing through
each succeeding filter might be 70, 50, 30, 15, 8, 4, and 2 per cent.
The densities produced by each intensity at each wavelength in a
specified time of exposure could then be measured on a densitometer,
and calibration curves could be plotted for important spectrum lines
spaced not more than 25 A apart over the spectrum range to be
covered. Curves of the type shown in Fig. 13.3 would be obtained.
Unfortunately, truly neutral filters are not available, and filters
called neutral show variations in transmission with wavelength that
may amount to 10 per cent or more in the visible region alone. In
recent years, however, it has been demonstrated that the rotatingsector disk, if operated under carefully controlled conditions, can be
made the equivalent of a neutral filter, and this is the most commonly
used method of varying light intensity.
Density
LoQiQ Intensity
—
Fig. 13.3. Calibration curves for different wavelengths. (Each curve is
plotted to the same scale but with different origin to prevent overlap.)
Various methods that have been used for imposing intensity calibration marks on photometric plates are listed in Table 13.1, where they
are grouped as single-exposure methods and multiple-exposure methods. If various parts of a single beam can be sent simultaneously
through different weakeners, it is unnecessary to use a steady source;
but if multiple-exposure methods are used, the light intensity must
not vary between one exposure and the next. The single-exposure
methods are in turn divided iiito those in which the light being measured can be used to produce its own calibration marks and those
which require an auxiliary source, which need not, however, be steady.
In general. Table 13.1 is laid out with the most convenient methods of
[§13.2
PHOTOGRAPHIC PHOTOMETRY
332
•Ji
-S
3^;i
3l
O
= ^fe1-f
^
S 3; g
3.S.S *
s^3r5
§•3
B g C
(« So =« c c;
o ^ o
Uo
U
IP .
IJII
Jl
are & B CJ - ^
> aj >
a 3 c
tH cS
s^
-
CJ o •—'
J3
.SP
il
O,
55
O
«
60
<
60
O
d a
< <
ca ^ rti
c!
'^ ^ 1
-Si S
Co
ca
tJD
£3 0 ) OJ
c .c
=
;^ "S .-a o. S 3
<
<
<
•p:J
C3
c4
o
X
EH
2:2
i S 2
1/J Q;
a
-
PHOTOGRAPHIC PHOTOMETRY
§13.2]
II
V
.2
^•3^
" i S 2
g § " S3
»-2 i
§:2
cro
-5"
S C ^
^O
:S-a" =^
0) tT =3
.2
iijll
V
si"j§s
fc« ^
C3 =
H
333
•S
O
O
>
1^
>
<<
C/:
^
1 IJIIII
<
S
fee
n
p
o
o.B"
3J
a;
.2 "
*+J "0
'C
rf
"o
CU
rg ^
-VTB
^ a
•|1
<- „
.tiS
= 2 ^ o
• ; £=2
'7^ «
CS
rt
a
O
O
111
N 5
.:s
«„
s --s
a,
tj-
-5
-5
i
-s ^ ^
>
o
o
ISb
:I1
111
O
334
PHOTOGRAPHIC
PHOTOMETRY
[§13.2
calibration at the top, so that in selecting a method it is necessary only
to run down the column until a method that will fit the circumstances
is reached.
It is desirable to use the spectrum being studied as its own calibration spectrum whenever possible, since this method reduces errors due
to the Eberhard effect (§ 7.13), results in a great saving of plate space
and time (since only one exposure is needed to produce all the calibration marks), and ensures constancy of exposure time. Also, a
number of determinations of each intensity ratio desired can be obtained from each pair of lines, since each appears at several densities.
When two lines being compared have the same wavelength, their
two calibration curves should be identical but displaced owing to the
different intensity scales, and the amount of this displacement gives
the intensity ratio of the two lines. This method enables one to
determine the intensity ratio from the whole course of a curve, as in
Fig. 13.4, rather than from a single point on it. When available, this
method is usually more convenient and precise than any other, and
always more rapid. It is listed in Table 13.1 as a single-exposure
autocalibration method.
Density
LogiQ Exposure
^
Fig. 13.4. Determination of intensity ratio for two spectral lines of same
wavelength. The two curves are plotted to the same scale, and their separation
gives log lA — |log iB = log.(iAAB).
When this method cannot be used, a number of separate exposures
of equal duration must be made, requiring use of an auxiliary source
unless the source of light being measured is steady during the time
required to take all necessary exposures./
,
Although the nonneutral or selective methods given in Table 13 1
are capable of giving results as accurate as the neutral methods, they
are less convenient because different transmission values must be
plotted at each wavelength. Under several of the methods listed
in Table 13.1 will be found two alternatives which differ only in that
one produces step spectra as shown in Fig. 13.5a and the other wedge-
f
§13.3]
PHOTOGRAPHIC
'^ wuiic^i
335
PHOTOMETRY
spectra,* shown in Fig. 13.5b. The step spectra are more convenient
for densitometer use since positions along the line can be determined
more readily. The wedge spectra are designed primarily for use with
the limiting density methods discussed below, which are somewhat
less precise than densitometer methods, though simpler.
13.3. Methods of Varjring Light Intensity.
1. Step sector or logarithm sector disk at spectrograph slit. This
method is widely used on account of its convenience. It has been
the subject of much controversy because the average intensity of light
falling on the plate is varied, rather than the actual intensity. The
investigations of Twyman,' of O'Brien,^ and especially of Webb,' have
(b)
Fig. 13.5.
(a) Short portion of a step spectrum,
spectrum.
(b) Short portion of a wedge
justified its use under controlled circumstances. Webb's results
indicate that an intermittent exposure is equivalent photographically
to a continuous exposure when the rate of flash is so great that each
grain of the emulsion receives on the average not more than one
quantum of light per flash. It is now realized that frequency of flash
was the critical variable that caused such lack of correlation among
the results of many earlier workers. Webb showed that for each
intensity, emulsion, and wavelength a critical frequency of flash exists
above which it is safe to use intermittent exposures as equivalent to
continuous. The critical frequency for ordinary conditions is of the
* A term derived from the use of wedge-shaped cells or filters varying in density
from top to bottom.
' F. Twyman and A. Harvey, Trann. Opt. Sof. (London), 33, 1 (19.'{l-32).
= B. O'Brien, Phys. Rev., 33, fi40 (1929): B. O'Brien and E. D. O'Brien, Phys. Rev.,
Z7, 471 (I9.SI).
5 J. H. ^Yebb, Jour. Opt. Soc. .4m., 23, 157 and 316 (1933).
336
PHOTOGRAPHIC PHOTOMETRY
[§13.3
order of ten flashes per second; to be safe, however, a rotating-sector
disk should be run as rapidly as possible, preferably at a speed greater
than 1200 rpm. The greatest reciprocity and intermittency failures
occur when the quanta are small (infrared region), the emulsion grain
small (slow plates, films, or papers), or the intensity high; but even
under such conditions O'Brien and Parks'* showed that suflSciently
high sector speeds could be used to give accurate results. Stroboscopic effects must of course be avoided if AC-operated sources are
used.
Convenient forms of step and wedge sectors are shown in Fig. 13.6.
These disks can be obtained commercially, or may readily be filed out
of thin disks of aluminum, appropriate counterweights being added
to balance any asymmetry. They may be mounted on the spindle
of a small motor and placed directly in front of the slit of a stigmatic
(a)
(b)
Fig. 13.6. (a) Step-sector disk (logarithmic), (b) Wedge-sector disk
(logarithmic).
spectrograph. The motor should be so fastened that any vibration
which it may produce will not affect the spectrograph.
Errors to be guarded against arise from running the disk too slowly
or from using it in such a way that all parts of the spectrograph collimator are not simultaneously filled with light during a flash. Under
these conditions the resolving power of the spectrograph and the
shape of the spectrum lines may be changed, and the intensity steps
may not be truly neutral. The cutoff edge of the disk should be
mounted accurately parallel to the spectrograph slit.
2. Step weakener or wedge at slit. To avoid any possibility of errors
due to reciprocity or intermittency failures, a step weakener or an
« B. O'lSrien and V. L. Parks, Phyn. Rev., 41, 387 (1932).
§13.3]
PHOTOGRAPHIC PHOTOMETRY
337
optical wedge can be used at the slit. Unfortunately, no substance
has been discovered t h a t reduces light uniformly throughout t h e
spectrum, so a truly neutral wedge does not exist. I t is, however,
possible t o measure t h e transmission of each step of such a weakener
by means of a thermoelectric or photoelectric photometer, used, in
conjunction with a quartz-mercury arc and monochromator. Figure
13.7 shows typical calibration curves for t h e various steps of a
platinum-on-quartz step weakener. I n making such a calibration it
is not safe to assume linearity of density with thickness. Monel or
aluminum evaporated on quartz gives weakeners t h a t are approximately neutral throughout the visible a n d ultraviolet.
80
•
1
70
60
Platinum on quartz
'a
Transmission
50
2
4o|
30
20
id
———^^
.
0
5000
4000
'
3000
J
Wavelength in Angstroms
Fig. 13.7. Calibration curves for a platinum-on-quartz step weakener.
steps are numbered in order of increasing density.
The
F o r the visible a n d photographic infrared regions, a weakener can
be produced by exposing a photographic plate, preferably one of fine
grain such as a contrast oi- positive plate. Absorption and scattering
produce t h e desired reduction of intensity, b u t both vary r a t h e r
rapidly with wavelength.
A step or wedge reducer m u s t be uniformly illuminated along its
length, and obtaining this uniformity m a y present dilBculties. Accepted m e t h o d s are discussed in § 6.6. I n selecting a m e t h o d of slit
illumination, it should be borne in mind t h a t the resolving power of
the spectrograph m a y be lowered if an image of the source is focused
on its collimator so t h a t it is not illuminated with coherent radiation.
I n order to make sure t h a t uneven illumination of t h e slit has n o t
338
. PHOTOGRAPHIC PHOTOMETRY
[§13.3
affected the intensity ratios, it is desirable to invert the step weakener,
wedge, or sector frequently so that such effects will tend to cancel out.
3. Step diaphragm before the slit. Diaphragms in which steps
equivalent to those of a step weakener are produced by geometrical
means have been described by several workers.* Diaphragms having
steps of various lengths are introduced into the beam, and by means of
astigmatic focusing the illumination on each part of the slit is made
proportional to the length of one of the steps. Wide departures from
spectral neutrality can sometimes occur with such diaphragms because of changes in the illumination of the spectrograph from one step
to the next. Though this method is suited only to use with a stigmatic
mounting, Frerichs^ has used step diaphragms with a Rowland mounting by placing the diaphragm at the position of the vertical astigmatic
focus outside the slit, as in Fig. 13.8. Only a limited spectral region
can be photographed at one time by this means, however, since the
position of the step diaphragm must be shifted for each new wavelength setting. This method is frequently used with spectrographs
using the Eagle mounting.
4. Step weakener or wedge at the plate. Concave grating mountings
and certain types of prism spectrographs are frequently so astigmatic
that the methods listed above cannot be used, since any point on the
slit is drawn out into a vertical line in the spectrum. In the PaschenRunge and Eagle mountings, for example, long astigmatic lines are
produced with central portions that may be uniform in intensity and
hence ideally suited for superposition of intensity-reducing devices.
This condition suggests placing a reducing wedge or weakener directly
in front of the plate and photographing the lines through it. This
method is practicable when narrow multiplets or Zeeman or Stark
patterns are to be measured, but a reducer that will cover more than
a few inches of plate is seldom available.
5. Step slit or wedge slit. With a stigmatic spectrograph having a
fairly long slit, one can use a slit having varying widths along its
length. In this method, when calibration's are made by means of a
source with a continuous spectrum, a spectrogram consisting of
several bands, having known relative intensities is obtained, just as in
the case of the step-weakener or step-diaphragm methods. The
= G. Hansen, Zeitschr. f. Phys., 29, 356 (1924); A. von Hippel, Ann. d. Physik, 80,
072 (1926).
« R. Prerichs, Zeitschr. f. Phys., 36, 524 (1926).
§13.3]
PHOTOGRAPHIC PHOTOMETRY
339
method has been successfully used by Ornstein and his collaborators'
(see General Reference 13.1) in the visible region and in the near
ultraviolet, where an incandescent lamp can be used to give the continuous spectrum, and by Thompson and Duffendack* in the ultra-
Fig. 13.8. Step diaphragm as used with a Rowland moimting by Frerichs.
The rotating diaphragm, D, is located on the tangent, T, to the Rowland circle,
R, at point P. The line A is the axis of rotation of the diaphragm, and as the
diaphragm rotates, a mask M rotates synchronously about axis A' in front of the
plate, P. This mask contains slits t h a t cause a fixed part of the spectral line
image to be photographed through a given step in the rotating diaphragm.
violet, using a hydrogen discharge tube as a continuous source. I t
can be used also with lines which are broader than the slit.
The method has limited application, because the slit must never
be so narrow that diffraction produces errors or so broad that overlapping of wavelengths becomes important. Also, the fact that a
' L . S. Ornstein, Phys. Zeitschr., 28, 688 (1927); Proc. Roy. Soc. (London), 37,
337 (1925).
' K . B. Thompson and O. S. Duffendack, Jour. Opt. Soc. Am..'23, 101 (1933).
340
PHOTOGRAPHIC PHOTOMETRY
[§13.3
continuous spectrum is being used to calibrate a plate on which line
spectra are being measured may cause large errors due to the Eberhard effect. The- form of wedge slit illustrated in Fig. 13.9 will be
found useful, the width of the slit varying logarithmically along its
length.
6. Multiple-exposure methods. If it is necessary to use a multipleexposure method, a steady source must be provided. An incandescent lamp is usually satisfactory for the region 10,000 to 3800 x\
(or to 2800 A if a quartz window is provided), a quartz-mercury arc
for the region 5800 to 2300 A, and a hydrogen discharge tube for
7000 to 900 A. Steady sparks between rotating electrodes also can
be used for calibration.
One of the simplest methods of intensity variation is that in which
the cross section of a collimated beam of light is altered by means of
Fig. 13.9. A wedge slit whose
width varies logarithmically
along its length.
(Magnilied
horizontally.)
Fig. 13.10. A rotating-sector
diaphragm. The open sectors
are inmic variable so that average intensity can be varied. ,
a diaphragm of variable aperture. This change may be made readily
when the beam in which the aperture is placed is of uniform cross
section for all wavelengths used and when all wavelengths are treated
alike by the spectrograph. These conditions are difficult to fulfill,
however, since most lenses have zonal variations of focus and transmission. A diaphragm of the shape shown in Fig. 13.10 may serve
with a spectrograph, but it may cut the resolving power in half if
placed so as to reduce the apparent width of the dispersing unit, and
such a system should be calibrated at selected wavelengths.
§13.3]
PHOTOGRAPHIC PHOTOMETRY
341
Wire-gauze screens^ are effective multiple-aperture diaphragms that
avoid the difficulties outlined in the preceding paragraph, but they
may introduce diffraction errors unless handled carefully as described
in the references. They should have their transmission factors
calibrated under the conditions of use, whereupon they are usually
found to be sensibly neutral. They should be kept in continuous
relative oscillating motion when used in multiple.
If the transmission factors of a series of (approximately) neutral
filters are known at various wavelengths over the spectral range desired, a method is available that is useful for revealing hidden errors
in other methods of calibration. The filters are introduced into the
beam, preferably where it has a large cross section, one at a time or
in groups, since they are additive in density. Care must be taken
to see that all filters are plane, because any prismatic effect may cause
some light to be thrown off the slit and thus introduce refraction
errors. Interreflection effects between slit jaws and filters should also
be avoided.
The inverse-square law gives a fundamental method of varying
light intensity. This method is very useful for testing the accuracy
of other methods directly, since it caii be used with a sufSciently
intense source that approximates a point. A mat surface, such as a
quartz disk with both sides rough-ground, is placed at the normal
position of the light' source being studied and is illuminated by an
intense source, such as a short section of mercury arc or ah incan^ descent lamp, placed on an optical bench at an adjustable distance
from the scattering surface. The intensities at the mat surface are
accurately proportional to the inverse square of the distance between
source and scatterer if the distance is large in comparison with the
effective source size (say 20X). I t is particularly necessary to keep
out stray light and to avoid atmospheric refraction and absorption
in long paths.
Variation of exposure time, so commonly resorted to because of its
simplicity, should be used only as a check on the other methods, or
when the reciprocity failure of the plate is determined directly, as in
the fluorescence method of Harrison and Leighton.^" Under no circumstances should it be trusted for use with a given type of emulsion
9 L. B. Ham, D. H. Fehr, and C. Bitner, Jour. Franklin Inst., 178, 299 (1914); G. R
Harrison, .Jour. Opt. Soc. Am., 18, 492 (1929).
i»G. R. Harrison and P. A. Leighton, Jour. Opt. Soc. Am., 20, 313 (1930); Phys.
Rev., 36, 779 (1930).
342
PHOTOGRAPHIC PHOTOMETRY
[§13.4
just becaus^ someone else found that this emulsion obeyed the
reciprocity law. Reduction of intensity by varying the excitation of
the source or by using polarizing apparatus, though of value in other
methods of photometry, has little application in photography.
Calibration exposures may be made with a spectrograph different
from that used for photographing the spectrum to be measured, but
great care must be taken to control scattered and false light and to
ensure that the Eberhard effect associated with differences in line
shape will not affect the results.
HETEKOCHBOMATIC
PHOTOMETRY
13.4. Uses of Heterochromatic Photometry. When we wish to
compare two beams of light having different wavelengths or to measure the relative intensities of two spectrum lines more than 25 A
t=t,
diV
^ 4 ^
\ '
\4t
\
"9r
Wavelength
Fig. 13.11. Three-dimensional plot of the relation between density, log intensity,
and wavelength.
apart, a new step must be added that is not involved in homochromatic photometry. This step is designedj-to take account of, the
variation in sensitivity of the emulsion with wavelength, whereas in
homochromatic photometry througKbut a range of spectrum we need
take account only of the variation of,pontrast with wavelength.
In homochromatic photometry each calibration curve is determined by producing known variations in the unknown but fixed
intensity of a light beam. The additional step in heterochromatic
photometry involves determining the intensities used for the different
calibration curves. I t is usually not convenient to take this step
§13.5]
PHOTOGRAPHIC PHOTOMETRY
343
by determining the intensity variation with wavelength of the source
used for cahbration. Instead, additional exposures are made to a
source whose intensity distribution is known throughout that part of
the spectrum being studied. This process is called plate standardization.
The basis for heterochromatic photometry is illustrated in Fig.
13.11. The three-dimensional plot in the figure gives the relation
between density, log intensity, and wavelength for a typical emulsion.
Calibration curves at the various wavelengths appear as plane sections
cut through the surface of this plot perpendicular to the wavelength
axis, and the process of standardization consists in determining the
relative intensity values that will put the calibration curves in their
proper positions along the log i axis (see General References 13.2
and 13.3).
13.5. Light Sources for Standardization. Commonly used standard sources are the blackbody (§ 8.5), whose spectral energy distribution is known theoretically, sources of continuous radiation
(§§ 8.6 and 8.7) having known emissivity, and sources which have
2S
26
27
.28
29
30
lltl t IlIillnilliiiil.oiliiiiI.inloiTUliilii
I
I
I
I
31
32
I
I I
III1III1I
I
33 34 35 36 37 3S 3940
ml
I
45
SO
55 60
iiKimiliuiiiiiilHIUIUMdllllMM i1 I I 1 T T I I MlUrilllllllllljIllillltllhli
I
II
I
I
Fig. 13.12. The quartz mercury-arc spectrum.
such steadiness that their spectral-energy distribution can be measured directly with a monochromator or thermopile or other nonselective radiometer.
A convenient source for use in standardizing plates in the visible
and ultraviolet regions is the quartz-mercury arc, operated at constant current and temperature so that the voltage drop across it and
the pressure within it remain constant throughout the exposure series,
as discussed in § 8.14. A quartz monochromator can be used to
separate the various mercury lines, which are intense and well
separated in groups that can be identified readily. The mercury
spectrum contains strong lines well spaced in the range 5800 to
2300 A, as shown in Fig. 13.12 and Usted in Table 9.1.
The transmission of the monochromator for various wavelengths
is first measured, preferably with another monochromator and a
photoelectric cell or thermopile. It is important to ensure that all
diaphragms and other stops in the second monochromator be adjusted exactly in the manner in which they will be used later in making
344
PHOTOGRAPHIC PHOTOMETRY
[§13.5
the intensity-distribution measurements. The slit of the first monochromator is made sufiiciently wide that all mercury lines or line
groups are flat-topped, and the light passing from this slit through the
entrance slit of the second monochromator is then measured at some
convenient wavelength. A similar reading for the same setting is
then made by moving the radiometer so as to intercept all the light
passing through the exit -slit of the second monochromator. The
ratio of the two deflections gives the transmission of the second
mojiochromator at that wavelength, if the radiometer has a linear
response. These readings are then repeated throughout the spectrum. Transmission factors for a typical quartz-prism monochromator are given in Table 13.2.
TABLE
13.2
M E A S U B E D T K A N S M I S S I O N F A C T O R S TOK A T Y P I C A L Q U A B T Z PRISM
Wavelength
2000 A
3000
4000
5000
6000
^
MONOCHKOMATOR
Per Cent
Transmissifm
.14
26
34
38
41
Once such a transmission curve has been determined for a given
monochromator, it can be used with thermoelectric or other spectrally
neutral radiometric equipment to measure the actual amounts of
energy in the; spectrum of a mercury arc or other steady source run
under constant conditions. Thisi spectrum can then be impressed
on the photographic plate that is to be standardized.
It may be necessary to reduce the intensity of the light from the
mercury arc by some method approaching neutrality as closely as
possible, to the point where it will produce densities in the range 0.3
to 1.5 during, the exposure time used on the plate being standardized.
The principal errors of heterochromatic photometry arise from the
difficulty of bridging this gap between the lowest intensities that can
be measured precisely with a neutral radiometer and the highest
intensities that can be recorded in a suitable time of exposure on the
photographic plate. This gap may in some cases exceed 1000 to 1
and can probably best be bridged by a combination of si rapidly
rotating sector disk and a specially calibrated screen.
§13.5]
PHOTOGRAPHIC PHOTOMETRY
345
A somewhat more direct method of standardization, especially
useful in the visible region, involves use of a blackbody operated at
a known temperature as measured by an optical or other pyrometer.
The energy distribution can be determined from Wien's law cast in
the following form:
logio /
= 28.532 - 5 logio X - antilog (7.7939 - logio X - logio T) (13.1) *
or from tables (§ 8.4). To ensure the blackbody character of the
radiation used, simply take the radiation from the inside of a V-shaped
hollow in the side of a graphite rod. A convenient blackbody is
described by Harrison,''^ and other forms are discussed in General
References 13.3 and 13.4.
The principal limitation of the use of the blackbody is that at all
practicable temperatures the intensity falls rapidly in passing from
the near infrared to the ultraviolet, and the radiation is usually not
of sufficient intensity to be used for standardizing plates at wavelengths shorter than 3000 A. Also, a continuous spectrum is produced, so that the Eberhard effect is likely to cause error unless the
spectrum is artificially broken up into "spectrum lines" of the approximate size and shape of those being measured.
A tungsten filament can be used as a standard source, especially if
it is of the ribbon type made for radiometric purposes. The coiled
filament is not so useful, because the temperature varies greatly over
the surface of an individual coil. Since tungsten is not a true blackbody, its color temperature must be used instead of its true temperature. The brightness temperature of the filament can be determined
by means of an optical pyrometer. Tables giving the conversion
from brightness temperature to color temperature or from brightness
temperature to true temperature to color temperature are given in
Chapter 8 and in the International Critical Tables. The actual
temperature of the filament is likely to vary considerably along its
length, and its brightness temperature should be determined on the
actual area which is used to produce the standard spectrum.
When a continuous spectrum is used for standardization, it is necessary to correct the actual emissivity values for the dispersion of the
spectrograph, since the radiation is spread thinner where the disper* In this expression, I is in watts per cm* of blackbody surface per angstrom of
spectral range, X is in angstroms, and T is in °K.
" G. E . Harrison, Jour. Opt. Soc. Am., 19, 290 (1929); H. B. Dorgelo, Phys. Zeitschr.,
26, 767 (1925).
346
PHOTOGRAPHIC PHOTOMETRY
[§13.6
sion increases. With line spectra this correction is not necessary,
since all of the energy of a spectrum line usually goes into the radiometer regardless of effective slit width.
APPARATUS AND METHODS FOR
PHOTOGRAPHIC PHOTOMETRY
13.6. Selection of the Spectrograph. For photographic photometry a spectrograph should be selected which will permit a fairly large
number of exposures to be taken on a single plate and which is as free
as possible from scattered light. The latter consideration makes the
Littrow mounting and those mountings in which light traverses a
prism more than once less satisfactory than other spectrographs. A
spectrograph having dispersion as great as is consistent with other
requirements should be selected, inasmuch as higher dispersion yields
higher precision in wavelength measurements on continuous spectra
and allows wider slits to be used with line spectra. Increased dispersion also reduces the intensity of the continuous background relative
to the lines, and diffraction errors of the type discussed below are
decreased as the dispersion and slit width are increased. In photographic photometry the slit is often made as much as 10 times as
wide as is suitable for wavelength measurements, to obtain flattopped lines. In general, concave grating spectrographs are somewhat more convenient and flexible for photographic photometry of
line spectra than are prism instruments.
For measuring relative spectrum line intensities, it is not necessary
to know the transmission of the spectrograph, since under proper
conditions of illumination, transmission losses affect unknown and
standard beams alike.
13.7. Selection of the Emulsion. From among those types of
emulsion which are sensitive inthe spectral regions desired, one having
good photometric characteristics should be chosen. Usually an emulsion of medium contrast and speed will be found most useful. /Fast
emulsions, which usually have low contrast, are undesirable for
photometry because large variations in intensity produce only small
variations in density. Also, large errors may result from the tendency
of a fast emulsion to fog, and from its high sensitivity to pressure,
temperature, and humidity effects. On the other hand, plates having
very high contrast can be used over only a narrow intensify range,
and one is likely to find it necessary to work on the under- and
§13.8]
PHOTOGRAPHIC PHOTOMETRY
347.
overexposed portions of the curve where the rapidly changing slope
increases the error of interpolation. Though it is not at all necessary
to restrict measurements to the straight-line portion of a calibration
curve, a fairly long straight-line portion with a slope of 45 deg is likely
to be found most convenient.
Certain types of emulsion have multiple coatings of different speeds.
Such emulsions should be avoided for photometric purposes, since
they often show a change in contrast in the middle of the calibration
curve. The advantage of having a number of points lying in a
straight line is then lost.
Although it is, of course, necessary to use dyed and coated emulsions
in certain regions of the spectrum, each additional treatment of an
eniulsion increases the possibility of variation over its surface, so
"clear-working" plates, showing little residual dye, should be used
when possible. Schumann plates are so nonuniform as to be undesirable for photographic photometry, and special methods must be
used in the extreme ultraviolet (§ 19.7).
Plates to be used for photographic photometry should always be
stored on edge in a cool place to reduce pressure marks and fogging.
No intensity marks or other important exposures should be made
within I in. of the edge of a plate.
13.8. Timing the Exposure. Usually it is not necessary to know
the exact duration of an exposure, but merely to ensure that all
exposures to be compared are of the same duration. There is a
certain amount of choice in the time of exposure to be employed
unless the light source is very weak, since it is always possible to
reduce the intensity of a strong source by changing the mode of
illumination of the spectrograph. If such a change is necessary, it
should be made to give a time of exposure of at least 1 min., (a)
because the response of the emulsion is somewhat more uniform to
a moderate rate of illumination, (b) because the effect of short-time
fluctuations in the intensity of the light source will then be averaged
out, and (c) because timing of the exposure can be more precise. The
most satisfactory exposure time is usually from 1 to 3 min.
When exposures of 5 min or longer are used, a hand-operated
shutter, or a card held in front of the slit, can be timed with an
ordinary watch. For exposure times between 1 sec and 5 min, a
magnetically operated shutter run by impulses from a clock beating
seconds, or from an electric clock, is useful. For exposures that must
be less than 1 sec in duration, a double-sector disk is effective. This
348
PHOTOGRAPHIC PHOTOMETRY
[§13.9
disk consists of two coaxial disks arranged to rotate at a known speed
ratio with a single sector cut out of each. One disk moves so slowly
that the spectrograph shutter can be opened or closed conveniently
during a single revolution. The operator opens the shutter while the
light beam is cut off by both disks. The slow disk then allows the
beam to strike the fast sector^ which shortly thereafter permits the
light to fall on the slit for a brief exposure. Before it can repeat this
operation, the slow sector has again cut off the light; and before the
slow sector opens again, the spectrograph shutter has been closed.
13.9. The Brush Effect. It is probably desirable in photometry
to let a longer time elapse between the last exposure on the plate and
the beginning of development than has elapsed between the first and
last exposures, because of a phenomenon known as the Brush effect}^
Although the existence of this effect has been the subject of much
controversy, the evidence seems to indicate that something of the sort
occurs. The effect involves a progressive change in the latent image
during the time between its production and its development. To
avoid possible errors from this cause, exposures that are to be compared should theoretically be allowed equal times to change. I t
would be undesirable, for example, to expose a plate to an unknown
spectrum one day and to the calibration and standardization exposures the next morning just before development, not only because of
possible Brush effect but also because the humidity and temperature
of the plate may affect its sensitivity to a slight extent.
13.10. Processing the Spectrogram. Photographic photometry
requires close control of development. A developer should be chosen
which will not stain the emulsion and which will keep chemical fog
and Eberhard effect (§ 7.13) to a minimum. It has been claimed that
Rodinal produces less Eberhard effect than other developers. The
most effective method of reducing the Eberhard effect is that suggested by Clark,!^ in which the plate is brushed continuously throughout development. The intimate contact of the brush removes the
developer from the regions where it has been depleted and mixes it
thoroughly. Rocking the trays is insufficient because it does not
remove the thin adhering layer of developer next to the emulsion.
Various workers'* have described devices for mechanical agitation
•2 C. F.'Brush, Phys. Rev., 3, 241 (1910).
1= W. Clark, Plwt. Jour., 65, 76 (1925).
" G. M. B. Dobson, I. O. Griffith, and D. N. Harrison, Photographic ^Photometry.
London: Oxford University Press, 1926, page 121.
§13.11]
PHOTOGRAPHIC PHOTOMETRY
349
during development, but the brush method is simpler and more
effective.
To obtain the greatest uniformity over the emulsion surface,
development should not be unusually rapid, because high speed may
result in uneven penetration of developer. On the other hand,
development should not be unduly slow because of tediousness in
carrying out the development process. Development should be
allowed to proceed fairly well toward completion, or so that it will
give, with a plate of moderate contrast, a characteristic curve whose
straight-line portion has a slope of about 45 deg.
Development should be carried out in the dark in developer of
carefully controlled temperature and for a time accurately measured
with a stop clock. After development, the plate should be processed
as outlined in § 7.10.
It is often desirable to reduce the time
required for processing plates when highspeed quantitative spectroscopic analysis
is involved, and many of the procedures
outlined can be simplified. Eastman spec, ^ ,
,
,
f ^
Fig. 13.13. Point of dentroscopic plates have been produced to g^^y j^^^ch determined by
permit development in a much shorter time eye.
than is required by the ordinary emulsion.
To test whether a photometric method is satisfactory, a sample
plate of the type to be used may be exposed in a darkroom to a light
placed as far away as possible, so as to get uniformity of illumination,
and then developed and fixed under the standard conditions chosen.
Test areas measured on a densitometer will then reveal the uniformity of response over the surface of the emulsion. When the
spectrum being measured, the calibration spectrum, and the standardization spectrum are taken on the same plate under conditions
which give uniform response over the entire plate in such a test,
photometric errors should be small.
13.11. Short-cut Methods of Photographic Photometry. Simple
methods of photographic photometry that use the emulsion as a null
indicator have been developed. I t is easy to locate by eye two
contiguous areas of equal density in two photometric strips varying
in density in opposite directions, as in Fig. 13.13. Since other variables are kept identical in the two exposures, equality of density
indicates equality of intensity. The minimum density at which a
line fades into the background can also be used as an indication of
PHOTOGRAPHIC
350
PHOTOMETRY
[§13.11
intensity. These methods when carefully controlled can be used to
give precision to within 5 to 10 per cent and make possible rapid
determination of the course of an absorption curve, for example, over
a long spectral range. Such methods are discussed in Chapter 14
under Absorption Spectrophotometry.
By means of optical comparators it is possible to bring into apparent juxtaposition the lines of two spectra as in Fig. 13.14 (§ 16.13)
and to adjust one of them until
p o i n t s of e q u a l d e n s i t y a r e
matched. This method is considerably superior to that involved
in determining the end point of a
line, since the minimum observable density has low sensitivity
because of the very low contrast
t
of the emulsion at this point in
the characteristic curve. When
(a)
(b)
two lines varying logarithmically
Fig. 13.14. Matching of two spectra
in intensity along their lengths
by means of an optical comparator.
In (a) are shown two spectrum lines
are brought into juxtaposition,
photograplied through a logarithmic
they can be compared along their
sector. The two rectangles represent
entire lengths.
two different fields of view of the com-
i-
k
Ju,
parator that can be translated horizontally and vertically so t h a t any part
of line 2 can be brought into coincidence with any part of line 1. In
(b), t shows the distance through which
line 2 has to be translated to match the
density of line 1. t may be read
directly from the comparator.
DENSITOMETRY
The density of the silver deposit
in any portion of a photographic
emulsion is most readily measured
with a densitometer, a photometric
device designed to determine the reduction in intensity of a beam of
light when sent through a restricted area of the emulsion. When the
area is small, as in the case of a spectrum line, the instrument may
be called a microdensitometer; it is sometimes incorrectly referred to
as a "microphotometer" (General References 13.2-13.5).
Fundamental features of a densitometer are a source of light that
provides a beam to measure the plate, provision for ensuring that this
beam passes through only that part of the emulsion which' is to be
measured, and a device to compare the intensity of the beam of light
after passing through the exposed emulsion to that of the same beam
passed through an unexposed area of emulsion. Most densitometers
§13.12]
PHOTOGRAPHIC PHOTOMETRY
351
utilize thermoelectric or photoelectric devices for measurement of the
beam, though a few less precise instruments use eye-match methods.
13.12. Densitometers. Densitometers may be classified as
direct-reading, in which the deflection is a function of the density of
the plate, and null instruments, which permit matching of an unknown
density by an equivalent known density. Direct-reading densitometers are of the recording type, in which the deflection of the
indicating device is recorded on paper or on a photographic plate or
film, and the spotting or nonrecording type, in which deflections are
read only at selected locations, as for example on the flat tops of
certain spectrum lines. Many densitometers are of the projection
type, in which an area of the emulsion immediately surrounding the
part being measured can be observed visually immediately before or
during measurement.
A densitometer may determine either diffuse or specular densities.
In measuring diffuse density, all the light passing through the emulsion up to a considerable angle is included in the emergent beam,
which is then compared to the incident beam. Specular density, on
the other hand, is determined by including only that part of the light
which continues in the direction of the original beam. Specular
density is thus always greater than diffuse density for a given silver
deposit. Most microdensitometers used in spectroscopic work measure specular densities.
The density of a silver deposit varies considerably with the color
of the light used to measure it. A densitometer that uses blue light,
for example, will give entirely different density readings for a given
deposit from one using red light. This effect makes it necessary to
keep as constant as possible, during the measurement of a given plate
or film, the temperature of the densitometer lamp filament and the
arrangement of all optical parts used. In photometric work, only
densities on a given plate are compared, so that this requirement offers
no difficulty, and changes made between plates are of no consequence.
Figure 13.15 shows the optical system of a typical microdensitometer. Two beams may be supplied, one with which to inspect the plate
to aid in locating the exact spot to be measured and a second to carry
out the measurement. Often the first beam is converted into the
second at the moment of measurement by insertion or removal of a
lens, and sometimes of a filter. Unless care is taken, light scattered
in the optical system of the densitometer will produce errors in
[§13.13
PHOTOGRAPHIC PHOTOMETRY
352
reading.!" j^^ jg necessary not only to take into account the error in
reading produced by the scattered hght but also that produced by
variations in the amount of scattered light which result when different
parts of the plate are brought into the beam. The effect is to reduce
the apparent density of areas surrounded by less dense areas, and is
one of the most common sources of error in densitometry. Every
densitometer when first set up should be tested for scattered light.
The test should be repeated whenever optical surfaces may have
become dusty, by setting on a very dense line and then seeing if the
deflection changes when most of the light passing through the surrounding plate is shielded.
Condensing
lens
—"To golv.
Lannp
Lens Slit
Projection
Slit Photocell
or
thermopile
Fig. 13.15. The optical system of a typical microdensitometer.
This effect is especially serious in instruments having mirrors or
large numbers of lens components in their optical systems. It is
possible to reduce the effect by cutting off any viewing beam during
actual measurement, or in a photoelectric instrument by using for
viewing light of a color which, though visible, does not affect the
photometer. It can also be minimized by using an extremely narrow
shutter to simulate a spectrum line of infinite density and taking a
zero reading with this over every area being measured, thus permitting
the scattered light to enter the photometer while the zero is being
taken.
13.13. Photoelectric Densitometers. The original photoelectric
microdensitometer was designed by Koch.^^' His optical design is
shown in Fig. 13.16. Light from the Nernst lamp L (in modern
instruments replaced by an automobile headlight lamp) passed
through the plate P to be measured and was focused on photocell Ci. An electrometer E was used to measure the voltage across
this cell, this voltage depending not only on the amount of light
passing through the plate but also on that falling on cells C2 and C3,
«K. Schwarzschild and \V. Villiger, Astrophys. Jour., 23, 287 (1906).
« P. P. Koch, Ann. d. Physik, 39, 705 (1912)
§13.13]
PHOTOGRAPHIC PHOTOMETRY
353
which were connected so as to form a variable compensating leak
across Ci. Two cells were used in parallel so that the electrometer
could work on either side of ground potential. A projection system
was used to throw an image of the electrometer thread on a recording
plate.
The Koch microdensitometer as improved by Goos'^ was manufactured commercially for many years by Kriiss of Hamburg. A
projection system for viewing, the plate during measurement was
added, and to eliminate errors due to scattered light a red filter was
put in the viewing beam and a photocell insensitive to red light was
used. This recording instrument, in which various ratios of record
Fig. 13.16. The optical and electrical arrangement of the original photoelectric
microdensitometer of Koch (1912).
motion to plate motion were provided, was in effect a recording
comparator.
In recent years photronic and other barrier-layer photocells have
come into wide use in microdensitometers, since they require no
batteries and can be connected directly to a galvanometer. The
Hilger densitometer shown in Fig. 13.17 uses such a cell. When
barrier-layer cells are used, it is important to put a suitable resistance
in series or in parallel with them so that the galvanometer will be
made approximately dead-beat when the cell is exposed to light corresponding to a medium-density reading. It is desirable to choose
cells of a type showing low fatigue effects and to remember that
photoconducting cells give linear current vs. light flux for low load
resistance.
The advantages of photoelectric densitometers are high sensitivity,
which permits use of short-period galvanometers; freedom from
thermal disturbances; limited spectral sensitivity, so that few errors
" F. Goes, Zeitschr.f. Inst., 41, 313 (1921); Phys. Zeitschr., 22, 468 (1921); F. Goes
and P. P. Koch, Zeitschr. f. Physik, 44, 855 (1927).
1
354
PHOTOGRAPHIC PHOTOMETRY
[§13.13
are introduced if the focusing lenses are not completely achromatic;
and output that can readily be fed into amplifiers. The characteristics of photoelectric photometers have been discussed in
Chapterl2.
The photoelectric cell responds more quickly than any thermoelectric device yet developed; and since ordinarily it has a higher
output impedance, its response can be amplified more effectively. It
13.17. The Hilger microdensitometer, Model H451 with HS34 reader unit
attached. (Courtesy Jarrell-Ash Company, Boston.)
can be used as a sensitive null indicator. In this form it provides the
basis for the most accurate type of densitometer yet devised, since
almost every factor producing irregularities can be balanced out.
Two beams of light may be used, one passing through the plate to be
measured and the other through a comparison wedge of graded
density, which should be as similar as possible to the plate and hence
may well be made of an exposed photographic emulsion. A single
§13.14]
PHOTOGRAPHIC PHOTOMETRY
355
beam can also be used, made to pass alternately through the plate to
be measured and the wedge.
In a well-designed densitometer every variable should be kept as
constant as possible. For example, the same photocell should be used
to measure both beams, if possible. Although two cells may be
balanced at one light intensity, they are likely to get out of balance
at other intensity levels. If great care is taken to match them in this
respect, it will usually be found that the sensitivity of one will change
with time more than that of the other. Cells are likely to vary in
sensitivity over different portions of their sensitive surfaces and to
have different wavelength responses, so that if the color of one measuring beam changes, they will be thrown out of balance. An ideal
photometer of this sort would be one in which a single light source,
light beam, and measuring device were used, the plate to be measured
and a similar plate whose density characteristics were known being
put alternately into the beam. Such a design has the advantage that
the precision would be about the same for high as for low densities,
readings being reproducible to 0.2 per cent or better.
A photomultiplier cell has been used in the microdensitometer that
is a part of the automatic comparator designed by Harrison (§9.11),
in which ten inches of plate can be scanned per minute, the opacity
curve being photographed on motion-picture film. The photomultiplier tube is to be considered as being a phototube plus a very
simple and convenient form of DC amplifier.
13.14. Thermoelectric Densitometers. Thermoelectric devices of
the sort discussed in Chapter 12 lend themselves to use in simple
and rugged densitometers. They are more suited to use in direct
deflection than in null instruments. On account of the low impedance of most thermoelectric devices, their output currents
are difficult to amplify electrically and are ordinarily fed to lowimpedance galvanometers, though the so-called G-M amplifier
(§ 12.10) has been highly successful. The prototype of thermoelectric densitometers is that of Moll,i' which has for many years
been manufactured commercially by Kipp and Zonen in Delft. It
was in the early years cheaper and simpler than its photoelectric
competitors, though not so rapid in action.
The advantages of using simple thermocouple-galvanometer combinations have been offset to a considerable extent in recent years by
' W. H. J. Moll, Proc. Phys. Soc. (London), 33, 207 (1921).
356
PHOTOGRAPHIC PHOTOMETRY
[§13.15
the introduction of the photronic cell, which is an equally simple and
more sensitive photometric device and is, moreover, free from thermal
disturbances. Thermoelectric photometers must be very carefully
shielded from air currents if they are to be kept free of drift. They
are sensitive to all wavelengths and so must be shielded from stray
light and heat rays. Focusing lenses used with them must be carefully achromatized, since most of the light received from an incandescent lamp lies in the near infrared region. Neglect of this fact
may greatly reduce the resolving power of a microdensitometer where
an image of the slit is thrown on the plate by an incompletely achromatized lens.
Densitometers have been constructed using the Boys radiomicrometer and the Nichols radiometer, and various types of thermocouples,
bolometers, and thermopiles. These densitometers, though formerly
the simplest to use, are in general giving way to the newer types of
photocells for densitometric purposes.
13.15. Operation of the Densitometer. Although the operator of
a densitometer may feel that he need make only one measurement for
each spot measured on the photographic emulsion, actually each
determination of the density d involves measuring four quantities.
(13.2)
where 7o is the incident light being used to measure the plate, I is
that transmitted by the portion of the plate being measured, and
G, G', and Go are the readings of the galvanometer corresponding to
intensities /o, / , and no light, respectively. It cannot always be taken
for granted that Go will remain constant from one reading to the next.
The difference in deflection G — Go, called the clear reading, should
be taken through a developed but unexposed portion of the plate
having any fog and stray light exposures on it that may be present in
the spectrum but having no exposure to the spectral radiation being
measured. G — Go is also sometimes called the "100 reading," since
it is convenient to adjust the intensity of the densitometer beam so
that the galvanometer scale reading is exactly 100 for a clear portion
of the plate near the exposed portion to be measured. The difference
G' — Go is the deflection on the exposed line. In a good densitometer
the light source remains constant and the detector does not drift, so
that the 100 reading and the zero reading remain constant. To
justify these two assumptions, it is necessary to use a very constant
§13.15]
PHOTOGRAPHIC PHOTOMETRY
357
light source, operated from a storage battery, a stabilized transformer,
or an electronic voltage regulator, and to use a recorder and amplifier
system with little drift.
G depends not only on the brightness of the densitometer beam
but also on the transmission of the unexposed portion of the plate,
which is likely to vary considerably over the plate surface. In consequence, one of the principal problems in photographic photometry
is to determine where to measure the clear plate readings G. Best
results are obtained when chemical and other controllable fog is kept
to a minimum so that the plate surface is as uniform as possible in
Fig. 13.18. The Baird Associates'nonrecording densitometer.
transmission. Neglect of this factor is probably responsible for more
of the errors of photographic photometry than any other cause.
If zero drift is low, the number of determinations of Go can be kept
small. When Go and G are kept constant, G' is the only quantity
that need specifically be determined for points on the plate which are
to be measured, and it is the continuous curve of G' values which
is given by a recording microdensitometer.
Some operators prefer to use blackening rather than density as an
indicator of plate response. Blackening is the difference between the
clear plate reading and the line reading, or G — G', when the clear
plate reading is set at 100, and Go is zero.
358
PHOTOGRAPHIC PHOTOMETRY
[§13.16
13.16. Recording Densitometers. The recording devices discussed in Chapter 12 can be used for densitometry. Ordinarily a
record is useful only when line contours or some similar features are
to be measured. For routine work in quantitative spectrochemical
analysis or for making intensity measurements on spectrum lines, it is
usually more convenient to make spot measurements than to spend
the time required for taking a complete record and then selecting from
each record the lines to be measured.
Fig. 13.19. The Leeds and Northrup recording microdensitometer.
In densitometers of the older type, the galvanometer deflections
were recorded by a light beam on photographic paper. This paper
was wrapped on a drum which was driven by clockwork or, in later
models, by synchronous motors provided with reduction gears. In
some of the higher-priced commercial instruments, which are now
less in fashion than formerly, complicated gear arrangements were
provided to move a long photographic plate on which the opacity
curve was recorded. This recording method is of importance only
when the comparator features are necessary so that intensities and
§13.18]
PHOTOGRAPHIC PHOTOMETRY
359
distances along the plate can be exactly determined simultaneously.
Modern practice tends more in the direction of rapid records on
paper, as in the Leeds and Northrup densitometer shown in Fig. 13.19.
13.17. Precision of Densitometers. The precision available in
well-designed densitometers is usually greater than that required for
photographic photometry. Instrumental errors can readily be kept
below 0.5 per cent for the direct-reading type and below 0.2 per cent
for the null type. Failure to reproduce readings to this precision is
usually caused by the difficulty of repeat setting on the same portion
of the plate. An advantage of the recording method is that no
uncertainty is introduced as to whether the exact peak of a narrow line
is set upon. To obviate this uncertainty, many spotting instruments
are provided with a transverse screw motion that can be quickly
locked in when the line is nearly in position. The line is then slowly
moved across the measuring light beam and the minimum deflection
of the galvanometer is read.
The Zeiss firm at one time manufactured a spotting microphotometer in which a given spectrum line would appear in position at the
slit when the proper key was pressed. Thus when the same four or
five lines were to be measured in a number of different spectra on
the same plate, it was easy to set on one line after another, then to
rack up the next spectrum and measure the same lines. The Applied
Research Laboratories manufacture a projection instrument with a
similar feature, shown in Fig. 13.20. This uses a contact slit.
When very short and narrow slits are used in the densitometer,
variations in deflection may be encountered due to graininess of the
emulsion. These effects can be reduced by lengthening the slit,
providing an oscillating motion of the plate parallel to the spectrum
line, using an astigmatic optical system, or by various other means.
13.18. Special Computing Densitometers. Since it sometimes
takes longer to determine with the aid of calibration curves the
intensity of the light which has fallen on a plate than to measure the
plate with a densitometer, it is desirable to have this translating done
automatically. Wouda'^ has described an instrument for the rapid
reduction of data from a calibration curve. The image of a long
straight-filament lamp is moved across a plot of the calibration curve
by the deflection of the galvanometer mirror of the densitometer. By
adjustment of the size of the plot or the optics of the projection
system, line intensities can be read off directly from the intersection
" J. Wouda, Zeitschr.f. Physik, 79, 511 (1932).
^
360
PHOTOGRAPHIC
PHOTOMETRY
[§13.18
of the filament image and this curve. However, the calibration
curve for each portion of the plate must be determined by the operator and set into the machine; and since from 2 to 20 calibration
curves may be required for reduction of a single plate, the region
which can be reduced from a single curve is likely to be limited.
Fig. 13.20.
The Applied Research Laboratories' projection microdensitometer.
A similar automatic device is that of Thompson,'-" which sends a
long line of light from the galvanometer mirror of the densitometer
through a slit cut to the predetermined shape of the calibration curve
for a portion of the plate. That section of this line which is transmitted through the slit lies in a position that indicates the intensity
of the line directly. This method suffers from the same limitations
2»N. Thompson, Proc. Phys. Soc. (London), 4S, 441 (1933); see also G. O. Langstroth and D. R. McRae, Jour. Opt. Soc. Am., 28, 440 (1938).
§13.18]
PHOTOGRAPHIC PHOTOMETRY
361
as W o u d a ' s device. T h e utility of these devices lies in the fact t h a t
the intensity curves obtained with t h e m from a continuous record
can be integrated directly by means of a planimeter to obtain t o t a l
intensities of lines t h a t are not flat-topped.
GENERAL REFERENCES
13.1.
L. S. Ornstein, W. J. H. Moll, and H. C. Burger, Objektive SpeJdral•photometrie. Braunschweig: Vieweg, 1932.
13.2. G. R. Harrison, "Instruments and Methods Used for Measuring Spectral
Light Intensity by Photometry," Jour. Opt. Soc. Am., 19, 267 (1929).
13.3. G. R. Harrison, "Current Advances in Photographic Photometry,"
Jour. Opt. Soc. Am., 24, 59 (1934).
13.4. W. E. Forsythe (Ed.), Measurement of Radiant Energy. New York:
McGraw-Hill Book Company, Inc., 1937, Chapters VIII and I X .
13.5. K. Henney and B. Dudley (Eds.), Handbook of Photography. New
York: McGraw-Hill Book Company, Inc., 1939.
13.6. G. M. B. Dobson, I. O. Griffith, and D. N. Harrison, Photographic
Photometry. London: Oxford University Press, 1926.
CHAPTER
14
Absorption Spectrophotometry
ABSORPTION SPECTKOPHOTOMETBY IS THE TECHNIQUE of determin-
ing the relationship between the wavelength, or frequency, of radiation
and its attenuation by absorption upon passage through a particular medium. The following discussion will be limited to absorption
spectrophotometry in the region from approximately 2000 to
10,000 A. These wavelengths correspond to the nominal limits
within which measurements can be made satisfactorily with usual
equipment for the visible and ultraviolet. Spectrophotometric
measurements at wavelengths longer than about 10,000 A (in the
infrared) and shorter than about 2000 A (in the vacuum ultraviolet)
require special techniques that are discussed respectively in Chapters
17 and 19.
Spectrophotometry in the range 2000 to 10,000 A is normally applied to the study of substances in the solid or liquid state or in
solution. The observable absorption may be continuous and fairly
uniform throughout the region in question, in which case it is called
general absorption, or it may exhibit one or more broad maxima and
minima, in which case it is called selective absorption.
The techniques to be described have wide application in organic
chemistry in the characterization of compounds, in the control of
isolation and purification procedures, in the determination of molecular structure, and in qualitative and quantitative analysis. Other
applications include the determination of the transmission characteristics of dyestuffs and filters, the study of cytochemical and
histochemical problems by microspectrophotometry, and the investigation of photochemical reactions.
LAWS OF ABSORPTION AND THEIR APPLICATION
TO ABSORPTION SPECTROPHOTOMETRY
14.1. Lambert's Law. The simplest case of absorption is that in
which a parallel beam of monochromatic radiation passes rectilinearly
362
§14.1]
ABSORPTION SPECTROPHOTOMETRY
363
through a homogenous absorbing medium. Under such circumstances, the intensity of the radiation is reduced by the same fractional
amount in equal succeeding portions of its path. Thus, if the intensity is reduced by half in the first centimeter, it will be reduced by half
again in the second centimeter (that is, to one-fourth of the original
intensity), and so on. The medium may be considered to be made
up of layers of infinitesimal thickness, dl, perpendicular to the path
of the radiation. Let / represent the intensity at any point of the
path, and a the fraction by which absorption reduces the intensity in
unit length,of path.* Then
- ^ = «/
(14.1)
The constant a is known as the absorption coefficient. I t is characteristic of the absorbing medium and is a function of the wavelength
of the radiation.
When Eq. (14.1) is integrated between the thickness limits 0 and x,
an expression known as Lambert's law results. It gives the ratio,
la/lx, of intensity of radiation before, to that after, passing through
the thickness x:
logeY = a^'
or
1^ = 106-""
(14.2)
Using base 10 instead of base e, Eqs. (14.2) become, respectively,
logio r = ^^'
or
^^ = -^0 lO""""'
(14 3)
where K = 0.43iSa: and a = 2.303i?'. The constant K is known as
the extinction coefficient;^ 1/K is the thickness of absorbing layer
necessary to reduce I^ to (l/10)7oOnly in the rare case of a neutral absorbing substance, which
exhibits the same absorbing power throughout the entire wavelength
region under consideration, are a and K independent of wavelength.
* There is considerable variation in the letter symbols used by various authors to
designate quantities of interest in absorption spectroscopy, and the symbols used here
should, therefore, not be regarded as representing a universally accepted standard
nomenclature. Insofar as practicable, the symbols used have been chosen to correspond with those employed widely in the literature, including scientific papers in
German and French as well as in English. Where confusion might arise from the use
of one symbol for two different quantities, a separate letter symbol has been used
for each quantity.
i R . Bunsen and H. K. Hosccie. Pojg. Ann. Physik, 101, 235 (1857).
364
ABSORPTION SPECTROPHOTOMETRY
[§14.2
I n all other instances, t h e preceding expressions are valid only for
strictly monochromatic radiation. A p p a r e n t deviations from L a m bert's law m a y , therefore, occur in practical measurements if t h e
radiation employed is insufficiently monochromatic. Other causes
of a p p a r e n t deviation from L a m b e r t ' s law are geometric factors, such
as obliquity or lack of parallelism of the beam of radiation, which
m a y introduce errors in the assumed p a t h length x; lack of h o m o geneity in the absorbing m e d i u m ; a n d losses b y reflection from surfaces in t h e p a t h of the beam of radiation.
14.2. B e e r ' s Law. If t h e absorbing medium is a substance in
solution, the attenuation of radiation on traversing a given p a t h
length depends on t h e concentration of t h e solution. Under suitable
conditions, the absorption by dissolved substances is closely proportional to the n u m b e r of molecules of solute per unit volume of solution, whence
a = fiG
•
(14.4)
and
A' = kc
'
(14.5)
where a and K are, respectively, t h e absorption and extinction
coefficients as previously defined, c is t h e concentration of t h e absorbing substance, expressed in suitable units, and yu and k are, respectively
the absorption and extinction coefficients per unit of concentration.
T h e above expressions are known as Beer's Law.''
Beer's law is based on t h e assumption t h a t t h e specific absorption
per molecule of t h e absorbing substance does not vary with t h e
concentration of t h e substance in solution. This condition is often
satisfied within t h e limits of experimental error (particularly for
highly dilute solutions), b u t it is b y no means universally applicable.
Changes in concentration m a y lead t o changes in t h e n a t u r e of t h e
molecular species in solution. Such changes include, on t h e one
hand, polymer formation and the formation of other kinds of molecular associations (including complexes with the solvent), and, on t h e
other h a n d , dissociation. If suclijchanges t a k e place and if t h e
specific molecular absorption is influenced thereby, yu and k become
functions of c instead of constants, a n d a p p a r e n t deviations from
Beer's law are observed.
\
L a m b e r t ' s a n d Beer's laws-n?ay be combined in single expressions
' A. Beer, Pogg. Ann. Physih, 86, 78 (1852).
§14.3]
ABSORPTION SPECTROPHOTOMETRY
365
for the attenuation of the monochromatic radiation that traverses a
soUition of absorbing material, thus:
logio Y = '^cz;
or
h = he-"'-'
(14.6)
or
i^ = /„io-*«
(14.7)
If the concentration c in the Lambert-Beer equations is expressed in
gram-molecular weights per liter (moles), the corresponding coefficients of specific absorption may be represented by E and e; the
latter is known as the molecular extinction coefficient.
14.3. Variables Measured in Absorption Spectrophotometry.
Spectrophotometric measurements of the absorbing characteristics of
substances lead to the evaluation, at one or more wavelengths, of
the transmittsivity T. where
r =^
n4.8)
the per cent transmission t, where
t= 100—
1,14.9)
the opacity 0, where
0 = Y
U4-10)
or the optical density D, where
I» = log,of°
(14.11)
Such evaluation may be accomplished either by direct determination
of the ratios Ix/Io or logio {lo/Ix) or by separate determinations of
the two light intensities or the logarithms thereof.
To determine the attenuation of the radiation that traverses a
medium, a suitable specimen is placed in the path of the radiation.
If the substance is a solid, the specimen may be a slab or block of
material with polished, plane-parallel faces, so placed that a parallel
beam of the radiation enters and leaves the specimen perpendicularly
to the faces. If the substance is a liquid or is in solution, it may be
contained in a cell with polished, plane-parallel end plates that are
substantially transparent to radiation in the region of interest. The
cell and its contents may then be placed in the path of the radiation
in the same manner as a solid specimen.
t
366
ABSORPTION SPECTROPHOTOMETRY
[§14.4
14.4. Elimination of Effects Due to Reflections and Absorption by
Cell Windows and by Solvents. In order for the foregoing equations
to be applicable to absorbing specimens, /o rnust be the intensity of
radiation that enters the absorbing medium and 7^ the intensity
incident upon the exit face. 7o is not usually the same as the intensity I incident on the entrance face of a solid specimen or cell, nor
is Ix the same as the intensity i of radiation leaving the exit face,
since the intensity is reduced at all faces by reflection whenever
changes in refractive index occur at these points. The intensity may
be reduced further by absorption in the cell windows in the case of
a cell containing a liquid or solution. In absorption spectrophotometry, I and i are the intensities that are readily measurable. One is
confronted, therefore, with the problem of how to determine 7o and
Ix (or IQ/IX) from measurements of 7 and i (or I/i).
In the case of solids, 7o/7i may be determined by measuring ii and
t2 for two specimens of unequal lengths Xi and x^. If Xi is greater
than X2, ii/ii is equivalent to lo/Ix- where x' corresponds to the
difference in path length Xi — x^, provided that the entrastce faces
and the exit faces of the two specimens reflect the same fractions y
and 6 of the radiation incident upon them.
t^ = ^ " = ^ . ' = ^
ii
e-"^
(14.12)
Ix'
The same method may be employed for absorbing liquids. The
intensities ii and t2 of the transmitted radiation are determined with
cells of different lengths Xi and x^ but otherwise identical. The ratio
ii/ii is then equivalent to lo/Ix; where x' = Xi ~- x^.
With substances in solution, it is also essential to eliminate effects
due to absorption by the solvent. If the method outlined for liquids
is employed, the resultant measurement of 12/11 yields a value of
Issllx' representing the attenuation of intensity by both solute and
solvent in a path length x'. To eliminate absorption by the solvent,
it is customary to employ two cells that are identical in every respect,
including length, one containing the solution and the other the
solvent. If the solvent has the same absorbing effect in the presence
or absence of the solute and if the reflection coefficients atf the cell
faces are the same in both instances, the ratio 7o/7i can be measured
directly. The first condition holds except for instances in which
interaction between solute and solvent modifies the absorbing characteristics ,of both. The second condition is not true insofar as the
§14.5]
ABSORPTION SPECTROPHOTOMETRY
367
inner faces of the cells are concerned when the index of'refraction of
t h e solution is different from t h a t of the solvent. However, t h e
reflection error introduced by differences in refractive index at t h e
inner faces is not usually as great as other experimental errors a n d
is commonly neglected.
Additional sources of error in the determination of la/Ix m a y arise
from fluorescent radiation emitted by the absorbing medium, scattering of radiation b y suspended particles in a liquid medium, multiple
reflections between the faces of a solid specimen or a cell, and reflections from the sidewalls of a cell or the sides of a solid specimen when
t h e beam of r a d i a n t energy is not strictly parallel. T h e conditions of
measurement should be so chosen as t o reduce such effects t o a
minimum.
14.5. P r e s e n t a t i o n of Data. All spectrophotometric d a t a are expressed in terms of two variables which it is convenient t o call t h e
intensity variable and t h e wavelength variable. T h e intensity variable
indicates, directly or indirectly, the power of the absorbing substance
t o a t t e n u a t e radiation a t a particular wavelength. I t m a y be expressed as transmissivity, per cent transmission, opacity, absorption
coefficient, extinction coefficient, and so on, as these terms have been
defined previously. T h e wavelength parameter designates the wavelength, or frequency, of radiation to which a particular attenuation
factor applies. I t m a y be expressed as wavelength in angstroms or
millimicrons, as frequency in oscillations per second, or as wave
n u m b e r in cm"^. T h e choice of the most appropriate terms in which
t o express these two variables depends on the application to be m a d e
of the d a t a .
I n t h e case of optical filters, the information usually desired is t h e
fraction of the incident intensity t r a n s m i t t e d a t various wavelengths
b y a filter of definite thickness and composition. T h e intensity
variable m a y therefore be expressed in transmissivity or per cent
transmission.
I n t h e use of spectrophotometric d a t a t o characterize chemical
compounds, it is desirable, for purposes of comparison, to express t h e
intensity variable in terms t h a t are independent of the specific conditions of m e a s u r e m e n t (such as length x of optical p a t h or concentration c of a solution). For solids or liquids, the intensity variable is
therefore conveniently expressed in t e r m s of the absorption coefficient a or extinction coefficient K. F o r solutions, the corresponding specific absorption and extinction coefficients ii and k per u n i t
368
ABSORPTION SPECTROPHOTOMETRY
[§14.6
concentratio'n may be employed. Molecular extinction coefficients
are convenient for comparing the absorption, of solutions on the basis
of molecular concentration or for computing the absorption of a
complex molecule in terms of its constituent chemical groups. Occasionally, the logarithm of the extinction coefficient or of the molecular
extinction coefficient is employed in order to permit presentation of
curves with widely different values on a single graph sheet or presentation of curves having specific shapes characteristic of the substances
regardless of the units of concentration and path length employed.
The wavelength variable for chemical compounds is usually expressed
in angstroms or millimicrons. Frequency or wave-number units
have some advantages for theoretical studies but are less generally
used.
Table 14.1 summarizes the various expressions employed for the
intensity variable in absorption spectrophotometry.
SELECTION OF APPARATUS FOR ABSORPTION
SPECTROPHOTOMETRY
14.6. Choice of Source of Radiation. For absorption spectrophotometry in the visible region, incandescent lamps are generally
used. An ordinary 60 to 100 watt lamp with flashed opal bulb
placed close to the slit of the dispersing system will give sufficient
intensity for the direct qualitative observation of spectra (without a
photometer) and has the advantage of giving uniform slit illumination
without the necessity of critical alignment. Where greater intensities
are required for quantitative observations, a lamp with clear bulb rnay
be employed in conjunction with a condensing lens for forming an
image of the filament on the slit. The shape, size, and brightness of
the filament should be such that the working portion of the slit-can, be
filled with uniform radiation of sufficient intensity for convenient
observation. The image of a ribbon filament most closely approximates the geometry of the usual slit, but concentrated coil filaments
of the type used in projection lamps give sufficiently good images for
many purposes. Ribbon-filament lamps are used in lieu of slits in
the collimating systems of cectain simple photoelectric spectrophotometers. Suitable sources for use with polarizing and other
types of photometers for the visible region are described in Chapter 8.
For abbreviated absorption spectrophotometry in the visible region,
using filters instead of a dispersing system, mercury arcs have the
TABLE 14.1
STJMMAHT or
Expression
EXPRESSIONS USED FOK TRANSMISSION AND ABSORPTION OP
RADIATION
Units usually
employed
Nomenclature
Principal
applications
T — transmissivity
la
t = 100^-
I in cm
/3 = coefficient of
Transmission charactertransmission
istics of filters and dyestuffs
t — per cent transmission
0 = opacity
Photographic blackening measurements.
Indication of absorption of solutions
a = absorption co- Absorption of solids or
X in cm or mm
/o
efficient
liquids (Continental
usage)
/o
X in cm or mm, c in fj. = absorption co- Absorption of solutions
grams per liter,
efficient
(Continental usage)
mg per ml, per
cent, or grams
per ml
X in cm or mm
K = extinction co- Absorption of solids or
efficient
liquids (British and
American usage)
X in cm or mm, c in k = extinction co- Absorption of solutions
grams per liter,
efficient
(British and American
mg per ml, per
usage)
cent, or grams
per ml
/« X in cm, c in moles e = molecular ex- Useful in computing the
tinction coper liter
combined absorption
efficient = Mk
of substances present
in known molar ratios
Used in compressing data
so that curves occupy
less space, etc.
Used in discussion of ab/o X in cm in medium 7 = absorption
index
under considersorption as derived
from classical electroation; X = —
magnetic theory
n
where Xo is for
vacuum and n
is refractive index; X in cm
369
D = logio^—
1
1
1,
logio t or
log, k
X
Z> = density
Photographic blackening measurements
370
ABSORPTION SPECTROPHOTOMETRY
[[14.6
advantage of providing several widely spaced emission lines, permitting isolation of more nearly monochromatic radiation than is
possible by the use of filters with continuous sources. This advantage
is, however, somewhat offset by the resulting restriction to a limited
number of wavelengths; hence incandescent sources are ordinarily
employed. There is usually no slit or stop in such a system to act
as a secondary source. Therefore, if the optical system is to provide
a nearly parallel beam of radiation, a lamp with concentrated filament
or emitter, such as a projection lamp, an automobile headlight lamp,
or a Western Union concentrated arc, must be used.
For photographic absorption spectrophotometry in the ultraviolet
region, a condensed spark between tungsten-steel electrodes is convenient and often adequate. The spectrum of this source is rich in
strong lines, but there are troublesome gaps (for example, at about
2650 A), and the intensity falls off abruptly below about 2350 A.
Aluminum electrodes give several strong groups of lines in the region
below 2000 A, in which tungsten is particularly deficient. They may
be used as a supplementary source when it is desired to measure
absorption in this region. Uranium electrodes give a spectrum very
rich in lines of almost uniform intensity and have been recommended
as a substitute for continuous sources.' They are costly, however,
and their spectrum is quite weak below 2200 A. With suitably cored
carbons, such as National Carbon Co. 6 mm C, carbon arcs can be
used satisfactorily as a substitute for the condensed tungsten spark.
They may require somewhat more frequent adjustment than sparks,
even when provided with automatic feeding mechanisms, and they
do not approximate point sources so closely.
Continuous sources have a definite advantage in ultraviolet
spectrophotometry when it is desired to study the finest details of
complex absorption spectra. Of such sources, the underwater "spark
is adequate, but hydrogen discharge tubes in quartz are more convenient and more readily available. Both of these sources permit
observation to the short-wave limit of transmission of usual ultraviolet spectrophotometric equipment. The General Electric' type
H-4 high-pressure mercury arc with outer glass envelope removed
yields a spectrum with :a> strong continuous background, in addition
to the mercury lines, down to about 2100 A.*
* Some lamps of this type will not operate satisfactorily without the outer glass
envelope Unless they are placed in an enclosure that will maintain them at-a comparatively high operating temperature.
§14.7]
ABSORPTION SPECTROPHOTOMETRY
371
For photoelectric absorption spectrophotometry, it is usually
essential t h a t t h e radiant o u t p u t of t h e source be free from fluctuations. I t is also desirable t h a t the spectrum of the source be continuous, since t h e presence of strong emission lines m a y cause such
characteristics as nonuniform wavelength response of the photocell,
scattered radiation, and limited resolving power of the dispersing
system, t o result in appreciable errors in the measurements. T h e
usual choices are, therefore, incandescent lamps for the visible region
a n d well-regulated hydrogen discharge tubes in quartz for t h e
ultraviolet.
For a more detailed discussion of sources, Chapter 8 should be
consulted.
14.7. Choice of Absorption Cells. T h e general characteristics t o
be considered in the choice of absorption cells for the spectrophotometry of liquids or solutions are as follows: transmission characteristics
of cell windows, p a t h length through absorbing medium {inside cell
length), precision and uniformity of inside cell length, flatness a n d
parallelism^ of window faces, perpendicularity of faces to the p a t h
of radiation, and convenience in manipulation.
Glass windows are sufficiently t r a n s p a r e n t for use in the region
10,000 to 3400 A. Below about 3400 A, quartz, fluorite, or lithium
fluoride must be used. Fused quartz is the usual window material
for cells for ultraviolet spectrophotometry. If it is desired to eliminate effects due to fluorescence of the windows and to obtain the
greatest possible transparency below 2200 A, crystal quartz* is
somewhat better.
T h e inside cell length should usually be chosen so t h a t it is possible
to obtain a density, logio (lo/Ix), of about 0.4 to 0.5 for photoelectric
spectrophotometry and of about 1.5 to 2.0 for photographic spectrop h o t o m e t r y , a t the absorption maxima for which precise data are
required. T h e possibility of diluting the absorbing material so as
to change its density per unit p a t h length within wide limits usually
permits considerable latitude in choice of cell length. For general
use, therefore, it suffices t o have available cells of a few convenient
inside p a t h lengths, say 1, 2, and 4 cm. F o r some samples, shorter
or longer p a t h lengths m a y be required.
T h e precision a n d uniformity of the inside cell length should be
such t h a t the over-all errors of the measurements are not increased
* Crystal-quartz windows must be cut perpendicular to the optic axis to avoid
effects caused by double refraction.
372
ABSORPTION SPECTROPHOTOMETRY
[§14.7
appreciably by path-length errors. Cells are available commercially
in which the inside-length tolerance is kept below ±0.15 per cent
of the total path, corresponding to a maximum density error of
about ±0.0023 at a density of 1.5 or of about ±0.00075 at a density
of 0.5.
The cell-window faces should be flat to within a few wavelengths,
at most, of the radiation used, and all faces should be parallel to
within a few minutes of arc.
The dimensions of cells perpendicular to the path of the radiation
should be sufficient to ensure that radiation will not strike the cell
walls and be reflected from them. This condition having been fulfilled, it is desirable to keep these dimensions to a minimum in order
that minimum volumes of material will be required.
For convenience in manipulation, cells with removable windows
have the advantage of being easy to clean. Certain designs may
Fig. 14.1. Typical types of cells for visible and ultraviolet absorption
spectrophotometry.
have disadvantages, such as variation in path length upon reassembly,
leakage, and contamination of cell contents by absorbing substances
from the gaskets. When gaskets are used, troublesome contamination can often be avoided by washing the gaskets repeatedly, before
use, with the solvent or liquid absorbing medium to be used in the
measurements. Spacer tubes of metal are sometimes used in demountable cells; glass or silica must be employed if there is danger
that metal may be attacked by the cell contents. Permanently
assembled cells with cemented-on windows can be used only with
fluids in which the cement is insoluble. Hence one-piece cells with
fused-on windows are preferable for genera,l applications. Permanently assembled cells with open tops or with two filler tubes are
somewhat easier to clean than those with a single filler tube, but the
single-tube type with a ground-glass stopper is useful when it is de-
§14.8]
ABSORPTION SPECTROPHOTOMETRY
373
sired to prevent evaporation of the cell contents during long periods
of use. Typical absorption cells are shown in Fig. 14.1.
A cell of variable length, such as a Baly tube (Fig. 14.2), is a convenience in preliminary explorations to determine the most convenient
concentration of a solution to employ with a particular cell of fixed
length. Baly cells are obtainable with mircometer adjustment
whereby the path length may be set to within about ±0.005 mm.
Other cells of special design, such as the notched echelon cell, are described in later sections dealing with special spectrophotometric
methods.
14.8. Choice of Spectrophotometric Method. I t is convenient to
classify spectrophotometric methods according to the radiation detector employed. If one uses this basis of classification, there are
three principal spectrophotometric methods from which to choose:
(1) visual, (2) photographic, and (3) photoelectric. Although
(a)
(b)
Fig. 14.2. Baly cells, (a) Sliding tube type, (b) Micrometer type.
thermoelectric methods are used in the infrared where no others are
available, their comparative insensitivity makes them undesirable for
regions where one of the three more responsive methods can be
employed.
The visual method is of use only in the region from about 4000 to
7500 A. For this region it has the advantage of simplicity and the
disadvantage of yielding results that may vary considerably from
observer to observer even under the best of operating conditions.
Particularly in those spectral regions in which the eye is least sensitive, some workers find it difficult to obtain satisfactory results by
this method.
The photographic method has the advantage of providing a permanent record which, at sufficiently high dispersion, may be made to
record the minutest details of complex absorption spectra. If the
photographic record is photometered visually, this method has some
374
ABSORPTION SPECTROPHOTOMETRY
[§14.9
of the disadvantages inherent in visual spectrophotometry, though
the situation is improved by the fact that the photometric procedure
can be carried out with light of high intensity and uniform spectral
quality. The pitfalls of visual photometry may be avoided by the
use of objective photometric devices (for example, photoelectric
densitometers; see Chapter 13). The time required to process the
emulsion is a further disadvantage of the photographic method.
Photoelectric methods may be subdivided into those which involve
point-by-point measurements at selected wavelengths and those
which permit automatic recording at all wavelengths within the range
to be investigated. A special case of the former is the absorptiometer
or colorimeter. The colorimeter is a device equipped with a simple
dispersing system, or with filters, that permits determination of the
concentration of substances by their absorption of approximately
monochromatic radiation at a selected wave band. Such devices are
useful in analytical chemical procedures. Point-by-point photoelectric spectrophotometers are convenient for routine measurements
in applications to analytical chemistry, organic chemistry, and biochemistry, but the data obtained with them ,may not show all the
important details of the absorption spectrum because of the missing
data between the observed points. Automatic-recording photoelectric spectrophotometers have the advantage of covering, continuously,
the entire range of wavelengths under investigation. They present,
however, maintenance and adjustment problems somewhat in proportion to their complexity. Photoelectric methods, in general, share
the advantage of yielding objective measurements that can be reproduced consistently with less skill and care than are usually required for
similarly consistent results with visual or photographic methods.
Table 14.2 summarizes the bases of choice among the various
methods of spectrophotometry (compare Table 12.1).
VISUAL ABSORPTION SPECTROPHOTOMETRY
'
/
14.9. General Considerations. Apparatus for visual absorption
spectrophotometry requires the following principal components:
(1) a suitable light source; (2) a means for separating the light,from
the source into two parallel beams (one traversing the absorbing
medium and the other serving as a comparison beam) and for bringing
these beams into juxtaposition so that they may be viewed ultimately as parts of the same photometric field; (3) a means of
514.9]
ABSORPTION SPECTROPHOTOMETRY
1
t
o ^
—
0
1
1
1
0)
O
O
a|
O
-3 a
cS-B
C
•"
C-i
QJ
O
O
S
s
i
a =>
a o
_«
"a
>
a
o
»
EH
fa
-a
OS
w
5
O!
eS-O
•O
fl
^
=3
o a
. S - 0 ^
alia ^ 1
u o S
O
,J3
o
<
io
t'
i =^
4J
O
O
1
o
o
o
0 M
fcH 0
^ o.g
"•^ ^ s a •
3
o
•"
5J
2 3 a
•g a « . 2
t- n
0
1
.0.
'3
o"
o
•a-s d S
^
^
•
^
•'^
^
la
0
§ =.3 g
.a
^ 2 S •»
iJ a -0 g rt 3
• ^
fe
<U 1
o
J
a
o
O
m a
4-'
T-H
1
o
o
o
OJ fc-
3 a
>a
r-
^
0
M
O
S o S
8 Sa
-0 a
•"
+^
y
U
ft
m
3
oj
=^a
.2 a. -S
e.S
^ '^5 g
a
>
a
" C3
•73 - g ^ O )
o
o
o
o"
d
65
H
lU
>•
rt
0
0
's'a
^ "..«
1
i a"^ a
Bog
rt
0
0
S
•» g S,
«
<
o
o
o
o"
1
o
o
<
<
1
T
o
o
o
o"
o
o
o
0
0
1
a
4J
O
•§^
s
a
cS
M3
0
a ^„ 0
|.a^
<
OH
a
o
0
O
0
PH
C- fcH OJ
ri 0
a;
C3
o
<
-a
P,
A
»
y
0
(..
• 2 i> a
.2
a
u
o
nS
a g<^ . . jj
a g = S £ 5?
QJ
-M
(U
•a
0
•^ ^ « s
PS
"3
«!
*-"
0
»_, a
a o-s
o
o
}> a.
ffs
sag)
S ^ r^
in
K
^
1 &4-I
c3
ia-°
"H,
m
'^^
^
.-
&< "^ -^ ??
O
T3
'3 i g = 2
O
>
°-^ » <Ufe
C O
4> >
-a c3.2
a ^ oE" -2 >— a
Safe
0)
3
-a
•§..2 S ^
a o
^
2 -« a t^ a
a ^ SJ3 22
o
3.'S
<5
V » 0
K4-I
o
C3
g >>
<
•
a en .. S
t
1
fc- 0
3
•s °
ci, a
o o
>; u
•8.2
rt
375
1
<U
376
ABSORPTION SPECTROPHOTOMETRY
[§14.9
changing, by known amounts, the intensity ratio of the two beams;
(4) suitable absorption cells (if liquids or solutions are used); and
(5) a spectrometer that permits isolation of a narrow spectral band
in both beams and presentation of this spectral region of the two
beams in a photometric field for visual comparison.
Any variation in the brightness of the different portions of the light
source from which the radiant flux is collected into the two parallel
beams may cause irregularities in the brightness of the different portions of the final photometric field. Consequently, directly illuminated diffusing screens or indirectly illuminated diffuse reflecting
surfaces are often used as light sources in order to ensure that the
entire area contributing flux to the photometer will be essentially
uniform. This method of illumination is wasteful of light and is
likely to result in low brightness of the photometric field, especially
when highly absorbing materials are to be measured. When the
highest intensity of effective illumination of the field is desired for
transmission measurements of specimens of high density, a nearly
uniform concentrated source, such as a Pointolite, Western Union
concentrated arc, or ribbon-filament lamp, may be used (§ 8.7).
The two light beams arfe separated, recombined, rendered parallel,
and brought to focus by suitable reflecting and refracting surfaces.
The design of this portion of the optical system is governed by the
applications to which the instrument is to be put as well as by mechanical and optical considerations. Control of the intensity of one or
both light beams is most commonly accomplished by the use of
polarizing prisms in tandem. The analyzer is mounted so that it
may be rotated with respect to the polarizer (or polarizers) by known
amounts indicated by a circulai; scale. Alternatively, fixed diaphragms of variable aperture or rotating sectors may be employed for
attenuating the beams. .
The photometer may be an integral part of the spectrometer^ as
in the original Konig-Martens instrument (§ 14.11), or it may be
separate frorn it. In the, latter case, it is cufetomary to use a constantdeviation type of spectrometer employing a Pellin-Broca prism
(Chapter 3). The light source, beam-splitting and focusing optics,
cell holders, and intensity-controlling device are, in this latter case,
usually mounted as a single unit which is called the photometer or
(somewhat incorrectly) the spectrophotometer.
Various forms of absorption cells may be used in studying the transmission of liquids and solutions. Inside path lengths of 2 to 4 cm
§14.10]
ABSORPTION SPECTROPHOTOMETRY
377
may be required for specimens that have low absorption coefficients.
Commercial visual spectrophotometers are usually designed so that
they may be used for reflectance as well as transmission measurements, and for measuring absorption by relatively opaque materials.
Since visuabspectrophotometers are intended for use in the visible
region of the spectrum only, all transmitting optical components,
except polarizing prisms, are of glass.
14.10. Advantages and Limitations of the Visual Method. The
visual spectrophotometer is easily set up and is free from the complications of photographic manipulation or of the operation and
maintenance of electrical measuring devices. As a consequence, it
has been used widely and is still employed considerably in industrial
control and research laboratories for measurements of transmission,
reflectance, and absorption by opaque samples within the range of
the visible spectrum.
The limitations of the visual method arise from certain characteristics of vision, among which the following are most important:
the matching of intensities by the eye involves subjective as well as
objective factors; under the same conditions of observation, different
eyes vary considerably in the precision with which they make it
possible for the observer to perceive differences in contrast; the
sensitivity of an observer to differences in contrast varies markedly
with the average brightness level of the visual field; and the visual
stimulus is a complex function of the wavelength of the exciting
radiation.
As shown in Fig. 14.3, the average observer is able to perceive
intensity differences of about 1 per cent (corresponding to density
differences of about 0.0043) at a brightness of 10 mL.* However,
at a brightness of 0.001 mL, an intensity difference of about 20 per cent
(corresponding to a density difference of 0.079) is required for discrimination. In order for the percentage density error to be small
in measurements of absorbing substances in the region of maximum
absorption, it is essential that the measurements be made with
samples having high density values (say 1.5 to 2.0). Such values
correspond to low transmissions and may result in reduction of
brightness of the field to the point at which sensitivity to contrast is
* The lambert, abbreviated L, is a unit of brightness equal to the average brightness
of a surface emitting or reflecting 1 lumen per square centimeter. The millilambert,
abbreviated mL, which equals 0.001 L, is generally used as the unit of brightness
except for very bright surfaces.
378
ABSORPTION SPECTROPHOTOMETRY
[§14.11
materially reduced (Fig. 14.3). This reduction is especially a p t t o
occur a t t h e two ends of t h e visible spectrum, where the luminosity
per unit of r a d i a n t energy is quite low.
I n general, it m a y be concluded t h a t (a) t h e visual method is useful
in some routine work, (b) more objective methods should be employed if the greatest precision is required, a n d (c) special caution
needs t o b e used when t h e visual m e t h o d is applied t o transmission
measurements of specimens having absorption bands a t t h e extreme
red a n d blue ends of the visible spectrum. I n the following four
sections, several typical visual spectrophotometers are discussed.
0.7
\
\
flB
B
0.4
U.i
\
\
0.2
0.1
0.00001
0.001
0.1
0.0001
0.01
1000
10
1.0
100
10,000
B= Brightness Level, Millilamberts
Fig. 14.3. Variation in discrimination of the eye for change in brightness,
A B, at different brightness levels, B.
14.11. Instraments Using the Martens Type of Polarizing
Photometer. T h e M a r t e n s t y p e of polarizing photometer (Fig. 1,4.4)
employs a Wollaston prism as a polarizer. When the two beams of
light t o be photometered are passed t h r o u g h such a prism, each beam
is split into two beams which diverge from each other and which are
plane-polarized in directions perpendicular t o each other. T h u s four
beams emerge from the Wollaston prism.
Of these beams, only two, polarized perpendicularly t o each other
and corresponding to the two original beams of light t o be p h o t o m e tered, are to be observed in the final photometric field. I t is neces-
§14.11]
ABSORPTION
SPECTROPHOTOMETRY
379
sary, therefore, to discard the undesired beams and to direct the
desired ones in such a manner that they will be brought together side
by side in the field of view. This result is accomplished by a biprism, which splits the four beams into eight but directs only the two
desired ones toward the photometric field, and by appropriate stops
ILLUMINATING DEVICE
(TOP VIEW)
SIDE VIEW
W
TOP VIEW
Fig. 14.4. Schematic diagram of Kiin^g-Martens Spectrophotometer
A,
light source; L, condensing lenses; '"C, a^bsorption cells; 6', entrance slits; Pi,
collimating lens and refracting prism; P2, dispersing prism; W, Wollaston
polarizing prism; B, biprism and telescope lens; G, analyzing prism; D, divided
circle; E, eyepiece lens. (By permission from Optical Methods of Chemical
Analysis by T. B. P. Gibb, Jr. Copyrighted 1942 by McGraw-Hill Book Co.,
Inc.)
which minimize the possibility that scattered light from the discarded
beams will traverse the system. The observer, looking into the eyepiece at the end of the optical train, sees a field of view divided in the
middle. One half is illuminated by plane-polarized light from one
of the original beams; the other half is illuminated by perpendicularly
plane-polarized light from the second of the original beams. A Glan-
380
ABSORPTION SPECTROPHOTOMETRY
[§14.11
Thompson or Nicol prism, placed a t a suitable position between the
biprism and the eye of the observer, is used as an analyzer.
As t h e analyzer is rotated continuously in one direction, the two
halves of the photometric field are extinguished alternately a t 90-deg
intervals. These positions of the analyzer are known as the extinction
points.
T h e two halves of the photometric field are m a t c h e d in
intensity a t intermediate positions between t h e extinction p o i n t s ,
known as the zero points.
If the original beams are of the same
intensity and if the optical system is symmetrical, zero points occur
halfway between extinction p o i n t s ; otherwise, this m a y not be
precisely t r u e .
Fig. 14.S. Polarizing spectrophotometer manufactured by the Gaertner
Scientific Corp., Chicago.
I n the Konig-Martens spectrophotometer,' t h e Wollaston prism
and biprism are mounted a t the position of the telescope lens of t h e
spectrometer. T h e analyzer prism is mounted in the eyepiece,
behind a slit t h a t gives a photometric field in which each half is
illuminated uniformly throughout with a mixture of colors corresponding to the spectral range under examination.
M a n y commercial instruments, such as those of Bausch & L o m b
and Gaertner (Fig. 14.5), employ an alternative arrangement in
which the photometer (together with accessories for illumination,
holding the cells, and other purposes) is a unit separate from t h e
spectrometer, the latter being of the constant-deviation t y p e . With
this arrangement, the photometer field is focused on the entrance slit
' H. J. McNicholas, Bur. Standards J. Research, 1, 79.S (1928).
§14.12]
ABSORPTION SPECTROPHOTOMETRY
381
of the spectrometer. The field viewed through the eye lens of the
spectrometer consists of two spectra, one immediately above the
other. The spectral range in view at any one time is limited by
laterally adjustable stops in the spectrometer eyepiece.
For transmission measurements, the Bausch & Lomb and Gaertner
instruments use a back-illuminated diffusing screen as a source.
For measuring the transmission of liquids, the Bausch & Lomb
photometer may be obtained in a modified form with cells of continuously adjustable depth similar to those used in Duboscq colorimeters. For reflection measurements, both the Bausch & Lomb and
Gaertner instruments use sphere illuminators.
14.12. Hilger-Nutting Polarizing Spectrophotometer. TheHilgerNutting polarizing spectrophotometer employs a photometer and
constant-deviation spectrometer arranged in tandem as in the Bausch
& Lomb and Gaertner instruments. The original Hilger-Nutting
photometer had one fixed polarizing prism system followed by a
rotatable analyzer. An improved model, described in two papers
by Dowell,^ utilizes a second, fixed analyzer-prism system following
the rotatable analyzer, which results in a more extended, or open,
density scale at the higher density values (§ 14.13), in the elimination
of a certain amount of stray light, and in the emergence from the
photometer of a beam with a constant plane of polarization instead
of one that rotates with changes in the setting of the rotatable
analyzer as in the case of instruments with a single analyzer. The
constant plane of polarization obviates the possibility that changes
in the photometer adjustments might cause changes in the transmission characteristics of the spectrometer as a result of reflections
at optical surfaces.
For transmission measurements, the Hilger-Nutting spectrophotometer ordinarily uses a Pointolite-lamp source, which gives a
comparatively bright photometric field. It may also be fitted with
a sphere illuminator, which gives a somewhat more uniform field but
one of much less brightness, not so well adapted to transmission
measurements at high densities. For reflectance measurements, the
sphere illuminator should be used. An attachment is provided
whereby opaque objects and standard reflecting surfaces may be
illuminated by a 500-watt Pointolite lamp, the angles between the
specimens and the light source on the one hand and between the
' J . H. Dowell, Jour. Sci. Inst, 8, 382-384 (1931); 10, 153-156 (1933).
382
ABSORPTION SPECTROPHOTOMETRY
[§14.13
specimens and the photometer beams on the other being adjustable.
To measure absorption by polished opaque specimens with high surface reflectance, a Pointolite illuminator is provided in which the light
from the source is reflected back and forth several times between two
samples of the specimen before entering the photometer.
14.13. Manipulation of Polarizing Spectrophotometers. For
transmission measurements, the specimen is placed in the path of one
of the photometer beams. If the absorbing specimen is a substance
in solution, a duplicate cell containing the solvent is placed in the path
of the other beam (§ 14.4). For reflectance measurements and for
measurements of the absorption of opaque substances, the light
entering one of the photometer beams is reflected from the specimen
and the light entering the other photometer beam is reflected from a
comparison surface {reference standard) having a nearly uniform
reflection coefficient in the range of the visible spectrum. Magnesium oxide is, aside from its fragility, an excellent material for a
reference standard. Its reflection coefficient is about 0.97 throughout
the visible region, it is a good diffuser, and its reflection characteristics
do not vary appreciably with time. If a less fragile surface is desired,
white glass or porcelain may be used and calibrated with respect t o
MgO. Further details regarding methods of illumination and
manipulation in reflectance measurements are given in the references
at the end of this chapter.
Several commercial photometers are provided with scales from
which opacity, reflectance, and density may be read directly.
To determine a complete spectral curve of transmission or reflectance, readings are taken at intervals throughout the spectrum, the
wavelength being adjusted by means of the spectrometer drum foi;
each new set of readings. The most appropriate wavelength interval
between readings depends on the accuracy with which it is desired to
determine the characteristics of the specimen and the nature of the
spectral transmission or reflectance curve. For specimens with pronounced selective absorption, determinations must be made at more
frequent intervals than for those having general absorption only.
Intervals of 50 to 500 A are commonly uSed. The width of the entrance and exit slits should be adjusted for the maximum purity of
spectrum consistent with sufBcient brightness of fleld for accurate
matching. Measurements should be carried out in a darkened room
after the eyes have become accommodated to the low light level
of the room.
I
§14.14]
A.BSORPTION SPECTROPHOTOMETRY
383
14.14. Other Visual Spectrophotometers arid Their Manipulation.
The Keuffel and Esser color analyzer^ (Fig- 14.6) employs a rotating
sector of fixed angle in the specimen beam and one of variable angle
in the comparison beam, both being mounted on the same shaft. A
motor rotates the sectors at sufficient speed to eliminate noticeable
flicker. The two photometer beams illuminate the entrance slit of
a constant-deviation spectrometer. The division in the photometric
field is produced by a biprism placed between the dispersing prism
Fig. 14.6. Keuffel and Esser color analyzer. (By permission from Measurement of Radiant Energy, edited by W. E. Forsythe. Copyrighted 1937 by
McGraw-Hill Book Co., Inc.)
and the spectrometer telescope lens, resulting in a field similar to that
viewed in the Konig-Martens spectrophotometer (§ 14.11) except
that the dividing line is horizontal instead of vertical. The variable
sector is adjusted by a special mechanical arrangement, while the
sectors»are in motion, until the two halves of the photometric field
match. The value of Ix/Ia or R/R^ corresponding to this adjustment
is read directly from a suitable scale. A sphere illuminator is used
for both transmission and reflectance measurements.
5 C. W. Keuffel, Jour. Opt. Soc. Am. and Rev. Sci. Inst., 11, 403 (1925).
384
ABSORPTION SPECTROPHOTOMETRY
[§14.15
For comparison of the intensities of light sources, it is advantageous
to be able to introduce two widely separated light beams into the
spectrometer. The Lummer-Brodhun" and Guild^ photometers accomplish this by the use of two collimating systems in conjunction
with a suitable optical system for causing the two collimated beams to
illuminate different portions of the final photometric field. The
Buckley and Brookes photometer^ uses a Lummer-Brodhun cube to
bring together two light beams before focusing'them on the slit of a
constant-deviation spectrometer.
Several means of varying the light intensities other than those previously described have been used. Variation of collimator slit
width has been employed with the Lummer-Brodhun instrument.
Adjustment of a rotating sector while in motion can be accomplished
by the use of a cylinder instead of a disk and by cutting away the
cylinder in such a way that moving it along its axis perpendicular to
the light beam changes the effective aperture.' The Guild instrument '' employs a series of interchangeable fixed sectors in combination
with movable neutral absorption wedges placed close to the collimator
slit. In the Buckley and Brookes photometer,* the brightness of the
source is varied by controlling the voltage at which it is operated.
With small concentrated sources, use may be made of the inversesquare law as a means of varying the intensity.* Finally, a variable
aperture may be placed at a suitable point in the optical train at
which there is a uniform parallel beam of radiation (§ 14.22).'
PHOTOGRAPHIC ABSORPTION SPECTROPHOTOMETRY
14.15. General Considerations. Photographic absorption spectrophotometry involves the use of homochroinatic photographic
photometry, which was discussed in detail in Chapter 13 and is considered here only in relation to certain special aspects concerned with
absorption measurement.
/
Photographic materials may be applied to the study of absorption
(a) as qualitative or semiquantitative recorders for indicating the
- O. Lummer and E. Brodhun, Zeitschr. Instrumenienk., 12, 132 (1892).
' J. Guild, Trans. Opt. Soc, 26, 74-94 (1925).
••
.•
8H. Buckley and F. L. C. Brookes, Jour. Sci. Inst. 7, 305-317 (1930).
9 F. L. Dunn, Rev. Sci. Inst., 2, 807-809 (1931).
* The inverse-square law holds strictly only for point sources. As a general working
rule, it may be assumed that deviation from the inverse-square law is less than 0.1 per
cent if the minimum distance from the source is 15 times its diameter.
§14.16]
ABSORPTION SPECTROPHOTOMETRY
385
positions and relative magnitudes of absorption maxima and minima,
(b) as calibrated responders to radiant energy for determining the
relative intensities, /o and 7x, of the radiation incident upon and
transmitted by an absorbing specimen at particular wavelengths, and
(c) as null indicators for determining the wavelengths at which the
spectra of radiation transmitted by an absorbing specimen match
comparison spectra photographed with predetermined reductions in
intensity level or time of exposure. One or the other of these applications is the basis of every method of photographic absorption spectrophotometry.
All the above applications make use of: (a) a suitable light source
(§ 14.6), (b) a means of introducing the specimen into the beam of
radiation between the source and the spectrograph slit (for example,
absorption cells, as discussed in § 14.7, if liquids or solutions are to be
examined), (c) a spectrograph, and (d) photographic plates or films
and appropriate means of processing them. In addition, the second
and third applications require a means of changing the intensity of
the radiation or the time of exposure by known amounts. This
requirement is most often met by a split-beam photometric device,
placed between the source and the spectrograph.
The determination of absorption spectra photographically with a
split-beam photometer is somewhat similar to the techniques described in the section on visual spectrophotometry. In visual
spectrophotometry, however, one normally selects a particular wavelength region and adjusts the photometer until the two halves of the
photometric field are matched in that region. With photographic
methods, it is customary to photograph a series of pairs of spectra,
each with a different setting of the photometer, and then to determine
the wavelengths, if any, at which match points occur for each spectrum pair (see Fig. 14.7). This procedure is an example of the null
•method. The same plate, with its several pairs of spectra, may be
examined also at specific wavelengths, the photometric settings that
correspond most closely to match points being determined for each
wavelength, and the corresponding values of Z^/Zo being estimated by
interpolation. This procedure is an example of the calibration
methodt,
14.16. Advantages and Limitations of the Photographic Method.
Photographic spectrophotometry is not so well adapted to use by
semiskilled technicians as is photoelectric spectrophotometry with
standard commercial instruments. In addition to the manipulation
386
ABSORPTION
[§14.16
SPECTROPHOTOMETRY
of the photometer and spectrograph, photographic processing and
examination and interpretation of the plates are involved. These
iiiiiiiiiiiitiiiiiiiiijittuiau iN^tnii
III mil
iiiiiiiiii
li:ll;iill(MailiMllllllliilllllll 'IIMir
itiiiiiiiitiiiiiiiiiMMiiiii iniiiiinil
I ' i i n n u m M i i m imui
ffiliilll
MUllllMlllUlilllllliiW l i r a ,
HI iii!i uMiinHiiiHi iiuiiiati
,111111 iiiiiitiiiiiiiiii!iiiini iiniinii
11 iiiiiratiiiiiiii I ii
iHiiiiiiniiiiijiiiiniPiH iimm
• .'iifiiiiiMiiiinu lihiiii]
i'li iiiiiiiiriiii
iiitiiniiiijiDjrni in
" 11 iiiima
I iiigiiiiHiiiii
"Uiir
mn
mm
III I
I
• • i i II nil
i m U H I H IIIIHpi: • I I H O l i i III tl< I
iiiiiiiii liiiii I
ii
hiiijii.i iiis III I I,
I'liiiiiiinii
I
iiiiiiiiiriiini
iiiiimiiiii
1IIIIIIINI1II
lillHIi I lli< III i '.
i i r i m i i i f l i i i III ]ii I
'' I'
III II I
n i l ii
NiiiiiiNi^ii
'•fWPIfl «'"f'I I . . •"•""!"*
i
iiii'
Fig. 14.7. Absorption Spectra of human serum photographed with a Spekker
split-beam photometer (§ 14.20). The dots indicate match points between
adjacent spectra (comparison and absorption) as determined visually. The
general shape of the absorption curve (corresponding to the presence of proteins
and nucleic acids in the serum) is indicated by the positions of the dots, but for a
more quantitative presentation these data would be plotted as an absorption
curve.
req\iire considerable skill and care if reliable results are to be obtained, so that commercial photoelectric spectrophotometers will
probably be preferred in many laboratories.
§14.17]
ABSORPTION SPECTROPHOTOMETRY
387
On the other hand, the photographic method ofTers certain advantages. A simple test plate, taken and processed in less time than is
required for a point-by-polnt photoelectric determination of an
absorption curve, often yields information of considerable value.
The fact that the entire spectrum is recorded on the plate often makes
it possible to observe details of absorption that might be overlooked
with point-by-point methods. Furthermore, spectrum details that
are unresolvable with simple photoelectric instruments may sometimes be observed in photographic spectrograms. Because of these
advantages, research laboratories find it useful to have available both
photographic and photoelectric absorption spectrophotometers, using
each for the particular applications to which it is best suited.
14.17. Spectrographs for Photographic Absorption Spectrophotometry. Prisms have been employed much more extensively than
gratings in spectrographs for photographic absorption spectrophotometry, partly as a result of the greater availability of prisms. This
situation may change as improvements take place in the production of
gratings and grating replicas. Gratings have several advantages,
including approximately constant dispersion as a function of wavelength. They also have disadvantages, of which the most important
for many spectrophotometric applications are (a) the astigmatic
images of the slit formed on the spectrum plate, except when the
Wadsworth type of mounting or special methods of illumination are
used, and (b) the necessity for removing overlapping orders, except
in the first order of wavelengths (Chapters 2 and 4).
Prism spectrographs for absorption spectrophotometry usually have
quartz optics, since the photographic method is largely employed for
the ultraviolet region from 2000 to 4000 A. Glass optics are, of
course, better for work in the visible region, where the dispersion of
quartz is low, and many suitable spectrographs with glass optics are
available (Chapters). Constant-deviation wavelength spectrometers
of the type used in certain visual absorption spectrophotometers
(see §§ 14.11 to 14.14) may usually be fitted with a camera in place of
the viewing eyepiece, thus converting them to spectrographs suitable
for photographic spectrophotometry in the visible region. Finally,
some spectrographs may be obtained with interchangeable optics of
various materials, including glass and quartz. Typical commercial
quartz spectrographs for absorption work are listed in Table 14.3.
The choice between a small or a medium quartz spectrograph depends largely on whether measurements are to be made with it only
388
ABSORPTION SPECTROPHOTOMETRY
[§14.17
in the ultraviolet or are to include the visible region. The so-called
"medium" size usually has about three times the theoretical resolving
power and linear dispersion of the "small" size. This higher resolving
power and dispersion are of distinct advantage in the range above
4000 A because of the low dispersion of quartz in this region. In the
ultraviolet, however, the situation is different. With a slit width of
0.02 mm, the effective resolution obtainable with a typical small
TABLE 14.3
SPECIFICATIONS OF TYPICAL COMMERCIAL QUARTZ SPECTROGRAPHS USEFUL IN
ABSORPTION SPECTROPHOTOMETRY
(All dimensions in centimeters)
Manufacturer
and type No.
Prism
Collimator
lens
Telescope
lens
Length of spectrum
2000 to 8000 A
Diam 2.6
19.7
Diam 2.5
fo
16.5
7.3
Small instruments
Adam Hilger
Gaertner
Cornu
1.7 X 2.2
Cornu
2.0 X 3.0
Diam
1.9
20.8
Diam 2.5
ID
18.5
/D
ID
7.2
Medium or intermediate instruments
Adam Hilger
Bausch & Lomb
Gaertner
Cornu
4.1 X 6.5
Cornu
4.0 X 6.5
Cornu
4.0 X 6.5
Diam
5.1
61.0
Diam 5.0
/o
67.5
Diam 5.0
fo
70.0
ID
Diam
fo
Diam
fo
Diam
SD
5.1
61.0
5.0
67.5
5.0
70.0
22.5
24.9
37.5
-
quartz spectrograph is about 1.2 A at 2500 A and 3.0 A at 3000 A;'
these values are entirely- acceptable for routine absorption spectrophotometry, and the dispersion from 1^50 to 3000 A is satisfactory
for most needs; For work in the region below 2300 A, where quartz
absorbs appreciably, small spectrographs are sometimes to be preferred because of shorter path length through their optical components, and the greater feasibility of selecting small quartz specimens
of high transmission in this region. Spectrography to 1850 A may
be accomplished readily with small instruments having specially
selected quartz optics.
The greater dispersion of medium instruments may be useful even
in the ultraviolet when techniques involving direct visual examination
§14.19]
ABSORPTION SPECTROPHOTOMETRY
389
of the plates are employed. Usually one may accomplish this
equally well by enlarging the plates or examining them by projection.
If, however, spectra are to be matched or densitometered by objective
devices such as photoelectric densitometers, which are unable to distinguish between true photographic effects and defects in the plate, a
large spectrum size on the negative is of advantage in reducing errors
that arise from grain size, dust specks, and scratches in the emulsion.
14.18. Photographic Materials and Processing. Photographic
emulsions of moderate speed, high contrast, and small grain size, such
as process plates or film, are best adapted to routine absorption work.
Contrast lanterfl-slide plates are quite satisfactory for use with small
spectrographs. For wavelengths longer than about 5000 A, especially sensitized emulsions must be used. Panchromatic process
plates are entirely suitable. So also are the various fine-grain, highcontrast spectroscopic plates sensitized for this spectral range (see
Chapter 7).
Below about 2300 A in the ultraviolet it becomes necessary either
to use special emulsions in which the ultraviolet absorption due to
gelatin is negligible or to sensitize the plates with a surface coating of
a suitable fluorescent substance. A thin coating of almost any light
machine oil,"^" swabbed on the emulsion with a piece of chamois or a
tuft of lint-free cloth, gives satisfactory sensitization for absorption
work below 2500 A. Before development of the plate, the oil must
be washed off carefully with a suitable solvent, such as benzene. Special ultraviolet-sensitized spectrographic plates may also be obtained
(see Chapter 19).
The precautions outlined in Chapter 7 as to uniformity of conditions of development, fixing, and washing should be observed. In
order to minimize sources of error in density comparisons, it is particularly important to avoid the presence of air bubbles on the emulsion during development and fixing, and to carry out all processing,
including drying, in an atmosphere as free from dust as practicable.
For further details regarding photographic techniques, see Chapter 7.
14.19. Semiquantitative and Plate-calibration Methods. Semiquantitative indications of relative absorption may be obtained by
photographing a series of absorption spectra with different exposure
times and constant intensity, different intensities and constant ex'»G. R. Harrison, Jour. Opt. Soc. Am. and Rev. Sci. Inst., 11, 113, 341 (1925).
390
ABSORPTION SPECTROPHOTOMETRY
[§11.19
posure time, or different thicknesses of the absorbing specimen (using,
for example, a Baly tube, as described in § 14.9). Such groups of
spectra can be made to blacken the plate in such a way t h a t the
outline of the less dense areas traces out an approximate absorption
curve of t h e specimen (Fig. 14.8). M u c h early work in absorption
spectrophotometry was done by such means.
Instead of photographing a series of separate absorption spectra,
one m a y t a k e a single spectrogram through a wedge-shaped or m u l t i step specimen so placed t h a t the light arriving at different portions of
the spectrograph slit (and thence proceeding to different positions on
the photographic plate) traverses different thicknesses of the specimen.''• 12 Alternatively, t h e plate m a y be caused t o move a t varying
Fig. 14.8. Absorption spectra of tyrosine at liquid-hydrogen temperatures,
photographed with different exposure times.
rates of speed, by means of a motor-driven cam, during the period of
exposure, so t h a t t h e time of exposure is a function of the position
of the spectrum on the plate,''' or the plate m a y be caused to move at
constant speed while the intensity of the light incident on t h e specimen
or the thickness of the absorbing specimen is changed.
F r o m these semiquantitative methods, it is b u t a step t o t h e
calibration of the plate (see C h a p t e r 13) and the q u a n t i t a t i v e determination of Ix/IoIf a series of comparison spectra is photographed
a t different intensities or exposure times, the blackening of the plate
" H. S. I hier and R. W. Wood, Atlas of Ah.wrptlnn Sprrlrn.
Washington. D . C :
Carnegie Institute, 1907.
•^ C. E. K. Mees, Allan of Ah.sorption Speclra.
I^ondon; Longmans, Green and Co.,
1909.
" E . R. IloUday, Joiir. Sri. Inst., 1 1 , IfiO (19.S7).
§14.20]
ABSORPTION SPECTROPHOTOMETRY
391
a t particular wavelengths by t h e comparison spectra m a y be compared with the blackening produced by the absorbing spectrum, and
the values of / x / / o niay be determined a t these wavelengths b y
interpolation. T h e density comparison m a y be made with a densitometer (Chapter 13) or by visual means. I n one of the earliest
methods of quantitative absorption spectrophotometry, Henri'* took
alternate spectrograms through equal p a t h lengths of solution and
solvent, varying the time of exposure through the solvent. H e t h e n
determined t h e wavelengths a t which adjacent pairs of spectra
matched each other in blackening and computed the ratios of Ix/Ia
from the relative times of exposure. Corrections were made for
recipTOcity law failure by application of an expression due t o Schwarzschild (Chapter 7).
A simple method of quantitative absorption spectrophotometry,
similar to t h a t of Henri and based on a simplified method of plate
calibration, is the following: A series of exposures of say 10, 20, 50,
and 100 sec duration is photographed through the specimen a n d a
series of comparison exposures is photographed on the same plate
in steps of 1 sec from 1 to 10 sec, inclusive. After development of
the plate, an enlargement is made. T h e absorption spectra are cut
out and brought into juxtaposition with t h e various comparison
spectra to determine approximate match points visually at various
wavelengths. T h e ratios Ix/Its a t each wavelength are determined b y
interpolation between the adjacent comparison spectra most closely
matching t h e absorption spectra, on t h e assumption t h a t t h e reciprocity law holds. Uncertainties regarding the reciprocity law m a y be
eliminated by v a r y i n g the intensity by one of t h e methods of C h a p ter 13.
14.20. Split-beam P h o t o m e t e r s . Split-beam photometers are
devices for splitting a beam of light from the source into two beams,
one of which traverses the specimen and the other a comparison cell
(if one is used), for bringing the two beams into juxtaposition on the
spectrograph slit and for reducing the intensity, or time of exposure,
of the comparison beam by known a m o u n t s . Such devices permit
the simultaneous photography of pairs of absorption and comparison
spectra under circumstances in which the relative exposures of t h e
two spectra m a y be controlled.
For photographic spectrophotometry in t h e spectral regions which
' V. Henri, Phys. Zeitschr., 14, .515-,516 (1913).
Fig. 14.9. Rotating sector photometer manufactured by the Gaertner Scientific
Corp., Chicago.
392
§14.20]
ABSORPTION SPECTROPHOTOMETRY
393
they transmit (roughly 3400 to 10,000 A), combinations of polarizing
split-beam photometers and wavelength spectrometers of the type
described in | § 14.13 and 14.14 may be used. The eyepiece of the
spectrometer is replaced by a camera, and a series of pairs of spectra is
photographed with different photometer settings corresponding to the
values of /^//o for which spectral match points are to be determined.
While polarization photometry is possible down to 2300 A with
appropriate polarizing prisms,*^ modern split-beam photometers for
the ultraviolet employ adjustable rotating sectors or variable apertures rather than polarization optics.
Fig. 14.10. Diagram of Spekker photometer. L, light source; Ki, R^^ /ig, xi4,
reflecting rhombs; Lu L2, Ls, Ln, collimating and focussing lenses; Si, fixed slit;
52, variable slit, adjusted by drum D; Ci, absorption cell; C2, comparison cell;
53, spectrograph slit.
The rotating-sector photometer, first employed by Henri^* and later
developed by Twyman,i* has been used extensively for photographic
spectrophotometry, particularly in combination with quartz spectrographs for studies in the ultraviolet region. A rotating sector of
fixed angular aperture (usually 180 deg) interrupts the light beam
passing through the absorbing specimen, and one of variable angular
aperture (usually 0 to 180 deg) interrupts the comparison beam.
These sectors may be mounted on separate shafts, as in the Hilger
photometer, or may be parts of a single sector disk mounted on one
shaft, as in the Bausch & Lomb and Gaertner photometers (Fig. 14.9).
The Judd-Lewis^^ and Spekker'* photometers (Figs. 14.10, 14.11)
'* E. W. Wood, Physical Optics. New York: The Macmillan Company, 1934.
i«See H. E. Howe, Pki/s. Rev., 8, e,7i (1916).
" S. Judd-Lewis, Trans. Chern. Soc. (London), 115, 312-319 (1919).
18 F. Twyman, Trans. Opt. Soc. (London), 33, 9-19 (1931-32).
394
ABSORPTION
SPECTROPHOTOMETRY
[§14.21
are adjustable-aperture devices t h a t are similar in principle though
somewhat different in optical design. Both control intensity by
varying t h e sizes of apertures in parallel beams of radiation t h a t are
uniform in intensity throughout their cross section. For this purpose, the Judd-Lewis uses mechanical vanes t h a t m a y be adjusted
by rotation so as to pass more or less of the beam, whereas t h e
Spekker uses a rectangular aperture adjustable by a micrometer
screw. T h e Spekker photometer has been employed extensively in
photographic absorption spectrophotometry. I t has a sufficiently
open scale at the higher densities to permit settings up to a density
of 2.0 with errors of probably less t h a n 1 per cent.
Fig. 14.11.
Spekker photometer, manufactured by Adam Hilger, Ltd., London.
(Courtesy Jarrell-Ash Company, Boston.)
T h e instruments described represent the principal types of splitbeam photometers used in pliotographic absorption spectrophotometry. Other methods may, of course, be used both for accomplishing
separation and juxtaposition of the beams and for varying the
intensity or time of exposure of one or both beams.
14.21. Multiple-beam Photometers. Split-beam photometers require a separate exposure for each pair of absorption and comparison
spectra. Several ingenious methods have been devised to permit t h e
§14.21]
393
ABSORPTIO^f SPECTROPHOTOMETRY
photography of several pairs of spectra at the same time, t h u s reducing the total exposure time.
T h e notched-echelon-cell photometer^' (Fig- 14.12) employs a pair
of cells in which the p a t h length through the cell contents varies as a
function of the height in a series of 10 steps. T h e ratio of the length
of each new step to t h a t of the one immediately preceding it is a
constant. T h e edges of the cells are cut at 45 deg to the faces so as
to act as totally reflecting prisms. One edge of the absorption cell
has a series of rectangular notches cut in it, each of which overlaps
Absorption cell
(Notched)
Spectrograph
slit
znmML
Plan
Elevation
^
Light source
Fig. 14.12. Diagram of the notched echelon cell photometer developed by
Adam HJger, Ltd., London.
half of two adjacent steps. When the cells are mounted and illuminated, light from the comparison cell and from the specimen cell
alternately impinges on the spectrograph slit, because of t h e action
of the notches in alternately blocking and transmitting each beam.
As a result, there appears on the photographic plate, after one exposure, a series of 10 pairs of spectra corresponding to the 10 different
p a t h lengths through the specimen and the solvent. An adjustable
rotating sector is used to reduce the intensity of the comparison
'" F. Twyman, Proc. Pkys. Soc. (London), 45, 1-19 (1933).
396
ABSORPTION SPECTROPHOTOMETRY
[§14.22
beams by a known amount. The cells used are difficult to make and
to maintain, and the instrument has not come into wide use.
O'Brien^" made use of a multiple cube to bring into juxtaposition
on a spectrograph slit 10 or 12 alternate beams of light from an
absorption cell and a comparison cell. The comparison beams were
reduced in exposure time by a rotating step sector with steps corresponding to the various beams. A multiple spectrograph slit, consisting of a series of apertures of different widths to provide variation
in light intensity of the comparison beam and a single aperture of the
maximum width for the specimen beam, is used by Kipp and Zonen
in a simple apparatus for photographic absorption spectrophotometry.
Other methods include the use of multiple-aperture diaphragms of
fixed aperture in combination with wedge or step cells, multipleaperture diaphragms of variable aperture, and so on.-'- ^^' ^'
14.22. Considerations Governing Alignment and Illumination.
Critical attention'to optical alignment is important (see Chapter 5).
Some spectrographs and photometers are equipped with optical
benches to simplify alignment procedures. If the fittings that support absorption cells are adjustable, care should be exercised to see
that the cells are centered in, and parallel to, the beams in which
they are placed.
Condensed spark sources, frequently used in photographic spectrophotometry, may deposit evaporated metal on near-by areas.
This debris must be cleaned, at regular intervals, from any optical
surfaces on which it accumulates. Spark and open-arc sources
should be retrimmed and realigned before the electrodes show appreciable wear. If a shutter is used to control exposures, it should be
placed between the source and the specimen so as to minimize the
possibility of photochemical effects on the absorbing material.
The possibility of errors resulting from synchronization between an
intermittent source and a rotating sector may be eliminated by using
a discharge tube or arc source operated on direct current. In a circuit with comparatively high capacitance operated from a 60-cycle
AC transformer, trains of sparks may occur only once in each half
cycle (that is, 120 times per second, corresponding to sector rotations
20 B. O'Brien, Phys. Rev., 37, 471 (1931).
^1 F. Twyman, L. J. Spencer, and A. Harvey, Trans. Opt. Soo. (London), 33, 37
(1932).
22 A. Harvey, Science Progress, 27, 650 (1933).
23 O. E. Miller, Res. Sci. Inst, 3, 30 (1932).
§14.23]
ABSORPTION SPECTROPHOTOMETRY
397
of about 9 to 30 deg). Under such circumstances, the illumination
of the different aspects of the sector throughout a single revolution
may be quite different. An averaging out to yield approximately
uniform integrated illumination of the sector openings during the
exposure may be assumed only if the sector rotation is not synchronized with the alternations of the AC supply and if many revolutions
of the sector take place during an exposure. Similar considerations
apply if an AC-operated discharge tube or arc is used (or any other
source operated from alternating current or low-frequency interrupted direct current).
Dirt on the spectrograph slit is especially troublesome in methods
of photographic photometry in which spectrograms arising from
different portions of the slit are to be compared. Lack of parallelism
of the slit jaws may also introduce errors in such instances.
14.23. Choice of Density in Specimens; Determination of
Match Points. The greatest accuracy in matching or measuring the
plates is attainable if the specimen has comparatively high density
values. On the other hand, the accuracy of some photometers (for
example, sector photometers) is somewhat decreased at the highest
density values for which they are calibrated. Good results are
generally obtained with specimens that have a maximum density,
within the spectral range to be measured, of about 1.5 for sector
photometry and of about 1.8 to 1.9 for photometry with the Spekker
instrument. Samples of higher density may be measured with the
notched echelon cell, the simple method described in § 14.19, and
certain other photometric methods.
The appropriate concentration and cell length to yield the optimum
maximum density may be computed in advance for solutions of substances whose concentration and absorption characteristics are
known. If either the concentration or the absorption characteristics
of the solute are unknown, one or more test plates must be taken to
determine the concentration and cell length to use. After some
experience, a rapid visual examination of the plate will suffice to
indicate, very closely, what changes in concentration or cell length are
required. Alternatively, test plates may be taken with several
different cell lengths.
The final plate in split-beam photometry Is usually taken with a
considerable range of density settings, at density intervals of about
0.1. For greatest precision in density determinations, intervals of
0.05 may be used; closer settings are of little assistance.
398
ABSORPTION SPECTROPHOTOMETRY
[§14.24.
The determination of the positions of match points on adjacent
pairs of absorption and comparison spectra may be accomplished by
observing the plate directly or under a magnifying glass, the match
points being marked with ink spots. This procedure is tedious and
difficult for many persons, even when appropriate viewing stands and
illumination systems are provided. Some workers prefer to use paper
positives, at 4 X to 5 X enlargement, for the visual determination of
match points. Others prefer to project the plates upon screens, at
many times enlargement, determining the match points on the projected images. If the plates are projected through tracing cloth or
paper, the match points may be spotted on the reverse side, so as to
yield a permanent record.
Better accuracies in visual matching may sometimes be attained by
determining, through interpolation, the densities at particular wavelengths that correspond to match values, rather than by determining
the wavelengths that correspond to match points for each particular
pair of spectra. Photoelectric densitometry may also be eniployed
(see Chapter 13).
14.24. Precision of Determination of Wavelengths and Densities.
Spectrographs for photographic absorption spectrophotometry are
usually equipped with built-in transparencies by means of which
scales of wavelengths (or frequencies) may be printed directly on the
plate while it is in the spectrograph. Ordinarily, a scale is printed
at the top of the plate, and again at the bottom, and the wavelengths
of positions on the intervening spectra are determined by placing a
straight-edge across the spectra and corresponding marks on the two
scales. This method involves errors of the positioning of the individual spectra on the plate, as well as, errors in the scale and its positioning, but it is sufficiently accurate for much absorption work. I t is
well to check the positioning of a built-in scale by comparing it with
the known wavelengths of a line source, siich as a mercury arc, before
relying on it for wavelength determiniition. When the greatest
wavelength accuracy is desired, a line source may be employed and
wavelengths may be determined from the lines and from interpolation
between them. If a continuous source is used, a line source may be
photographed on the same plate for wavelength calibration!
When match points are determined visually, the precision attainable is primarily a function of the ability of the eye to distinguish
between different brightness levels. The precision of concentration
determinations is about ± 1 per cent at a density of 1.0 or ±0.5 per
§14.25]
ABSORPTION SPECTROPHOTOMETRY
399
cent at a density of 2.0, if a method is used in which densities are
determined at particular wavelengths. In general, an error of
about ± 1 per cent may be expected if high density values are used.
Determination of match points by photoelectric densitometers may
decrease the detectable density increment, AZ), by a factor of 10
or more.^* With such an increase in the precision of match-point
determination, other errors, such as those of photometer calibration
and adjustment, usually become limiting.
PHOTOELECTRIC ABSORPTION
SPECTROPHOTOMETRY
14.25. General Considerations. Photoelectric and thermoelectric photometry are discussed in detail in Chapter 12. Photoelectric
photometry is used extensively in absorption spectrophotometry of
the ultraviolet and visible; thermoelectric photometry, in the infrared.
Photoelectric absorption spectrophotometry usually makes use of
photocells as radiation receivers either to measure the intensity of
radiation in specimen and comparison beams alternately at a succession of wavelengths, or to determine when an equality of intensity has
been established, at each of a series of wavelengths, between specimen
and comparison beams in optical null methods of photometry. In
either of these methods, two important problems arise: that of
achieving sufficient freedom from scattered light in the dispersing
system so that the photocells (which usually have .markedly different
sensitivities at different wavelengths) are not unduly influenced by
scattered radiation; and that of providing a sufficiently low threshold
of response in the photocell-amplifier-indicator system (in terms of
radiant power required to produce a signal equal to the background
noise level) to permit measurements to be made with a high degree
of spectral purity at low light levels.
Scattered light is a particularly troublesome source of error when
measurements are made in a spectral region for which the photocell
sensitivity is low and in the presence of scattered radiation for which
its sensitivity is high. Insofar as practicable, therefore, photocells
should be chosen to have at least as high sensitivity in the range to be
measured as outside this range. To approximate this condition, it is
possible to use two or more photocells of different characteristics to
2* F. Twyman and G. F . Lothian, Proc. Phys. Soc. (London), 4S, 643 (1933).
400
ABSORPTION SPECTROPHOTOMETRY
[§14.26
cover the spectral range 10,000 to 2000 A (see Chapter 12) and to
supplement them with filters to modify their effective response characteristics within certain regions. Below 2500 A the situation is
particularly difficult, since both photocell response and available
light intensity are usually low in this range.
Errors resulting from scattered light may be reduced by using a
suitably designed double monochromator (Chapters 3 and 4).
Photoelectric absorption spectrophotometry requires a light source,
a dispersing system, a specimen holder, and a photocell and accessory
electrical components. The possibility of making measurements with
narrow bands of comparatively high spectral purity is facilitated by
use of sources of high steradiancy, dispersing elements of large area
and high angular dispersion, and sensitive photocell-amplifier systems
with low noise level. Choice of an appropriate light source was discussed in § 14.6. The dispersing system is usually a monochromator;
various suitable monochromators are described in the following
discussions of individual types of spectrophotometers as well as at
the ends of Chapters 3 and 4. Except in certain optical null methods
and automatic recording instruments, the specimen holder is usually
a movable mounting by means of which absorption and comparison
cells can be shifted alternately into the light beam between the photocell and the monochromator exit slit. Photocells, amplifiers, and
measuring instruments are discussed in Chapter 12.
14.26. Point-by-Point Instruments for Relative Intensity Measurements. With instruments of this type, separate determinations
of the relative intensities of the specimen and comparison beams are
made at each wavelength for which measurements of /j//o or
logio (lo/Ix) are desired.
The Beckman spectrophotometer (Fig. 14.13) is an example of a
commercial instrument in this category designed to cover the range
10,000 to 2000 A. It employs a quartz prism of the Littrow type
with a concave mirror of 50 cm focal length for collimation. The
relative aperture of the system is about / / l l . Two light sources are
used: a 32-candle-power, 6-volt automobile headlight bulb, for
measurements from 10,000 to 3200 A, and a hydrogen discharge tube,
for extension of the short-wave limit to about 2200 A. The wavelength scale is calibrated to 2000 A, but the practical working limit
of the instrument with present light sources is somewhat less when
substances having considerable absorption in this region are being
measured. Filters are used to reduce stray light in various spectral
§14.26]
ABSORPTION SPECTHOPHOTOMETRY
401
regions. I t is claimed t h a t stray light can be kept below 0.2 per cent
throughout most of t h e 10,000 to 2000 A range. Two photocells are
employed, one for t h e region above 6250 A and one for t h a t below.
T h e photocell circuit consists of a two-stage direct-coupled amplifier.
An o u t p u t meter indicates when the potential developed by t h e
phototube current has been balanced by an opposing potential from
a slide-wire potentiometer in the input circuit. T h e potentiometer is
calibrated in per cent transmission from 0 t o 110 a n d in densities
from 0 to 2.0.
Fig. 14.13. Beckman photoelectric quartz spectrophotometer, manufactured
by National Technical Laboratories, South Pasadena, Calif. This instrument is
also obtainable with glass optics.
T o operate t h e instrument, the wavelength at which transmission
is to be determined is first selected by means of a calibrated dial t h a t
controls t h e angle of t h e prism relative t o t h e coUlmated beam. W i t h
the slide wire adjusted to correspond to 100 per cent and the beam
passing through t h e comparison cell, t h e slit widths a n d the sensitivity
are then adjusted until a null reading is obtained on t h e o u t p u t
meter. Next, t h e cell containing the absorbing specimen is shifted
into the b e a m a n d t h e potentiometer is adjusted until a null reading
is again obtained. T h e per cent transmission, or density, at t h e
wavelength in question may then be read directly from the slide-wire
402
ABSORPTION SPECTROPHOTOMETRY
[§14.26
scale. In such a reading, it is assumed, of course, that the photocell
current is a linear function of the intensity of radiation incident on it
and that the slide wire is uniform. The accuracy with which I^ can
be determined is about ± 1 per cent of Zo or better throughout most
of the spectrum, provided sufficiently large slit widths are used. To
achieve high accuracy in transmission measurements, it is often
necessary to use slit widths corresponding to spectral band widths of
about 50 to 100 A at half-maximum intensity. In order to measure
samples that transmit less than 10 per cent {D greater than 1.0), the
sensitivity may be increased tenfold to permit use of the full potentiometer scale for transmission measurements from 0.1 to 10 per cent
{D from 3.0 to 1.0).
The Coleman double-monochromator spectrophotometer employs
two transmission gratings in tandem as the dispersing system and a
storage-battery-operated, 32-watt incandescent lamp as the light
source, to cover the spectral range 10,000 to 3500 A. The use of
two gratings reduces the stray light to less than 1 per cent on the
average. The potential developed by the photocell current in an
adjustable decade resistor is balanced against an opposing potential
from a potentiometer. Null settings are determined with the aid of
an electronic amplifier as in the case of the Beckman instrument.
Adjustment (at any particular wavelength) for 100 per cent reading
on the slide wire, with the comparison beam incident on the photocell,
is accomplished by means of the decade resistor. The slits are of
fixed width, corresponding to spectral band widths of 300, 150, 75,
or 50 A. The accuracy of determination of /^ is ± 1 per cent of /o or
better, depending on the slit widths used, except at the extremes of
the spectral range. The potentiometer and electronic amplifier are
external to the monochromator. I t is possible to use a pH meter
to supply these components.
The Coleman single monochromator spectrophotometer (Fig. 14.14)
is of somewhat simpler design, employing a single transmission
grating as the dispersing component and a barrier-layer photocell,
connected to a sensitive galvanometer, as the radiation detector. A
wider band width must be used than in the case of the double monochromator under similar conditions, but the instrument is considerably simpler and more compact. The photocell response may be
read directly on the galvanometer or determined by means of a
built-in potentiometric circuit.
r
§14.26]
ABSORPTION SPECTROPHOTOMETRY
403
The Cenco Spectrophotelometer (Fig. 14.15) makes use of a concave
replica grating in a modified Eagle mounting as the dispersing system,
an incandescent lamp as the light source, and a barrierjayer photocell
and galvanometer as the radiation-sensitive receiver and indicating
system. Fixed exit slits of 200, 100, or 50 A band width are used.
I
Fig. 14.14. Photoelectric spectrophotometer for the region 3500 to 10,000 A.
Manufactured by Coleman Instrum'ents, Inc., Maywood, 111.
The value of /i/'/o at a particular wavelength is determined from
successive readings of the galvanometer with the specimen and comparison beams incident on the photocell. The spectral range covered
is 10,000 to 8500 A.
The commercial instruments described are typical of those used
routinely in chemical, biochemical, and biological laboratories. In
404
ABSORPTION SPECTROPHOTOMETRY
[§14.26
order t o obtain sufficient transmitted light for accurate transmission
measurements, they all require entrance and exit slit widths corresponding t o a lower order of magnitude of spectral p u r i t y (50 t o 300 A
band width) t h a n is usually achieved in photographic spectrophotometry (1 t o 5 A band width). Although t h e use of a wide band eliminates t h e possibility of observing fine details of absorption or of
obtaining t r u e measurements of substances having very narrow a b sorption bands, the convenience, moderate cost, a n d ease of operation
Fig. 14.15. Spectrophotelometer employing a concave replica grating in a
modified Eagle mount for photoelectric absorption spectrophotometry in the
range 3,500 to 10,000 A. Manufactured by Central Scientific Company, Chicago.
of such instruments m a k e t h e m extremely useful in routine analytical
work.
M a n y point-by-point photoelectric spectrophotometers h a v e been
designed a n d used by individual research workers. T h e G e r m a n
investigators (Pohl, K u h n , Smakula, and others) were pioneers in t h e
application of this method of spectrophotometry. A n u m b e r of
interesting designs of instruments, each with certain novel features.
§14.27]
ABSORPTION SPECTROPHOTOMETRY
405
have been described in recent years.^^''^ The type of spectrophotometer developed by Hogness and his associates,'* which is capable of
measuring I^ with an accuracy of 0^2 per cent of Jo with slit widths
limiting the transmission band to 10 A or less, is an example of what
can be done by careful attention to details when low cost, compactness, and simplicity are not essential.
14.27. Photoelectric Null Methods: Nonrecording. Photoelectric cells may be employed in null methods of spectrophotometry to
determine when two beams of radiation or two halves of a photometric
field are matched, using any of the usual optical methods of varying
the intensity or intermittent exposure of one or both beams (polarizing
prisms, rotating sectors, diaphragms). Several such arrangements
have been suggested and used by individual workers.'^' ^' The
optical null method has the advantage of involving no assumptions as
to the linearity of photocells, amplifiers, and indicating instruments.
It has the disadvantage of being limited in its precision of determining
IX/IQ by the accuracy of the optical device used to balance the two
beams.
The Hilger photoelectric spectrophotometer (Fig. 14.16) employs
two gas-filled photocells connected in opposition to a Lindemann
electrometer. A collimated beam of light from the monochromator
is partially reflected into one photocell by a quartz plate placed at an
angle to the beam. The remainder of the light passes through a
rotating sector of the cylindrical type (which may be adjusted while
in motion) to the second photocell. The value of logio (lo/Ix) at a
particular wavelength is determined from the two sector settings required to achieve a null reading of the electrometer with, say, a
^ D. H. Follett, Proc. Phys. Soo. (London), 46, 490 (1934).
2" S. Jacobsen, H. E. Bent, and A. J. Harrison, Rev. Sci. Inst., 11, 220 (1940).
" F. P. Zscheile, Jr., Jour. Phys. Chem., 28, 95 (1934).
28 E. S. Miller, Plant Physiology, 12, 667 (1937).
29 W. C. Bosch and K. D. Coleman, Phys. Rev., 57, 941 (1940).
'" D. L. Drabkin, Proc. Seventh Summer Conference on Spectroscopy, 1939. New
York: John Wiley & Sons, Inc., 1940.
SI W. C. Bosch'and B. B. Brown, Jour. Opt. Soc. Am., 29, 466 (1939).
32 M. Barnard and P. McMichael, Jour. Opt. Soc. Am., 21, 588 (1931).
33 S. Schlaer, Jour. Opt. Soc. Am., 28, 18 (1938).
3«T. R. Hogness, F. P. Zscheile, Jr., and A. E. Sidwell, Jour. Phys. Chem., 41, 371
(1937).
^ W. Deck, Helvetica Phys. Acta, 11, 3 (1938).
*> H. von Halban and H. Giegel, Zeitschr. phys. Chem., 96, 214 (1920).
" H . von Halban and K. Seidentopf, Zeitschr. phys. Chem., 100, 208 (1922).
406
ABSORPTION SPECTROPHOTOMETRY
[§14.28
solution cell and a solvent cell in the beam. The monochromator
employed is of large aperture (approximately//4 t o / / 6 , depending on
the wavelength) and of high dispersion (equivalent to a train of four
60-deg prisms), and has interchangeable optics of quartz, glass, and
rock salt for work in different spectral regions. Because of the large
monochromator aperture and the sensitivity of the photocell-electrometer circuit, comparatively narrow spectral band widths may be
used. The precision of density determinations (limited by the
accuracy of the sector) is said to be about 0.004 for densities near 0.43By using suitable photocells for different regions, a spectral range of
12,000 to 1850 A can be covered.
P.
=4^
Fig. 14.16. Optical system of null type of photoelectric photometer developed
by Adam Hilger, Ltd., London. S, slit; Li and L2, collimating and focusing
lenses; Ci and d, absorption and comparison cells; Q, quartz plate; Pi and P2,
photocells; S, rotating sector; M, motor.
14.28. Automatic Recording Photoelectric Spectrophotometers.
The earliest successful automatic recording spectrophotometer is that
of Hardy,^^"^' of which a commercial model is manufactured by the
General Electric Company (Fig. 14.17). It is designed for transmission or reflectance measurements in the visible region. It employs a
method developed from that used by Dobson and Prefect,*^ in which
the absorption (or reflection) and comparison beams are allowed to
illuminate a single photocell alternately in such a manner that a
fluctuating current is set up in the photocell circuit unless the two
beams are of equal intensity. Rotating mechanical shutters were
«8 A. C. Hardy, Jour. Opt. Soc. Am., 25, 305 (1935).
' '
2' A. C. Hardy, Jour. Opt. Soc. Am., 28, 360 (1938).
« J. L. Michaelson, Jour. Opt. Soc. Am., 28, 365 (1938).
*i K. S. Gibson and H. J. Keegan, Jour. Opt. Soc. Am., 28, 372 (1938).
^G. M. B. Dobson and D. S. Prefect, Photoelectric Cells and Their Applications.
Pages 79, 174, 185. London; The Physical Society, 1930.
§14.28]
ABSORPTION SPECTROPHOTOMETRY
407
408
ABSORPTION SPECTROPHOTOMETRY
[§14.28
used to alternate the beams in early instruments employing this principle. In order to obtain smoother transition from one beam to the
other, Hardy's system makes use of a rotating Rochon prism to
alternate the beams, these having been polarized perpendicularly to
each other by prior passage through a fixed Wollaston prism. When
the beams are unbalanced, the alternating current generated in the
photocell circuit is applied, through a suitable amplifier, to a control
circuit that causes a motor to rotate a second Rochon prism in the
beams until they are balanced. The direction of rotation of the lightbalancing motor depends on the phase of the alternations of the
unbalanced beams and is always such as to compensate the unbalance.
The recording pen is coupled to this balancing mechanism.
The principle of this optical null method is the same as that of the
Kbnig-Martens type of photometer (§ 14.11), except that balancing
is achieved and recorded automatically instead of manually. A
double-prism glass monochromator is used. The spectrum band is
shifted continuously by a motor-driven cam mechanism while the
wavelength axis of the record chart is correspondingly shifted with
respect to the recording pen. A curve of absorption or transmission
throughout the visible range is plotted automatically in from 2 to
5 min. The slit widths of the commercial instrument normally correspond to a spectral band width of 100 A, but special instruments
may be obtained for operation at 40 A band width. The precision of
determination of Ix is considerably better than ± 1 per cent of /oErode and Jones^' have described methods for adapting the Hardy
spectrophotometer to measurements in the ultraviolet, using a quartz
prism in a Wadsworth mounting as the dispersing system. They
have employed both rotating aluminized reflection sectors and
rotating polarizing prisms for chopping the beams. In the former
case, adjustment of relative intensities was accomplished by a vane
photometer; in the latter case, by polarizing prisms.
The automatic recording spectrophotometer designed by Harrison**
and improved by Harrison and Bentley*^ (Pig- 14.18) will measure, at
high speed, transmissions or densities throughout the range, 9700 to
2300 A with spectral band widths of 1 A or less throughout most of
this range. Thus an effective resolution is attained which compares
« W. R. Erode and C. H. Jones, Jour. Opt. Soc. Am., 31, 743 (1941).
" G. R. Harrison, Proc. Sixth Summer Conf. on Spectroscopy, 1938, page 91. New
York; John Wiley & Sons, Inc., 1939.
« G. R. Harrison and E. P. Bentley, Jour. Opt. Soc. Am., 30, 290 (1940).
§14.28]
ABSORPTION SPECTROPHOTOMETRY
409
favorably with that achievable under optimum conditions in photographic spectrophotometry. The precision of measurement of Z^,
about ± 1 per cent of 7o, is limited, in all but the extreme ends of the
spectral range, by the precision of the mechanical and optical components of the photometer rather than by the sensitivity of the
photocell-amplifier system.
The 3-meter concave grating employed (which is ruled, 15,000 lines
to the inch, on an aluminized glass blank) gives a plate factor of 5.5 A
per millimeter, almost linear with respect to wavelength. The en-
14.18. Harrison automatic recording photoelectric spectrometer for the
visible and ultraviolet regions.
trance and exit slits are adjustable in width from 0 to 2 mm. A
110-volt coil-filament incandescent lamp is used from 9700 to 3400
A and a high-pressure mercury arc (G.E. H-6) from 3400 to 2300 A.
The radiation receiver is an 11-stage photomultiplier tube, connected
to a "memory unit" incorporating a 0.01 mf condenser, through an
appropriate amplifier.
In operation, the monochromatic light beam from the exit slit is
passed alternately through the absorption and comparison cells by
an oscillating biprism, while a sector rotates in front of the entrance
\
410
ABSORPTION SPECTROPHOTOMETRY
[§14.29
slit in synchronism with the rotation of the drum on which the
recording chart is wrapped. With the specimen beam incident on
the photocell, a potential corresponding to the beam intensity is
developed across the condenser in the memory circuit. This potential is stored while the comparison beam is shifted to the photocell and
the variable-width portion of the sector is rotated before the entrance
slit. As the intensity of the comparison beam increases, the potential
developed in the photocell circuit increases also. When it equals the
stored potential (that is, when the intensity of the comparison beam
has been matched to that of the specimen beam), a thyratron triggering circuit causes a point to be marked on the recording chart at the
appropriate density or transmission value. A separate motor drive
automatically shifts the wavelength of the transmitted band, and the
corresponding position of recording on the chart, at a maximum speed
of 100 A per second.
The system is inherently low in scattered radiation. To decrease
further the effects of stray light, filters are interposed automatically
in the incident beam when measurements are being recorded in the
various wavelength regions. The optical null method used involves
a time delay, less than 0.05 sec in normal operation, between evaluation of the intensities of the specimen and comparison beams.
H. Cary^^ has developed an automatic recording photoelectric
spectrophotometer for the visible and ultraviolet regions which is
manufactured by the Applied Physics Corporation (Fig. 14.19). It
employs a double-prism monocliTomator to obtain high dispersion
and freedom from scattered radiation. The light beam is chopped
at 90 cps, and the radiation receiver is a photomultiplier tube. The
output of the photomultiplier is fed through an electronic amplifier
to a strip-chart recorder of the type used with infrared spectrophotometers. Except in the extreme ultraviolet, half-intensity band widths
as small as 1 A may be used with this instrument.
These examples indicate the trend in instrument design, which is
toward both automatic recording and improved effective spectral
resolution.
14.29. Abbreviated Absorption Spectrophotometry and Fluorimetry. In many routine analytical chemical procedures, it is desirable
to be able to determine the concentration of a single substance in a
solution by means of such optical properties as color or fluorescence.
'Jnd. Eng. Chen., 39, 75A (1947).
§14.29]
ABSORPTION SPECTROPHOTOMETRY
411
Various photoelectric instruments employing principles of spectrophotometry in simplified form are available for such measurements.
In the case of the determination of concentration by application of
the property of color, use is made either of the natural color of the
substance itself or, if the substance is colorless, it is treated with
appropriate reagents with which it forms a color complex. Trans-
h,,
•
Fig. 14.19. Gary automatic recording photoelectric spectrophotometer for the
visible and ultraviolet regions. Manufactured by Applied Physics Corp., Pasadena, Calif.
mission measurements are then made at or near the wavelength of
maximum absorption, and the concentration is determined by application of the Lambert-Beer law or by use of a calibration curve
made by determining the transmission values of a series of dilutions of
a solution of known concentration.
412
ABSORPTION SPECTROPHOTOMETRY
[§14.29
Though it is entirely feasible to employ spectrophotometers of the
types previously described for quantitative analyses by this method,
it is often customary to use simpler devices, variously known as
photoelectric colorimeters, abbreviated absorption spectrophotometers, or
absorptiometers for this purpose. These employ optical filters
(§ 14.30) or simple dispersion systems for isolating the spectral band
with which transmission measurements are made. In general, the
band width employed is large (from 200 to 400 A or even more);
hence it is preferable to base the determination upon a calibration
curve rather than to assume that the Lambert-Beer law applies.
Fig. 14.20.
Typical pliotoelectric absorptiometer (colorimeter) of tlie filter type.
Manufactured by E. Leitz, Inc., New York.
Many designs of absorptiometers are available (Fig. 14.20). Further information regarding the technique of photoelectric absorptiometry will be found in General Reference 14.3.
The concentration of a substance that fluoresces in solution when
excited by ultraviolet radiation may be determined by photoelectric
measurement of the intensity of the fluorescent light."" In comparatively dilute solutions, the concentration is approximately pro" See J. R. Loofbourow and R. S. Harris, Cereal Chem., 19, 151 (Wii) for a discussion of typical commercial fluorimeters.
§14.29]
ABSORPTION SPECTROPHOTOMETRY
413
portional to the intensity of fluorescence for constant intensity of
exciting radiation:
c = AI/
(14.13)
wiiere c is the concentration, in suitable units, ^ is a constant, and
If is the intensity of fluorescence. If relative values of 7/ are measured with, for example, a barrier-layer photocell and galvanometer,
and if the photocell galvanometer response is proportional to 7/, then
[provided Eq. (14.13) is valid]:
c= {R-
Ri) ^-IZL^
(14.14)
. ri — ra
where c is the concentration in an unknown sarnple, R and Rh are,
respectively, the galvanometer readings with the sample and solvent
alone in the cell, and ri and r2 are calibration readings of the galvanometer corresponding to known concentrations Ci and Cj. In
order for Eqs. (14.13) and (14.14) to be approximately valid, it is
necessary that the solution concentration be such, that only about
10 per cent of the exciting radiation is absorbed in the solution.
Excitation of fluorescence in fluorimeters is usually produced by
radiation in a spectral band including the 3650 A mercury line.
High-pressure mercury arcs of the type described in Chapter 8 are
convenient sources of such radiation, and glass optics transmit it
freely. Quartz-mercury arcs and quartz optics are used in certain
instruments to extend the range of the exciting radiation below
3400 A. It is necessary to eliminate visible light as effectively as
possible from the exciting radiation, since the fluorescence to be
measured is primarily in the visible range. This reduction is accomplished by suitable primary filters (§ 14.30). In general, the more
effective the primary filter is in eliminating visible radiation, the less
effective it is in transmitting exciting radiation.
It is also necessary to eliminate response of the photocell to the
exciting radiation insofar as practicable, by the use of secondary
filters (§ 14.30) and usually by placing the photocell at right angles to
the exciting beam. Secondary filters for absorbing stray exciting
radiation may convert an appreciable fraction of the absorbed radiation into fluorescent light in the visible range. I t is therefore usual
to follow the ultraviolet-absorbing filter by one or more additional
filters to absorb the fluorescence excited in the first filter. Even with
the best of primary and secondary filter combinations, stray visible
414
[§14.30
ABSORPTION SPECTROPHOTOMETRY
and ultraviolet light usually fails on the photocells in sufficient quantities to cause appreciable readings with a "blank" (cell containing
solvent only) in the beam. Hence a correction must be made for the
reading with the blank [Ri in Eq. (14.14)].
14.30. Optical Filters. As indicated in § 14.29, optical filters are
frequently used in absorptiometers and fluorimeters for isolating desired spectral regions. They are also used extensively to obtain
approximately monochromatic radiation for photochemical or photobiological investigations, and to reduce the effects of stray radiation or
undesired spectral orders in dispersing systems.
The three principal types of optical filters depend for their operation, respectively, on selective absorption, interference, and selective
scattering. Of these, selectively absorbing filters are most widely
used.
Selectively absorbing filters may be solids, liquids, or gases, the
first being the most convenient to use. Many types of colored glasses.
200
240
280
320
360
400
440
480
520
560
600
640
Wavelength, Millimicrons
Fig. 14.21. Transmission curves of typical glass filters. Manufactured by the
Corning Glass Company, Corning, N. Y.
having different selective-absorption characteristics, are available for
use as optical filters in the visible region, and to a somewhat lesser
extent such filters are also applicable to the isolation of spectral
regions in the ultraviolet and near infrared. Transmission curves of
a few typical glass filters are illustrated in'Fig. 14.21. Glass filters
manufactured by the Jena Optical Works, Corning Glass Company,
Eastman Kodak Company, and others are available in wide variety
and detailed data on their transmission characteristics will be foimd
in the catalogues of their respective manufacturers. Solid filters may
also be made by imbibing absorbing materials in gelatin or adsorbing
§14.30]
415
ABSORPTION SPECTROPHOTOMETRY
them on cellophane. These are often mounted between protecting
covers of glass. Such filters are obtainable from distributors of
photographic and stage-lighting equipment.
Liquid filters (including solutions) provide a greater variety of
transmission characteristics than do selectively absorbing glasses.
They must, of course, be used in suitable cells with windows transparent to the radiation to be transmitted. Gaseous filters are occasionally employed to isolate spectral bands in the ultraviolet but are
Transparent
dielectric
30
n
25
•\
320
Cover
plate
H- 10
f^
Reflecting
metallic films
/
/
w)iV
v
400 450 500 550 600 650
(Q)
Wavelength, Millimicrons
(b)
Fig. 14.22. Interference filters, (a) Method of construction,
transmission curves.
(b) Typical
seldom used in the visible region. Again, they require the use of
suitable filter cells. Transmission data for liquids, solutions, and
gases suitable for use as filters will be found in compilations of absorption spectra, including General References 14.7 and 14.8.
Interference filters depend on the same principle as the Fabry
and Perot etalon, described in Chapter 20. They consist of two
partially reflecting evaporated metal films, separated by a transparent
spacer of evaporated dielectric, together with glass protecting covers
(Fig. 14.22a). For a given optical path length through the dielectric,
the interference arising from multiple reflections between the metallic
films is such as to result in maximum transmission (about 25 to
35 per cent) at particular wavelengths, on each side of which the
transmission falls rapidly to nearly zero (Fig. 14.22b). Interference
filters are available commercially for maximum transmission at any
wavelengths within the range 3700 to 7000 A. Those with transmission maxima near the ends of this range have two transmission
bands within this region, the unwanted one of which is removed by
416
ABSORPTION SPECTROPHOTOMETRY
[§14.30
the use of auxiliary glass.filters. Interference filters must be used
with a collimated beam of radiation. T h e wavelength position of t h e
transmission band m a y be shifted over a considerable range b y
varying t h e angle of incidence of t h e collimated b e a m on the filter.
Christiansen filters, based on selective scattering, have long been
used for isolating comparatively narrow spectral regions in t h e infrared, visible, a n d near ultraviolet (see General Reference 9.1), a n d their
use has been extended recently t o t h e ultraviolet region from 8000
to 2000 A.^* I n the Christiansen filter, small irregular chips of a solid
30
20
2500
LiX
2600
2700
2800
X in A
Fig. 14.23. Transmission curve of a Christiansen filter*' for the ultraviolet,
consisting of quartz chips suspended in a mixture of decahydronapthalene and
cyclohexane.
are suspended in a medium of different dispersion b u t with a refractive
index t h e same as t h a t of the chips a t t h e wavelength for which'
maximum, transmission is desired. This combination results in
minimum scattering, and hence maximum transmission, a t t h e
specified wavelength. T h e transmission decreases rapidly uj)on,
d e p a r t u r e from t h e optimum wavelength as ,a result of the increasing
difference in t h e refractive indices of t h e two media and increased
scattering of radiation. Christiansen filtqi-s must be used in a, collimated light beam, and the position of their transmission m a x i m u m
is sensitive t o temperature, especially if the suspending medium' is a
liquid, A typical transmission curve of such a filter for the, u l t r a violet region is shown in Fig. 14.23.
" R. L. Sinsheimer and J. R. Loofbourow, Nature, 160, 674 (1947).
'
§14.32]
ABSORPTION SPECTROPHOTOMETRY
417
ABSORPTION OF SUBSTANCES IN SOLUTION
14.31. General Remarks Regarding Solvents. The properties
that must be considered in choosing a solvent are the solubihty in it.
of the material to be investigated, its transmissivity for radiation in
the region to be explored, and its volatility.
It is desirable that the material to be investigated be soluble easily
in the solvent up to the highest concentration required for spectra.
Otherwise, errors may arise from the failure of samples to go entirely
into solution. If the maximum extinction coefficient of the substance is known, the required concentration for a given density at
the maximum may be computed from the Lambert-Beer law [Eq.
(14.6)]. Often this concentration is exceedingly small; hence it
sometimes happens that substances listed in the usual handbooks as
insoluble in particular solvents may be dissolved in them in sufficient concentrations for absorption spectrophotometry.
The solvents most commonly employed in absorption spectrophotometry are distilled water, methyl alcohol, ethyl alcohol, chloroform, hexane, cyclohexane, carbon tetrachloride, and ethyl ether.
Unless specially purified, many of these solvents contain sufficient
concentrations of such impurities as ketones, aldehydes, and benzene
to cause appreciable absorption in the ultraviolet. The extent to
which they must be purified for practical use depends on the spectral
region of interest. In general, the greatest care in solvent purification
is required for work in the region below 2400 A, where various factors
previously discussed combine to make measurements difficult.
Highly volatile solvents, such as ethyl ether, introduce possibilities
of error as a result of concentration changes brought about by
evaporation. Therefore such solvents should be used only if other
solvents will not serve, and then only with special precautions against
evaporation.
14.32. Absorption by Solvents; Purification. As indicated in
Table 14.4, many commercial solvents of the best grades are satisfactory for measurements to 2450 A. For work at shorter wavelengths than those listed as the useful limit, only especially purified
solvents should be used. Some of the solvents listed in the table and
purified for chemical work show absorption obviously attributable to
traces of impurities. This fact is indicative of the different requirements as to purity for laboratory reagents and for solvents for absorption spectroscopy. Merely because a reagent is ideal for chemical
418
ABSORPTION SPECTROPHOTOMETRY
[§14.32
purposes does not necessarily mean, therefore, that the particular
impurities which are most troublesome in spectroscopy have been
removed to the extent that might be desired.
TABLE 14.4
APPBOXIMATE WAVELENGTH LIMIT OF USEFULNESS OF VARIOUS
COMMERCIAL SOLVENTS IN 2-CM CELLS*
Solvent
Distilled water (from laboratory supply)
Ethyl ether
Hexane, practical, Eastman Kodak Co
Methyl alcohol
Ethyl alcohol, 95%
Ethyl alcohol, absolute, U.S.P
Hexane, synthetic, Eastman Kodak Co
Chloroform
Cyclohexane, pure, Eastman Kodak Co
Carbon tetrachloride
Cyclohexane, practicaJ, Eastman Kodak Co
Shortest useful
wavelength
2000 A
2250
2250
2300
2350
2350
2450
2450
2650
2700
2750
Laboratory-distilled water redistilled twice in all-glass stills is
usually sufficiently free from absorbing substances for work to 1850 A.
The usual methods for preparing conductivity water may be employed if the greatest purity is desired.
Ethyl alcohol (95 per cent) may be purified by adding 25 cc of
12 N sulfuric acid per liter, refluxing for several hours, and distilling.*^
The distillate is then treated with 20 grams of potassium hydroxide
and 10 grams of silver nitrate per liter, refluxed, and distilled. This
method yields a solvent sufficiently good for work to 2000 A. If this
product is dried over aluminum amalgam, the transmission at shorter
wavelengths is somewhat improved!
Ethyl ether distilled once from alkali has sufficiently good transmission characteristics for most work in which this solvent is required.
An outline of useful methods for purifying cyclohexane, hexane,
carbon tetrachloride, chloroform, ethyl alcohol, and methyl alcohol
is given in General Reference 14.2.
'
* These data, which are based on experience of the'authors and on the assumption
that the solvent should transmit at least 50 per cent of the radiation at useful wavelengths, agree approximately with those in General Reference 14.1, in which data on a
number of additional commercial solvents will be found.
" P. A. Leighton, R. W. Gary, and L. T. Schipp, Jour. Am. Chem. Soc, S3, 3017
(1931).
§14.33] .
ABSORPTION SPECTROPHOTOMETRY
419
14.33. Relation of Absorption to Chemical Constitution. As was
pointed out in Chapter 11, the electronic spectra of soHds, liquids, and
substances in solution as ordinarily observed by absorption spectrophotometry in the 10,000 to 2000 A region do not show the fine
structure observed in the absorption spectra of substances in the
gaseous state. Rotational transition lines are not resolved at all,
and the vibrational band structure can only be identified clearly in
special instances. The lack of detail in the spectroscopic data,
together with the complexity of the substances usually studied by
absorption spectrophotometry, has, in general, made it impracticable to work out relationships of structure to absorption in a
manner analogous to analyses of the absorption spectra of simple
molecules in the gaseous state. Some progress in this direction has
been made, however,^" and advances may be expected to be more
rapid as techniques of measurement and analysis improve.
Prom the empirical point of view, a number of generalizations
regarding absorbing groups (designated as "chromophores";
see
§ 11.3) and their relation to absorption have been worked out, especially insofar asdyestuffs are concerned ^^'^^ (General Reference 14.5).
Certain of these generalizations that are particularly useful to the
practical worker in applying absorption spectrophotometry to
problems of organic chemistry are summarized below:
1. To a close approximation, electronic absorption spectra associated with unsaturated linkages occur in the region X > 2000 A,
whereas those associated with saturated linkages occur in the region
X < 2000 A.
" For a summary to about 1940, see G. E. K. Branch and M. Calvin, The Theory of
Organic Chemistry, Chapter V. New York: Prentice-Hall, Inc., 1941.'
*i R. Nietzki, Verhandl. des Vereins zum Beforderung des Gemeibejleisses, 58, 231
(1879).
^ J. B. Cohen, Organic Chemistry for Advanced Students. Part II, Chapter II. New
York: Longmans Green & Co., Inc., 1939.
^ T . Forster, Zeitschr. f. Electrochemie, 45, 548 (1939).
" H. Niimiya, Chem. Rev. Japan, 3, 2240 (1937).
^ B. Beilenson, N. I. Fisher, and F. M. Hamer, Proc. Roy. Soc. (London), A163,
138 (1937).
^'F. Pruckner and A. Stern, Zeitschr. f. phys. Chemie, A180, 25 (1937).
" G. N. Lewis and M. Calvin, Chem. Rev., 25, 273 (1939).
** G. Scheibe and W. Friimel, Hand- und Jahrbuch der chem. Physik, 1936, Vol. 9,
page 142.
420
ABSORPTION SPECTROPHOTOMETRY
[§14.33
2. The principal groupings contributing to absorption are the
following :^»-58 > C = C < , > C = 0 , > C = S , > C = N — — N = N — ,
—N=0,>S=0,—N^
— N ^ = N — , > C = C = 0 , and conjugated
chains or rings made up of > C = C < , > C = N — , or — N = N —
groups (compare Table 11.3).
3. The absorption maximum associated with a particular group is,
in general, shifted toward longer wavelengths (lower frequencies)
when a substituent is linked to the group; this effect is increased
with increasing atomic weight of the substituents.^'' ^*
4. Separation of unsaturated groups by two or more single-bond
linkages usually reduces the influence of such groups on each other
to such an extent that the total absorption may be considered as the
sum of the absorption bands attributable to each of the structures so
isolated.69' eo
5. The mutual influence of unsaturated groups on absorption is
considerable in structures involving conjugated double bonds.
Marked absorption is found in structures with several conjugated
double bonds in which there are many possibilities for resonance
(as in benzene, for example). In general, increase in the number
of conjugated double bonds results in an increase in absorption and
a shifting of the absorption bands toward longer wavelengths (lower
frequencies) .*'"**
6. The influence of substituents, especially those of higher atothic
weight on an absorbing group or a conjugated double bond structure,
is usually to increase the absorption.^^ *^"^
7. Changes in the substituents of, or linkages to, absorbing groups
as a result of salt f ormation,*^'^' association,'^' '^ and so on, may cause'
^' V. Henri, Etudes de Photocfiemie, Paris, 1919.
"»W. Aumiiller, H. Fromherz, and C O. Stro'ther, Zeitschr. f. phys. Chemie, B37,
30 (1937).
" L. B. Arnold, Jr. and G. B. Kistiakowsky, Jour. Anf. Chem. Soc, 54, 1713 (19,32).
^^ K. W. Hausser, R. Kuhn, A. Smakula, and K. H. Krenchen, Zeitschr. f. phy-s.
Chemie, B29, 371 (1935).
'
'
«' K. W. Hausser, R. Kuhn, and G. Seitz, Zeitschr. f. -phys. Chemie, B29, 391 (1935).
" R. Kuhn and A. Deutsche, Ber. 65, 43 (1932).
^ R. Kuhn and M. Hoffer, Ber. 65, 651 (1932).
I* K. V. Auwers and R. Hugel, Zeitschr.}. phys. Chemie, A178, 315 (1937). ,'
" R. K. Callow, Biochem. Jour., 30, 906 (1936).
«8E. C. C. Baly and E. K. Eubank, Jour. Chem. Soc. (London), 87, 134,7 (1905).
" «' H. Ley, Zeitschr. f. phys. Chemie, 94, 405 (1920).
§14.34]
ABSORPTION
SPECTROPHOTOMETRY
421
changes in absorption. Salt formation may arise as a result of the
presence of acids or bases in solution, as for example when these are
used to adjust pH. Association may be influenced by the nature
of the solvent and the concentration of the solution.
8. Changes in the kinds or positions of unsaturated groups in
molecules as a result of molecular rearrangements may cause marked
changes in absorption.'"'^^ When such effects arise from changes in
32,000
30,000
28,000 ^
26,000 .1
24,000 ^
22,000 1
20,000 ^
18,000 .9
16,000 c
14,000 S
12,000 ..
10,000 3
8,000 ^
6,000
i
4,000
2,000
or .< , -,
V^:JO
2200 2300 2400 2500260027DO 28002900
Wavelength, A
Fig. 14.24.
Effect of pH of the solution on the absorption spectrum of barbituric
acid."
the pH of the solution, they are easily brought about and are useful
in the identification of substances (Fig. 14.24). The colorimetric
determination of pH is based on this phenomenon.
14.34. Photochemical Effects; Fluorescence. The radiation used
to determine the absorption of substances may give rise to photo'" W. L. LewschiD, Ada Physicochimka U.R.S.S., 1, 685 (1935).
'1 G. Scheibe, Kolloid Zeitschr., 82, 1 (1938).
'2 W. Stenstrom and M. Reinhard, J. Phys. Chem., 29, 1477 (1925).
" C. S. Hicks, Jour. Chem. Soc. (London), 128, 643 (1926).
'* E. R. Holiday, Biochem. Jour., 24, 619 (1930).
'*F. F. Heyrothand J. R. Loofbourow, Jour. Am. Chem. Soc., S3, 3441 (1931);
56, 1728 (1934).
'« J. R. Loofbourow and M. M. Stimson, Jour. Chem. Soc. (London), 844 (1940);
1275 (1940).
422
ABSORPTION SPECTROPHOTOMETRY
[§14.34
chemical effects that have a marked influence on the absorption. If
such changes take place to an appreciable extent while the absorption
spectrum is being determined, they may lead to errors in the determinations. There is more likelihood of errors arising from this source
in ultraviolet spectrophotometry than in visible spectrophotometry;
even though some substances, such as riboflavin, are photochemically labile in the visible region, instances are quite rare in
which photochemical changes suflicient to influence the validity of
measurements are.encountered in absorption spectrophotometry in
this region.
Those methods of absorption spectrophotometry which involve the
least total irradiation of the specimen with photochemically active
radiation are freest from the possibility of error as a result of photochemical change. At first sight, it appears that photoelectric
methods, in which the specimen cell follows the exit slit of a monochromator, are superior in this regard to photographic methods, in
which the cell is placed before the entrance slit of a spectrograph.
This superiority is not necessarily actual, however, if a point-by-point
method is used, involving many determinations not carried out at
high speed.
In most instances the amount of radiant energy required to cause
appreciable change in absorption is considerably greater than that to
which the specimen is exposed during measurements with any of the
usual spectrophotometric methods. Whether errors caused by
photochemical effects are being encountered may be checked by
successive absorption determinations on the same specimen.
Many substances that absorb in the ultraviolet exhibit appreciable
fluorescence when exposed to such radiation. When the absorption
spectra of such substances are determined by photographic methods, ,
some of the fluorescent radiation enters the slit of the spectrograph
and causes blackening of the photographic plate in the spectral
regions to which the fluorescence corresponds. As a result, absorption spectra determined photographically 'sometimes show what
appears to be negative absorption at wavelengths longer than that of
the absorbing region. Since the fluorescent Jight is radiated in all
directions, this effect is unlikely to be noticed unless the specimen cell
is placed close to the slit.
,'
Data regarding the' absorption of particular substances are given
in the General References,
§14.34]
ABSORPTION SPECTROPHOTOMETRY
423
GENERAL REFERENCES
Methods and Theory
14.1.
14.2.
14.3.
14.4.
14.5.
W.R.Brode, Chemical Spectroscopy, id ed. New York: John Wiley
& Sons, Inc., 1943.
P. Twyman and C. B. AUsopp, The Practice of Absorption Spectrophotometry, 2d ed. London: Adam Hilger, Ltd., 1934.
Thomas R. P. Gibb, Jr., Optical Methods of Chemical Analysis. New
York: McGraw-Hill Book Company, Inc., 1941.
W. West in Physical Methods of Organic Chemistry, Vol. 2, Ed. by A.
Weissberger. New York: Interscience Publishers, Inc., 1946.
E. A. Braude,^Mn. Rep. Chem. Soc. (London), 42, 105 (1945).
Compilations of Data
14.6.
R. A. Morton, TJie Applications of Absorption Spectra to the Study of
Vitamins, Hormones, and Coenzymes. London: Adam Hilger,
Ltd., 1942.
14.7. E. P., Carr, M. L. Sherrill, and V. Henri, in International Critical
Tables, Vol. 5. New York: McGraw-Hill Book Company, Inc.,
1929; W. C. Holmes, idem., Vol. 7.
14.8.
V. Henri, in Tables Annuelles de Constantes et Donnees Numeriques.
New York: McGraw-Hill Book Company, Inc., (1910-1936).
14.9. F. EUinger, Tabulae Biologicae, 12, 291 (1937); 16, 265 (1938),
Den Haag: Junk.
14.10. E. S. Miller, Quantitative Biological Spectroscopy. Minneapolis: Burgess Publishing Companj', 1934.
14.11. W. R. Erode, "The Absorption Spectra of Vitamins, Hormones and
Enzymes," Advances in Enzymology, 4, 269 (1944).
14.12. John R. Loofbourow, "Physical Methods for the Identification and
Assay of Vitamins and Hormones," Vitamins and Hormones, 1,
109 (1943).
CHAPTER
15
Qualitative Spectrographic Analysis of Materials
To
DETERMINE WHICH METALLIC ELEMENTS ABE PRESENT i n
a
given sample of material, analysis by the emission spectrum gives
a method that is usually more direct, more rapid,'more specific, more
complete, and probably easier to use than any other yet developed.
It does not require the analyst to guess in advance which elements
are likely to be present, to select methods of procedure that may or
may not turn out to be justified, or to separate the constituent materials into chemical groups. All elements that are readily detectable
spectrographically can be found in a single operation.
Qualitative spectroscopic analysis as discussed in the present chapter refers to analysis by means of the emission spectrum. Absorption
spectrophotometry (Chapter 14) can be used as a method of analysis
for certain molecules, atoms, and ions, but it is somewhat less specific,
though often more sensitive, than the emission method. Both methods should be considered before any given problem of analysis is
undertaken. Emission analysis for qualitative purposes is one of the
most widely used fields of spectroscopy, yet its development has been
neglected somewhat in recent years.' This neglect is due partly to
the great amount of attention that has been given to quantitative
analysis, a more difficult field, but it also arises from the fact that
qualitative emission analysis is relatively so satisfactory that little
attention has been given to improving it.
.
,
Qualitative analysis with the spectrograph is a relatively simple
process. A small sample of the material to be analyzed is placed' in
an electric arc, sparky or other suitable source of excitation in sucli a
way that the molecules of the sample will be dissociated into their
constituent atoms, which are then stimulated to emit light. This
light is sent through the spectrograph, which separates the various
wavelengths and records these individually as spectrum lines on a
photographic plate. Each chemical element emits a well-known
424
§15.1]
QUALITATIVE ANALYSIS OF MATERIALS
425
group of lines whose wavelengths are recorded in printed tables of
the types listed in Tables 9.7 and 15.2 and given in Appendices I and
II. Lines in the spectrum can be positively identified, when their
positions have been determined, as having come from a specific
element. Numerous lines are emitted by each kind of atom, but at
most two lines are needed to give positive identification.
Spectrographic analysis gives a permanent record so that results
are readily available for future reference. This feature also makes it
possible for unskilled personnel to carry out routine parts of an
analysis and to prepare spectrograms from which an expert spectrographer can later make an accurate analysis. The method is
especially valuable in cases in which the operator does not know what
elements to look for. It is also much more specific than most chemical wet methods, for so definite are the wavelengths emitted by the
atoms that misidentification of elements is almost impossible.
The spectrograph can be used for qualitative analysis of substances
difficult to handle by wet methods, such as glasses and slags. It can
also be used to detect in alloys minute impurities that are sufficient
in amount to affect crystal structure but are in concentrations too low
to be found by wet methods. It can be used to differentiate between
two chemically similar substances. For instance, neodymium and
praesodymium are so much alike as to defy chemical separation, but
they resemble each other as little spectroscopically as silver resembles
calcium. The spectrographic method can be used to resolve doubts
remaining from wet analyses, as in the case where the brown coloration of a solution produced by hydrogen sulfide may be open to
various interpretations. I t can also be used effectively to follow the
course of a chemical reaction or separation.
The spectrographic method is especially suited for use with samples
of which only small amounts are available. In some cases 0.1 mg of
sample is sufficient for a complete qualitative analysis, although
10 mg is desirable if available. The sensitivity of the method varies
from element to element as discussed below, but amounts of certain
elements as small as 10~* gram can be detected, in concentrations of
less than 1 part in 100 million.
The spectrographic method cannot be used satisfactorily to determine negative elements or ions such as CI"" and SO J that may be
present in a sample, since these are not stimulated to emit light in
ordinary sources unless the radicals contain metallic atoms. Since
most molecules are broken down in the source, chemical combinations
426
QUALITATIVE ANALYSIS OF MATERIALS
[§15.1
cannot usually be determined by emission analysis. Speetrographic
analysis of nonmetallie elements such as gases require special techniques which are discussed in § 15.7.
The visual spectroscope can be used for certain types of qualitative
analysis in the manner originated by Kirchhoff and B.unsen/ but it
serves merely as a simple approximation of the general spectroscopic
method and is applicable only in special cases^
15.1. Sensitivity of Detection of Various Elements. It is commonly stated that only some 70 of the known chemical elements can
be detected spectrographically. Actually, any type of atom can be
detected and identified through the radiation it emits under proper
excitation, and therefore any atom can be detected spectrographically
with proper techniques. In general, however, speetrographic methods for nonmetals, particuarly for the halogens and gases, are as yet
more complicated and difficult than are the equivalent chemical wet
methods. Therefore speetrographic analysis is ordinarily restricted
to the metallic elements and some of the metalloids. The 20 elements
commonly not considered susceptible to such analysis are the permanent gases, the halogens, sulfur and selenium, and a few of the rare
heavy metals.
If we examine in detail the list of "nonspectroscoplc" elements, it is
easy to see why these are difficult to detect: their atoms either are
relatively difficult to excite to emit radiation, especially radiation in a
readily accessible spectral region; or they emit so many spectrum
lines that none are especially intense. These difficulties can sometimes be overcome by methods discussed in § 15.7.
Neutral atoms have ionization potentials lying between the 3.1 volts
of cesium and the 25.4 volts of helium, as discussed in Chapter 10.i
The metals almost all have ionization and excitation potentials lying,
between 5 and 10 volts, whereas those of the gases and halogens lie
above this value (see Table 10.1, Column I).,
It is desirable to distinguish between the concentrational sensitivity
and the absolute sensitivity of detection by spectrum lines. Figure 15.1 shows the intensities of two spectrum lines plotted as fiinctions of the percentage concentration of the element producing them
in a sample, or matrix, which is otherwise kept uniform. It will be
noted that over the range plotted these lines have d.ifferent slopes and
therefore produce different concentrational sensitivities. Also, the
' G. R. Kirchhoff and R. Bunsen, Ann. d. Physik, 110, 160, (1860); Phil. Mag., 20,
89 (1860); ibid., 22, 329 (1861); Ann. Chim. Physique, 62, 452 (1861).
§15.1]
QUALITATIVE ANALYSIS OF MATERIALS
427
curves intercept t h e axis of minimum detectable intensity a t different
points, so t h a t t h e y produce different absolute sensitivities. I n
making q u a n t i t a t i v e analyses, concentrational sensitivity is of great
importance, whereas in qualitative analyses, where detection of t h e
minimum possible concentration is desired, t h e absolute sensitivity is
of importance.
T h e absolute sensitivity of analysis for an element depends on
which spectrum line is selected for its detection. I n t h e following
discussion, it is assumed t h a t the most sensitive lines of each element
are used.
Intensity
of
Spectral
Lines
Absolute
sensitivities
%
Line B
Concentration
Fig. 15.1. Line intensity curves as a function of percentage composition,
showing the distinction between concentrational sensitivity and absolute sensitivity of detection.
Table 15.1 gives a list of the present approximate limiting concentrations t h a t have been reached for the various elements. Boron, for
example, has been determined spectrographically t o below 1 part in
10^, whereas sulfur has been detected only t o one p a r t in 10^. These
factors are, of course, only very approximate a n d can be expected t o
change as further progress is made in developing special methods for
improving t h e sensitivity of analysis for a n y given element. Absolute
sensitivity can be increased b y various methods discussed in § 15.3,
which have been tried with some elements a n d not others. Chemical
preconcentration methods can be used also, reducing t h e limiting
concentration factor b y as much as 10* in some cases. I n theory,
there should be no limit to absolute sensitivity, since a single atom,
if kept in t h e source for a sufficiently long time, could be excited over
TABLE 15.1
APPROXIMATE SENSITIVITY LIMITS OF DIRECT SPECTHOGRAPHIC DETECTION
OF THE E L E M E N T S *
Element
Least detected
ppm by weight
Ag
Al
As
Au
0.5
1
100
10
B
Ba
Be
Bi
0.1
1
2
10
Element
Least detected
ppm by weight
Mg
Mn
Mo
1
10
10
Na
Nb
Nd
Ni
0.1
80
90
5
Os
•
C
Ca
Cb
Cd
Ce
CI
Co
Cr
Cs
Cu
0.5
80
0.2
90
10
40
0.2
5
0.5
Dy
90
Er
Eu
10
10
.
F
Fe
1
0.1
Ga
Gd
Ge
4
90
7
Hg
Ho
50 90
In
IT
K
5
100
0.1
100
P
Pb
Pd
Pr
Pt
10
0.2
10
90
50
Ra
Kb
Rh
Ru
70
10
10
S
Sb
Sc
Se
Si
Sm
Sn
Sr
10
20
1
1
80
300
2
0.5 .
Ta
Tb
Te
- TiTl
V
/
50
90
500
50
10
0.3
10
W
/
La
Li
10
0.5
Y
Yb
4
3
Zn
Zr
2
20
* Many elements listed as having been detected to lower limits than others are less
readily detectable but have been more intensively studied.
428
§15.2]
QUALITATIVE ANALYSIS OF MATERIALS
429
and over again to emit light until enough was accumulated to give a
detectable spectrum line. In practice, however, most of the atoms
wander from the excitation stream before they are excited even once.
The limit on absolute sensitivity is set by the background intensity,
which overwhelms the weak light produced by the small number of
atoms which are excited. Any method that will reduce the relative
background intensity can be expected to increase absolute sensitivity.
The minimum amount of material that can be detected spectrographically is known for very few elements. For materials not diflBcult to detect, it appears to lie between 10~* and 10~^ gram. Greatly
increased absolute sensitivity of detection usually results whenever
sources of excitation are improved specifically with this sensitivity
in mind. The main objective is to prevent the atoms of interest from
wandering from the excitation stream or from combining into molecules that do not readily dissociate, and to keep other atoms and
molecules from emitting radiation.
The statement is sometimes made that 10'* gram of lead is the
least that can be detected spectrographically. This figure corresponds to 10^' lead atoms. An average spectrograph will project
onto the plate only about 1 quantum of radiation of a given wavelength out of every 10,000 such quanta emitted by the source. Simple
calculations based on these facts show that only 1 atom out of each
200 put into the source emits even one quantum of the line 4580 A
commonly used for lead detection before it is lost from the stream of
excitation.
15.2. Sensitive Lines and Ultimate Lines. Since some spectrum
lines are stronger than others because of a high probability of occurrence of the appropriate transition in the atom, and since some are
more easily excited than others, it is not surprising that certain lines
should be found more sensitive for detecting small quantities of
material than others. Usually, though not always, the strongest lines
of a spectrum are the most sensitive. De Gramont,^ Hartley,' and
others have made careful empirical studies to determine the sensitive
lines of each element. In particular, de Gramont has published
tables of so-called raies ultimes, or ultimate lines (General Reference
15.5). The raie ultime is the last line of an element to disappear as the
2 A. de Gramont, Compt. Rend., 144, 1101 (1907); ibid., 145, 1170 (1907); Rev. Met.,
19, 90 (1922).
*W. N. Hartley, Jour. Chem. Soc. (London), 41, 90 (1882); Phil. Trans., 175, 50
(1884).
[§15.2
QUALITATIVE ANALYSIS OF MATERIALS
430
<
s
L
•1
-*
-1
<
<
3
I
<
a
i
o
o
<
• < *
J
>
o
o
o
f.
ry
1
;i —
D
O
n
>o
-
Mi
•^
D
e
z
?
r
_
o
if
Q.
*
<
^^
a
o
o
o
5.
3
5
(J
a
O
o
lO
X
8
•n
CO
E
H
d
o
_
m
tTn^
O
>
>:
a
u
0
8
y
e
y
-
o
1o
in
fo
i
,
8
^^-
N
y
o
O
(0
/I
n
is
0
w
55
3
"CL.—0 .
B!
< t .
>-
c
C
M
^CO .M
t ~
3
,
1
>
in
^ f (1)
•p 0
z
J.
u
u
<u
•53
<o
8
a>
B
^
in
•g
•0
z
d
«
8
1
V
D
o
o
-
•n
r~-
„
z
z
-
1
•
5
<u
r/>
dS
w
(>-.
^n
o
o
o ,
a>
f»
bjn
<
:r
f
A,
«,
ja
X)
8
(
u
X
c
o
i
<
s. _
8
in
crt
a
o
o
o
n
-
u
6
O
O
o
.•ea
§15.2]
QUALITATIVE ANALYSIS OF MATERIALS
431
q u a n t i t y of the element b u r n e d in a sample is decreased t o the vanishing point. D e G r a m o n t ' s tables are to some extent out of date, since
the ultimate line of an element depends strongly on t h e spectroscopic
techniques used. Several lines m a y t e n d t o disappear together as
the limit of sensitivity is reached. These usually are lines belonging
to the same multiplet, which differ only slightly in n a t u r a l intensity
(that is, by not more t h a n a factor of 2 or 3). T o be sure of the
presence of a specific element in a given sample, it is in a n y case
desirable to identify a t least two lines originating from it, so several
lines are ordinarily listed as " u l t i m a t e . "
Tables of sensitive lines are very useful, since if the most sensitive
lines of a n element do not appear in a spectrum being studied for
qualitative analysis, t h e element can be marked absent. This
absence should be understood as being qualitative, however, and
merely means t h a t within th6 limits of possible detection the element
was not found.
Appendices 1 and 2 contain lists of sensitive lines
of t h e elements. Figure 15.2 shows a chart, due t o Hasler and
Kemp,^ of the distribution of sensitive lines according t o wavelength.
Comprehensive tables of spectral wavelengths are listed in Table 9.7,
a n d abbreviated tables useful for routine spectrochemical analysis are
listed in Table 15.2.,
TABLE 15.2
ABBREVIATED TABLES FOB SPECTBOCHEMICAL ANALYSIS*
1. H. Kayser, "Emission Spectra of Elementary Substances," International Critical Tables, Vol. V (New York: McGraw-Hill, 1929), page 276.
Contains about 12,000 lines in the range 350 to 31,000 A, in arc and spark,
for 86 elements, with precision of from 1, to O.OOl A, and intensity range
1 to lb. Contains also table of air lines.
2. H. Kayser and R. Ritschl, Tabelle der Hauptlinien der Linienspektren
alter Elemente, 2d ed. (Berlin: Springer, 19S9). Contains about 27,000 lines
in the range 90,850 to 33 A, in arc, spark, and discharge, for 88 elements,
given to 0.1 and 0.01 A, with intensity range 0 to 1000.
3. Handbook of Chemistry and Physics. Cleveland: Chemical Rubber
Publishing Co.
4. "W. H. Brode, Chemical Spectroscopy (New York: Wiley, 1939). Contains a number of tables: (1) persistent lines by wavelength, listing 400 lines
in the range 7950 to 1850 A, given to 0.1 A; (2) persistent lines by elements,
with 400 lines in the range 9000 to 1600 A, given to 0.1 A for 71 elements;
(3) principal lines by wavelength, containing about 4300 lines for the more
* M. F. Hasler and J. W. Kemp, Jour. Opt. Soc. Am., 34, 21 (1944).
* Reprinted by permission from G. R. Harrison, Jour. App. Phys., 10, 76 (1936).
432
QUALITATIVE ANALYSIS OF MATERIALS
[§15.2
TABLE 15.2 (Continued)
common elements, with the gases and some other nonmetals omitted, in
wavelength range 8000 to 2000 A, given to 0.5 A, with intensity range 2 to 10;
(4) principal lines by elements, containing about 6000 lines in the range 8000
to 2000 A, in arc and spark, given to 0.5 A, with intensity range 1 to 10;
(5) spark spectrum of air, containing 300 lines in the range 7952 to 2288,
given to 0.1 A, with intensity range 1 to 10; (6) principal lines by elements,
discharge spectrum, containing about 300 lines in the range 8000 to 2500 A,
given to 0.1 A, with intensity range 1 to 10.
5. D. M. Smith, Visual Lines for Spectrum Analysis (London: Hilger,
1928).
6. A. Gatterer and J. Junkes, Atlas der Restlinien (Castel Gandolfo, Italy:
Specola Vaticana, 1937). Contains about 2400 lines in the range 8100 to
2200 A, in arc and spark, for 30 elements, given to 0.01 A; also some band
spectra.
7. W. Gerlach and E. Riedl, Die chemische Emissions-Spektralanalpse,
Part I I I (Leipzig: Voss, 1936). Contains about 3000 lines in the range 5900
to 2140 A, in arc and spark, for 57 elements, given to 0.1 A. Gives control
lines and possible interfering lines.
8. J. A. Hannum, "Wavelength of the Principal Lines in the Emission
Spectra of the 'Elements," Handbook of Chemistry, N. A. Lange, ed. (Sandusky,
Ohio: Handbook Publishers, Inc., 1937), page 863. Contains about 5600
lines in the range 90,850 to 124 A for 83 elements, with accuracy to 1 A.
Sensitive and ultimate lines are indicated.
9. G. R. Harrison, M.I.T. Wavelength Tables (New York: Wiley, 1939).
Table of 500 sensitive lines of the elements according to elements and again in
wavelength order, in range 9237 to 2025 A, in arc, spark, and discharge, for
85 elements, mostly given to 0.001 A, with intensity range 1 to 9000. (Reproduced in part in Appendices 1 and 2.)
10. W. J. Crook, Metallurgical Spectrum Analysis (Stanford University
Press, 1935). Contains 9540 lines in the range 5800 to 2780 A, in the arc, for
64 elements, given to 0.1 A, with intensity range 1 to 10.
11. W. J. Crook, Table of Arc Spectrum Lines Arranged in Order of Wavelengths (Stanford University Press, 1933).
12. J. M. Eder and E. Valenta, Atlas typischer Spektren (Vienna: Holder,
1911). Contains about 35,000 lines in the range 7950 to 1850 A, in arc,
spark, and flame, for 75 elements, given to 0.01 A, with intensity range 1 to
100. Contains also some band heads. Note Rowland scale.
13. F. Exner and E. Hascheck, Die Spektren,'der Elementen bei normalem
Druck, Vol. I (Leipzig and Vienna: Deuticke, 1911). Contains about
20,000 lines in several tables (not mutually exclusive), in the range 7070.to
2135 A, in arc and spark for 77 elements, given to 0.01 A, with intensity/range
1 to 1000. Note Rowland scale.
14. F. Twyman and D. M. Smith, Wavelength Tables for Spectrum Analysis
(London: Hilger, 1931). Contains a number of tables: (1) Sensitivedines in
spark spectra of solutions (Hartley; PoUok; PoUok and Leonard; Leonard
and Whelan). Contains about 1300 lines, arranged by elements, in the range
6725 to 2200 A, for 42 elements, given to 0,1 A, with intensity range 1 to 10.
§15.2]
QUALITATIVE ANALYSIS OF MATERIALS
433
TABLE 15.2 {Concluded)
(2) Sensitive lines in spark spectra of solids and fused substances (De Gramont). Contains 330 lines, arranged both by elements and by wavelengths,
in the range 7948 to 2138 A, for 83 elements, with accuracy to 0.1 A. Most
sensitive lines indicated. (3) Sensitive arc lines of 50 elements (in R.U.
powder) arranged by elements. Contains 367 lines in the range 6717 to
2288 A, given to 0.01 A, with intensity range 1 to 10. Most sensitive lines
are indicated. (4) Sensitive lines in flame spectra arranged both by elements
and by wavelengths. Contains 98 lines in the range 7699 to 2484 A, for
'io elements, given to 0.1 A. (5) Sensitive lines in arc and spark spectra,
arranged by elements. Contains about 1000 lines in the range 7000 to
2000 A, for 50 elements, given to 0.01 A, with intensity range 1 to 10.
15. F. Lowe, Atlas der letzen Linien der wichtigsten Elementen'{J^eipz\g and
Dresden: Steinkopf, 1928). Table of persistent spark lines of the more important elements.
16. G. Scheibe, Physikalische Methoden der analytischen Chemie, Part I,
W. Bottger, ed. (Akademische Verlagsgesellschaft, Leipzig, 1933), p. 67.
Contains about 1100 lines in the range 7950 to 1862 A, in arc and spark, for
62 elements, given to 0.01 A, with intensity range 0 to 40. Includes some
air lines.
17. J. C. Boyce and J. M. Maclnnes, Wavelengths of the Extreme Ultraviolet
Lines from Gas Discharges (Princeton, N. J.: Palmer Physical Laboratory,
1930). Contains about 2300 lines in the range 2500 to 115 A, in discharge
tube, for 10 elements, given to 0.01 and 0.001 A, with intensity range 000 to
10. Contains air lines and some bands.
18. J. C. Boyce and H. A. Robinson, "Wavelength Identification Lists
for the Extreme Ultraviolet,'-' Jour. Opt. Soc. Am., 26, 133 (1936). Contains
about 2600 lines in the range 2000 to 38 A, in discharge, for 14 elements,
given to 0.001 A, with intensity range 00 to 50.
Meggers a n d Scribner^ have formulated a rule for predicting theoretically the ultimate lines of a n y element: "A raie nltime in a n y
spectrum originates with a simple interchange of a single electron
between s a n d p states, usually preferring configurations in which only
one electron occurs in such a s t a t e . " This rule usually gives the
strongest line characteristic of the spectrum, even though it m a y not
involve the normal state of the a t o m as a lower level, b u t does not
necessarily give t h e raies uliimes.
Excitation a n d other factors can
greatly affect the result.
I n some elements those lines which would be expected t o be most
sensitive, or ultimate, lie in relatively inaccessible spectrum regions.
This is t h e case with rubidium and cesium, for example, which from
a n excitation standpoint should be the most readily detectable of all
elements. Their theoretically ultimate lines lie in t h e infrared, b u t
" W. F. Meggers and B. F. Scribner, Jour. Res. Nat. Bur. Standards, 13, 657 (1934).
434
QUALITATIVE ANALYSIS OF MATERIALS
[§15.3
in the absence of fast plates for this region, weaker Hnes in the blue
region are usually more sensitive. The improvement in infrared
sensitivity of" photographic emulsions in recent years makes it possible that the infrared lines will ultimately be found the most sensitive.
The ultimate lines of oxygen, nitrogen, sulfur, and the other gases
lie in the vacuum ultraviolet region. Equipment for taking spectrograms in this region is not available in many laboratories, however.
Special methods of excitation may be used to produce other sensitive
lines in easily accessible regions, as discussed in § 15.7.
Certain elements can be detected with greater sensitivity in the
spark thannn the arc, whereas others require use of special sources
to be detected at all. The data given in most tables of sensitive lines
apply when the most suitable source is used in a given case. The
final determination of ultimate lines must in the long run be made
empirically by the individual investigator using his own apparatus.
1S.3. Improvement of Sensitivity Limits. If the ultimate lines of
an element do not appear on a spectrogram, the analyst is likely to
conclude that the element in question is not present in significant
amounts. This absence may be due in part, however, to reduction
in the exciting power of the source by prevalence of ions produced
from atoms of low ionization potential, such as sodium or other
alkalies. As smaller impurity concentrations become significant, it
becomes increasingly necessary to introduce factors that will lower
detection limits.
It is a common misapprehension to suppose that the absolute sensitivity limit is usually set by the light-gathering power of the spectrograph. Although sensitivity to light is the limiting factor when
only a small sarnple of material is available, most qualitative analyses
are made on samples available in considerable amounts. Under such
circumstances a spectrograph of high apertjire, which often involves
low dispersion, and an emulsion of high speed, which almost always
has low contrast, are seldom best for reaching^ the limit of sensitivity.
When plenty of material is available, the lower detection limit is set'
by the ratio of intensity of the line being searched for to that of the
background spectrum, a sort of "signal-to-rioise ratio." When plate
sensitivity or time of exposure are increased, both faint lines, and
background increase, and no improvement in detection results.
Many spectrographic analyses are carried out with instruments
having dispersion too low to give the best results, not only for qualita-
§15.3]
QUALITATIVE ANALYSIS OF MATERIALS
435
tive sensitivity but also for quantitative precision. If a spectrograph
of low resolving power is used, the background intensity is increased
relative to the line intensity over what would be obtained with an
instrument of higher dispersion. To take an extreme case, it was
found that a 35-ft concave grating in a Wadsworth mounting, plate
factor 3.2 A per millimeter, could be used to photograph in two
minutes 10 tin lines in concentrations too small to be detected with
a standard medium quartz spectrograph, having a plate factor of
16.2 A per millimeter, in any exposure time, however long. A good
general rule when extreme sensitivity is desired is to use a spectrograph with resolution and dispersion as great as is consistent with
obtaining a suitable exposure in a reasonable time with the amount
of material available for burning.
It is often possible to increase the intensity of a faint line relative
to that of its neighboring background by the use of some of the following procedures:
1. By using a spectrograph of high resolution and dispersion, to
decrease continuous background relative to line intensity.
2. By selecting the optimum slit width for the lines being sought.
3. By selecting excitation conditions such that the line intensity
is increased relative to the background intensity.
4. By reducing factors in the source that produce continuous
background or bands.
5. By reducing scattered light and chemical fog.
6. By using the moving-plate technique discussed in § 15.7 to
separate lines that have maximum intensity at different times in the
burning life of the arc or spark.
7. By using fractional distillation of the sample in the source
(§15.6).
8. By treating with pure reagents the material being analyzed,
to change the negative radicals to another form in which background
intensity will be reduced and line sensitivity increased.
9. "By introducing buffer materials in the arc, or by using around
the arc an atmosphere that suppresses bands and continuous background more than it reduces line intensity.
10. By using the carrier-distillation method of Scribner and Mullin
(§ 15.8) to accelerate fractional distillation in a certain part of the
burning cycle and aid in sweeping into the arc stream the atoms to
be excited.
436
QUALITATIVE ANALYSIS OF MATERIALS
[§15.4
By a combination of these means it is often possible to extend the
sensitivity hmit of analysis for a particular element by a factor of
100,000 or more.
1S.4. Identification of Elements. I t is ordinarily not necessary to
determine wavelengths accurately for qualitative analysis, since
master spectrograms can be prepared with the spectrograph used, to
show the locations of all important lines. For certain elements the
lines will lie in definite patterns that can be identified at sight. Once
a master spectrogram has been prepared, spectrograms taken with
the same instrument for other samples can be superposed on this on
a viewing box, and lines that coincide on the two can readily be
marked. The Judd-Lewis comparator shown in Fig. 9.4 is a device
for bringing the master and sample plates to optical coincidence
without putting them in mechanical contact.
When master plates taken with the instrument being used are not
available, it is still possible to identify lines by comparison of patterns. Standard charts of the type described in § 9.3 will be found
useful for this purpose, if proper allowance is made for differences in
dispersion, or for variation in dispersion with wavelength, when the
master chart is taken with a prism or grating instrument and the
sample spectrogram is taken with an instrument of the other type.
The Spekker Steeloscope/ shown in Fig. 3.4, is a spectroscope of
moderate dispersion with an eyepiece that has been marked for setting
on the positions of sensitive lines of several metallic elements. It
furnishes a simple and convenient means of making quick optical
estimates of such elements as nickel, chromium, and copper, in steel
or other samples, and can be used for some quantitative as well as
qualitative work.
The R.U. powders and charts discussed in § 9.3 are also of considerable aid m qualitative analysis, sinqe the operator can take
spectrograms with his own instrument of samples of R.U. powder
and quickly identify from the charts the various lines in which he is
interested. On spectrograms he should be' able to find the more
important sensitive lines of each "spectroscopic" element.
;
Incorrect identification of an element is tinlikely if more than' one
line is used to establish, its presence or absence. Errors due to
inhomogeneity of-the sample-are much'more-likely, and here it is
necessary to have* recourse to those established procedures which
ensure t h a t the sample is truly representative.
' F. Twyman, General Reference 15.3, page 275.
§15.5]
QUALITATIVE ANALYSIS OF MATERIALS
437
15.5. Light Sources and Handling of Material. The original
method of qualitative spectroscopic analysis introduced by Bunsen
and Kirchhoff ^ in 1859 involved spraying liquids or powders into the
gas fed to a Bunsen burner. Although flame excitation is often used,
particularly for analysis by the method of Lundegardh/ the flame
does not supply the excitation needed to make more than about
30 metallic atoms detectable. Properly designed flame photometers
are extremely simple to use. Photography of a flame spectrum offers
little advantage over visual observation, however, since the excitation
is so low that many of the lines are confined to the visible. The excitation can be increased somewhat by using an oxyhydrogen or oxyacetylene flame.
Except in the case of modern flame photometers, emission spectroscopic analysis is ordinarily accomplished with the arc or spark
as source. The arc gives greater sensitivity in most cases, burns more
of the sample, involves a simpler electric circuit than the spark, and
requires no dangerously high potential. The spark requires somewhat less attention than the arc and may be used when very small
samples are to be consumed and extreme sensitivity is not required,
as for example in determining constituents in the metallic coating of
a watch case. Circuits for arcs, sparks, and other sources are discussed in Chapter 8.
Certain parts of an arc or spark provide greater sensitivity than
others for the detection of various elements.* ' Lines arising from
molecules that may readily be dissociated appear most strongly in
the outer layers of an arc, whereas lines arising from un-ionized atoms
may appear near the positive electrode, and from ionized atoms
near the negative.
If the material to be analyzed can be obtained in rods or chunks,
"self-electrodes" should be used. These may be roughly shaped and
held in an arc holder such as that shown in Fig. 8.5. Alternatively,
the sample may be placed in a hollo wed-out electrode of pure graphite,
as shown in Fig. 15.3. The sample electrode is usually made positive
because the positive becomes hotter than the negative, thite difl'erence
in temperature assisting in vaporizing the material. However, increased sensitivity of detection may result from making the sample
electrode negative.
To avoid contamination, the sample to be analyzed should be
' H . LundegSrdh, Zeitschr. f. Physik, 66, 109 (1930).
' Pee L. Strock, General Reference 15.7, page 49.
438
QUALITATIVE ANALYSIS OF MATERIALS
[§15.5
handled as little as possible. So sensitive is the spectrographic
method that an arc operated in a room in which a commutatorequipped electric motor is running may show copper lines even though
no copper is present in the sample. It is usually desirable to use a
sample of 10 mg or larger both for convenience in handling and so
that it may truly be representative of the master sample.
Very pure graphite electrodes, wrapped in cellophane until used
and handled only with scrupulously clean tongs, are useful for holding
(c)
(d)
Fig. 15.3. Several forms of hollowed-out graphite electrodes.
samples to be analyzed. Such "spectrographic carbons" can be
obtained from several companies or may be prepared in the laboratory.' Carbons of i^-in. diameter are most convenient for the lower
electrode, a^d may be bored out with a j-in., drill kept clean for the
purpose. The upper electrode may well be ,J in. in diameter or less
and somewhat pointed. A 5-amp arc is suitable for this size pi
electrode.
A disadvantage of the graphite electrode is the prevalence of
cyanogen bands, which are due to the molecule CN, when it is used.
' See J. S. Owens, General Reference 15.13, page 17.
§15.5]
QUALITATIVE ANALYSIS OF MATERIALS
439
These bands, some of which are shown in Fig. 15.4, are likely to
interfere with sensitive lines in various parts of the spectrum, especially in low-dispersion spectrograms. The bands are especially strong
when the sample material has been burned out of the electrode. This
condition can readily be detected by the appearance of strong violet
light in the arc. Provision should be made to turn off the arc whenever this condition appears or to bar the light from the spectrograph
while the arc is being reloaded.
|
Copper or silver electrodes are sometimes used instead of graphite
in exciting samples, but their use, which largely eliminates the
cyanogen bands, greatly reduces sensitivity. This reduction results
both because graphite electrodes become hotter than metallic ones
and because the carbon arc stream reaches a higher level of excitation
than that of a metallic arc.
To reduce the likelihood of contamination, electrodes may be preburned for a few seconds before the sample is introduced. Merely
Fig, 15.4. The cyanogen bands in the near ultraviolet (3500-4200 A) produced
by graphite electrodes burning in air.
because a given element is not found in the spectrum of the empty
electrode, however, one is not justified in assuming that it is not
present in the electrode material, since sensitivity of detection is
likely to be increased in such circumstances by the addition of
metallic ions to the arc stream, as when the sample is added. Thus,
certain highly purified graphite electrodes show no vanadium when
burned in the arc, but when a small amount of pure metallic salt,
known to contain no vanadium, is added, vanadium lines may appear.
Electrodes should always be cut or shaped with tools which have been
carefully cleaned with alcohol or ether and wiped with clean filter
paper.
Any material can be made to support an electric arc if it can be
heated to the point of volatilization into the arc stream. When a
nonconducting powder is to be studied, it can be mixed with a conducting material that gives few spectrum lines. Powdered graphite
is useful for this purpose. Ammonium sulfate can also be used, since
its component atoms do not give lines in the visible or ultraviolet
regions under arc excitation. Slags, refractories, and inorganic materials can be ground up with pestle and mortar, preferably of agate or
440
QUALITATIVE ANALYSIS OF
MATERIALS
[§15.5
other hard substance, and mixed with the conducting material, which
may be moistened with pure dilute HCl. Biological materials may
be ashed in a furnace or by digestion in pure acids, and the ash
introduced into the arc for analysis.
Strock^ has described the cathode glow method of analysis of
Mannkopf and Peters,'^^ in which use is made of the very high sensitivity of excitation that exists in the small region of the electric arc
near the negative electrode when this is used to hold the sample.
This so-called "Glimmschicht method" is widely used. High ab-
(a)
(b)
Fig. 15.5. Hollow-cathode source of McNally, Harrison, and Rowe'^ for
spectrochemical analysis, (a) .Assembled, (b) Parts.
solute sensitivity is also obtainable with the high-voltage AC arc, as
described by Duflendack" and his coworkers. A transformer giving
about 1200 volts is connected to the electrodes, and 1 amp or more
is sent through the spark, so it becomes an incipient arc. This source
must be handled very carefully on account of the danger of shock.
The hollow-cathode tube, discussed in §§ 8.19 and 15.7, probably
gives the most sensitive means of detecting small quantities of mate" R . Mannkopf and C. Peters, Zeitschr. f. Physik, 70, 444 (1931).
" O. S. Duffendack and K. B. Thonapson, Proc. Am. Soc. Testing Materials, 36,
310 (1936).
12 J. R. McNally, Jr., G. R. Harrison, and E. Rowe, Jour. Opt. Soc. Am., 37, 93
(1947).
fl5.6]
QUALITATIVE ANALYSIS OF MATERIALS
441
rial. Since the sample can be evaporated into the excitation column
over and over again, it does not escape excitation so readily as in
other sources. McNally, Harrison, and Rowe'^ found t h a t with t h e
hollow cathode they could go down to extreme limits of concentration
even in detecting the halogens and sulfur. Their source is shown
in Fig. 15.5.
A very convenient form of spark electrode for qualitative analysis
is shown in Fig. 8.10. This is applicable when electrodes can be
formed of the sample t o be analyzed. I t is desirable t h a t t h e spark
should wander somewhat in order t o prevent overheating of a portion
of the electrode, b u t this wandering should be held in a line parallel
t o t h a t connecting t h e spark with the slit so as not to get off the
spectrograph line-of-sight. Very small electrodes can be used with a
spark, especially if t h e amount of current flowing is kept small; for
example a satisfactory spark can be formed between two pieces of
t h e hair spring of a watch.
T h e spark is likely t o be less satisfactory t h a n the arc in reaching
a high sensitivity because continuElectrode (e.g.,go)d)
ous background is usually more
intense in the spark. Another disadvantage is the likelihood of exHole in shield
for emergent
citation of "air lines," which do not
radiation
appear in sources having lower exElectrode
citation. These m a y be partially
suppressed by t h e addition of inductance t o t h e spark circuit, shown
in Fig. 8.11. T h e inductance shown
in t h a t figure should be adjusted t o
Fig. 15.6. Arrangement
for
suppress air lines as m u c h as possparking from metal to liquid
sible without reducing excitation surface.
below the needed level.
Liquids can be handled directly by using t h e m t o wet graphite
powder placed on the graphite holder in an electric arc or spark, or
t h e y can be used with one of the various methods for sparking from
metal to liquid surfaces"- " (Fig. 15.6). Also, as in the L u n d e g a r d h '
method, liquids can be introduced into a flame t h a t can then be
excited electrically, if necessary.
IS.6. Moving-Plate and Fractional Distillation M e t h o d s . I t was
shown by Mannkopf and Peters'" t h a t when a sample is held in a
'' F. Twyman and C. S. Kitchen, Proc. Roy. Soc. (London), A133, 72 (1931).
442
QUALITATIVE ANALYSIS OF MATERIALS
[§15.6
deep hole drilled in the lower terminal of a graphite arc, the various
constituents are distilled into the arc a t different times. T h e y used
this fractional distillation method t o increase the sensitivity of quantitative analysis, b u t it is also extremely useful for qualitative analysis, especially when combined with the moving plate technique.
Preuss,'* who took several successive exposures during the burning
time of the arc, mentions t h a t fractional distillation was useful for
separating lines which happened t o lie close together. T h e plate
can be moved continuously parallel t o the spectrum lines during t h e
exposure,'^ by h a n d or with a small motor provided with a suitable
reducing gear, and in this way resolution in time is effected even more
conveniently. Figure 15.7 shows such a plate, taken with a medium
Fig. 15.7.
Moving plate spectrogram of a brass sample in a carbon arc.
The elapsed time was 4 min.
q u a r t z spectrograph. T h e plate was moved uniformly a t such a
speed t h a t 4 min was required t o cover the distance from the wavelength scale at t h e t o p to t h e b o t t o m of the photograph. At the t o p ,
j u s t after the arc was started, lines of zinc, cadmium, lead, and copper
p r e d o m i n a t e . After a b o u t half a minute, zinc, cadmium, and lead
disappear, the copper spectrum is greatly enhanced, and iron lines
a p p e a r . These last for another minute or more, there is a flash of
calcium a n d aluminum lines, and finally there is nothing left except
weak copper lines, a carbon line, and the cyanogen bands.
" K. Preiiss, Chem. Erde, 9, 3G5 (19.S5); Zeit.irhr. angew. Min., 1, O."! (1937).
'' See I). Uiehardson, General Reference 15.13, page 04.
\
§15.6]
443
QUALITATIVE ANALYSIS OF MATERIALS
T h e a d v a n t a g e s of using the moving plate are again apparent on
inspection of Fig. 15.8, in which the intensities of Unes from elements
having different ionization and excitation potentials are plotted
against time as 50 m g of brass turnings is burned in pure graphite
electrodes. If the spectrogram from which these d a t a were taken had
been obtained from a single 10 sec exposure, t h e lines of both elements
would have been lost a t concentrations 100 times greater t h a n those
reached by spreading the light from the arc out on a time scale.
Similar sensitivity could have been attained, of course, by photo-
0.9
Line 1
0.8
Line 2
0.7
0.6
05
0.4
0.3
\
_^-, -
\
0
/
\
\
w
y
0.2
0.1
/
\
__———
—
J
20
—
'
—
'
—
30 40
\\
-
•
50
—
J
60
-
\
'
' f "
•
'
70
80
90
100
-
• • —
^^"^^
120
Time in Seconds
Fig. 1S.8. Intensity-time curves for different elements in the same sample.
graphing the arc for 10 sec at just the proper time, b u t the best time
t o take this exposure varies from element t o element and from sample
to sample, and the moving-plate technique gives the opportunity t o
select the optimum portion of the exposure for each element.
T h e moving-plate method is especially valuable for biological
material, in which the character of the matrix is often somewhat
indeterminate. T h e conditions of excitation a t any instant are the
complex result of a n u m b e r of different boiling points, diffusion rates,
and excitation and ionization potentials. If fairly high concentrations of a material h a v i n g a low ionization potential a n d a low boiling
point are present, the excitation conditions in the arc are depressed.
-444
QUALITATIVE ANALYSIS OF MATERIALS
[§15.7
Scribner and Mullin'^ have added a further development to the
use of fractional distillation in the arc, known as the carrier-distillation
method. To the sample they add a material that is found empirically
to speed up the "washing" into the arc stream of the impurities being
detected. They first convert the sample chemically to a form having
high volatility, so that the rate of distillation in the arc will be increased. They then mix a carrier material of intermediate volatility
to the amount of about 2 per cent with the powdered sample. This
carrier may be gallium oxide, as they recommend, silver chloride, as
used by Kent,i' or some other material found empirically to be
suitable. Use of such material results in a speeding up and spreading
out of the fractional distillation process, which is very effective in
extending sensitivity limits when the moving-plate method or an
equivalent technique is used.
High-speed plates are sometimes recommended for high sensitivity
of detection, because it is thought that weak lines can be detected most
readily in this way. Weak background intensity is correspondingly
increased, however; and since what is wanted is increased density of
the line over that of the rapidly increasing background, high contrast
or, better still, rapidly increasing contrast is a more desirable plate
property. This observation suggests that slow plates, such as
process plates, are desirable for qualitative work when limiting sen. sitivities are to be reached, and .puch plates are indeed found to give
a real gain in detection sensitivity. The use of a high contrast
developer such as Eastman D-11 is also helpful.
1S.7. Analysis for Elements Difficult to Detect. The 20 elements
considered least susceptible to spectrographic analysis, either qualitative or quantitative, are the permanent gases, the four halogens, sulfur
and selenium, polonium, actinium, protoactinium, thorium, and
uranium. In recent years it has been found possible to analyze for
thorium, uranium, and the other heavy elernents by using spectrographs of high resolution and by using various innovations of excitation and handling.
/
t
The gases are difficult to detect spectrographically, not only; because they are hard to excite but also becaUse their principal lines lie
below 2000 A.^^ This vacuum ultraviolet region has been studied
"= B. F. Scribner and H. R. Mullin, Jmir. Res. Nat. Bur. Standards., 37, 369 (1946).
" R. Kent III, unpublished report.
18 But see R. A. Wolfe and O. S. Dilffendack, General Reference 15.14, page 66;
also footnote 19.
§15.7]
QUALITATIVE ANALYSIS OF MATERIALS
445
extensively in scientific laboratories, and it is usual to find lines of
nitrogen, oxygen, and other gases appearing on almost all vacuum
spectrograms. These elements can readily be detected by the
methods of vacuum spectroscopy, especially in hot spark and other
sources having high excitation (Chapter 19). Gatterer and Frodl ^'
have investigated their excitation with high-frequency sources and
have developed methods for their detection down to low concentrations.
Sulfur and selenium show no ultimate lines in most tables because
their sensitive lines lie at wavelengths shorter than 2000 A. Harrison
and Merrill ^^ developed three different ways of analyzing for these
two elements. For sulfur the three lines that theoretically should be
the most sensitive lie between 1807 and 1826 A, while for selenium
they lie between 1960 and 2062 A. The standard type of quartz
spectrograph does not quite reach them, because of absorption by its
quartz optical train and because of the absorption of light of these
wavelengths by oxygen in the air and by the gelatin in the photographic emulsion. It is also necessary to overcome the effect of the
high excitation potentials of sulfur and selenium, which are above the
ionization potentials of the metals forming the usual matrix. The
latter difficulty can be overcome by using arcs of high current density
or sparks with high average excitation. The former difficulty was
overcome by flushing an ordinary spectrograph continuously with a
stream of commercial tank nitrogen, allowing this to leak out the slit
against the arc or spark placed directly in front of it, and by using
specially sensitized plates. When a lithium fluoride optical train or a
diffraction grating was used instead of a quartz prism, an increase of
sensitivity resulted. By this means, nickel was successfully analyzed
for sulfur down to concentrations below 1 in 10^.
Harrison and Merrill carried out similar analyses by using infrared
lines of sulfur which, though not the ultimate lines, were found very
sensitive when used with modern infrared plates. Sulfur was also
analyzed in the ordinary spectrum range by using the hot spark and
detecting the stronger lines of SII and SHI. The first of these
three methods was found most convenient and practical, but the
hollow-cathode method was found later to detect S and Se at still
lower concentrations.
" A. Gatterer and V. Frodl, Ric. Speitroscop., 1, 201 (1946).
20 See G. R. Harrison and D. P. Merrill, General Reference 15.16.
446
QUANTITATIVE ANALYSIS OF MATERIALS
[§15.8
Konovalov and Frisch^' used the hollow cathode to detect nitrogen
and argon to a few tenths of 1 per cent. McNally, Harrison, and
Rowe'2 have developed a method of spectroscopic analysis of the nonmetals with emphasis on analysis for fluorine, and their paper gives
references to previous papers on the analysis of gases spectroscopir
cally. They studied the relative sensitivities of fluorine lines in the
vacuum ultraviolet region and in the visible, and found that certain
visible lines appeared more sensitive than the theoretical ultimate
lines of fluorine. By using a specially designed hollow-cathode
source, they were able to detect as little as 10""^ gram of fluorine at a
concentrational sensitivity of less than 1 in 10^. Chlorine and sulfur
were readily detected in amounts as small as 0.2 /xg, and 1.0 ixg respectively, in samples weighing 20 mg.
Fluorine can be determined spectrographically by adding calcium
salts to the sample, if necessary, and using the band head of calcium
fluoride at 5291 A, since the CaF molecule resists dissociation in the
arc. Papish, Hoag, and Schnee^^ obtained an absolute sensitivity
limit of 10 Mg of fluorine by this method.
15.8. The Qualitative Analysis. Figure 15.9 shows a typical
master plate marked with the sensitive lines of the principal elements
obtained with R.U. powder (§ 9.3). The extra trouble of preparing
such master plates is worth while for any spectrograph that is to be
used extensively for analytical work.
It is convenient in making a qualitative analysis to put the elements
sought Into four categories: major constituent, minor constituent,
trace, and absent, corresponding roughly to the ranges 100 to 1 per
cent, 1 to 0.01 per cent, 0.01 per cent to the minimum detectable,
and less than this. Such categories and limits are only approximate
and will vary greatly from element to element. Often a great deal
of semiquantitative Information can be obtained from the relative
intensities of the lines, particula,rly when several similar samples are
being compared In which the amount of pne element relative, to
another Is changing. The temptation to estimate amounts without
some sort of control .should be resisted, even if the estimate is no
closer than a factor of 10, since the inten'slty of a line depends on
many factors In addition to the concentration of the element being
studied. A typical qualitative analysis report is given in Table 15.3.
=ii V. A. Konovalov and S. E. Frisch, Jour. Tech. Phys. {U.S.S.R.), 4, 523 (1934).
22 J. Papish, L. E. Hoag, and W. E. Snee, hid. Eng. Chem., Anal, ed., 2, 203 (1930).
itt
73^9oCa
J>
w
a
3.
IB
ti
?3
I
o
>
B
— .— '
K
r
c
•
QUANTITATIVE ANALYSIS OF MATERIALS
448
[§15.8
TABLE 15.3
REPORT OF SPECTKOCHEMICAL QUALITATIVE ANALYSES
^Xx*-t..CZ..oc^^-<_i^
For:^
Element
1
Ag 7 n Al
M
As
Au
B
Ba
/*n/
Be
Bi
Ca
M
Cb
Cd
U
Ce
Co
Cr
Cs
P^
Cu
/y
Dy
Er
Eu
Fe
/yr\y
Ga
Gd
Ge
Hf
Hg
Ho
In
J>v
Ir
K
M
La
Li
yfyU
Lu
M g />n/
Mn
Mo
2
Sample
4
3
i>n-
/yri-h /yn
5
/yr\M
M
M
/iX-
/yn
/yn/
/yyv /yn.
M
-»v
M
M
/yn,
/Ou
M
M
/V
/yn.
/rru
/>n- /yrv />n-'
M
A
/A/
A•
M
/V
/yyv
M
tt/
/rn/ /yyv
/tW
'
/yri/
•»t^
/yn/
/yyu
-
M = constituent probably major
m = constituent probably minor
tr = constituent probably trace
^aJ^
Date: OT^CLA^CA.
^f
EleSample
ment
1
2
3
4
5
Na
A/
M
M
M
/^
Nd
Ni
tk,
/yyu
Os
P
/y>^- /yyu /y/v /yn/ /yyu
P b XA/ ;tA/
Pd
Pr
tk/ tfu
Pt
Ra
Rb
•
Re
Rh
Ru
Sb
tA/
Sc
Si
/ni/ /m/ /yrt. M M
Sm
Sn
/yyu
Sr
/yrf /yyu A- i^ /yyu
Ta
Tb
Te
Til
Ti
/yyi/ XA/
yyyi/
Tl
Tm
U
V
t)i/ ^
W
,
!
1
Yb
,
Yt
Zn /yyu /yyi/ />n/ zi. fiu
Zr
XA.
X/t„
t^
•
QUANTITATIVE ANALYSIS OF MATERIALS
449
GENERAL REFERENCES
15.L
15.2.
15.3.
15.4.
15.5.
15.6.
15.7.
15.8.
15.9.
15.10.
15.11.
15.12.
15.13.
15.14.
15.15.
15.16.
15.17.
15.18.
15.19.
W. R. Erode, Chemical Spectrnscopy, 2d ed. New York: John Wiley
& Sons, Inc., 1942.
R. A.Sawyer, Experimental Spectroscopy. New York: Prentice-Hall,
Inc., 1944.
F. Twyman, The Spectrochemical Analysis of Metals and Alloys.
Brooklyn: Chemical Publishing Company, Inc., 1941.
D. M. Smith, Bibliography of Spectrochemical Analysis, 2d ed. London: British Non-Ferrous Metals Research Association, 1940.
F. Twyman and D. M. Smith, Wavelength Tables far Spectrum Analysis, 2d ed. London: Adam Hilger, Ltd., 1931.
T. R. P. Gibb, Jr., Optical Methods of Chemical Analysis. New York:
McGraw-Hill Book Company, Inc., 1942.
L. W. Strock, Spectrum Analysis with the Carbon Arc Cathode Layer.
London: Adam Hilger, Ltd., 1936.
W. Gerlach and E. Schweitzer, Die chemische Emissions-speJdral• analyse,!.
Leipzig: Voss, 1930.
W. Gerlach and E. Schweitzer, Foundations and Methods of Chemical
Analysis by the Emission Spectrum. London: Adam Hilger, Ltd.,
1931. (Translation of 15.8.)
W. Gerlach and W. Gerlach, Die chem^ische Emissions-spektralanalyse,
II. Leipzig: Voss, 1933.
W. Gerlach and W. Gerlach, Clinical and Pathological Applications of
Spectrum Analysis. London: Adam Hilger, Ltd., 1934. (Translation of 15.10.)
W. Gerlach and E. Riedl, Die chemische Emissions-spektralanalyse.
Leipzig: Voss, 1936.
Proe. Fifth Summer Conf. on Spectroscopy, 1937. New York:
John Wiley & Sons, Inc., 1938.
Proc. Sixth Summer Conf. on Spectroscopy, 1938. New York:
John Wiley & Sons, Inc., 1939.
Proc. Seventh Summer Conf. on Spectroscopy, 1939. New York:
John Wiley & Sons., Inc., 1940.
G. R. Harrison, "Practical Possibilities in Spectrographic Analysis,"
Metals and Alloys, Nov. 1936.
W. F. Meggers, "Principals and Principles of Spectrochemical Analysis," Spectrochimica Acta, 3, 5 (1947).
W. F. Meggers and B. F. Scribner, Iridex to the Literature in Spectrochemical Analysis. Philadelphia: American Society Testing Materials, 1939.
Jour. Applied Phjsics, Nov. 1939, Vol. 10.
CHAPTER
16
Quantitative Spectrochemical Analysis
Q U A N T I T A T I V E ANALYSIS O F MATERIALS BY EMISSION SPECTROSCOPY
is frequently referred to as spectrochemical analysis, less frequently as
spectrum analysis, the latter term being likely to be confused with the
analysis of spectra. Since 1930 the development of spectrochemical
methods has been rapid, and they can now be made so precise, can be
carried out so rapidly and simply, and are used in so many routine
industrial applications that they must be considered as well tested and
established.
Spectrographic emission methods of quantitative analysis can be
readily used whenever the elements to be determined are metallic or
metalloidal, when the chemical combinations in which thfey are
present in the sample need not be determined, and when the combined
concentrations of the elements of interest are less than about 5 per cent
of the entire sample. Outstanding advantages of the method are its
rapidity, the smallness of the sample that can be analyzed, the
simplicity of the operations needed, the sensitivity available for determining very low concentrations of material, and the precision obtainable. Precision to within 2 per cent of the amount of element
being determined is no longer unusual, and this precision is to ,a
considerable degree independent of the actual concentration. Thus
it is usually as easy to determine spectrdgraphically the difference
between 0.0010 and 0.0011 per cent of lead in gold, for example, as
that between 0.10 and 0.11 per cent. Ten miUigrams of sample may
suffice for the quantitative determination of from 1 to 30 elements,
and if necessary as little as 0.01 mg may serve.
Probably the greatest usefulness of spectrochemical methods arises
in carrying out routine procedures where similar analyses are to be
made on hundreds or thousands of samples, as in production composition control. Here the time and effort consumed in preparing standard samples and in setting up special procedures are soon repaid in
450
QUANTITATIVE SPECTROCHEMICAL ANALYSIS
451
savings of time in individual tests, aiid in consequent reductions of
operating costs in plants and laboratories. In most of the metals
industries and in many chemical laboratories, spectrochemical analysis is now a routine control process.
As chemical wet methods grow less precise at lower concentrations
owing to the small quantities of material available, spectroscopic
methods gain in advantage. Quantitative analyses can be made
spectrographically under almost all conditions in which qualitative
spectrographic analyses can be made, the difference being largely the
control of the conditions of spectrography. The situations in which
quantitative spectrographic analysis is most useful are similar to those
discussed in Chapter 15 for qualitative analysis.
Simple quantitative spectrographic methods serve for determination of some 70 of the chemical elements, including all metallic and
metalloidal elements. Special methods, somewhat less convenient
than those svifficing for metallic elements, have been developed for
sulfur, selenium, and some of the halogens (§§ 15.7 and 16.14). Satisfactory spectrographic methods have not yet been developed for the
analysis of gases in gross quantities, but tentative steps in this direction were discussed in § 15.7.
Quantitative determination of molecular constituents in a sample
can sometimes be carried out with high sensitivity and excellent
precision by the methods of absorption spectrophotometry and
Raman spectroscopy. These are discussed in Chapters 14, 17 and 18.
Many points of importance in quantitative analysis were examined
in Chapter 15, and the contents of that chapter should be carefully
kept in mind in connection with the problems now to be discussed.
THE BASIC PROCEDURfi OF
QUANTITATIVE ANALYSIS
Though many procedures of quantitative analysis are discussed in
the literature and are given the names of their proposers, practically
all are variations of one basic procedure. Since the relative numbers
of atoms present in a sample are to be determined by means of the
radiation which they are caused to emit and since the quantity of this
radiation can depend on many factors besides the number of atoms
emitting it, a null method is required. Elimination of the influence of
factors that need not be determined can most readily be done by
keeping as many of them constant as possible.
QUANTITATIVE SPECTROCHEMICAL ANALYSIS
452
[§16.1
16.1. Basic P r o c e d u r e . T h e sample t o be analyzed may be regarded as consisting of the following: (a) one or more major constituents, which together constitute the matrix. These influence t h e
excitation conditions in t h e source, such as t h e equivalent t e m p e r a t u r e
of burning, b y the number and character of t h e ions they provide.
(b) F r o m 1 t o 70-odd atomic constituents, t h e elements which are t o
be quantitatively determined, present in such small quantities t h a t
variations in their concentrations will n o t significantly affect t h e
excitation conditions.
Suppose t h a t we h a v e a sample of palladium in which we wish t o
determine quantitatively the a m o u n t of platinum. This can be done
Pt Pd
PdPt
OT CO
f
O o*
o
W m
5 1 CO
O O
O SS
11
n a*
//
PtP
oM s*
Q»
o a*a
<N
11
Pt
Ptpd
11
Pt Pt Pt Pt
I
I //
Pt Pt Pt Pt
i //
II
III
IV
Fig. 16.1. Determination of platinum in palladium sample. 'The Pt lines
(X 2650) shown in the spectrum a (the unknown) are seen to have an intensity
between that of standard sample III and standard sample IV.
b y preparing a series of otherwise identical palladium samples in
which t h e a m o u n t of platinum present is controlled. These standard
samples are t h e n burned in a uniform m a n n e r in an electric a r c , o r
other source of excitation, and t h e unknowri sample is burned in, as
closely t h e same manner as possible. F r o m t h e resulting spectrogram
t h a t s t a n d a r d sample is selected which has jproduced P t hnes of most
nearly the same density as those produced from the sample being
analyzed, t h e " u n k n o w n . " This situation is shown in Fig. 16.1.
Here t h e two standards I I I and I V are seen t o bracket the unknown
sample a.. A new series of standards intermediate in P t content
between I I I and I V can then be made u p and burned, until one is found
§16.2]
QUANTITATIVE SPECTROCHEMICAL ANALYSIS
453
whose Pt lines duplicate in intensity those of a. The unknown can
then be assumed to have the same concentration of P t as this standard
sample.
In practice it is not necessary to match the density of the line in a
standard sample exactly to that of the unknown. A working curve
of concentration against intensity of light emitted under controlled
conditions can be plotted, on which the unknown concentration can
then be interpolated from measured densities. I t is essential, however, that standard and unknown samples be treated as nearly as
possible alike. The precision of an analysis will usually depend on
how closely this duplication can be controlled.
16.2. Sources of Excitation. Arc, spark, and flame excitation, as
discussed in Chapters 8 and 15, can all be used for quantitative
analysis, and any one of these may serve best for a particular type of
sample. In the electric arc more of the sample is likely to be consumed, and hence the sample can more easily be made truly representative of the material being analyzed. The arc produces very bright
spectrurn lines, involves simple electrical circuits, and produces no
confusing air lines that sometimes interfere with spark analysis.
Somewhat higher excitation, which is sometimes needed, can be
obtained with the spark than with the arc in air. The spark requires
less attention than the arc, consumes a smaller amount of material,
may produce less continuous background and fewer interfering bands,
and vaporizes volatile material less rapidly. With special methods,
such as those of Ramage' and of Lundegardh,^ flames of various types
have certain advantages, especially when biological samples are to be
consumed. The flame photometer has been referred to in Chapter 15.
For certain types of samples the AC arc method' developed by
Duffendack and his collaborators gives very high sensitivity, and the
Glimmschicht or cathode-layer method described in § 15.5 gives
equally high sensitivity under other conditions. The hollow cathode
(§§ 8.19 and 15.7) forms an extension of the cathode-layer method
and probably gives the highest sensitivity of all sources when an
extremely small amount of material is available for excitation. The
reason for this sensitivity is that when the cathode layer is curved
iH.'Ramage, Nature, 123, 601 (1929); ibid., 137, 67 (1936); H. Ramage, J. H.
Sheldon, and W. Sheldon, Proc. Roy. Soc. (London), B113, 308 (1933).
^ H. Lundegardh, Zeitschr.f. Phy.^., 66, 109 (1930); see also General Reference 16.3.
' O. S. Duffendack and K. B. Thompson, Proc. Am. Soc. Testing Materials, 36,
310 (1936).
454
QUANTITATIVE SPECTROCHEMICAL ANALYSIS
[§16.4
into a complete cylinder, an atom has more difficulty in wandering
from the stream of excitation t h a n in other sources.
T h e accuracy of quantitative analysis is usually limited by fluctuations in the source. Causes of these fluctuations are differential
evaporation, cathode spot wandering, current fluctuations which are
caused by these a n d other effects, and irregular train length in the case
of spark sources. For this reason special sources, such as the controlled spark, discussed in Chapter 14, are of special importance in
spectrochemical analysis.
16.3. Form and Preparation of the Sample. In preparing samples
for burning, it is i m p o r t a n t to minimize handling and to simplify
preparation procedures. If the material to be analyzed is solid and
can be obtained in small, uniform sticlss or bars, these can be used
directly as "self-electrodes." A common source of error is inhomogeneity of t h e sample. When a sample is not uniform, it m a y be
ground t o a powder, mixed thoroughly, and packed in a hollow electrode of graphite or some other conducting material. T h e sample
m a y also be dissolved and the resulting solution used to moisten
graphite powder, which can then be packed into a graphite electrode.
Sometimes a spark can be made to pass directly to a liquid surface
(Fig. 15.6).
Graphite, copper, silver, and gold are used as container electrode
materials, principally because of their high melting points and
because they can be readily obtained in spectroscopically pure form.
Graphite is m u c h more widely used t h a n the other three because of
its desirable electrical and thermal characteristics. Somewhat higher
excitation is obtained in an arc between graphite electrodes t h a n
between metallic electrodes. Pellets of graphite or other material
can be compressed in briquetting a p p a r a t u s , to give exact amounts of.
uniformly dense material for burning.'
16.4. Standards for Comparison. By far the greatest part of the
work involved in making a spectrographic quantitative analysis lies
in preparing t h e standards, which must be similar in general constitution or matrix to the sample to be analyzed b u t which contain known
a m o u n t s of t h e elements being determined. /Fortunately, one standard sample will serve as a container of known amounts of a number
of elements. There is no limit to this number except t h a t the total
q u a n t i t y of varying elements must be kept low enough to avoid
affecting the matrix conditions. This requirement usually means
below 1 per cent of the entire sample, though in certain cases as much
§16.4]
QUANTITATIVE SPECTROCHEMICAL ANALYSIS
455
as 5 per cent is permissible. A spectrograph with sufficient resolution
m u s t be used, of course, so t h a t the lines of the various elements d o
not interfere.
One per cent of zinc in a sample of sodium chloride will give m u c h
less light t h a n 1 per cent of zinc in a sample of gold, for example,
because the excitation conditions set by the prevalent sodium ions in
t h e first case are much lower t h a n those set by the prevalent gold ions
in the second case. I n analyzing for zinc in N a C l it would be necessary to make standard samples in which the matrix was N a C l , a n d
varying amounts of zinc were added. I n analyzing gold for zinc it
would be necessary to have a matrix of metallic gold t o which varying
a m o u n t s of zinc were added, and to have any sodium ions present in
q u a n t i t y in the unknown gold sample also present a n d constant in
concentration in the s t a n d a r d samples. Elements of low boiling point
and ionization potential, such as the alkali metals, exert very powerful
effects on the matrix, and small variations in concentration of these
m a y produce large changes in the light emitted by a minor constituent
in the same sample. A sample of gold containing 1 per cent zinc a n d
no copper can be expected to emit zinc lines as strongly as a similar
sample containing 1 per cent of zinc and 1 per cent of copper, b u t if
as much as 5 per cent of copper is introduced, or 1 per cent of sodium,
the intensity of the zinc lines can be expected to be different.
Metallurgical standard samples can be made either by mixing
alloys in the induction furnace,^ or by dissolving the samples in acid
to produce salts which can then be ground and thoroughly mixed.^
Another procedure is to mix metallic powders with a basic matrix
buffer material, such as graphite or a salt, such as ammonium sulfate
or ammonium chloride, which produces few spectrum lines. Biological samples to be examined for traces of metals can usually be
digested with acids or ashed in a furnace to remove organic constituents, and then added to some matrix base t h a t can be reproduced in
preparing standards.
One of the simplest methods of preparing a s t a n d a r d is to t a k e
some of the actual material being analyzed and a d d to it known
a m o u n t s of the elements to be determined. This procedure involves
extrapolation back toward zero concentration in successive approximations, however, which is much less precise t h a n if a pure sample
* Sse General References 16.1 and 16.2.
* See General References 16.1 and 16.3.
456
QUANTITATIVE SPECTROCHEMICAL ANALYSIS
[§16.5
of similar material can be obtained and used as the matrix for the
addition of t h e elements of interest. E v e r y spectrographic laboratory
in which q u a n t i t a t i v e analyses are to be m a d e should build u p a
collection of standard samples as rapidly as possible.
16.5. Burning of the Sample. T h e role of t h e excitation source is
threefold: t o vaporize the sample a t a rate as controlled as possible;
t o dissociate all molecules present into their constituent atoms or
atomic ions; a n d to excite these atoms or ions to emit radiation.
When very low concentrations are involved, it is important t o cause
every a t o m of the element of interest to e m i t as much light as possible, a n d the methods of improving sensitivity discussed in § 15.3
m a y be invoked.
F u n d a m e n t a l in control of the source, however, is the necessity of
treating t h e s t a n d a r d sample and the unknown sample exactly alike
Such equal t r e a t m e n t can be attained satisfactorily only when both
samples have the same matrix, are subjected t o the same conditions
of excitation, and are caused to emit radiation in the same way.
16.6. Selection and U s e of the Spectrograph. Quantitative
analyses can be made with a visual spectroscope, b u t this method has
fallen into disuse except for rapid, short-cut analyses of fairly high
concentrations. T h e Spekker Steeloscope (Fig. 3.4 and § 15.4) is
furnished by Hilger fitted with the I n s t a eyepiece for this purpose.
Usually a spectrograph is better t h a n a visual spectroscope because
it gives greater precision and permits use of t h e ultraviolet spectrum,
a ' region where suitable lines for analysis of m a n y elements lie.
Moreover, a p e r m a n e n t record is given. A glass-prism spectrograph can be used, b u t a spectrographer limited to the visible region
alone is decidedly handicapped. A q u a r t z spectrograph of medium
dispersion will serve for quantitative, analyses of elements in the first
three rows of t h e periodic table, which^haye comparatively simple
spectra. M o s t analyses are m a d e with prism, spectrographs having
a focal length of 6 ft or more, or with gratings of focal length 2 t o
6 meters. Ferrous materials and others containing a matrix whose
spectrum is complex require axdispersion of 0.15 m m / A or more.
Where only a limited spectral region can be covered, more elements
can be analyzed for by using the range 2400 t o 4000 A t h a n any other
region of similar extent, but provision should be m a d e if possible for
covering the range 2000 to 8000 A.
I t is of particular importance for q u a n t i t a t i v e analysis t h a t background light and scattered light be kept t o a minimum. T h e spec-
§16.7]
QUANTITATIVE SPECTROCHEMICAL ANALYSIS
457
trograph should of course cover all spectral regions that include the
lines chosen for the analysis. It also should have sufBcient dispersion
so that the encroaching background will not limit to values greater
than those present in the samples being analyzed the sensitivity to
concentrations of the elements being determined (see § 15.3). It
should produce a spectrum plate on which as many as 10 or 20 exposures to various standard and unknown samples can be taken.
It is usually not difficult to produce sufficient radiation from a
sample to obtain a satisfactory spectrogram with an exposure time
of 5 min or less. If the exposure time is less than 30 sec, it is sohietimes desirable to lengthen it by artificial means, such as the use of
diaphragms or rotating disks to reduce the speed of the spectrograph,
since a burning time of 1 min or longer rnay be necessary to ensure
the consumption of a truly representative sample.
A stigmatic spectrograph is especially suitable for analytical purposes, because with it simple methods of photometry can be used.
Rotating-disk photometry requires uniform illumination of the slit,
which can best be carried out by one of the methods discussed in
§§ 6.6 and 13.2-13.3. It is desirable also to use a spectrograph that
produces straight spectrum lines. A fairly large concave grating in
a Wadsworth stigmatic mount (§ 4.7) will be found especially flexible
and convenient for analytical work.
16.7. Selection of Lines for Quantitative Analysis—^The Working
Curve. If very low concentrations of a minor constituent are to be
determined, it is usually necessary to use its ultimate lines for analytical purposes. For higher concentrations, lines that lie in a more
convenient region of the spectrum may be selected, but those lines
should be chosen which have a straight-line working curve in the concentration range which is of interest. Figure 16.2 contains a typical
working curve, in which the percentage concentration of a minor constituent is plotted logarithmically against the logarithm of the ratio
of the intensity of one of its spectrum lines to that of a line of the
matrix. At very low concentrations, all lines show a linear variation
of intensity with the number of atoms present. This linear variation
when plotted logarithmically gives a straight line having a slope of
45 deg. At the lowest concentrations, theory can thus be used to
check the actual course of the working curve. At higher concentrations, the values of which depend on the line under consideration, the
concentrational sensitivity of the curve becomes less, and at very high
concentrations ah actual decrease in intensity with increasing con-
458
QUANTITATIVE SPECTROCHEMICAL ANALYSIS
[§16.8
centration may arise from self-reversal. The working curve is then
curved downward.
Two spectrum lines should be used for a concentration determination of a given element. One of these is a line of the element itself,
chosen to give high, and if possible uniform, concentrational sensitivity
so that the working curve will be a straight line with a 45-deg slope.
The second line is one selected to lie as close in the spectrum to the
first as possible, to simplify photometric problems; and similar in
intensity to the first line, to simplify the comparison of their intensities. It may be a line of an element of the matrix material, of
some impurity known to be present in the samples in constant
amounts, or of an element especially added in constant amounts to
all samples to furnish a spectrum line whose intensity remains as
nearly constant as possible in all exposures. The purpose of this line
LoQio Percentage Composition
'~
Fig. 16.2. Typical working curve. 7o is the intensity of a spectrum line in the
matrix and Ix that of a line due to the element under analysis.
is to furnish a standard of comparison' that will indicate any changes
in source or photometric conditions which might affect the intensities
of lines of the working element. The intensities of the working and
control lines should also vary similarly with excitation conditions.'
Thus if the working curve is plotted as a function of the intensity
ratio of the unknown line to the control line, instead of the unknown
line intensity alone, variations between unknown and standard spectra
are to some extent nullified. Other functions can be plotted to give
the working curve, but none are so convenient as those used, in
Fig. 16.2.
16.8. The Calibration Curve. If the working and control lines
have about the same wavelength, which usually requires that they lie
not more than 25 Asfrom each other in the spectrum, the ratio of their
§16.9]
QUANTITATIVE SPECTROCHEMICAL ANALYSIS
459
intensities can be determined from a single calibration curve (§ 13.2).
If they have widely differing wavelengths, which is usually not the
case, different calibration curves must be used for the two lines, and
the methods of heterochromatic photometry are required (§ 13.4).
Ordinarily, however, homochromatic photometry suffices for analytical purposes, since a line of the matrix or of an added internal standard
can almost always be found near a suitable line of a minor constituent.
Even when a single calibration curve can be used for the two lines,
it is not satisfactory, in general, to plot the working curve in terms
of the densities of the two lines rather than the logarithms of their
intensities, because the calibration curves are almost never linear.
This procedure may sometimes be justified in special cases over a
limited density range, but can be assumed a priori to be correct only
when the densities of the two lines are equal. The latter is the basis
of the length-of-line method discussed below (§ 16.13),
I
I I
ill
111
I
1 I
'd,
I
:]•
I
.1 I
i
T
Pb 2833
Fig. 16.3. Spectrogram showing Pb lines at 2833 A in a dried sample of
condensed milk.
The methods of photographic photometry, discussed in Chapter 13,
are so straightforward, and the necessary apparatus so readily available, that there is little excuse for permitting photometric errors to
influence the results of a quantitative analysis. They can quite
readily be kept below 2 per cent, and errors due to nonhomogeneity of
sample, variation of excitation and burning in the light source, and
other errors of handling are usually found to be much greater.
16.9. A Typical Analysis. Every quantitative spectrographic
analysis should be preceded by a qualitative analysis (Chapter 15).
Only thus can a satisfactory appraisal be made of the matrix conditions, of various constituents not of interest that might interfere with
the determination of those which are of interest, and of the approxi-
460
QUANTITATIVE SPECTROCHEMICAL ANALYSIS
[§16.9
m a t e concentration ranges of the elements of interest. A qualitative
analysis will also give some indication of t h e most satisfactory excitation conditions t o b e used.
We give as a n example d a t a
obtained in t h e determination of
lead in a sample of condensed
Density
milk. After a preliminary qualitative analysis, 50 m g of spectrographically pure graphite
powder was placed in a porcelain
evaporating dish known t o h a v e
no lead contamination, a n d on
Logio Intensity
^
this 50 mg of condensed milk
Fig. 16.4. Calibration curve for analysis
was pipetted direct from , t h e
at \ = 2833 (Pb line).
freshly opened can. T h e dish
was covered loosely with a glass plate in such a w a y t h a t condensed
droplets of moisture would not fall back into the sample, and the
escape of steam was permitted.
T h e dish was placed in an
evaporating oven a n d heated
slowly until all moisture had
escaped; it was t h e n baked at
a t e m p e r a t u r e such t h a t only
charred residue in t h e graphite
remained, care being taken to
ensure t h a t a t no t i m e was the
t e m p e r a t u r e greater t h a n the
boiling point of lead or of any
LoQio percent Pb
»lead salts t h a t might be
Fig. ,16.5. Working curve for Pb analysis.'
formed. T h e ash-graphite mixt u r e was t h e n placed in an agate mortar, carefully ground to a fine
powder, a n d thoroughly mixed. , T e n milligrams of the resulting powder was then packed into the cup of a prebufned graphite electrode
of f-in. diameter with a | - i n . hole drilled in its end to a depth of f in.
This electrode was made the positive terminal of an electric arc, and
was b u r n e d a t 5 a m p against a negative electrode consisting of a
J-in. diameter pencil of pure graphite. T h e spectra shown in Fig. 16.3
were obtained with a Hilger Littrow q u a r t z spectrograph, using 6b-sec
exposures. T h e plate was measured on a densitometer; t h e calibration a n d working curves obtained are shown in Figs. 16.4 and 16.5.
§16.9]
QUANTITATIVE SPECTROCHEMICAL ANALYSIS
461
Many typical analyses involving samples of different types will be
found described in the literature (see General References 16.1-16.7).
SPECIAL METHODS
The basic method described in the preceding sections was gradually
evolved between 1890 and 1940 and is subject to many variations.
Such investigators as de Gramont^ worked for many years to develop
methods that would be sufficiently reproducible to give quantitative
results. Because of the large number of variables that govern the
intensity of the spectrum lines produced by a given quantity of material, many investigators thought it impossible to gauge accurately
the quantity of a substance in a mixture by means of its spectrum.
Early workers suggested four different methods for measuring the
quantity of an impurity present: (1) the length of the lines produced
in a spark, higher concentrations producing lines which extended
further from the electrode;' (2) the number of the lines of an element
that appeared on the plate, only the ultimate lines remaining when
the concentration was below certain limits f (3) the intensities of the
lines; and (4) the time taken for disappearance of the lines of a
volatile element as it was burned out of the arc. Any of these
methods can be used qualitatively, but only the third has been
developed to give precise quantitative results.
Much of the credit for the development of satisfactory methods of
quantitative analysis should go to de Gramont,* who spent years
convincing his colleagues that the method could be made reproducible.
Work in America was given an impetus in 1922 by a classical paper by
Meggers, Kiess, and Stimson,^ entitled "Quantitative Spectroscopic
Analysis of Materials." This outlined a straightforward method of
comparison of samples, relying on constancy of excitation conditions.
The next major step forward resulted from the work of Gerlach and
Schweitzer,'" who introduced the method of internal standards and
« A. de Gramont, Compt. Rend., 144, 1101 (1907); ibid., 159, 6 (1914); ibid., 171,
1106(1920); Rev. Met., 19,90 {im^).
' J . N. Lockyer, Phil. Trans. (London) (I), 163, 253 (1873); ibid., 164, 479 (1874);
J. N. Lockyer and W. C. Roberts, ibid., 164, 495 (1874); A. Occhialini, Rendiconli
R. Accad. Lincei, 9, 573 (1929).
n V . N. Hartley, Phil. Tram. (London) (I), 175, 50 (1884); A. G. Leonard and
P. Whelan, Proc. Roy. Soc. Dublin {N.S.), 11, 23 (1908).
1 W. F. Meggers, C. C. Kiess, and F. S. Stimson, Nat. Bur. Standards Sci. Paper
444 (1922).
'" W. Gerlach and E. Schweitzer, General Reference 16.3.
462
QUANTITATIVE SPECTROCHEMICAL ANALYSIS
[§16.10
the concepts of homologous pairs and fixation pairs.
This method,
though now seldom used in the form envisioned by its authors, has
made a very definite contribution to the basic method.
16.10. The Method of Internal Standards. Gerlach and Schweir
tzer developed this method in an a t t e m p t to avoid the necessity of
maintaining a group of standard samples containing known concentrations of impurities. T h e y sought a method in which the comparison with s t a n d a r d samples of a given element-matrix combination
could be done once and for all in a single laboratory, so t h a t from
published lists anyone could determine concentrations by merely
matching the relative intensities of - constituent and matrix lines,
called homologous pairs.
Since the intensities of the lines of a major
constituent of t h e matrix do not vary greatly with concentration,
whereas those of t h e minor constituents do vary greatly, it should be
possible to select strong lines of the minor constituent t h a t a t low
concentrations will be equal in intensity to weak lines of the matrix.
Equality of intensity would be indicated b y equality of density on
the plate. Gerlach and Schweitzer found t h a t the equality of intensity of such a pair of lines would be maintained under widely varying
conditions of excitation.
A second concept introduced by Gerlach and Schweitzer was t h a t
of t h e fixation pair, another pair of lines of t h e matrix selected so as
to be equal in intensity under the excitation conditions used b u t
extremely sensitive in intensity ratio to variations of excitation
conditions. T h u s these lines could be used t o indicate the a t t a i n ment of correct excitation conditions.
M a n y variants of the Gerlach and Schweitzer method have been
described, b u t of these t h e principal contribution t h a t remains in the
basic method is the procedure of producing t h e working curve b y
using the intensity ratio of a pair of lines of t h e minor constituent a n d
matrix rather t h a n the intensity of the former line. Pairs of lines are
seldom used t o d a y either as homologous or fixation pairs, since mor^e
direct methods of determination are available.
T h e advantages of t h e method of homologous pairs are t h a t 'it
furnishes a null method of photometry, since it is easy to judge when
the densities of a pair of lines are equal, a n d t h a t it avoids the necessity of preparation of standard samples whenever an analysis is to be
made. I t s great disadvantage is t h a t it is largely a theoretical
method," since it makes the assumption t h a t matching of the fixation
and homologous pairs has been carried out with a fairly small n u m b e r
§16.11]
QUANTITATIVE SPECTROCHEMICAL ANALYSIS
463
of spectrograms, whereas in t h e case of an actual analysis a great
many spectrograms are likely t o be required before satisfactory
matching is obtained.
Gerlach and Schweitzer's greatest contribution is the method of
internal standards, involving use of a line of the matrix or an added
material as a control in the basic method.
16.11. M e t h o d s for High Concentrations. T h e upper limit of
concentration at which ordinary methods of q u a n t i t a t i v e spectrographic analysis become uncertain is usually given as 5 per cent,
though in some cases 10 per cent concentration can be reached when
only one variable element is involved. I n certain cases great care
must be t a k e n when even 1 per cent of variable material is exceeded,
for example when alkali metals or other elements are involved which
have ionization potentials widely different from t h a t of the matrix
material. T w o special methods are available, however, t h a t can be
used to make analyses a t any concentration. T h e y are not very
precise b u t have the advantage of simplicity.
B a r r a t t ' i made use of t h e fact t h a t a working curve of a sort can
be established with a series of samples of known concentrations over
the entire range of 1 to 100 per cent. His method is of value when
only one component of a system is being varied. Like other highconcentration methods, it suffers in comparison with wet gravimetric
methods, whose precision increases with the concentration of material
available, whereas t h a t of the spectrographic method remains essentially constant a t all concentrations.
T h e B a r r a t t method, which can be used either photographically or
visually, rests on t h e measurement of the relative intensities of a
spectrum line produced by two different sources, using a standard
specimen in one source a n d a specimen of t h e substance t o be analyzed
in the other. E v e n a t such high concentrations as 50 per cent, a
definite relationship exists between the ratios of the quantities of the
elements in the specimens and the relative intensities of their spectrum
lines if proper lines are chosen. However, this relationship is by no
means t h e simple one t h a t holds at low concentrations, and it is
necessary t o use exactly similar conditions of excitation for t h e two
sources. Particular care must be taken in selecting lines to avoid
errors due to self-reversal and to excitation differences. A series of
alloys or mixtures must be prepared for use in t h e s t a n d a r d exposures,
but once these have been obtained, interpolation is not difficult.
" See General Reference 16.1, page 169.
464
QUANTITATIVE SPECTROCHEMICAL ANALYSIS
[§16.11
Barratt's apparatus consists of a spark gap having two sparks in
series, with means for adjusting the sparks to equal lengths. Light
from these two sources is sent through a polarizing head that is used
to decrease the intensity of one beam relative to the other, and then
through a visual spectroscope or a spectrograph. A working curve
similar to that shown in Fig. 16.6 is set up by measuring with the
instrument, for a number of standard samples, the intensity ratios of
lines' from the two sources. From this curve one can determine the
2.0
Log-ifi
1.0
•2
-I
O
<• I
+2
Logio Percentage Cpmposiiion
Fig. 16.6. Working curve obtained with the Barratt method. Ix is the intensity of a line in the unknown element observed in one spark gap, and /» is the
intensity of the same line from a standard sample observed simultaneously in
the other spark gap.
percentage composition of an unknown sample excited in the spark.
To obtain precise results, it is necessary to provide means whereby
the substance to be examined can be made to produce radiation that
is truly representative of its condition; production of such radiation is '
often difficult. The precision claimed is about 5 per cent of the,
quantity of material present.
A second method of measuring high concentrations involves dilution, so that high concentrations become low. If we have samples
containing about 20 per cent of cadmium in, zinc, we can introduce
this cadmium-zinc alloy as an impurity in a'new matrix of graphite
powder, into some metal such as zinc or copper, or into a spectrographically neutral salt such as ammonium sulfate. The concentration of the zinc having been reduced from 20 per cent to, say, 0.2 per
cent by a dilution factor of 100, zinc is now determined in'the ordinary way. The limitation on this method is that in diluting the
§16.13] QUANTITATIVE SPECTROCHEMICAL ANALYSIS
465
sample we also dilute the precision of the results. For this reason,
chemical wet methods are usually found more precise than spectrographic methods on concentrations greater than 5 per cent.
16.12. Methods for Extremely Low Concentrations. Quantitative spectrographic analyses can be carried out at any concentration
at which qualitative analyses can be made, the ultimate lines being
used at the lowest concentrations. The table of sensitivity limits
given in § 15.1 lists the lowest concentrations that have been obtained
in quantitative spectrographic analyses for the various elements. As
pointed out there, it is often possible, by giving attention to certain
factors, to extend sensitivity limits in a given case by several orders
of magnitude below those obtained in ordinary analyses.
If, for example, one is interested in the determination of boron in
steel, burning in the ordinary arc of a small chunk of steel will make
possible a determination of boron in concentrations down to perhaps
1 part in 10^. I t has been possible to make analyses down to concentrations of a few parts in 10* by (a) more efficient feeding of the boron
from the sample into the excitation stream; (b) using the method of
fractional distillation discussed in Chapter 15; (c) introducing a
carrier as described by Scribner and MuUin^^ (§ 15.6); (d) selecting
proper excitation conditions in the source to bring out the lines of
boron and make them more sensitive; (e) converting the boron in the
source into a form more likely to be volatilized at the proper part of
the burning cycle, as revealed by a moving-plate study (§ 15.6); and
(f) using- a spectrograph with such high resolution that the background does not encroach rapidly on the line (§ 15.6). Most of the
methods of increasing sensitivity discussed in Chapter 15 were, in
fact, originally developed for use in quantitative analysis at very low
concentrations.
16.13. High-precision, Rapid, and Short-cut Methods. When
truly representative homogeneous samples are prepared, spectrographic analyses can be carried out with a precision of ± 5 per cent
of the minor constituent, from about 5 per cent concentration down
to the lowest concentrations determinable. However, special attention, particularly to the preparation of the sample, the constancy of
excitation conditions, and the accuracy of the photographic photometry, will result in increased precision. Probably the greatest uniformity claimed by any set of workers is that described by Duffendack
12 B. F. Scribner and H. R. MuUin, Jour. Res. Nat. Bur. Standards, 37, 369 (1940).
466
QUANTITATIVE SPECTROCHEMICAL ANALYSIS
[§16.13
and his collaborators, in which reproducibility to 1.4 per cent was
found in a long series of measurements m a d e under carefully controlled conditions.
When t h e highest precision is not required, analyses can be m a d e
quickly b y using t h e length-of-line method of photometry, described
in § 13.11, t h u s avoiding the necessity of using a densitometer. A
rotating disk with logarithmic opening is used, and the length of a
line is t h e n a measure of the logarithm of its intensity. Lines of a
pair being compared usually lie close together in the spectrum.Since the length of each of the two lines is proportional to the logar i t h m of its intensity, t h e difference in their lengths is a direct measure
of t h e difference in the logarithms of their intensity values.
1.00
This difference equals the loga[0.75
r i t h m of the intensity ratio of t h e
[o.50 LOQIQI
two lines, which is plotted as one
0.25
of the variables in the working
Lo 00
curve. T h u s in plotting the curve
'
^
it is possible, to use directly the
Fig. 16.7. Determination of inten- difference in position of the end
sity ratio of two lines by the difference
. ,
» ,,
,
,.
in their lengths obtained with a POi^it^ ^^ ^^^ ^ ^ ^ l^^ies, as mealogarithmic sector.
sured in Fig. 16.7. This method
should not be used where precision greater t h a n ± 10 per cent is desired. T h e limitations on precision
arise from t h e difficulty of determining the end point of a line. M e t h ods of etching and copying have been suggested for sharpening this
end point, b u t all share the basic difficulty t h a t the end point is always determined on a portion of the calibration curve of the plate t h a t
has very small slope, and therefore its location is actually indefinite.
Fagioli'' designed an a t t a c h m e n t for a reading comparator t h a t
splits the field of view into two sections parallel to the spectrum lines, '
these sections being movable relative to one another. T h e field is
split b y p u t t i n g two glass plates 4 m m thick between the objective a n d
the plate, with an arrangement for tilting these symmetrically in
opposite directions by means of a handle./ By this means t h e two
lines can be p u t side by side in such a position t h a t their densities
m a t c h all along their lengths, and the motion necessary to accomplish this matching is a measure of their intensity ratio.
" O . Fagioli, Nuovo Cimmlo. 13, 11 (1936).
§16.15]
QUANTITATIVE SPECTROCHEMICAL ANALYSIS
467
I n much analytical work, semiquantitative results are sufficiently
precise, and direct eye comparison of spectrum lines can often be
made to give useful values. T h e densities of the lines used to produce
the working curve can be judged, and the intensities of t h e corresponding lines in the samples to be determined can be quickly assigned to
locations between two known densities. I n this way, after some
experience, it is possible t o make a surprisingly accurate interpolation
t h a t m a y well come within d=25 per cent of the a m o u n t of material
present. T h e lines m u s t be photographed under controlled conditions, however.
16.14. Methods for Special Elements. When analyses for elements not ordinarily considered susceptible to spectrochemical
analyses must be m a d e spectrographically, special methods m u s t be
used. T h e methods of preparation and excitation required are
similar to those discussed in § 15.7, but the prime need is careful
consideration of the theoretical basis of the controlled excitation of
atoms to emit radiation.'^^ T h e primary problem involved in working
quantitatively with such trace elements as sulfur or chlorine is the
production under controllable conditions of spectrum lines a t the
concentrations desired.
16.15. Photoelectric Methods of Analysis. Increasing use of
spectrochemical methods in industry has caused much attention to
be given to the development of more rapid methods t h a n those
involving photography. Sawyer and his collaborators,^^ in addition
to improving the precision of the analytical process, have done much
to speed it up by shortening the time of plate processing and in other
ways. T o bring still closer the day of the a u t o m a t i c analytical
recorder, photoelectric methods of photometry have been applied to
problems of q u a n t i t a t i v e analysis (see § 12.17).
T h e application of the Geiger-Miiller counter to spectrochemical
analysis has been investigated by Duffendack and M o r r i s , ' ' by
H a n a u and Wolfe,'^' and b y Nahstoll and Bryan.'^ M a n y workers have
" W . F. Meggers, Jour. Opt. Soc. Am., 31, 39 (IWl); G. R. Harrison, Metals and
Alloys, Nov. 1930.
i^H. B. Vincent and R. A. Sawyer, Metal Progress, 36, 35 (1939); Jour. Opt. Soc.
Am., 32, 686 (194.2); R. A. Sawyer and H. B. Vincent, Jour. Opt. Soc. Am., 31, 47
(1941); H. H. Grossman, R. A. Sawyer, and H. B. Vincent, ibid., 33, 185 (1943).
i«0. S. Duffendack and W. E. Morris, Jour. Opt. Soc. Am., 32, 8 (1942).
I ' R . Hanau and R. A. Wolfe, Jour. Opt. Soc. Am., 37, 989 (1947); ibid., 38, 377
(1948). ^
i» G. A. Nahstoll and F. R. Bryan, ibid., 37, 990 (1947).
468
QUANTITATIVE SPECTROCHEMICAL ANALYSIS
[§16.15
investigated the use of photocells and amplifiers, and photomultipliers, in the direct measurement of the intensities of strong spectrum
lines. Dieke and Crosswhite have described the adaptation of a large
grating spectrograph in a Wadsworth mounting to photoelectric
measurement of emission spectra, as discussed in § 12.17.
Vertical
deflection
Fe Cu Mn Cr
(0)
Fe Cu Mn Cr
- (b)
Vertical
deflection
Fe Cu Mn Cr
(c)
Fe Cu Mn Cr
(d)
Fig. 16.8. Oscilloscope patterns obtained by Dieke and Crosswhite in the
analysis of steel samples for copper, manganese, and chromium. A weak iron ,
line serves as a reference, (a) Pattern fori electrolytic iron, (b) Chrome-molybdenum steel: Cu, 0.06%; Mn, 0.65%; Cr, 0.9.1%. (c) Manganese steel:
Cu, 0.12%; Mn, 1.4%,; Cr, 18%. (d) Stainless steel: Cu, 0.59%o; Mn, 4 . 1 % ;
Cr, 2 0 % .
'
,
.
- /
.
Dieke and Crosswhite have also described '•^ a photoelectric method
of spectrochemical analysis in which an oscilloscope is used as the
indicating device, with a 931 A or 1P28 RCA photomultiplier tube
placed at each spectrum line to be measured. Sensitivity was found
ample for reasonably strong lines. In analyzing a steel sample for
copper, manganese, and chromium, for example, suitable strong lines
" G. H. Dieke and H. M. Crosswhite, Jour. Opt. Soc. Am., 36, 192 (1946).
§16.15]
QUANTITATIVE SPECTROCHEMICAL ANALYSIS
469
of each element were selected, together with a n iron line t o serve as
control line. A rotating switch was used t o connect t h e o u t p u t of
each t u b e successively to the oscilloscope for 0.4 sec. T h e time scale
of the. oscilloscope was synchronized with a rotating switch so t h a t
the repeated traces of each line always fell on the screen in the same
position. A small condenser was inserted in the circuit to increase
the time constant, t h u s eliminating the influence of short-period
fluctuations.
Any changes in light intensity caused t h e whole p a t t e r n
to fluctuate up and down on the screen, so the traces were evaluated
in terms of the deflections for the iron line. T h e p a t t e r n s obtained
with four different steel samples b y Dieke a n d
Crosswhite are shown in Fig. 16.8.
/
I n a second m e t h o d introduced b y Dieke
,•' ^
and Crosswhite^'' t h e y caused the iron line to
/ ,•''
produce a ho izontal deflection and the other
/'y ,,''''
^^
lines vertical deflections on the oscilloscope.
ji^''"'''
Traces were t h e n obtained as shown in Fig.
'
16.9. I n these t h e slope of each line gives a
^'^- ^f;^"
?f/"°:
^
.
J. 1
scope pattern obtained
measure Oi the concentration of the cor- with standard element
responding material. A suitable transparent
o° horizontal axis and
scale can be a t t a c h e d t o the screen to make ^k'lown elements (Cu,
.
.
Mn) on vertical axis,
direct q u a n t i t a t i v e determinations possible. An AC source is used.
A photoelectric spectrometer with auxiliary The sample is the same
equipment to integrate the photocurrents pro- ^^^%f^ ^'g^^J° P""^""duced by spectrum lines, designated t h e
Quantometer, has been described by Hasler and Dietert.^" Commercial instruments of this sort are furnished by the Applied Research
Laboratories. I n this instrument, which is basically a small grating
spectrograph, u p to 12 electron-multiplier tubes are employed t o
detect t h e appropriate spectrum lines for t h e various elements t o b e
determined. While t h e sample is burned in t h e source the photocurrents generated in t h e various multipliers are amplified and t h e n
in effect used to drive small motors attached t o mechanical counters.
T h e source is kept in operation until the counter operated by t h e
photocurrent from t h e " s t a n d a r d " or control line reaches a preset
reading. T h e readings of the counters attached to the other tubes
then correspond to the. percentages of the various elements being
measured. Various p a r t s of the Quantometer are shown in Fig. 16.10.
2« M. F. Hasler and H. W. DieLert, Jour. Opt. Soc. Am., 34, 751 (1944).
470
QUANTITATIVE SPECTROCHEMICAL ANALYSIS C§16.15
Another direct-reading spectrochemical installation, which is
manufactured commercially by Baird Associates, has been described
by Saunderson, Caldecourt, and Peterson.^' A 2-meter concave
grating in an Eagle mounting is used, and electron multiplier phototubes are mounted behind appropriate exit slits in the focal plane to
measure the intensity of a spectral line from each element being
determined. The currents from the multiplier phototubes are col-
^ig. 16.10.
The Applied k e s e a r c h Laboratories' Quantometer,
lected in condensers, which are then discharged through resistors
whenever they reach a certain voltage. The number of discharges
occurring during a controlled period of operation is made to operate
a "clock" for each element, which is calibrated directly in percentage
concentration.
Although such devices give analytical results very quickly and are
sufficiently precise for most purposes, their complexity and cost make
^' J. L. Saunderson, V. J. Caldecourt, and E. TV. Peterson, Jour. Opt. Soc. Am., 35,
681 (1945); R. O'B, Carpenter, E. DuBois, and J. Sterner, ibid., 37, 707 (1947).
§16.15]
QUANTITATIVE SPECTROCHEMICAL ANALYSIS
471
t h e m of value principally in large installations, as in foundries, where
thousands of analyses must be made per week. Development of the
iconoscope or the image dissector tube to the point of usefulness for
precision measurements m a y make possible recording analytical
installations t h a t do not require the complexity of a n additional
phototube and circuit for each spectrum line studied.
GENERAL REFERENCES
16.1.
16.2.
16.3.
16.4.
16.0.
16.6.
16.7.
F. Twynian, The Spectrochemical Analysis of Metals and Alloys.
Brooklyn: Chemical Publishing Company, Inc., 1941.
D. M. Smith, Metallurgical Analysis hy the Spectrograph. London:
British Non-ferrous Metals Research Association, 1933.
W. Gerlach and E. Schweitzer, Foundations and Methods of Chemical
Analysis hy the Emission Spectrum. London: Adam Hilger, Ltd.,
1931.
H. Lundeg&rdh, Die quantitative Spektralanalyse der Elemente, I (1929);
II (1934). Jena: Fischer.
W. R. Erode, CAemicaZ Sjoerfroscop?/, 2d ed. New York: John Wiley
& Sons, Inc., 1942.
H. A. Sawyer, Experimental Spectroscopy. New York: Prentice-Hall,
Inc., 1944.
H. Mark, in Spektroskopische und radiometrische Analyse, Teil I.
Leipzig: Akademische Verlagsgesellschaft, 1933.
See also General References 15.4 to 15.19, inclusive.
CHAPTER
17
Spectroscopy of the Infrared Region
T H E INFRARED REGION OF THE SPECTRUM may be taken to include
the wavelength range from 0.75 n to about 1 mm and may be subdivided, on the basis of the instrumental techniques appropriate to
each subdivision, into the photoelectric infrared, the near infrared,
and the far infrared.
The photoelectric infrared covers the approximate range 0.75 to
3 IX. The lower boundary is set by the wavelength-sensitivity curve
of the human eye. Except for use of the eye, radiation of the these
long wavelengths can be detected and measured in the same way a:s
visible radiation, and the use of spectrographs suitable for the visible
region, both prism and grating, can be extended well into the photoelectric infrared. Photographic emulsions and photoelectric cells
can be made with a usable sensitivity over much of the region. At
the longer wavelengths, however, photoelectric processes—including
that of the photographic emulsion—begin to lose their sensitivity.
Not far beyond -3 n, the detection of radiation is best accomplished
through its heating effect, and spectrometric techniques are modified
accordingly.
^
The near infrared might be called the "prism infrared," because
prism materials transparent to the region are readily obtainable.
Most near infrared spectrometers' have prism optics, though gratings
are also used occasionally. The long wavelength limit to the near
infrared is set by the transmission of readily obtainable prism materials and lies at about 25 fj., where the absorption of potassium bromide
gets prohibitively large. To be sure, materials are known that are
transparent beyond 25 /i, but for reasons discussed later, the reflection
grating is usually^pteferred at th^e wavelengths.
The far infrared extends from 25 n to the ill-defined borderland
between the^infrared and microwaves (radar waves) in the neighborhood of 1 mm. Throughout the far infrared the most generally used
472
§17.1]
SPECTROSCOPY OF THE INFRARED REGION
473
dispersing element is the echelette grating, that is, a grating whose lines
have been so ruled that radiation of a given wavelength is largely
concentrated in one order (see § 2.5). Other techniques such as
residual rays may be applied in special cases. Far infrared radiation,
in default of more sensitive methods, must be detected by its heating
effect. Because the energy of infrared sources is greatly reduced in
the far infrared and the problem of extraneous radiation is much more
acute, the precautions required to obtain accurate radiation measurements are more elaborate than those used at shorter wavelengths.
It is an interesting coincidence that this subdivision of the infrared
agrees in a general way with the division of molecular energy levels
into electronic, vibrational, and rotational levels, as discussed in
Chapter 11. The lower electronic levels and a few higher vibrational
overtones lie in the photoelectric infrared, most vibrational fundamentals lie in the near infrared, and the frequencies of molecular
rotation lie in the far infrared. The fact that there is no natural
boundary for atomic or molecular energy levels in the 0.75 ju region
emphasizes the artificiality of segregating the photoelectric infrared
from the visible portion of the spectrum. The ensuing discussion of
the techniques of infrared spectroscopy will be concerned only with
the near and far infrared, the techniques of the photoelectric infrared
being more closely similar to those considered in previous chapters.
17.1. Radiation Sources and Filters for Infrared Spectroscopy.
All but a small fraction of infrared spectroscopic studies are concerned
with absorption spectra, for which a source of continuous infrared
radiation is needed. Incandescent solid bodies at temperatures of
1000 to 1500°C meet this requirement best. Such bodies emit radiation roughly in accordance with the Planck blackbody equation (see
Chapter 8) and therefore emit radiation throughout the infrared.
The fact that the intensity of the radiation varies somewhat rapidly
with wavelength is an inconvenience, especially at the longer wavelengths. When quasi-monochromatic radiation is wanted, the easiest
procedure usually is to select the desired wavelengths from the
continuum of an incandescent source. Among the many devices
that have been used for this purpose are selective filters, selective
reflectors and focal isolation.
A wide variety of substances have been used for infrared filters.
Powder filters, as described by Pfund,' and Christiansen filters^
1 A. H. Pfund, Phys. Rev., 36, 71 (1930); Jour. Opt. Soc. Am., 23, 375 (1933).
2 R. B. Barnes and L. G. Bonner, Phys. Rev., 49, 732 (1936).
474
SPECTROSCOPY OF THE INFRARED REGION
[§17.1
(§ 14.30) are particularly flexible with respect to the wavelength
ranges to which they are applicable. Soot-blackened paper is sometimes used for its transmissivity at long waves and opacity in the
near infrared region. In general, filters are more valuable for the
elimination of undesired radiation, especially short-wave radiation,
than for the isolation of a narrow range of wavelengths.
It has been known for many years^ that crystals possess reflecting
powers approaching those of metals at characteristic wavelengths in
the infrared region. For any given crystal the high reflectivity is
confined to one or two narrow characteristic bands in the spectrum.
Radiation corresponding to these bands can be isolated from a continuous source by the successive reflection of the continuous radiation
from plane surfaces of the crystal.^ Because the reflectivity of a
crystal at other than the characteristic wavelengths is very small,
three or four successive reflections leave a residue of only the characteristic wavelengths, whence the name residual rays.
A selective reflector of a different type has been suggested by
White.^ The reflector makes use of the fact that a plane grating gives
specular reflection of wavelengths longer by a factor of roughly 1.5
than the grating spacing, and disperses shorter wavelengths at angles
considerably different from the specular. Such a grating can thus
be used as a selective "cutoff" reflector, the cutoff wavelength being
about 1.5 times the grating space. Echelette gratings are usually
employed for this purpose because they put very little short-wave
radiation into the specularly-reflected central image. This type of
filter, for which replica gratings are quite suitable, has been successfully applied to the reduction of scattered radiation in infrared
spectrometers. Reduction by as much as a factor of 10 has been
realized with one filter, and with several in series the possibilities are
even more favorable (compare Figs. 5 and 10 given by White*).
Another ingenious method for_ selecting a narrow range of wavelengths in the far infrared makes use of the transparency of quartz
beyond 50 /x. The refractive indices of quartz below 8 fi and jibove
30 fi are so markedly different (for example, 1.5 at 3 n, 1.4 at 5 n, and
varying from 2.13 at 33 /x to 1.94 at 300 iJ) that the conjugate foci
of a quartz lens have greatly differing values for the near and the far
5 E. F. Nichols, Wiedemanns Annalen, 60, 401 (1897).
* See, for example. General Reference 17.5, page 383.
* J. U. \Yhite, Jour. Opt. Soc. Am., 37, 713 (1947); replica grating filters may be
purchased from the Perkin-Elmer Qorp., Glenbrook, Conn.
§17.1]
SPECTROSCOPY OF THE INFRARED REGION
475
infrared. By placing a small source rich in far infrared radiation at
one conjugate focus for a lens of assumed index 2.2 and an opaque
screen at the other conjugate focus, Rubens and Wood ^ were able to
bring the long waves to a focus at the screen, while the short waves
were not convergent. Perforation of the screen at the focal point
allowed the passage of the focused radiation, after which its approximate wavelength was measured interferometrically to be about 107 juThis procedure of "focal isolation" would appear to be useful for study
of the borderland between infrared and microwaves, particularly
with lenses of other materials in addition to quartz.
The two most commonly used infrared sources are the Nernst
glower and the Globar, but occasionally other sources have some
special applicability. The Welsbach lamp and a quartz-jacketed
high-pressure mercury arc have been used in the region beyond 50 n.
In the photoelectric infrared it is occasionally of advantage to utilize
a tungsten filament lamp as a source, though necessarily within the
region over which the envelope is transparent.
The Nernst glower (§ 8.8) was originally developed by its inventor
as an incandescent light. It is an excellent infrared source because
of its high emissivity and its simplicity of construction and of operation. The glower consists of a filament prepared ' by sintering a
finely powdered mixture of various oxides, particularly those of
zirconium, thorium, and cerium, held together with a binder. The
filament is maintained at incandescence electrically, for which purpose
platinum leads are attached to it with appropriate sealing techniques.
When the filament is operated at a temperature above 1500°C, its
emission curve resembles that of a blackbody fairly closely. At lower
temperatures, the short-wave end of the emission curve is quite
irregular because of the selective emission of the metallic oxides.
The filament of the Nernst glower has a large negative temperature
coefficient of electrical resistance and is customarily heated by flame
or other external means to lower its resistance at the start of operation.
At a temperature of several hundred degrees centigrade, this resistance
is low enough to pass sufficient current at the operating voltage to
bring the lamp to incandescence. It is then necessary to ballast the
lamp by a voltage regulator or series resistance to keep it from burning
" General Keference 17.6, page 533.
' H. D. Griffith, Phil. Mag. (6) 50, 263 (1925); R. A. Friedel and A. G. Sharkey, Jr.,
Rev. Sci. Inst., 18, 928 (1947).
476
SPECTROSCOPY OF THE INFRARED REGION
[§17.1
out. Typical operating data for a Nernst filament used as an infrared
source are shown in Table 17.1.
TABLE 17.1
DATA ON NERNST GLOWER AND GLOBAB INFRARED SOUHCES
Specifications
Length of element, centimeters
Diameter of element, millimeters
Operating potential drop, volts
Current, amperes
Power, watts
Approximate temperature, °C
Wavelength of radiation peak, microns
Nernst lamp
Globar
2-5
1.5
About 100
0.2-0.5
20-50
1600-1700
1.4
5-10
6
50-60
3-5
150-250
1200
1.8
The Globar (§ 8.8) is a rod of carborundum (silicon carbide). Its
temperature coefficient of electrical resistance is negative but rather
small. It conducts sufficiently in the rod sizes customarily used as
infrared sources to require no external preheating. Ballasting is not
necessary with a Globar to prevent its burning out, but some sort of
voltage regulation is desirable to maintain constancy of radiation.
The Globar surface is very rough, which accounts in part for the
excellence of its radiance. The attachment of electrical leads to the
Globar is not a critical problem because of its ruggedness. Usually
the ends of the rod are metallized and fitted into metallic sockets
which serve as electrodes. Operating data are shown in Table 17.1.
The choice between the Nernst glower and Globar as an infrared
source depends on the spectral region in which radiation is wanted. *
The advantages of the Nernst glower are its low operating wattage
and high intensity in the short-wave region. The Globar is more
useful at longer wavelengths (>10 /u), where it has considerably more
energy relative to its short wavelength peak than the Nernst glower.
The latter is physically smaller, and an enlarged image of it may be
needed to fill completely the slits of a spectrometer, especially when
these are opened wide a t the longer wavelengths. The large amount
of heat dissipated by the Globar usually necessitates a water-cooling
arrangement.
* The Globar is made by the Carborundum Corporation, Niagara Falls, N. Y.
The Nernst glower can be obtained from the Stupakoff Ceramic and Manufacturing
Company, Latrobe, Pa., or from National Technical Laboratories, South Pasadena,
Calif.
§17.2]
SPECTROSCOPY OF THE INFRARED REGION
477
17.2. Prism Spectrometers for the Infrared. The optics of both
prism and grating infrared spectrometers are fundamentally the same
as those of spectrometers described in Chapters 3 and 4. However,
all infrared spectrometers are monochromators and all use mirror
optics. The monochromator arrangement is dictated by the nature
of infrared detectors, and mirror optics by the opacity of glass in the
infrared and the wide spectral range to be covered, which makes it
infeasible to construct suitable lenses of any material whatever.
Prism instruments are the most widely used.. They are quite
satisfactory in the near infrared, are relatively simple to operate, and
when well designed give adequate resolving power for most purposes.
Grating instruments must be used in the far infrared because of the
lack of suitable prism materials. They have the advantage of high
resolving power and are used in the near infrared when high resolution
is essential. Some of the complications inherent in grating instruments are considered below.
Table 17.2 shows the wavelength ranges over which various prism
materials are reasonably transparent. Rock salt is the most widely
used of these. It is transparent to about 15 /x, but below 5 n its
TABLE 17.2
INFRARED PRISM AND WINDOW MATERIALS
Material
Glass
Quartz
Lithium fluoride
Calcium fluoride (fluorite)
Sodium chloride (rock salt)
Potassium chloride (sylvine)
Potassium bromide
Silver chloride*
Thallium bromoiodide (KRS-5)
Region of usable
transparency
(prisms)
Wavelength of
residual rays,
microns
Up to i.Sfi
3.5M
6M
9M
15M
21M
27M
28M
40M
9; 21
20
33
52
63
82
120
' Windows only.
dispersion is poor, and other prism materials must be used if the best
performance of a prism instrument is to be obtained at the short
wavelengths. Fluorite and lithium fluoride are excellent here. For
the range from 15 to 25 n, potassium bromide is suitable, ,and from
478
SPECTROSCOPY OF T H E I N F R A R E D R E G I O N
[§17.2
25 to 40 n, thallium bromoiodide.' All these materials are commercially available in blanks for window or prism fabrication.*
The traversal of the infrared spectrum by a prism monochromator
is accomplished by the rotation of the dispersing element in such a
way that the dispersed wavelengths are brought successively to a
focus at the exit slit. The two most widely used optical arrangements
for this purpose are the Wads worth and the Littrow mountings (see
Chapter 3). In the Wadsworth mounting, the prism and the plane
mirror, the latter usually lying in such a position as to constitute an
extension of the prism base, are rotated together about a vertical axis
through the center of the prism base. Under these conditions the
angular deviation of a monochromatic beam traversing the prism at
minimum deviation is exactly nullified by reflection from the mirror,
although lateral displacement and inversion of the beam result. The
lateral displacement, which is equal to the length of the prism edge
for a 60-deg prism, is practically independent of the wavelength.
Accordingly, when optical arrangements have been made for radiation
of one wavelength to traverse the prism at minimum deviation and
come to a focus at the exit slit, the whole spectral range of the prism
may be scanned at minimum deviation merely by rotation of the
prism-mirror combination about the above-mentioned axis.
In the Littrow mounting, the plane mirror is placed roughly normal
to the dispersed beam emerging from the prism. For one particular
wavelength, which varies of course with the orientation of the mirror,
the emergent beam is reflected exactly upon itself and retraces its path
through the prism. The return paths of slightly longer and shorter
waves, however, differ from their initial paths, and consequently
further dispersion occurs. The spectrum may be scanned simply by
rotation of the Littrow mirror about a vertical axis. If the prism!
remains fixed with respect to the ihcoming beam, however, only one
wavelength passes through the prism at minimum deviation.
The two chief advantages of the Littrow mounting are the doubled
dispersion obtained from a given prism train, and the compactness
and economy that result from the use of the, same concave mirror- as
a collimator and as a focusing mirror. To illustrate the latter point,
several optical systems frequently used in infrared spectrometers are
8 0 . F. Tuttle and P. H. Egle, Jour. Chem. Phys., 14, 571 (1946); E. K. Plyler,
Jour. Chem. Phys., 15, 885 (1947).
*Purchasable from the Harshaw Chemical Cortfpany, Cleveland, Ohio, or from the
Optavac Company, 112 Bickford Street, Boston 30, Mass.
§17.2]
SPECTROSCOPY OF THE INFRARED REGION
479
shown in Fig. 17.1. ' Figure 17.1a shows an arrangement using the
Wadsworth mounting. Si, 1S2 are the entrance and exit slits; C, the
collimating mirror; F, the focusing mirror; and P, M„, and A, respectively, the prism, plane mirror, and axis of rotation (perpendicular
to the plane of the sketch) of the Wadsworth system. C and F can
be either spherical or off-axis parabolic mirrors. Parabolic mirrors
avoid astigmatism and spherical aberration but are more difficult to
make than the spherical. It is to be noted that two are required.
Figure 17.1b depicts a widely used form of the Littrow mounting
in which the double-duty mirror OP is an off-axis paraboloid. The
beam returning from the Littrow mirror ML is indicated by dotted
lines, which show that the return beam is made to deviate slightly
from the initial path. With the help of a small plane mirror, M2,
the return beam is deflected just before it reaches the entrance slit,
Si, and instead is brought to a focus at the exit slit, S2. The spectrum
is scanned by rotation of ML about a vertical axis at A.
When an on-axis paraboloid is used in the Littrow arrangement,
the rather large angle between the collimated beam and the beam
from the slit Si. results in considerable astigmatism. An ingenious
variant (Fig. 17.1c) of the Littrow scheme was suggested by Pfund '
to avoid astigmatism from an on-axis paraboloid. The essential
feature is the Pfund mirror, Mp, a plane mirror with a small central
hole. This mirror is placed immediately behind the entrance and
exit slits. Si and S2, which are located respectively just above and
just below the optic axis, OA, of the on-axis paraboloid mirror C
OA passes through the center of the small hole in Mp, the hole being
large enough so that the entering beam diverging from Si and the
return beam focused on S2 have free passage. The two beams are
not shown separately, since their projections upon the plane of the
drawing coincide. The light losses associated with two reflections
from Mp and with the hole in Mp can be made small and are a low
price to pay for the high optical quality gained by the on-axis use of C.
The Pfund mirror introduces additional stray radiation, the effect of
which may be minimized by blocking out the unused central portion
of C so that radiation from this portion does not reach the exit slit S2.
The complete optical scheme for a prism spectrometer includes
some means for focusing the radiation to be studied on the entrance
slit. A concave spherical mirror usually is sufficient for this purpose.
' A. H. Pfund, Jour. Opt. Soc. Am., 14, 337 (1927).
480
SPECTROSCOPY OF THE INFRARED REGION
[§17.2
(b)
Fig. 17.1. Typicar optical systems for infrared prism spectrometers,
Wadsworth mounting, (b) Littrow mounting with off-axis paraboloid,
Littrow mountingVwith Pfund pierced-mirror arrangement.
(a)
(c)
§17.2]
SPECTROSCOPY OF THE INFRARED REGION
481
F o r absorption spectra, the incandescent source is focused on t h e slit
with a mirror whose focal length allows adequate separation of source
from mirror and ample room for absorption cells, shutters, and the like
t h a t must be placed in t h e beam. T h e beam emerging from the exit
slit of t h e spectrometer is commonly focused on t h e t h e r m a l detector
b y means of a n elliptical mirror. This t y p e of mirror concentrates a
reduced image of the exit slit on the detector with great efEciency
but is quite expensive t o m a k e ; in m a n y instruments a spherical
mirror of appropriate aperture would serve a b o u t as well.
A prism 85 in. high, having a 3-in. base, is a size commonly used in
infrared spectrometers. W i t h a 60-deg rock-salt prism of this size
used in the optical arrangement shown in Fig. 17.1b, a resolving
power of about 200 should be realized a t 5 ju and a b o u t 400 a t 14 /i.
These values correspond t o wave-number separations of 10 cm""^ and
2 cm~^ respectively. T h e resolving power of a spectrometer a t short
wavelengths is governed by the excellence of the optics (optical
quality of the off-axis paraboloid, alignment, and similar factors),
whereas a t t h e longer wavelengths t h e optical arrangement is n o t so
critical. There t h e limiting factor is the small a m o u n t of energy
available from the source, which forces t h e use of wide slits and a
consequent reduction in resolving power.
T h e wavelength calibration of a prism instrument consists of a plot,
on an adequately large piece of graph paper, of the prism-table or
Littrow-mirror orientation as a function of the wavelength brought t o
focus on the exit slit. T h e orientation m a y be expressed in any convenient fashion, which means t h a t the ordinate is usually in t e r m s of
screw turns, counter readings, fiducial-mark numbers, or some other
index of prism-table setting. T h e calibration curve is ordinarily
determined ^^ with t h e help of the precise wavelengths for sharp
vapor-absorption bands in certain simple molecules such as ammonia
(3 and 8-to-12 ix regions), water (6 n), carbon dioxide (4f a n d 15 11),
a n d methanol (20 n region). T h e wavelengths of t h e absorption
bands in these substances were originally measured "• ^^ with grating
spectrometers (described in §17.3).
Until quite recently the setting up of a prism spectrometer for
infrared involved both a long apprenticeship on the p a r t of t h e
"
222
"
^
See, for example, D. S. McKinney and E. A. Friedel, Jour. Opt. Soc. Am., 38,
(1948).
R. A. Oetjen, Chan-Lan Kao, and H. M. Randall, Rev. Sci. Inst, 13, 515 (1942).
A. Borden and E. F. Barker, Jour. Chem. Phys., 6, 553 (1938).
482
SPECTROSCOPY OF THE INFRARED REGION
[§17.3
operator, preferably in a laboratory where infrared research was in
progress, a n d t h e construction of an i n s t r u m e n t on a made-to-order
basis in an instrument shop. There are now on the market, however,
several prism instruments t h a t can be set u p a n d p u t into operation
by those who h a v e had no previous infrared experience. Figures 17.2
a n d 17.3 show infrared prism spectrometers m a r k e t e d respectively b y
t h e Perkin-Elmer Corporation, Glenbrook, Conn., and the National
Technical Laboratories, South Pasadena, Calif. Both of these ins t r u m e n t s use t h e Littrow arrangement shown in Fig. 17.1b, high-
Fig. 17.2. Perkin-Elmer Model 12C infrared spectrometer.
(Courtesy Perkin-Elmer Corp.)
speed t h e r m a l detectors, and a commercial pen recorder. T h e
resolution attained by both instruments is good for prism instruments
of this size a n d is satisfactory for most industrial control uses a n d for
m a n y research purposes.
17.3. Grating Spectrometers. Infrared wavelengths are of such
size t h a t t h e o p t i m u m grating spacing is of the order of 0.01 m m in
t h e near infrared and 0.1 m m or more in the far infrared. I t is
therefore possible to make transmission gratings for the latter region
b y winding wire of the appropriate diameter upon a flat frame. Such
gratings, which were once extensively used, h a v e been superseded b y
li
§17.3]
SPECTROSCOPY OF THE INFRARED REGION
483
ruled gratings of the echelette t y p e (see § 2.5). Echelette gratings,
because of their ability to concentrate most of the available radiation
I
r
Fig. 17.3. National Technical Laboratories Model rR-2 infrared spectrometer.
This instrument is often called the Beckmann infrared spectrometer. (Courtesy
National Technical Laboratories.)
into one order, have opened u p spectral regions t h a t could h a v e
scarcely been studied otherwise.
484
SPECTROSCOPY OF THE INFRARED REGION
[§17.4
It was said earlier that gratings are used in the near infrared when
high resolution is indispensable, and throughout the far infrared
because of the lack of prism materials. The optical arrangement of
the grating spectrometer proper does not differ greatly from the
Littrow mounting for prism instruments shown in Fig. 17.lb and c, in
which the prism and Littrow mirror combination is replaced by a plane
grating. Grating instruments of these two types have been described respectively by Randall '^ and J. D. Hardy." In the former,
with which much work on rotational absorption of small polyatomic
molecules has been carried out, the off-axis paraboloid is large (a
semicircle of 24 in. diameter and of 36 in. focal length) and fills a
grating 10 X 22 in. This large size makes the instrument highly
effective in the region beyond 50 n, where the radiation from the
source is extremely weak.
A difficulty in the use of gratings in the infrared is the overlapping
of different spectral orders, which is especially troublesome because
of the high source intensity of radiation at X/2, X/3, and so on, compared to that at X. The radiation must therefore be purified in some
way before it enters the spectrometer. This purification can be
satisfactorily accomplished in the near infrared by the use of a lowdispersion prism monochromator in front of the grating monochromator; but at wavelengths beyond 25 n, filters opaque to short
waves must be used. Thin plates of fused quartz are suitable for
elimination of radiation between 10 and 50 ju, and hard paraffin,
metallic blacks, specially prepared powder filters, and soot-blackened
paper have also been used. The reflection filter of White ^ mentioned
in § 17.1 can also be employed with great effectiveness.
At the present time, grating spectrometers for the infrared are not
available commercially, although there are numerous custom-built
instruments in research laboratories in the United States. Most of
these are equipped with gratings ruled on the engine at the University
of Michigan. Figure 17.4 is a reproduction of a record of the far
infrared absorption spectrum of deuterium oxide vapor made with
the grating instrument described by Randall i^ and located at the
Randall Laboratory of Physics at the University of Michigan.
17.4. The Measurement of Infrared Absorption. Because infrared spectra are nearly always studied as absorption spectra, the
technique of absorption measurements in this region will be consid" H . M. Randall Rev. Sci./Insf., 3, 396 (1932).
" J. D. Hardy, Phys. Rev., 38, 2162 (1931).
§17.4]
SPECTROSCOPY OF THE INFRARED REGION
485
ered in some detail. Methods for the measurement of spectral
radiation in the infrared region, including amplification and recording
of detector output, were discussed in Chapter 12. The special
features of infrared absorption measurement other than detection,
amplification, and recording will be described here.
Infrared-absorption measurements are carried out on substances in
any of the three states of aggregation. The techniques used in
handling the samples are not elaborate. The absorption cell is
placed in the optical path somewhere between the source and the
entrance slit, usually just in front of the latter. The cell windows
are ordinarily of polished rock salt several millimeters thick unless the
Fig. 17.4. A record of the spectrum of deuterium oxide ("iieavy water")
vapor in the spectral region 34|U to 38^. (Courtesy Prof. H. M. Kandall and
Dean II. .\. Sawyer.)
spectral region under investigation lies beyond 15 n, in which case
potassium bromide or some other suitable material is used. The cell
windows must be large enough to admit the entire coUimated beam
from the source, which means that in practice they are 2 to 5 cm in
diameter.
In the study of gases, the windows are fastened on the ends of a
glass or metal cylinder with an appropriate cement. The sample is
admitted to the cell through side tubes in the cylinder, and the gas
pressure is adjusted to give the optimum absorption. In general,
absorption spectra of gases are obtained at several pressures, because
the optimum pressure for the resolution of fine structure in one spec-
486
SPECTROSCOPY OF THE INFRARED REGION
[§17.4
tral region is usually too high or too low for best resolution in other
regions. If t h e vapor pressure of t h e substance under study is low
at room temperatures, sufficient absorption m a y be obtained by
increasing t h e length of the absorption cell or b y raising the cell
t e m p e r a t u r e . Neither of these expedients is entirely satisfactory, the
first because of t h e awkwardness of long cells a n d t h e latter because
of t h e difficulty of sealing rock salt t o other materials over a wide
t e m p e r a t u r e range. T h e length of gas absorption cells in general use
varies from a b o u t 10 to 30 cm, and the gas pressures used may run
from 3 m m or even less for strongly absorbing substances like fluorocarbons t o half a n atmosphere for weak absorbers such as hydrogen
chloride or water vapor.
T h e difference in densities between a liquid and its vapor under
standard conditions being nearly a thousandfold, a thousandfold
shorter p a t h length should be sufficient in a liquid to produce the same
infrared absorption. F r o m the figures given above, we should expect
t h a t liquid cells should have a length (usually called "thickness") of
0.1 m m or less. Actual practicable cell thicknesses run from 0.01 to
1.0 m m , t h e average being around 0.05 m m . Liquid cells usually
consist of a metallic gasket or spacer of a thickness equal to the desired p a t h length, which is placed in sandwich form between the two
rock-salt plates serving as cell windows. Liquid is p u t into the cell
either t h r o u g h a n orifice in the gasket or before the cell is assembled.
I n t h e latter procedure, one window is laid flat a n d t h e gasket is
placed upon it. A drop or so of the liquid to be studied is placed in
the shallow receptacle so formed, and the other window is then p u ^
on top of the gasket in such a way t h a t all air is squeezed out. T h e
resulting sandwich is held together b y metallic clamps. Highly
volatile liquids cannot be handled in this way, a n d must be placed in
cells b y t h e former technique. An absorption cell for volatile
liquids has been described by Gildart and Wright."^^
I n q u a n t i t a t i v e analysis of liquid mixtures b y their infrared spectra,
it is highly i m p o r t a n t t o have a cell of reproducible thickness, because
for highest accuracy the spectra of unknown and standard samples
should be measured in cells of the same thickness. T h e best means
of ensuring reproducible cell thickness is to employ a cell of a fixed
thickness t h a t is never changed, and t o use this cell for all analyses of
'^^L. Gildart and N. Wright, Rer. Sci. Inst., 12, '^04 (1911). Additional references
to construction of absorption cells will be found in this paper. See also General
Reference 17.9.
§17.4]
SPECTROSCOPY OF THE INFRARED REGION
487
a given kind of mixture. A disassembled cell will never regain
exactly the same thickness after reassembly, and therefore a cell
should not be taken apart if it is to be used again for quantitative
analyses based on working curves (§ 17.6) obtained previously. A
typical cell design is shown in Fig. 17.5.
Since the density of solids is of the same order as that of liquids, the
path length in solid media for optimum infrared absorption is about
0.1 mm. If a solid can be fabricated by rolling, melting, deposition,
and similar methods into a thin sheet or plate of this thickness, it
offers no difficulty. The optical quality of the sheet for visible
Fig. 17.5. Liquid absorption cell for infrared spectrometer.
wavelengths is unimportant. Solids that cannot be made into sheets
or plates can sometimes be studied in the form of a thin powder layer
deposited on a rock-salt plate by sedimentation. This technique has
been used by Pfund,^ Wright,^® and others. Solids are also occasionally examined in solution, but this procedure has limitations. It can
almost never be used for study of substances soluble only in water;
even for substances soluble in organic solvents, it requires the use of
two or more solvents so that the spectral regions of absorption of one
solvent can be examined in a different solvent that is transparent in
those regions. The particular solvents chosen for a given solid depend
'«N. Wright, Jour. Biol. Chem., 120, CUl (1937).
488
SPECTROSCOPY OF THE INFRARED REGION
[§17.4
on the spectrum expected for the soUd and on its solubiUties. Often
it is useful to examine a solid in the form of' a suspensoid or "mull"
in an inert liquid medium such as Nujol." As for solutions, the
regions of absorption of the medium interfere with the measurement
of the spectrum of the suspended solid (see General Reference 17.1,
page 11).
It is customary to present the results of an infrared absorption
study as a graph of per cent transmission plotted against wavelength
in microns or wave number in cm—i-. Some infrared spectrometers
produce such a record automatically and hence may properly be
called infrared spectrophotometers. Most instruments, however, produce records from which transmission curves can be computed. At
every point on these records, there must be explicit or implicit indication of the wavelength and the deflections corresponding to (a) the
transmission of the sample, (b) 100 per cent transmission, and (c) zero
transmission. AH of these except the last vary greatly through the
spectrum, but they must be known if the per cent transmission of the
sample as a function of wavelength is to be obtained.
Wavelength is generally indicated on the record by some kind of
fiducial mark made at convenient intervals during the recording.
These marks are related to wavelength when the instrument is
calibrated (see § 17.2) and their reliability may be checked easily with
reference to the well-known absorption bands of water and carbon
dioxide, or other convenient standards. The deflection for zero
transmission can be indicated as often as desired by the insertion of an
opaque shutter in front of the absorption cell, but the 100 per cgnt
transmission deflection is rather more complicated to record. This
deflection varies considerably with wavelength in approximate accord
with the blackbody curve. At any particular wavelength, moreover,
it is a function of slit width, source temperature, and other factors such
as scattered radiation within the instrument and absorption and
reflection in the optical path outside the absorbing substance. By
careful regulation of these factors, it is possible to obtain a 100 per cent
transmission curve throughout the spectrum which is sufficiently
reproducible for most purposes except those of quantitative chemical
analysis.
When the spectral record of a particular substance has been'obtained, it still has to be translated into a per cent transmission vs.
" R. B, Barnes, E. P. Williams, et al, Ind. Eng. Chem., Anal, ed., 19, 620 (194.7).
§17.4]
SPECTROSCOPY OF THE INFRARED REGION
489
wavelength curve with the help of the 100 per cent transmission curve.
The translation step is time-consuming, and various methods for its
elimination have been suggested. One of these consists of a device
for forcing the instrument to give a horizontal straight line as the
100 per cent transmission curve. Inasmuch as the slit widths have
to be changed several times during the scanning of the spectrum
anyway, a straight line for the 100 per cent transmission curve might
be obtained by varying the slit width continuously as the spectrum
is scanned, the rate of variation being governed by a cam so as to
compensate for the blackbody curve and other variables. The
difficulty here is the exceedingly fine control of slit width needed at
the short wavelengths near the peak of the blackbody curve, where
the curve is very steep and slit widths are very narrow. If electronic
amplification is used, control can also be exercised by continuous
variation of amplifier gain as the spectrum is scanned. A combination
of these two devices has been described by White.^*
The more basic and more satisfactory way to produce a record of
per cent transmission is to provide a mechanism within the spectrometer by which the actual transmission of the sample is compared
automatically and continuously with 100 per cent transmission as the
spectrum is scanned. Such a mechanism converts the spectrometer
into an automatic-recording infrared absorption spectrophotometer.
The fundamental principles of infrared absorption spectrophotometers are essentially the same as those for the visible and ultraviolet
(Chapter 14^, the differences in detail arising from differences in
sources and detectors. Various arrangements have been described
by Hardy and Ryer,"^' Wright and Herscher,^** and others.^i
The optical system of a commercial instruments^ having several
features in common with those of Hardy and Ryeri' and Wright and
Herscher-" is shown in Fig. 17.6. Radiation from a Globar source is
received on two identical spherical mirrors so located that the two
source-to-mirror beams make an angle of about 135 deg. The two
collimated beams then pass through two identical cells, one holding
the sample and the other the reference standard with respect to
which the transmission of the sample is to be measured. By defini1* J. U. White, Jour. Opt. Soc. Am., 36, 362A (1946).
18 J. D. Hardy and A. I. Ryer, Phy.i. Rev., 55, 1112 (1939).
2» N. Wright and L. W. Herscher, Jour. Opt. Soc. Am., 37, 211 (1947).
21 R. F. Wild, Rev. Sci. Inst., 18, 436 (1947).
^ W. S. Baird, H. M. O'Bryan, George Ogden, and Dorothy Lee, Jour. Opt. Soc.
Am., 37, 754 (1947).
490
SPECTROSCOPY OF THE INFRARED REGION
[§17.4
tion, the transmission of the reference cell is 100 per cent and the
instrument functions so as to determine the ratio of the transmission
of the sample to that of the reference. For example, the per cent
transmission ol a. substance in solution would be determined with
respect to a reference cell containing the solvent at a thickness equivalent to that of the solvent in the sample cell.
The beams pass through the two cells at right angles to each other
and proceed to a point of intersection at which a rotating-sector
interrupter is located. One sector of this interrupter is a plane
Fig. 17.6. The optical system of the Baird Associates' recording infrared
spectrophotometer. (Courtesy Baird Associates.)
mirror; and when this sector is at the beam/intersection, the beam
from the sample cell is reflected onto the entrance slit of the spectrometer while the beam from the reference/cell/is occulted. The other
sector is simply an opening through Which both beams pass, the
reference beani proceeding to the entrance slit and the sample beam
passing harmlessly off to one side! From the entrance slit onward,
the optical path followed by radiation of any given wavelength is
identical for bothtbeams.
§17.4]
SPECTROSCOPY OF THE INFRARED REGION
491
When the radiation a t a particular wavelength, say 10 n, is incident
on the exit slit of the spectrometer, this radiation will remain constant
with time if the 10-/U radiation from the source is constant and if t h e
reduction in intensity which results from passage through the two
absorption cell systems is the same for both. However, when t h e
sample cell absorbs more 10-;u radiation t h a n the reference cell, t h e
intensity of this radiation at the exit slit will "flicker" with the frequency of the rotating-sector interrupter. T h e amplitude of t h e
flicker depends on the absorption of the sample, a n d could in principle
be measured directly t o determine the per cent transmission. Complications of the sort discussed in § 14.28 and in Wright a n d Herscher's
article are avoided, however, if a null method is used. By reducing
t h e intensity of radiation in the reference beam to m a t c h the smaller
intensity of the sample beam, one can determine the latter from t h e
a m o u n t of reduction required.
I n the instrument show in Fig. 17.7, the reference beam is capable
of reduction by means of a comb-shaped shutter, the teeth of which
are triangular in shape. T h e shutter is driven by a motor whose
source of current is t h e bolometer on which the flickering beam
ultimately falls. T h e bolometer current is first amplified by an
AC amplifier tuned t o the flicker frequency (§ 12.8), the value of
which is of course determined by the sector interrupter. This current
causes the shutter motor t o move the shutter into the reference beam
until the flicker disappears. Simultaneously, the a m o u n t of motion
of the shutter is recorded. By adjustment of the size and shape of the
teeth in the shutter, it is possible to establish a linear relationship
between this motion a n d the per cent transmission, so t h a t the record
of .shutter motion is also a record of per cent transmission.
T h e spectrum is scanned by rotation of the Littrow mirror (§ 17.2),
marked WAVELENGTH MIRROR in Fig. ,17.6. T h e same motor t h a t
drives the mirror also drives the recorder drum, giving thereby a
direct relationship between wavelength and drum orientation. As
the spectrum is scanned, t h e drum rotates and a recording pen connected to the shutter traces out the per cent transmission curve on
a paper chart properly ruled in transmission and wavelength coordinates. A typical curve is shown in Fig. 17.8.
T h e use of AC amplification has the advantages mentioned in
§ 12.8. I n addition, it enables a traversal of the spectrum without
interruption from one end to the other. Ordinarily it is necessary t o
stop an infrared spectrometer while the slit width is being changed t o
492
SPECTROSCOPY OF THE INFRARED REGION
[§17.4
compensate for change in the radiation curve. In the Baird instrument the sHt-width change can be made continuously, because any
energy changes produced by changing slit width affect both sample
and reference beams equally, and therefore the intensity ratio of
sample beam to reference beam is not affected. Inasmuch as the
sample-beam and reference-beam intensities are balanced by the
Baird Associates'