PDF (5.23 MB)

Contact us
My IOPscience
Comparison of atom probe tomography and transmission electron microscopy analysis of
oxide dispersion strengthened steels
This content has been downloaded from IOPscience. Please scroll down to see the full text.
2014 J. Phys.: Conf. Ser. 522 012028
View the table of contents for this issue, or go to the journal homepage for more
Download details:
IP Address:
This content was downloaded on 05/02/2015 at 09:05
Please note that terms and conditions apply.
Electron Microscopy and Analysis Group Conference 2013 (EMAG2013)
IOP Publishing
Journal of Physics: Conference Series 522 (2014) 012028
Comparison of atom probe tomography and transmission
electron microscopy analysis of oxide dispersion
strengthened steels
AJ London1 , S Lozano-Perez1 , S Santra2 , S Amirthapandian2 , BK Panigrahi2 , CS
Sundar2 and CRM Grovenor1
Department of Materials, University of Oxford, Parks Road, Oxford, OX1 3PH, UK
Indira Gandhi Centre for Atomic Research, Kalpakkam - 603 102, TN, India
E-mail: [email protected]
Abstract. Oxide dispersion strengthened steels owe part of their high temperature stability to the nanoscale oxides they contain. These yttrium-titanium oxides are notoriously difficult to characterise since
they are embedded in a magnetic-ferritic matrix and often <10 nm across. This study uses correlated
transmission electron microscopy and atom probe tomography on the same material to explore the kind of
information that can be gained on the character of the oxide particles.
The influence of chromium in these alloys is of interest, therefore two model ODS steels Fe-(14Cr)0.2Ti-0.3Y2 O3 are compared. TEM is shown to accurately measure the size of the oxide particles and atom
probe tomography is necessary to observe the smallest sub-1.5 nm particles. Larger Y2 Ti2 O7 and Y2 TiO5
structured particles were identified by high-resolution transmission electron microscopy, but the smallest
oxides remain difficult to index. Chemical data from energy-filtered TEM agreed qualitatively with the
atom probe findings. It was found that the majority of the oxide particles exhibit an unoxidised chromium
shell which may be responsible for reducing the ultimate size of the oxide particles.
1. Introduction & Background
Ferritic oxide dispersion strengthened (ODS) steels are being considered as structural materials for fission
and fusion reactors due to their low activation, high temperature stability and radiation resistance [1]. The
1–10 nm yttrium-titanium rich oxides that hinder dislocation motion at high temperature also harmlessly
recombine radiation defects and trap inert gases away from grain boundaries [2]. Understanding
the particle character is critical because the size distribution and dispersion influence the mechanical
properties [3] and their chemical nature determines their thermal and radiation stability [4].
Many methods have been used to study these nano-scale oxides [5, 6, 7] but only transmission electron
microscopy (TEM) and atom probe tomography (APT) provide sufficiently spatially resolved chemical
and structural information to distinguish the smaller oxides from larger oxides located at the surface,
grain boundaries and in the bulk.
The methodology of this study includes both TEM and APT of the same specimens to give a more
convincing and complete characterisation of the same volume of material. The focus of the TEM
examination was the spatial distribution, size and the structure of the oxides. With atomistic sensitivity
and sub-nanometer resolution, APT was used to study the chemistry of individual oxide particles and
their chromium content.
Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution
of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Published under licence by IOP Publishing Ltd
Electron Microscopy and Analysis Group Conference 2013 (EMAG2013)
IOP Publishing
Journal of Physics: Conference Series 522 (2014) 012028
Part of the uncertainty relating to the identity of the oxides undoubtedly comes from the many
different alloy compositions and processing routes used to produce different ODS alloys. In this study,
the alloys have identical processing history and a systematic variation in chemistry to determine the
influence of chromium on the oxide particles. Chromium is a key component of structural steels,
providing enhanced strength and corrosion resistance. The model ODS steels are nominally Fe-0.2Ti0.3Y2 O3 and Fe-14Cr-0.2Ti-0.3Y2 O3 (all wt%) and are produced by the Indira Gandhi Centre for Atomic
Research (India).
Titanium has been attributed to having a strong refining influence on the oxides [8] but the effect
of chromium has been less fully investigated. Chromium rich shells around the Y-Ti-oxides have been
observed by TEM [9] and APT [10]. By comparing model alloys with and without a chromium addition
the effect of chromium can be evaluated.
2. Materials & Methods
A pre-alloyed powder containing Fe-(14Cr)-0.2 Ti was mechanically alloyed with 0.3 Y2 O3 powder (all
in wt%) using a horizontal high-energy Simoloyer attritor mill. Milling under an argon atmosphere took
4 hrs using a ball-to-powder ratio of 10:1. Consolidation was achieved using hot extrusion at 1050 ºC
and 1150 ºC for the Cr-free and Cr containing alloys respectively.
Bulk material were cut to 0.5x0.5x20 mm and electropolished into needles using perchloric acid by
standard methods [11]. TEM samples were also prepared using a FIB lift-out technique.
200 keV TEM observations were made using a Philips CM20 and a JEOL 2200MCO FEG TEM
fitted with a monochromator, two spherical aberration (Cs) correctors and an W-filter. The largest (120
mm) condenser aperture was used and the beam was focused to give an an image with a diameter slightly
bigger than the CCD to maximise intensity. The convergence and objective aperture half-angles were 6
mrad and 20 mrad, respectively. Low-loss energy-filtered series collected from the 10–80 eV energy-loss
range included the YN2,3 (26 eV), CrM2,3 (42 eV) and FeM2,3 (54 eV) edges. Steps of 1 eV, using a 2 eV
slit and 6 s acquisition time per image were used and images were hardware binned by 4 to 1024×1024
pixels. The EFTEM series were processed after acquisition using multivariate statistical analysis (MSA)
in order to reduce statistical noise, as described elsewhere [9].
Atom probe analyses were performed using a LEAP-3000X HR microscope. During analysis,
samples were maintained at 50 K and a 532 nm wavelength laser with a pulse width of 10 ps, repetition
rate of 200 kHz, pulse energy of 0.3 nJ and spot size of less than 10 mm were used to promote field
evaporation. Oxide clusters in the data were identified using the method of maximum separation of Y,
YO, TiO and O ions. A maximum separation of core ions of 0.7–1.1 nm was used with a Nmin between
8–15. The concentration line profiles were calculated in MATLAB for every cluster in the data scaled
to each cluster’s size. In this way profiles of large and small clusters can be combined on the same plot.
Previously, representative clusters had to be chosen for analysis but using this method all the clusters can
be investigated.
3. Results & Discussion
The particle radii were measured using diffraction contrast (Fig. 1a), Fresnel imaging and EFTEM
(Fig. 1b) techniques. All of these gave consistent results and the size distributions are shown in figure 2b.
The TEM imaging of electropolished atom probe needles leads to a bias in counting medium sized
particles as they are more obvious in the thicker shank of the tip (leftmost Fig.1a), but very large particles
are physically too large to exist in the end of the tip. EFTEM provides a wider field of view but is limited
to thin regions (< 100 nm). The APT data is in the form of a point cloud and the particles exist as more
diffuse clusters. It is common to report the radius of gyration (Rg ) of these clusters to compare their size.
Trajectory aberrations can influence the lateral dimensions of the particles and oxygen spreading can
increase their size perpendicular to the analysis direction, Z. But the size along Z is relatively unchanged
in-spite of these aberrations [12], therefore only the component of Rg along Z is reported. The Rgz is
much smaller than the radius measured by TEM as shown in the size distributions in figure 2b. Because
Electron Microscopy and Analysis Group Conference 2013 (EMAG2013)
IOP Publishing
Journal of Physics: Conference Series 522 (2014) 012028
Figure 1. (a) TEM image of a needle with accompanying APT data showing the Y and YO ions with
yttrium iso-concentration surfaces to highlight the oxides. Two tilt angles are used to show clearly the
oxides in the top and bottom grains. The grain boundary (GB) is visible in TEM and APT, 200 nm from
the tip. All images are shown at the same scale.
(b) EFTEM image of the YN2,3 edge showing yttrium rich oxides in a Fe-14Cr ODS alloy. With a 2 eV
energy slit and 6 s acquisition using MSA, oxides as small as 1.5 nm in diameter were observed.
only 37% of the atoms are detected in the LEAP some reduction of size is expected, especially for the
smaller clusters. The extent of the clusters (in Z) was measured to be 3 times Rgz and may therefore
give a more representative size measurement. In the Fe-14Cr alloy, the number density of particles as
measured by TEM was 3.0–4.4x1022 per m3 but ten times greater than this when measured by APT
(Fig.2b), suggesting that the APT is indeed more sensitive to the numerous smaller oxide particles.
Faceted particle edges parallel to {110}matrix were observed when viewing down [002]matrix . No
structural determination was possible for the <2 nm particles, but larger particles were structurally
consistent with the bulk structures of Y2 Ti2 O7 and Y2 TiO5 . The composition of the oxides from APT in
both alloys shows a range of Y:Ti ratios consistent with either bulk stoichiometry but the oxygen content
is only 50–60 at%. Fe is present in the clusters because of trajectory aberrations [12]. Fe & Cr are
not expected in the oxide core by energetic considerations [13] and Fe has not been found in extracted
particles [5].
In the Fe-14Cr alloy, a Cr-rich shell was apparent using EFTEM in >10 nm particles and in most
clusters observed using APT. As shown in figure 2a the Cr ions have a shell-like profile (11.5 at% in the
bulk) but the CrO ions take a profile similar to the O and FeO ions. We do not believe that there is either
oxidised Fe or Cr in the particles, but that oxygen atoms originating from the particles combine with
Fe or Cr from the matrix and evaporate as FeO or CrO molecular species. Similarly partially-oxidised
Cr-rich shells have been reported in similar alloys [10].
The addition of Cr to the alloy reduces the average size and distribution of sizes of the Y-Ti oxides
(Fig. 2b). The addition of Cr could increase the rate of nucleation of the Y-Ti oxides or reduce the rate of
coarsening in order that the average size is reduced even with the increased consolidation temperature.
The smallest clusters contain fewer than 100 atoms making the evaluation of shell size uncertain. For
the larger clusters, the shell size increases with cluster radius. Quantification of the size of the Cr-shell
will help to clarify the mechanism of its formation. This shell should not be ignored as it reduces the
oxide-metal surface energy, could increase thermal stability and has implications for the response of the
oxide particles to radiation.
Electron Microscopy and Analysis Group Conference 2013 (EMAG2013)
IOP Publishing
Journal of Physics: Conference Series 522 (2014) 012028
APT Cr−free
TEM Cr−free
No. Density m−3
100 0
100 points along analysis direction, 0 = top, normalised to cluster size
Radius nm
Figure 2. (a) Average line profiles along the analysis direction from atom probe (Cr-free = 200 and
Cr-alloy = 500 clusters total). Each line profile was normalised to the size of the oxide clusters so that
different sized clusters could be compared on one plot. Core ions=YO+Y+TiO.
(b) Size distributions of oxide clusters in the same alloys measured by TEM and APT (Rgz ).
4. Conclusions
TEM and APT have been used to characterise model ODS steels and investigate the oxides present in
detail. For larger particles diffraction contrast was suitable to determine their size and EFTEM imaging of
the YN2,3 edge showed yttrium rich oxides as small as 1.5 nm in diameter. The smallest most numerous
(< 1.5 nm) oxides were only visible using atom probe tomography but the size measurement by APT
requires careful calibration of the reconstructed data and analysis to reduce the influence of trajectory
aberrations. Low-loss EFTEM images showed Y-rich particles with a Cr-rich shell and the bulk oxides
Y2 Ti2 O7 and Y2 TiO5 were identified by structural measurements. Y-Ti-oxide clusters in the APT data
had compositions consistent with these two bulk oxides but were oxygen deficient. From the profile of
the CrO ions collected by APT, it is more likely the Cr shell is metallic or only partially oxidised. The
influence of the Cr on the oxides is to reduce their average size and increase their number density.
Dr. Tom Black for funding part of this project, Prof. A.Kirkland and & Dr. Judy Kim for the support on
the 2200 microscope, Paul Bagot & Oxford FIM group (University of Oxford).
Baluc N, Schäublin R, Spätig P and Victoria M 2004 Nuclear fusion 44 56
Odette G, MJ A and BD W 2008 Annual Review of Materials Research 38 471–503
Ukai S, Miztua S, Fujiwara M, Okuda T and Kobayshi T 2002 J. Nucl. Sci. and Tech. 39 778–788
Williams C, Smith G and Marquis E 2012 Scripta Materialia 67 108–111
Wu Y, Haney E, Cunningham N and Odette G 2012 Acta Materialia 60 3456–3468
Mathon M, de Carlan Y, Averty X, Alamo A and de Novion C 2006 ASTM Spec. tech. pub. 1475 120
Ortega Y, de Castro V, Monge M, Munoz A, Leguey T and Pareja R 2008 J. Nuc. Mat. 376 222–228
Ratti M, Leuvrey D, Mathon M and De Carlan Y 2009 J. Nuc. Mat. 386 540–543
Lozano-Perez S, de Castro Bernal V and Nicholls R 2009 Ultramicroscopy 109 1217–1228
De Castro V, Marquis E, Lozano-Perez S, Pareja R and Jenkins M 2011 Acta Mater. 59 3927–3936
Miller M 2000 Atom Probe Tomography: Analysis at the Atomic Level (Plenum Pub Corp)
Oberdorfer C and Schmitz G 2010 Microscopy and Microanalysis 17 15
Murali D, Panigrahi B, Valsakumar M, Chandra S, Sundar C and Raj B 2010 J. of Nuc. Mat. 403 113–116