224423 - Stanford University

PHYSICAL REVIEW B 74, 224423 共2006兲
La-alloying study of CeCoGe2: Magnetic susceptibility and specific heat measurements
C. R. Rotundu* and B. Andraka
Department of Physics, University of Florida, P.O. Box 118440, Gainesville, Florida 32611-8440, USA
共Received 31 July 2006; revised manuscript received 4 October 2006; published 18 December 2006兲
We report measurements of magnetic susceptibility and specific heat for polycrystalline Ce1−xLaxCoGe2 for
x = 0, 0.1, 0.2, 0.3, 0.5, and 1. CeCoGe2 is considered as a concentrated J = 5 / 2 Kondo system. We have found
significant preferential orientation in the magnetic susceptibility implying magnetic anisotropy. Magnetic susceptibility and specific heat results do not corroborate the single impurity model and imply the importance of
crystalline electric fields and magnetic correlations. Low-temperature specific heat and susceptibility data for
x = 0.2 suggest a possibility of a two-dimensional antiferromagnetic quantum critical point.
DOI: 10.1103/PhysRevB.74.224423
PACS number共s兲: 75.30.Mb, 75.20.Hr, 75.40.⫺s, 75.30.Cr
Lately, there have been reports of unusual heavy fermion
behavior among Ce-based alloys, such as CeNi9Si4 共Ref. 1兲
and CeCoGe2 共Ref. 2兲 described by the J = 5 / 2 Kondo
model.3 This model seems to provide a good consistent description of the specific heat and magnetic susceptibility
from room temperature down to the lowest temperatures of
reported measurements, approximately 1 K. In general, lowtemperature properties of Ce-based heavy fermions can be
accounted for to some degree by either the J = 1 / 2 or 3 / 2
Kondo models. This is because the sixfold degeneracy of 4f 1
moments is lifted by crystalline electric fields 共CEF兲, resulting in a ground-state doublet or quartet. This latter possibility
is rare and can occur for a cubic environment of Ce only. A
lack of CEF effects in CeNi9Si4 and CeCoGe2 has been explained by a somewhat unique hierarchy of relevant energy
scales, T0 ⬎ ⌬CEF Ⰷ TRKKY, where T0, ⌬CEF, and TRKKY stand
for the energy scales of Kondo interactions, CEF, and Ce-Ce
intersite exchange coupling, respectively. Furthermore, all
three energy scales have to be small 共T0 ⬇ 100 K兲 to result in
enhanced values of the electronic specific heat coefficient ␥.
We have further examined the applicability of the J
= 5 / 2 Kondo model to one of these exotic systems,
CeCoGe2.2 Our choice was motivated by the fact that a related compound, CeNiGe2, belonging to the same orthorhombic crystal structure, with almost identical lattice parameters, exhibits strong anisotropy in electrical resistivity
and magnetic susceptibility implying the importance of CEF
effects.4–7 The importance of CEF’s was also concluded from
a previous investigation of Co-deficient material,
Here we report on the alloying study of CeCoGe2, in
which La was partially substituted for Ce. Such an alloying
study of CePb3 proved the relevance of the J = 1 / 2 Kondo
model to Ce-based heavy fermions. In particular, bulk properties of CexLa1−xPb3, specific heat, and magnetic susceptibility showed almost perfect scaling9 with the Ce concentration x. The search for a similar scaling of these properties in
CexLa1−xCoGe2 was among the objectives of the reported
Polycrystalline samples of Ce1−xLaxCoGe2 with x = 0, 0.1,
0.2, 0.5, and 1 were obtained in the same manner as
CeCoGe2 described in Ref. 2, by arc-melting elements under
argon atmosphere, followed by vacuum annealing at 900 ° C
for three weeks. The starting constituents were Ce and La
from Ames, Co and Ge from Alfa Aesar and JohnsonMatthey, 99.995% and 99.9999% purity, respectively. Mass
losses during the arc-melting were less than 0.4% of the total
mass of ⬃0.8 g each sample. At least two different samples
were synthesized and characterized by x-ray powder diffraction and magnetic susceptibility, to minimize the effects of
possible sample mishandling and verify the reproducibility
of data. Magnetic susceptibility was measured using a Quantum Design SQUID magnetometer. Specific heat was measured by a standard relaxation method with a homemade 3He
calorimeter down to ⬃0.3 K.
A. X-ray diffraction
CeCoGe2 and CeCoGe2 belong to the orthorhombic
CeNiSi2-type10 crystal structure, space group Cmcm. All
x-ray diffraction lines for all samples were indexable to this
crystal structure. However, we have observed some discrepancies between calculated and measured intensities. These
discrepancies were not only for mixed alloys but also for the
end compounds. This disagreement can be due to atomic
disorder but also to preferential orientation reported in the
x-ray powder diffraction for the isostructural polycrystalline
CeNiGe2.11 The lattice constant b for CeCoGe2 共and
CeNiGe2兲 is much larger than the lattice constants a and c.
The corresponding values for the Co compound are 4.262,
16.781, and 4.215 Å for a, b, and c parameters, respectively.
In CeNiSi2, platelike grains with the 共010兲 plains preferentially oriented parallel to the x-ray-diffraction slides were
reported.11 The discrepancy between the measured and calculated intensities for CeCoGe2 is also consistent with 共010兲
planes preferentially parallel to the x-ray slides.
The results of the lattice constant measurements for
Ce1−xLaxCoGe2 are shown in Fig. 1. As expected, La being
larger than Ce expands the lattice. However, this expansion
is not isotropic. The lattice constants a and c are expanded
more than b, such as the ratios b / a and b / c are reduced by
La. Therefore, if there is some anisotropy in magnetic and
transport properties, we would expect this anisotropy to be
reduced by the La alloying.
©2006 The American Physical Society
PHYSICAL REVIEW B 74, 224423 共2006兲
FIG. 1. Upper panel: Volume of the unit cell vs La concentration
x. Full symbols are the data points from Ref. 1. Lower panel: Ratio
of lattice constants b / c and b / a vs x of Ce1−xLaxCoGe2 for x = 0,
0.1, 0.2, 0.3, 0.5, and 1.
B. Magnetic susceptibility
All samples used for magnetic susceptibility measurements were small buttons with similar dimensions in all three
directions. These buttons were checked first for possible
preferential orientation. A direction parallel to the axis of the
arc-melter 共perpendicular to the surface of the hearth of the
arc-melter兲, which we denote a.m., was identified for all
samples. Figure 2 shows the magnetic susceptibility for
CeCoGe2 and LaCoGe2, upper and lower panels, respectively, for the two directions of the magnetic field: H 储 a.m.
and H ⬜ a.m. Before discussing the preferential orientation,
we point to a rather good agreement between the susceptibility reported by Mun et al.2 and our measurements for both
The magnetic susceptibility of CeCoGe2 has larger values
when magnetic field is parallel to the a.m. direction. This
result, which cannot be explained by demagnetization effects
共estimated to be less than 0.5%兲, was confirmed on all other
CeCoGe2 samples. Furthermore, all studied concentrations
containing Ce showed systematically larger values in the
magnetic susceptibility for the field direction parallel to a.m.
This result suggests preferential orientation of arc-melted
Ce1−xLaxCoGe2 samples and significant magnetic anisotropy.
A similar preferential orientation in the magnetic susceptibility, i.e., larger susceptibility values for the field parallel to the
a.m. axis, was previously reported for polycrystalline
CeNiGe2.11 Subsequent susceptibility investigations on
single crystals confirmed strong magnetic anisotropy.4–7
We have also observed preferential orientation in the susceptibility of LaCoGe2. However, there is an important difference between the two compounds. All studied LaCoGe2
samples had the susceptibility for H 储 a.m. smaller than that
FIG. 2. Magnetic susceptibility of CeCoGe2 and LaCoGe2 for
two directions of magnetic fields: H 储 a.m. and H ⬜ a.m. H
= 1000 Oe.
for H ⬜ a.m. There is no reason to suspect different preferred
grain orientation in arc-melted CeCoGe2 and LaCoGe2.
Rather, these results imply that the magnetic anisotropy in
CeCoGe2 is different from that in LaCoGe2. Thus, our magnetic susceptibility data suggest magnetic anisotropy in
CeCoGe2, similarly to the case of CeNiGe2. Furthermore,
there seem to be two magnetic subsystems, of Ce and Co,
that show different magnetic anisotropy. This existence of
magnetic anisotropy in CeCoGe2 is obviously inconsistent
with the single impurity J = 5 / 2 Kondo interpretation. The
Kondo model is a single-parameter model, thus isotropic.
Therefore, the magnetic anisotropy in 共Ce, La兲CoGe2 alloys
implies the importance of crystalline electric-field effects.
Figures 3–6 show the evolution of the magnetic susceptibility between x = 0 and 0.5. The high-temperature susceptibility, when normalized to a mole of Ce, is similar for all
alloys above 150 K. The effective moment obtained directly
from the straight-line fits of the inverse of the hightemperature susceptibility to temperature is between 2.8 and
2.9␮B. Subtracting LaCoGe2 data results in a reasonable
agreement with the expected moment for trivalent Ce,
2.54␮B. The absolute value of the negative Curie-Weiss temperature ⌰, implying predominance of antiferromagnetictype interactions, decreases from 145 K for x = 0 to 110 K
for x = 0.5 for H parallel to a.m. This Curie-Weiss temperature showed even smaller and less systematic variation for
directions perpendicular to a.m. Thus, these hightemperature susceptibility results imply rather good scaling
with the Ce concentration, suggesting that single-ion parameters are weakly affected by the La doping.
PHYSICAL REVIEW B 74, 224423 共2006兲
FIG. 5. Magnetic susceptibility of Ce0.5La0.5CoGe2 cooled in
zero field 共zfc兲 and in 1000 Oe 共fc兲 for two orientations of the field.
FIG. 3. Magnetic susceptibility of Ce1−xLaxCoGe2, for x = 0, 0.1,
0.2, 0.3, 0.5, and 1; magnetic fields H ⬜ a.m. and H 储 a.m., respectively. H = 1000 Oe.
gued to be due to a small fraction of Co atoms displaced
from their correct crystallographic positions. However, we
were not able to fit this upturn to the Curie expression. Furthermore, this upturn in several studied LaCoGe2 samples
was approximately 20 times smaller than in the Ce compound. The low-temperature increase in mixed alloys is
clearly non-Curie-type. Susceptibility curves of mixed alloys
have pronounced shoulders, whose temperature again increase with x. These shoulders do not seem to directly cor-
The low-temperature susceptibility, on the other hand,
shows strong concentration dependence. La-containing alloys exhibit a rapid increase in ␹ at low temperatures. The
temperature corresponding to the rapid rise in ␹ increases
with x from about 20 K for x = 0.1 to over 50 K for x = 0.5.
Quite possibly these low-temperature increases are related to
a tail already observed in the pure compound below
10– 15 K. This low-temperature upturn in CeCoGe2 was ar-
FIG. 4. Magnetic susceptibility of Ce1−xLaxCoGe2 divided by
Ce mol, for x = 0 and 0.1; magnetic field H ⬜ a.m. The arrows indicate positions of broad maxima discussed in the text.
FIG. 6. Low-temperature magnetic susceptibility
Ce1−xLaxCoGe2 divided by Ce mol, for x = 0.1, 0.2, and 0.3.
PHYSICAL REVIEW B 74, 224423 共2006兲
respond to a weak maximum in ␹ of the pure Ce compound
between 50 and 60 K 共Fig. 4兲. This weak maximum in
CeCoGe2, previously suggested to be due to a single-ion
physics, seems to move to lower temperatures when 10% of
Ce is replaced by La 共Fig. 4兲. Also, all mixed 共Ce, La兲CoGe2
alloys exhibit some discrepancies between field-cooled 共fc兲
and zero-field-cooled 共zfc兲 ␹, which increase with x. The
discrepancy is particularly strong for x = 0.5, which also
shows a sharp cusp in the zfc ␹, typical of spin glasses 共Fig.
5兲. Almost identical cusps were observed for two external
fields applied at 100 and 1000 Oe.
These magnetic susceptibility data suggest a possible alternative explanation of the 50– 60 K broad maximum in ␹
of CeCoGe2 in terms of magnetic origin, probably due to a
short-range magnetic order or two-dimensional magnetic
correlations. Preferential orientation in magnetic susceptibility suggests strong magnetic anisotropy, similarly to the related CeNiGe2. CeNiGe2 is a layered material for which
room-temperature electrical conductivity in ac planes is almost two orders of magnitude larger than along the b direction. Magnetic susceptibility of CeNiGe2 is also anisotropic.
However, there is a controversy whether the easy magnetization axis is along the b direction4,6 or is in the ac plane.5,7
Can similarly strong anisotropy in exchange interactions lead
to a short-range order at 50– 60 K in CeCoGe2? An increase
of the susceptibility below 10– 15 K would be due to a precursor of a long-range magnetic order. Weakening of the anisotropy in La-doped alloys would result in threedimensional ordering or spin-glass freezing. Finally, CEF
excitations could also lead to a structure in the susceptibility
similar to that observed in CeCoGe2. For instance, the susceptibility of tetragonal CeAuSb2, which shows strong twodimensional character, is remarkably similar12 to that for
CeCoGe2. The susceptibility of CeAuSb2 has been accounted
for by a CEF model. However, strong sensitivity of this
structure in CeCoGe2 to doping 共Fig. 4兲 argues rather for its
magnetic origin.
Very interesting is the evolution of the susceptibility below 10 K for nonmagnetic alloys. Figure 6 shows this susceptibility for x = 0.1, 0.2, and 0.3. For x = 0.2, the susceptibility has approximately a linear temperature dependence,
␹ = ␹0 − aT. On the other hand, relatively strong curvatures
are seen in x = 0.1 and 0.3 with opposite signs. We will return
to this observation while discussing the specific heat data.
C. Specific heat
Low-temperature specific heat was investigated between
0.3 and 10 K. The results for x = 0, 0.3, and 0.5 in the form of
⌬C / T, where ⌬C is a difference between the measured specific heat and the specific heat of LaCoGe2 divided by Ce
concentration, versus x are shown in Fig. 7. The values for
CeCoGe2 are in good agreement with the report of Mun et
al.,2 yielding an enhanced Sommerfeld coefficient ␥
⬇ 120 mJ/ K2 mol. Our data suggest a possibility of a small
increase of ⌬C / T below 1.5 K, as reported previously for
Co-deficient samples8 and interpreted in terms of a precursor
of a ferro-or ferrimagnetic transition. However, this upturn in
our data is much weaker than that found in the Co-deficient
FIG. 7. Specific heat of Ce1−xLaxCoGe2 divided by Ce mol, for
x = 0, 0.2, and 0.5. The inset presents ⌬C / T vs ln T for x = 0.2.
material. ⌬C / T increases with T above 2 K. Compositions
x = 0.2 共not shown兲, 0.3, and 0.5 exhibit a decrease of ⌬C / T
with T in the whole studied temperature range, from
0.3 to 10 K. Thus, this temperature variation seems to correlate with some aspects of the susceptibility. An increase of
⌬C / T with T for the pure Ce compound reflects the existence of a weak maximum in the specific heat at much higher
temperatures, 50– 60 K.2
There is a monotonic increase of the low-temperature
⌬C / T values with x, such that ⌬C / T for x = 0.5 exceeds
550 mJ/ K2Ce mol at 0.3 K. Interestingly, two of the studied
concentrations, x = 0.2 共see inset to Fig. 7兲 and 0.5, have
⌬C / T approximately proportional to ln T. This variation is a
hallmark of non-Fermi liquids. The slopes of ⌬C / T versus
ln T are very different for the two compositions, suggesting a
different origin of these behaviors. A logarithmic temperature
variation and large values of ⌬C / T for x = 0.5 are surprising
since some kind of magnetic ordering or spin-glass freezing
takes place at 35 K. In this respect, Ce0.5La0.5CoGe2 is similar to two other systems, UCu5 共Ref. 13兲 and
U共Pt0.94Pd0.06兲3,14 in which low-temperature heavy fermion
states develop from high-temperature magnetic states. It is
also interesting to note that logarithmic temperature variation
of C / T has been observed over some temperature range for
these two systems. There is no satisfactory explanation of
these behaviors.
On the other hand, the nearly logarithmic variation of C / T
with T for x = 0.2, together with already discussed magnetic
susceptibility linear in temperature, implies a possibility of a
proximity to a two-dimensional AFM quantum critical
point.15 Obviously, these temperature variations of the specific heat and susceptibility could also be due to a large degree of disorder existing in mixed alloys. However, we
would like to point out a somewhat similar evolution of magnetic properties in previously investigated CeCo1−yNiyGe2.16
Alloys corresponding to y ⬎ 0.7 show three-dimensional
magnetic ordering. The y = 0.7 composition has lowtemperature magnetic susceptibility and specific heat consistent with two-dimensional quantum critical AFM behavior.
PHYSICAL REVIEW B 74, 224423 共2006兲
The described alloying results present considerable challenge to the established J = 5 / 2 single Kondo ion interpretation of CeCoGe2. First of all, no single-impurity 共even approximate兲 scaling is observed in the magnetic susceptibility
and specific heat data for Ce1−xLaxCoGe2. Obviously, deviations from the single-impurity scaling are expected if the
relevant single-impurity parameters are altered by the La alloying. This might be particularly relevant for Ce compounds
involving 3d transition metals as ligands, since 3d orbitals
are believed to be very sensitive to changes in local environment. On the other hand, a good concentration scaling has
been recently reported in a noncubic system involving a transition metal, Ce1−xLaxNi9Ge4.17 High-temperature susceptibility of Ce1−xLaxCoGe2 scales well with Ce, suggesting that
relevant single-ion parameters are not altered in any drastic
*Author to whom correspondence should be addressed. Email address: [email protected]
H. Michor, St. Berger, M. El-Hagary, C. Paul, E. Bauer, G.
Hilscher, P. Rogl, and G. Giester, Phys. Rev. B 67, 224428
E. D. Mun, B. K. Lee, Y. S. Kwon, and M. H. Jung, Phys. Rev. B
69, 085113 共2004兲.
3 V. T. Rajan, Phys. Rev. Lett. 51, 308 共1983兲.
4 M. H. Jung, N. Harrison, A. H. Lacerda, H. Nakotte, P. G. Pagliuso, J. L. Sarrao, and J. D. Thompson, Phys. Rev. B 66,
054420 共2002兲; M. H. Jung, A. H. Lacerda, P. G. Pagliuso, J. L.
Sarrao, and J. D. Thompson, J. Appl. Phys. 91, 8522 共2002兲.
5 Y. Okada, Y. Inada, A. Galatanu, E. Yamamoto, R. Settai, and Y.
Onuki, J. Phys. Soc. Jpn. 72, 2692 共2003兲.
6 M. Ohashi, G. Oomi, K. Ishida, I. Satoh, T. Komatsubara, T.
Kawae, and K. Takeda, J. Alloys Compd. 408–412, 84 共2006兲.
7 A. P. Pikul, D. Kaczorowski, Z. Bukowski, T. Plackowski, and K.
Gofryk, J. Phys.: Condens. Matter 16, 6119 共2004兲.
V. K. Pecharsky and K. A. Gschneidner, Jr., Phys. Rev. B 43,
8238 共1991兲.
manner across this pseudoternary system. Preferential orientation in magnetic susceptibility for all studied alloys implies
the susceptibility to be anisotropic, similarly to CeNiGe2.11
The preferential orientation of crystallites is supported by
direct x-ray-diffraction measurements. This magnetic anisotropy implies the importance of crystalline electric fields and,
therefore, is clearly inconsistent with the J = 5 / 2 Kondo
model, which is istropic. A possibility of a two-dimensional
antiferromagnetic critical point should be furthe explored on
single crystals of Ce1−xLaxCoGe2, when they become available.
This work has been supported by the Department of Energy, Grant No. DE-FG02-99ER45748.
C. L. Lin, A. Wallash, J. E. Crow, T. Mihalisin, and P. Schlottmann, Phys. Rev. Lett. 58, 1232 共1987兲.
O. P. Bodak and E. I. Gladyshevskii, Kristallografiya 14, 990
11 V. K. Pecharsky, K. A. Gschneidner, Jr., and L. L. Miller, Phys.
Rev. B 43, 10906 共1991兲.
12 A. Thamizhavel, T. Takeuchi, T. Okubo, M. Yamada, R. Asai, S.
Kirita, A. Galatanu, E. Yamamoto, T. Ebihara, Y. Inada, R. Settai, and Y. Onuki, Phys. Rev. B 68, 054427 共2003兲.
H. R. Ott, H. Rudigier, E. Felder, Z. Fisk, and B. Batlogg, Phys.
Rev. Lett. 55, 1595 共1985兲; B. Andraka, J. Dapprich, M. Baldus,
P. Kumar, and G. R. Stewart, Phys. Rev. B 45, 7481 共1992兲.
14 J. S. Kim, B. Andraka, and G. R. Stewart, Phys. Rev. B 45, 12081
15 A. J. Millis, Phys. Rev. B 48, 7183 共1993兲.
B. K. Lee, J. B. Hong, J. W. Kim, Kwang-hyun Jang, E. D. Mun,
M. H. Jung, S. Kimura, Tuson Park, J.-G. Park, and Y. S. Kwon,
Phys. Rev. B 71, 214433 共2005兲.
U. Killer, E.-W. Scheidt, G. Eickerling, H. Michor, J. Sereni, Th.
Pruschke, and S. Kehrein, Phys. Rev. Lett. 93, 216404 共2004兲.