arXiv:0803.0293v1 [cond-mat.str-el] 3 Mar 2008
X-ray absorption and x-ray magnetic dichroism study on Ca3CoRhO6and Ca3FeRhO6
T. Burnus,1Z. Hu,1Hua Wu,1J. C. Cezar,2S. Niitaka,3,4H. Takagi,3,4,5C. F. Chang,1N. B.
Brookes,2H.-J. Lin,6L. Y. Jang,6A. Tanaka,7K. S. Liang,6C. T. Chen,6and L. H. Tjeng1
1II. Physikalisches Institut, Universit¨ at zu K¨ oln, Z¨ ulpicher Str. 77, 50937 K¨ oln, Germany
2European Synchrotron Radiation Facility, BP 220, 38043, Grenoble, France
3RIKEN, Institute of Physical and Chemical Research, 2-1, Hirosawa, Wako, Saitama 351-0198, Japan
4CREST, Japan Science and Technology Agency (JST), Kawaguchi, Saitama 332-0012, Japan
5Department of Advanced Materials Science, University of Tokyo,
5-1-5, Kashiwanoha, Kashiwa, Chiba 277-8581, Japan
6National Synchrotron Radiation Research Center, 101 Hsin-Ann Road, Hsinchu 30077, Taiwan
7Department of Quantum Matter, ADSM, Hiroshima University, Higashi-Hiroshima 739-8530, Japan
(Dated: March 3, 2008)
Using x-ray absorption spectroscopy at the Rh-L2,3, Co-L2,3, and Fe-L2,3 edges, we find a valence
state of Co2+/Rh4+in Ca3CoRhO6 and of Fe3+/Rh3+in Ca3FeRhO6. X-ray magnetic circular
dichroism spectroscopy at the Co-L2,3 edge of Ca3CoRhO6 reveals a giant orbital moment of about
1.7µB, which can be attributed to the occupation of the minority-spin d0d2 orbital state of the high-
spin Co2+(3d7) ions in trigonal prismatic coordination. This active role of the spin-orbit coupling
explains the strong magnetocrystalline anisotropy and Ising-like magnetism of Ca3CoRhO6.
PACS numbers:78.70.Dm, 71.27.+a, 71.70.-d, 75.25.+z
The quasi one-dimensional transition-metal oxides
Ca3ABO6 (A = Fe, Co, Ni, ...; B
Ir, ...)have attracted a lot of interest in recent
years because of their unique electronic and mag-
netic properties.1,2,3,4,5,6,7,8,9,10,11,12,13The structure of
Ca3ABO6contains one-dimensional (1D) chains consist-
ing of alternating face-sharing AO6trigonal prisms and
BO6octahedra. Each chain is surrounded by six paral-
lel neighboring chains forming a triangular lattice in the
basal plane. Peculiar magnetic and electronic behaviors
are expected to be related to geometric frustration in
such a triangle lattice with antiferromagnetic (AFM) in-
terchain interaction and Ising-like ferromagnetic (FM) in-
trachain coupling. Ca3Co2O6, which realizes such a situ-
ation, shows stair-step jumps in the magnetization at reg-
ular intervals of the applied magnetic field of Ms/3, sug-
gesting ferrimagnetic spin alignment. It has a saturation
magnetization of Ms= 4.8µBper formula unit at around
4 T.14Studies on the temperature and magnetic-field de-
pendence of the characteristic spin-relaxation time sug-
gest quantum tunneling of the magnetization similar to
single-molecular magnets.15An applied magnetic field
induces a large negative magnetoresistance, apparently
not related to the three-dimensional magnetic ordering.11
Band-structure calculations using the local-spin-density
approximation plus Hubbard U (LSDA+U) predicted
that the Co3+ion at the trigonal site, being in the high-
spin (HS) state (S = 2), has a giant orbital moment of
1.57µBdue to the occupation of minority-spin d2orbital,
while the Co3+ion at the octahedral site is in the low-spin
(LS) state (S = 0).16An x-ray absorption and magnetic
circular dichroism study at the Co-L2,3 edge has con-
firmed this prediction.17Both studies explain well the
Ising nature of the magnetism of Ca3Co2O6.
= Co, Rh,
Ca3CoRhO6 and Ca3FeRhO6 have the same crys-
tal structure as Ca3Co2O6, but different magnetic and
electronic properties:Neutron diffraction and mag-
netization measurements also indicated intrachain-FM
and interchain-AFM interactions in Ca3CoRhO6 like
Ca3FeRhO6 reveal a single transition into a three-
dimensional AFM.5,18Although Ca3CoRhO6has a sim-
ilar magnetic structure as Ca3Co2O6, it exhibits consid-
erable differences in the characteristic temperatures in
the magnetic susceptibility. The high-temperature limit
of the magnetic susceptibility shows a Curie-Weiss be-
havior with a positive Weiss temperature of 150 K for
Ca3CoRhO6,5while 30 K was found for Ca3Co2O6.2,3
The measured magnetic susceptibility undergoes two
transitions at Tc1
= 90 K and Tc2
Ca3CoRhO6, and at Tc1 = 24 K and Tc2 = 12 K
for Ca3Co2O6,3,5,7,8,12,18which were attributed to FM-
intrachain and AFM-interchain coupling, respectively.
In contrast, Ca3FeRhO6 has an AFM ordering below
TN = 12 K.5,18,19Unlike Ca3Co2O6, there is only one
plateau at 4 T and no saturation even at 18 T in the
magnetization of Ca3CoRhO6at 70 K.7A partially dis-
ordered state in Ca3CoRhO6 has been inferred by the
previous work of Niitaka et al.8
susceptibility data on
= 25 K for
In order to understand the contrasting magnetic prop-
erties of Ca3CoRhO6 and Ca3FeRhO6, and, partic-
ularly, the type and origin of the intrachain mag-
netic coupling of these quasi 1D systems, the va-
lence, spin, and orbital states have to be clarified.
However, these issues have been contradictorily dis-
cussed in previous theoretical and experimental stud-
ies. The general-gradient-approximated (GGA) density-
functional band calculations20suggest a Co3+/Rh3+
state in Ca3CoRhO6, while LSDA+U calculations with
inclusion of the spin-orbit coupling favor a Co2+/Rh4+
state and, again, a giant orbital moment due to the
occupation of minority-spin d0 and d2 orbitals.21Neu-
tron diffraction experiments on Ca3CoRhO68,22suggest
the Co3+/Rh3+state. However, based on the magnetic
susceptibility5and x-ray photoemission spectroscopy23
the Co2+/Rh4+state was proposed. For Ca3FeRhO6, the
Fe2+/Rh4+state was suggested in a magnetic suscepti-
bility study,5whereas M¨ ossbauer spectroscopy indicates
a Fe3+state,19and thus Rh3+.
In order to settle the above issues, in this work we
first clarify the valence state of the Rh, Co, and Fe ions
in Ca3CoRhO6and Ca3FeRhO6 using x-ray absorption
spectroscopy (XAS) at the L2,3edges of Rh, Co, and Fe.
We reveal a valence state of Co2+/Rh4+in Ca3CoRhO6
and of Fe3+/Rh3+in Ca3FeRhO6. Then, we investigate
the orbital occupation and magnetic properties using x-
ray magnetic circular dichroism (XMCD) experiments at
the Co-L2,3 edge of Ca3ChRhO6. We find a minority-
spin d0d2occupation for the HS Co2+ground state and,
thus, a giant orbital moment of about 1.7µB. As will
be seen below, our results account well for the previous
Polycrystalline samples were synthesized by a solid-
state reaction and characterized by x-ray diffraction to
be single phase.5The Rh-L2,3 XAS spectra were mea-
sured at the NSRRC 15B beamline in Taiwan, which is
equipped with a double-Si(111) crystal monochromator
for photon energies above 2 keV. The photon-energy res-
olution at the Rh-L2,3 edge (hν ≈ 3000–3150 eV) was
set to 0.6 eV. The Fe-L2,3XAS spectrum of Ca3FeRhO6
was measured at the NSRRC Dragon beamline with a
photon-energy resolution of 0.25 eV. The main peak at
709 eV of the Fe-L3 edge of single crystalline Fe2O3
was used for energy calibration. The Co-L2,3XAS and
XMCD spectra of Ca3CoRhO6were recorded at the ID8
beamline of ESRF in Grenoble with a photon-energy res-
olution of 0.2 eV. The sharp peak at 777.8 eV of the
Co-L3edge of single crystalline CoO was used for energy
calibration. The Co-L2,3XMCD spectra were recorded in
a magnetic field of 5.5 T; the photons were close to fully
circularly polarized. The sample pellets were cleaved in
situ in order to obtain a clean surface. The pressure was
below 5×10−10mbar during the measurements. All data
were recorded in total-electron-yield mode.
III. XAS AND VALENCE STATE
We first concentrate on the valence of the rhodium
ions in both studied compounds. For 4d transition-metal
oxides, the XAS spectrum at the L2,3 edge reflects ba-
sically the unoccupied t2g- and eg-related peaks in the
Ohsymmetry. This is due to the larger band-like char-
acter and the stronger crystal-field interaction of the 4d
Photon Energy (eV)
Ca3FeRhO6 and a schematic energy level diagram for Rh3+
4d6and Rh4+4d5configurations in octahedral symmetry.
The Rh-L2,3 XAS spectra of Ca3CoRhO6 and
states as well as due to the weaker intra-atomic interac-
tions as compared with 3d transition-metal oxides, where
intra-atomic multiplet interactions are dominant. The
intra-atomic multiplet and spin-orbit interactions in 4d
elements only modify the relative intensity of the t2g-
and eg-related peaks. Fig. 1 shows the XAS spectra
at the Rh-L2,3 edges of Ca3FeRhO6 (dashed line) and
Ca3CoRhO6(solid line). The Rh-L2,3spectrum shows a
simple, single-peaked structure at both Rh-L2 and Rh-
L3edges for Ca3FeRhO6, while an additional low-energy
shoulder is observed for Ca3CoRhO6. Furthermore, the
peak in the Ca3CoRhO6spectrum is shifted by 0.8 eV to
higher energies compared to that of the Ca3FeRhO6.
The single-peaked spectral structure for Ca3FeRhO6
indicates Rh3+(4d6) with completely filled t2g orbitals,
i.e.only transitions from the 2p core levels to the
eg states are possible.The results are in agreement
with M¨ ossbauer spectroscopy.19The shift to higher en-
ergies from Ca3FeRhO6 to Ca3CoRhO6 reflects the in-
crease in the Rh valence from Rh3+to Rh4+as we
can learn from previous studies on 4d transition-metal
compounds.24,25,26,27Furthermore, for Ca3CoRhO6 the
spectrum shows a weak low-energy shoulder, which is
weaker at the Rh-L2edge than at the Rh-L3edge. This
shoulder can be attributed to transitions from the 2p core
levels to the t2gstate, reflecting a 4d5configuration with
one hole at the t2g state. Such spectral features were
found earlier for Ru3+in Ru(NH4)3Cl6.24,28Detailed cal-
culations reveal that the multiplet and spin-orbit inter-
actions suppress the t2g-related peak at the L2edge for
a 4d5configuration.24,25,26,27Thus, we find a Rh4+(4d5)
state for Ca3CoRhO6. Having determined a Rh3+state
in Ca3FeRhO6and a Rh4+state in Ca3CoRhO6, we turn
to the Fe-L2,3 and the Co-L2,3 XAS spectra to further
confirm the Fe3+state and the Co2+state, as expected
for charge balance.
700 705710 715720 725730
Oh: ∆CF = 0.9 eV
∆10 = 0.9 eV; Vmix = 0.4 eV
∆10 = 0.9 eV; Vmix = 0.4 eV
D3h: ∆10 = 0.9 eV
spherical (∆CF = 0)
Oh: ∆CF = 1.0 eV
Oh: ∆CF = 1.6 eV
FIG. 2: (color online) Experimental XAS spectra at the Fe-
L2,3 edge of (a) Fe2O3 (Fe3+), (g) Ca3FeRhO6, and (j) FeO
(Fe2+), taken from Park,29together with simulated spectra
(b, c) in Oh, (d) spherical, and (e, f) D3h symmetry for Fe3+
and simulated spectra in (h) D3h and (i) Oh symmetry for
Figure 2 shows the experimental Fe-L2,3XAS spectra
of (g) Ca3FeRhO6, along with those of (a) single crys-
talline Fe2O3as a Fe3+reference and of (j) FeO, taken
from Ref. 29, as a Fe2+reference. Additionally, calcu-
lated spectra for different symmetries using purely ionic
crystal-field multiplet calculations24,30,31,32are shown.
It is well known that an increase of the valence state
of the 3d transition-metal ion by one causes a shift of
the XAS L2,3 spectra by about one eV towards higher
energies.33,34,35The main peak of the Fe L3structure of
the Ca3FeRhO6lies 0.8 eV above the main peak of the
divalent reference FeO and only slightly lower in energy
than the one of Fe2O3 (Fe3+). This indicates trivalent
iron ions in Ca3FeRhO6. The slightly lower energy shift
of Ca3FeRhO6relative to Fe2O3can be attributed to the
weak trigonal crystal field in the former as compared an
octahedral field in the later, as we will show below.
The experimental spectra of the reference compounds,
curve (a) for Fe2O3and curve (j) for FeO, can be well un-
derstood using the multiplet calculations. For Fe2O3we
find a good simulation taking a Fe3+ion in an octahedral
symmetry with a t2g–egsplitting of 1.6 eV, which is de-
picted in curve (b) in Fig. 2. For FeO, a good match with
the experiment can be found for the Fe2+in an octahe-
dral environment with a splitting of 0.9 eV, see curve (i).
The weaker crystal field in FeO, compared with Fe2O3,
is consistent due to its larger Fe–O bond length.
In order to understand the experimental Fe L2,3spec-
trum of Ca3FeRhO6, we first return to the Fe2O3spec-
trum.When we reduce the t2g–eg splitting from 1.6
eV (curve b) via 1.0 eV (curve c) to 0.0 eV (curve
d), we observe that the the low-energy shoulder be-
comes washed out, while the high-energy shoulder be-
comes more pronounced.30Going further to a trigonal
crystal field, the high-energy shoulder looses its intensity
as shown in curve (e) for a splitting of 0.9 eV between d±1
(dyz/dzx) and d0/d±2(d3z2−r2/dxy/dx2−y2). The experi-
mental Fe-L2,3XAS spectrum of Ca3FeRhO6in Fig. 2(g)
can be well reproduced with this trigonal crystal field of
0.9 eV and in addition a mixing parameter Vmix = 0.4
eV, which mixes the d±2 with the d∓1 orbitals; the re-
sult for this Fe with the 3d5high-spin configuration is
presented in curve (f).
We note that curve (f) has been generated with the
Fe in the trivalent state. As a check, we have also tried
to fit the experimental spectrum of Ca3FeRhO6using a
divalent Fe ansatz. However, the simulation does not
match, as is illustrated in curve (h), in which we have
used the same trigonal crystal field splitting of 0.9 eV
and mixing parameter of 0.4 eV. To conclude, the Fe-L2,3
and Rh-L2,3XAS spectra of Ca3FeRhO6firmly establish
For the Ca3CoRhO6 system, the Rh-L2,3 XAS spec-
tra suggest that the Rh ions are tetravalent, implying
that the Co ions should be divalent.
Co2+/Rh4+scenario we have to study explicitly the va-
lence of the Co ion. Fig. 3 shows the Co-L2,3XAS spec-
tra of Ca3CoRhO6 together with CoO as a Co2+and
Ca3Co2O6as a Co3+reference.17Again we see a shift to
higher energies from CoO to Ca3Co2O6by about one eV.
The Ca3CoRhO6 spectrum lies at the same energy po-
sition as the CoO spectrum confirming the Co2+/Rh4+
scenario21and ruling out the Co3+/Rh3+scenario.20The
result is fully consistent with the above finding from the
Rh-L2,3edge of Ca3CoRhO6and in agreement with pre-
vious results from x-ray photoemission spectroscopy.23
To confirm this
XMCD AND ORBITAL
After determining the valence states of Rh, Fe, and
Co ions we turn our attention to the orbital occupation
and magnetic properties of the Co2+ion at the trigonal-
prism site. This is motivated by the consideration that
Co2+ions may have a large orbital moment,36whose size
depends on details of the crystal field, while the high-spin
Fe3+(3d5) and low-spin Rh3+(4d6) ions in Ca3FeRhO6
Intensity (arb. units)
Ca3CoRhO6 (40 K)
Photon Energy (eV)
FIG. 3: The Co-L2,3 spectra of (a) Ca3Co2O6 (Co3+), (b)
CoO (Co2+), and (c) Ca3CoRhO6. The simulated spectra
of high-spin Co2+(3d7) in trigonal prismatic symmetry are
shown in (d) for a d0d2 and in (e) for a d2d−2 minority-spin
FIG. 4: Scheme of the two possible 3d occupations for a high-
spin Co2+ion in trigonal prismatic symmetry, ignoring the
five up spins. (a) The d0d2 minority-spin occupation allows
for a large orbital magnetic moment, whereas (b) for d2d−2
the orbital moment vanishes.
FIG. 5: (color online) (a) Measured soft x-ray absorption
spectra with parallel (µ+, red dotted curve) and antiparal-
lel (µ−, black solid curve) alignment between photon spin
and magnetic field, together with their difference (XMCD)
spectrum (µ+− µ−, blue dashed curve); simulated XMCD
spectra for (b) d0d2 (olive curve) and (c) d2d−2 (magenta
curve) minority-spin occupation of the high-spin Co2+.
have a closed subshell without orbital degrees of freedom
and thus no orbital moment.
In trigonal-prism symmetry the 3d orbitals are split
into d±1, d0, and d±2 states, see Fig. 4. In terms of
one-electron levels, the d±1orbitals lie highest in energy,
while the lower lying d0, and d±2usually are nearly de-
generate. For a Co3+d6system, it is a priori not obvious
from band structure calculations to say which of these
low lying orbitals gets occupied.
inclusion of the spin-orbit interaction, can become cru-
cial. Indeed, for Ca3Co2O6, it was found from LDA+U
calculations16and confirmed by XMCD measurements17
that the spin-orbit interaction is crucial to stabilize the
occupation of the d2 orbital, thereby giving rise to gi-
ant orbital moments and Ising-type magnetism. For a
Co2+d7ion, however, the situation is quite different. As
we will explain below, the double occupation of the d0d2
orbitals is energetically much more favored than that of
the d2d−2: the energy difference could be of order 1 eV
while the d0and d±2by themselves could be degenerate
on a one-electron level. The consequences are straight-
forward: the double occupation of d0d2, see Fig. 4(a),
should lead to a large orbital moment of 2µB (neglect-
ing covalent effects) and Ising type of magnetism with
the magnetization direction fixed along the chains.7,21In
contrast, the d2d−2, see Fig. 4(b), would have given a
quenched orbital moment.
In order to experimentally establish that the Co2+ion
has the d0d2configuration, we have performed an XMCD
study at the Co-L2,3edges of Ca3CoRhO6. Fig. 5 shows
the Co-L2,3 XMCD spectrum of Ca3CoRhO6 taken at
50 K under 5.5 T. The spectra were taken, respectively,
with the photon spin parallel (µ+, red dotted curve) and
Details, such as the
antiparallel (µ−, black solid curve) to the magnetic field.
One can clearly observe large differences between the two
spectra with the different alignments. Their difference,
µ+− µ−, is the XMCD spectrum (blue dashed curve).
An important feature of XMCD experiments is that there
are sum rules, developed by Thole and Carra et al.,37,38
to determine the ratio between the orbital (morb= Lz)
and spin (mspin = 2Sz) contributions to the magnetic
here, ∆L3 and ∆L2 are the energy integrals of the L3
and L2 XMCD intensity. The advantage of these sum
rules is that one needs not to do any simulations of the
spectra to obtain the desired quantum numbers. In our
particular case, we can immediately recognize the pres-
ence of a large orbital moment in Fig. 5(a), since there is
a large net (negative) integrated XMCD spectral weight.
Using Eq. (1) we find morb/mspin = 0.63.
Co2+ion is quite ionic, mspin is very close to the ex-
pected ionic value of 3µB. For example, our LDA+U
calculations yield 2.72µB for the Co2+ion (2.64µB for
LDA) and Whangbo et al. obtained 2.71µB from GGA
calculations.20Using a value of 2.7µB for the spin mo-
ment, we estimate morb= 1.7µB, in nice agreement with
our LDA+U result of 1.69µB, for the d0d2minority-spin
To critically check our experimental and previ-
ous LDA+U results21regarding the d0d2 orbital oc-
cupation and the giant orbital moment, we explic-
itly simulate the experimental XMCD spectra us-
ing a charge-transfer configuration-interaction cluster
calculation,24,31,32which includes not only the full atomic
multiplet theory and the local effects of the solid, but
also the oxygen 2p–cobalt 3d hybridization. The results
of the calculated Co-L2,3 XAS and XMCD spectra are
presented in Figs. 3(d) and 5(b), respectively. We can
clearly observe that the simulations reproduce the ex-
perimental spectra very well.
are those which indeed give the d0d2orbital occupation
for the ground state. The magnetic quantum numbers
found are morb= 1.65µB and mspin= 2.46µB, yielding
morb/mspin= 0.67 and a total Co magnetic moment of
4.11µB. With the Rh in the S = 1/2 tetravalent state,
the total magnetic moment per formula unit should be
around 5µB. This is not inconsistent with the results
of the high-field magnetization study by Niitaka et al.7:
they found a total moment of 4.05µB, but there the satu-
ration of the magnetization has not yet been reached even
under 18.7 Tesla. This can now be understood since the
magnetocrystalline anisotropy, associated with the active
spin-orbit coupling, is extremely strong and makes it dif-
ficult to fully magnetize a powder sample as was used in
We also have simulated the spectra for the d2d−2sce-
nario. These are depicted in Figs.
3(e) for the XAS
FIG. 6: (color online) Top panel: Occupation number of the
d0, d2, and d−2 orbitals as function of the d0–d±2 splitting
∆02 [Fig. 4(b)]. Middle panel: Orbital and spin moments
(morband mspin) as function of ∆02. Bottom panel: J(J +1),
L(L + 1), and S(S + 1) as function of ∆02.
and 5(c) for the XMCD. It is obvious that the experi-
mental spectra are not reproduced. The simulated line
shapes are very different from the experimental ones and
the integral of the simulated XMCD spectrum yields a
vanishing orbital moment. We therefore can safely con-
clude that the ground state of this material is not d2d−2.
For completeness we mention that the magnetic quantum
numbers found for this d2d−2ansatz are morb= 0.03µB
and mspin= 2.86µB, yielding morb/mspin= 0.01 and a
total Co magnetic moment of 2.89µB.
V.STABILITY OF THE d2d0 STATE
Ca3CoRhO6 has the Co2+d7ion in the doubly oc-
cupied d0d2orbital configuration and not in the d2d−2,
it is interesting to study its stability in more detail. As
already mentioned above, for a Co3+d6ion, the d0and
d±2states can be energetically very close to each other.
For a Co2+d7ion, however, the d0d2 and d2d−2 states
are very much different in energy. This is illustrated in
the top panel of Fig. 6, in which we have calculated the
occupation numbers of the d0, d2, and d−2orbitals as a
function of ∆02, the one-electron level splitting between
the d0 and d±2 orbitals. The d0d2 ground state which
gives the best simulation to the experimental XAS and
XMCD spectra was obtained with ∆02 ≈ 0.4 eV. We
can observe that the d0d2situation is quite stable for a
wide range of ∆02 values, certainly up to 0.8 eV. With
a transition region between ∆02 = 0.8–1.2 eV, we find
a stable d2d−2situation only for ∆02values larger than
1.2 eV. (For the d2d−2 simulations above we have used
∆02= 1.4 eV.) This is a very large number indeed, and
it can be traced back to the multiplet character of the
on-site Coulomb interactions: an occupation of d2d−2
means a much stronger overlap of the electron clouds
as compared to the case for a d0d2. This results in a
higher repulsion energy, which is not a small quantity in
view of the atomic-like values of the F2and F4Slater
integrals determining the multiplet splitting.31,40
In the middle panel of Fig. 6 we also show the ex-
pectation values for morband mspinwhen varying ∆02.
Again we clearly observe that the large orbital-moment
situation is quite stable. To quench the orbital moment
establishedthatthe ground stateof
one would need much higher ∆02 values. Important is
that the spin state does not change here. Bottom panel
of Fig. 6 demonstrates that the high-spin state of the
Co2+ion is not affected by ∆02: the expectation value
?S2? remains constant throughout at a value consistent
with an essentially S = 3/2 state. Obviously, the L2and
J2quantum numbers are strongly affected by ∆02.
To summarize, the Rh-L2,3, Co-L2,3and Fe-L2,3XAS
measurements indicate Co2+/Rh4+in Ca3CoRhO6 and
Fe3+/Rh3+in Ca3FeRhO6. The magnetic properties of
Ca3FeRhO6 are relatively simple as both the HS Fe3+
and LS Rh3+ions have a closed subshell and thus no
orbital degrees of freedom and no orbital moment. The
weak intrachain AFM coupling between the HS Fe ions
can be understood in terms of the normal superexchange
via the intermediate non-magnetic O–Rh–O complex.
For Ca3CoRhO6, the combined experimental and the-
oretical study of the Co-L2,3 XAS and XMCD spectra
reveals a giant orbital moment of about 1.7µB.
large orbital moment is connected with the minority-spin
d0d2 occupation for HS Co2+(3d7) ions in the pecu-
liar trigonal prismatic coordination. The high FM or-
dering temperature in Ca3CoRhO6, compared with that
of Ca3Co2O6, can be attributed to the distinct octahe-
dral sites (which mediate the Co–Co magnetic coupling):
the magnetic Rh4+ion (S = 1/2) in the former and the
nonmagnetic Co3+ion (S = 0) in the latter.
We would like to thank Lucie Hamdan for her skillful
technical and organizational assistance in preparing the
experiment. The research in K¨ oln is supported by the
Deutsche Forschungsgemeinschaft through SFB 608.
1H. Fjellv˚ ag, E. Gulbrandsen, S. Aasland, A. Olsen, and
B. C. Hauback, J. Solid State Chem. 124, 190 (1996).
2S. Aasland, H. Fjellv˚ ag, and B. Hauback, Solid State Com-
mun. 101, 187 (1997).
3H. Kageyama, K. Yoshimura, K. Kosuge, H. Mitamura,
and T. Goto, J. Phys. Soc. Jpn. 66, 1607 (1997).
4H. Kageyama, K. Yoshimura, K. Kosuge, M. Azuma,
M. Takano, H. Mitamura, and T. Goto, J. Phys. Soc. Jpn.
66, 3996 (1997).
5S. Niitaka, H. Kageyama, M. Kato, K. Yoshimura, and
K. Kosuge, J. Solid State Chem. 146, 137 (1999).
6A. Maignan, C. Michel, A. C. Masset, C. Martin, and
B. Raveau, Euro. Phys. J. B 15, 657 (2000).
7S. Niitaka, H. Kageyama, K. Yoshimura, K. Kosuge,
S. Kawano, N. Aso, A. Mitsuda, H. Mitamura, and
T. Goto, J. Phys. Soc. Jpn. 70, 1222 (2001).
8S. Niitaka, K. Yoshimura, K. Kosuge, M. Nishi, and
K. Kakurai, Phys. Rev. Lett. 87, 177202 (2001).
9B. Mart´ ınez, V. Laukhin, M. Hernando, J. Fontcuberta,
M. Parras, and J. M. Gonz´ alez-Calbet, Phys. Rev. B 64,
10E. V. Sampathkumaran and A. Niazi, Phys. Rev. B 65,
11B. Raquet, M. N. Baibich, J. M. Broto, H. Rakoto, S. Lam-
bert, and A. Maignan, Phys. Rev. B 65, 104442 (2002).
12V. Hardy, M. R. Lees, A. Maignan, S. H´ ebert, D. Flahaut,
C. Martin, and D. Mc K. Paul, J. Phys.: Condens. Matter
15, 5737 (2003).
13X. Yao, S. Dong, K. Xia, P. Li, and J.-M. Liu, Phys. Rev.
B 76, 024435 (2007).
14A. Maignan, V. Hardy, S. H´ ebert, M. Drillon, M. R. Lees,
O. Petrenko, D. Mc K. Paul, and D. Khomskii, J. Mater.
Chem. 14, 1231 (2004).
15V. Hardy, D. Flahaut, M. R. Lees, and O. A. Petrenko,
Phys. Rev. B 70, 214439 (2004).
16H. Wu, M. W. Haverkort, Z. Hu, D. I. Khomskii, and L. H.
Tjeng, Phys. Rev. Lett. 95, 186401 (2005).
17T. Burnus, Z. Hu, M. W. Haverkort, J. C. Cezar, D. Fla-
haut, V. Hardy, A. Maignan, N. B. Brookes, A. Tanaka,
H. H. Hsieh, et al., Phys. Rev. B 74, 245111 (2006).
18M. J. Davis, M. D. Smith, and H.-C. zur Loye, J. Solid
State Chem. 173, 122 (2003).
19S. Niitaka, K. Yoshimura, K. Kosuge, K. Mibu, H. Mita-
mura, and T. Goto, J. Magn. Magn. Mater. 260, 48 (2003).
20M.-H. Whangbo, D. Dai, H.-J. Koo, and S. Jobic, Solid
State Commun. 125, 413 (2003).
21H. Wu, Z. Hu, D. I. Khomskii, and L. H. Tjeng, Phys. Rev.
B 75, 245118 (2007).
22M. Loewenhaupt, W. Sch¨ afer, A. Niazi, and E. V. Sam-
pathkumaran, Europhys. Lett. 63, 374 (2003).
23K. Takubo, T. Mizokawa, S. Hirata, J.-Y. Son, A. Fuji-
mori, D. Topwal, D. D. Sarma, S. Rayaprol, and E. V.
Sampathkumaran, Phys. Rev. B 71, 073406 (2005).
24F. M. F. de Groot, Z. W. Hu, M. F. Lopez, G. Kaindl,
F. Guillot, and M. Tronc, J. Chem. Phys. 101, 6570 (1994).
25Z. Hu, H. von Lips, M. S. Golden, J. Fink, G. Kaindl,
F. M. F. de Groot, S. Ebbinghaus, and A. Reller, Phys.
Rev. B 61, 5262 (2000).
26Z. Hu, M. S. Golden, S. G. Ebbinghaus, M. Knupfer,
J. Fink, F. M. F. de Groot, and G. Kaindl, Chem. Phys.
282, 451 (2002).
27R. K. Sahu, Z. Hu, M. L. Rao, S. S. Manoharan,
T. Schmidt, B. Richter, M. Knupfer, M. Golden, J. Fink,
and C. M. Schneider, Phys. Rev. B 66, 144415 (2002).
28T. K. Sham, J. Am. Chem. Soc. 105, 2269 (1983).
29J.-H. Park, Ph.D. thesis, University of Michigan (1994).
30F. M. F. de Groot, J. C. Fuggle, B. T. Thole, and G. A.
Sawatzky, Phys. Rev. B 42, 5459 (1990).
31A. Tanaka and T. Jo, J. Phys. Soc. Jpn. 63, 2788 (2004).
32See the “Theo Thole Memorial Issue”, J. Electron Spec-
trosc. Rel. Phenom. 86, 1 (1997).
33C. T. Chen and F. Sette, Physica Scripta T31, 119 (1990).
34C. Mitra, Z. Hu, P. Raychaudhuri, S. Wirth, S. I. Csiszar,
H. H. Hsieh, H.-J. Lin, C. T. Chen, and L. H. Tjeng, Phys.
Rev. B 67, 092404 (2003).
35T. Burnus, Z. Hu, H. H. Hsieh, V. L. J. Joly, P. A. Joy,
M. W. Haverkort, H. Wu, A. Tanaka, H.-J. Lin, C. T.
Chen, et al., Phys. Rev. B 77, 08xxxx (2008), (accepted).
36G. Ghiringhelli, L. H. Tjeng, A. Tanaka, O. Tjernberg,
T. Mizokawa, J. L. de Boer, and N. B. Brookes, Phys.
Rev. B. 66, 075101 (2002).
37B. T. Thole, P. Carra, F. Sette, and G. van der Laan, Phys.
Rev. Lett. 68, 1943 (1992).
38P. Carra, B. T. Thole, M. Altarelli, and X. Wang, Phys.
Rev. Lett. 70, 694 (1993).
39Parameters (in eV). U3d,3d = 5, U2p,3d = 6.5, ∆ = 4,
= 0.65, Vionic
= −0.2, Vpdσ = −1.024, Hex = 0.045.
The Slater integrals were reduced to 80% of their Hartree-
Fock value. ∆ionic
= 0.4 (d0d2 scenario), ∆ionic
40E. Antonides, E. C. Janse, and G. A. Sawatzky, Phys. Rev.
B 15, 1669 (1977).