The electronic structure of CaCu2 O3 : Spin ladder or 1-dimensional spin chain?

Esther Bordas, Coen de Graaf,∗ and Rosa Caballol

Departament de Quı́mica Fı́sica i Inòrganica, Universitat Rovira i Virgili
Plaça Imperial Tàrraco 1, 43005 Tarragona, Spain

Carmen J. Calzado
Departamento de Quı́mica Fı́sica, Universidad de Sevilla
c/Prof. Garcı́a González s/n, 41012 Sevilla, Spain

Quantum chemical calculations on embedded cluster models have been performed to extract
accurate estimates of the magnetic coupling J and hopping parameters t of CaCu2 O3 . It is shown
that this copper oxide compound is best described as a quasi one-dimensional spin chain with
weak interchain interactions within and between the Cu2 O3 planes. This magnetic structure is not
reflected in the hopping parameters, since we find a large interplane hopping. Hence, the use of
the simple second-order expression that relates J, t, and the on-site repulsion U (J=-4t2 /U ) is not
justified in all cases.

PACS numbers: 75.30.Et, 74.25.Jb, 75.10.Dg, 75.50.Ee

I. INTRODUCTION

The spin ladders form a group of compounds with a

wide variety of interesting phenomena. In relation to
the high critical temperature superconductivity, the cop-
per oxide spin ladders received a lot of attention since
the publication of the crystal structure of SrCu2 O3 and
derivates by Hiroi and co-workers1 in 1991. The even-
legged ladders in this series of compounds show spin-gap
behavior and finite spin-spin correlation length, while
the odd-legged ladders behave as isolated spin chains.2,3
Beside these planar copper oxide spin ladders, other
compounds with similar characteristics have been de- FIG. 1: Crystal structure of CaCu2 O3 . Thick grey lines con-
scribed in the literature being (La,Ca,Sr)14 Cu24 O41 and nect the cluster atoms used to compute the two different in-
terplane interactions. The c, 1 and c, 2 interaction paths are
LaCuO2.5 the most important ones.4–6 The first one ex-
schematically depicted on the left- and right-side, respectively.
hibits Cu2 O3 spin ladder planes similar to SrCu2 O3 com-
bined with CuO2 spin chain layers, whereas in the latter
compound the ladders are oriented in such a way that
planes) was argued to be of less importance being highly
large interladder coupling can be expected which gives
frustrated. The Néel temperature(TN ) is significantly
rise to a three-dimensional magnetic network with long-
higher than in Sr2 CuO3 (5.4 K) and Ca2 CuO3 (11 K),
range order below ∼ 110 K.7
for which the interchain magnetic interactions are of the
Recently, Kiryukhin and co-workers discussed the mag- order of -1 meV.9,10
netic properties of the structurally related CaCu2 O3 .8 Based on x-ray absorption spectroscopy combined with
The buckling of the spin ladder Cu2 O3 planes in this density functional theory (DFT) calculations, Kim et al
compounds reduces the magnetic interactions along the make an attempt to quantify more precisely the mag-
rungs of the ladder and it was argued that CaCu2 O3 is netic interactions in CaCu2 O3 .11 The DFT calculations
actually not a spin ladder, but should be considered as are performed within the local density approximation +
a quasi one-dimensional spin 1/2 chain. Consequently, a U (LDA+U , where U is the on-site repulsion energy for
phase transition is observed at ∼ 25 K, where magnetic two electrons on the same Cu atom). By fitting the re-
ordering sets in. The dominant magnetic interaction is sulting band structure, the hopping parameters were de-
along the legs of the ladders (Jleg ) in the b direction of termined along the rung (trung ∼ 250 meV) and in the
the crystal (see Fig. 1). The high temperature magnetic c-direction between different spin ladder planes (tc ∼ 125
susceptibility was fitted with the theoretical expression meV). From the simple superexchange expression J=-
for a 1D spin-1/2 chain based on the Bethe ansatz with 4t2 /U , the authors estimate the corresponding magnetic
a J-value of -170 meV. Weaker interactions of about 10 coupling parameters. Using U ∼ 3-5 eV, they arrive
meV were assumed along the rung (Jrung ) and along the at Jrung ∼ -50 meV and Jc ∼ ±20 meV. With a ra-
c-direction between different ladder planes (Jc ). The in- tio Jrung /Jleg = 0.3, a pseudoladder magnetic structure
terladder interaction within the spin ladder planes (a-b is proposed and the authors ascribe the disappearance of

Typeset by REVTEX
 2

Madelung potential on a dense grid in the cluster region

with a standard deviation less than 0.1 meV. The choice
of formal ionic charges is consistent with the assignment
of an integer number of electrons to the cluster to en-
sure overall charge neutrality. Furthermore, there exist
by now substantial evidence in the literature that the
electronic structure parameters considered here do not
critically depend on the value of the charges to calculate
the Madelung potential.22–24 This observation only holds
for ionic transition metal compounds. In case of materi-
als with covalent bonds such as CuGeO3 , the embedding
procedure with formal charges does not lead to meaning-
ful results and alternative embedding schemes should be
applied.25,26 In the results section, we will shortly come
back to the influence of the Madelung potential on the
properties studied here.
To avoid artificial polarization of the electronic charge
FIG. 2: Schematic representation of the two-center clusters distribution due to the cluster atoms towards the point
in the a-b plane. charges, total ion potentials (TIPs)27 replace the point
charges on the boundary of the cluster. These TIPs
account for the Coulomb and exchange interactions be-
the spin gap to the relatively large magnetic interaction tween the cluster atoms and the atoms directly around
between different ladder planes. it. The Ca2+ TIP corresponds to the large core potential
Quantum chemical calculations on embedded clusters of Durand and Barthelat28 and the Cu2+ TIP is a modi-
are used here to further quantify the magnetic interaction fication of the original large core potential to include the
and hopping parameters in CaCu2 O3 . The methodology 3d shell in the potential.29 The geometry of the cluster
solves as accurate as possible the exact (non-relativistic) has been taken from experiment.8
Hamiltonian within the material model defined by the The validity of the embedded cluster model to ex-
embedded cluster. In this way, the important electron tract electronic structure parameters in ionic transition
correlation effects present in many transition metal com- metal compounds has been established in two ways. In
pounds can be treated very accurately, and electronic the first place, several studies have been published that
structure parameters can be derived without any further contrast embedded cluster results with periodic calcula-
assumption. This ab initio computational strategy has tions. In all cases, the calculated values are very simi-
been applied very succesfully over the last ten years and lar given that the approximation to the N -electron wave
has reproduced, clarified and even predicted the mag- function is identical in both approaches.30–33 Secondly,
netic interaction and hopping parameters in many tran- cluster size effects have been studied before in other
sition metal compounds (see Refs. 12–21 and references cuprates (Li2 CuO2 , Sr2 CuO3 and La2 CuO4 ) and nickel
therein). compounds (NiO, KNiF3 and K2 NiF4 ) by comparing re-
sults obtained with two, three, four and even five mag-
netic centers. In none of these cases significant effects
II. COMPUTATIONAL INFORMATION were observed.10,12,13,34
Two different computational schemes have been ap-
The clusters used in the calculation of the electronic plied to approximate the exact N -electron wave function
structure parameters contain two or four copper atoms of the cluster electrons. The first scheme is the complete
and all oxygens coordinating them. Figure 2 shows the active space self consistent field (CASSCF), which simul-
two-center clusters that are used to determine the inter- taneously optimizes the spatial extent of the orbitals and
actions along the rung, the leg and between ladders in the wave function expansion in the space spanned by the
the a-b plane. The clusters used to extract information N -electron configurations that can be constructed by dis-
about the interactions between magnetic centers in dif- tributing the unpaired electrons in all possible ways over
ferent a-b planes are depicted in Fig. 1. The four-center the active orbitals. For the undoped clusters, the ac-
cluster contains four Cu atoms placed on the same lad- tive orbitals are mainly localized on the copper centers
der in a rectangular geometry and all the oxygens coor- although they show some delocalization onto the neigh-
dinated to them. All clusters are embedded in optimized boring oxygen ions. The two-center clusters CAS con-
point charges (OPCs) that represent the Madelung po- tains two orbitals and two electrons. In the four-center
tential due to the rest of the crystal. This Madelung cluster, there are four magnetic orbitals and four elec-
potential is calculated by an Ewald summation assuming trons in the CAS. For the doped clusters, we used active
formal ionic point charges, i.e. 2+ for Cu and Ca and 2- spaces with one electron less. The active orbitals turn
for O. The optimized point charges reproduce the exact out to be more delocalized with large contributions from
 3

the neighboring O-2p orbitals.45

TABLE I: CASSCF and IDDCI magnetic coupling J and hop-
The CASSCF approach accounts for the direct ex- ping parameters t (in meV) for CaCu2 O3
change K between the two magnetic centers and the ki-
netic exchange -4t2 /U . However, this description of the d(Cu-Cu) Magnetic coupling Hopping
electronic structure leads to a severe overestimation of CASSCF IDDCI CASSCF IDDCI
U , and consequently too small magnetic coupling param- inter 2.8 Å 3.0 24.3 -124 -143
eters. Secondly, we include the remaining electron cor- rung 3.3 Å -1.9 -11.3 -250 -244
relation effects such as spin polarization, ligand to metal c, 1 3.5 Å 0.02 0.09 -16 -2.8
charge transfer configurations, etc. (see Ref. 35 for a leg 4.1 Å -24.0 -134.1 -720 -622
detailed discussion) with the difference dedicated con- c, 2 4.8 Å 0.09 0.75 -122 -134
figuration interaction (DDCI). This method is specially diag 5.3 Å -0.04 -0.16 -14 -30
designed to obtain accurate relative energies of different
electronic states,36 and has been proven to give magnetic
interaction parameters in close agreement with experi- magnitude of the coupling. Our final IDDCI estimate
ment. The molecular orbital set is obtained by means of the coupling is -134 meV (1610 K). The experimental
of the iterative DDCI (IDDCI) scheme37 to avoid any estimate based on the fitting of the temperature depen-
possible bias toward one of the electronic states. In this dence of the magnetic susceptibility (χ(T )) is -160 ± 25
approach, an average density matrix is constructed from meV (1950 ± 300 K).8 Our ab initio estimate is on the
the DDCI density matrices of all electronic states. The lower limit of this range, but in the fitting of the ex-
vectors that are obtained from the diagonalization of this perimental data only the interaction along the leg was
average density matrix serve as input for a new DDCI cy- considered, the influence of other magnetic interactions
cle. The procedure is repeated until convergence, what on χ(T ) was not accounted for.
usually happens within 4-6 iterations. The interaction along the rung is much weaker because
The basis set to expand the one-electron functions have of the distorted Cu–Or –Cu bond (Or is the oxygen lo-
the following characteristics: The Cu (21s, 14p, 10d, 4f ) cated on the rung) with an angle of 123◦ . The buckling of
primitive set is contracted to (6s, 5p, 4d, 1f ) functions the ladder planes makes the ratio Jrung /Jleg calculated
and the O (14s, 9p, 4d) primitive set is contracted to (4s, with IDDCI less than 0.1, whereas it is close to 1 in the
3p, 1d) functions.38,39 Calculations have been performed structurally related spin ladder compound SrCu2 O3 . In a
with molcas5.440 and casdi.41 computational experiment, we gradually restore the lin-
earity of the Cu–Cu linkage along the rung by varying the
Cu–Or –Cu angle at fixed distances. The embedding is
III. RESULTS kept frozen. As expected, Jrung increases monotonicaly
with increasing bond angle and Jrung /Jleg approaches
A. Magnetic coupling 0.3, as proposed in Ref. 11, for 140◦ . For angles as large
as 170◦ , the ratio is 0.7, still significantly smaller than 1.
Table I lists the magnetic coupling and hopping param- To study the stability of the results with respect to
eters ordered by increasing Cu–Cu interatomic distance. the Madelung potential, we varied the value of the em-
Beside the usual in-plane interactions along the leg, the bedding point charges around the formal ionic value of
rung and between copper atoms situated on different lad- ±2 by increasing and decreasing them with 20%. In line
ders (Jinter and tinter ), we also investigated two different with previous findings,22–24 Jrung is found to be rather
interplane interactions. The interaction between copper stable with the variation of the Madelung potential. It
atoms in different planes separated by 3.5 Å along the c- varies between -13.2 meV and -9.5 meV for the reduced
axis (see Fig. 1, on the left) is labelled with the subscript and increased point charges, respectively.
c,1. A second possible interaction pathway is shown on The interladder interaction in the Cu2 O3 planes is fer-
the right in Fig. 1 and is referred to with the subscript romagnetic and about twice as large as the rung coupling.
c,2. Although the copper atoms are further separated in However, it has recently been mentioned that IDDCI pos-
space, the relative orientation of the CuO4 units with an sibly slightly overestimates ferromagnetic couplings,22,43
oxygen atom connecting both metallic centers could be and the present value of 24 meV should be taken as an
more favorable for magnetic interaction and hopping pro- upper limit. The remaining magnetic coupling within the
cesses than along the c, 1-path parallel to the c-axis. To ladder planes Jdiag has been calculated from a four-center
complete the discussion, we also mention the in-plane in- cluster. Calzado and Malrieu showed that the energy
teractions along the diagonal of the Cu4 O12 plaquettes.42 eigenvalues of the spin wave function no longer give suffi-
The corner sharing CuO4 squares along the legs of the cient information and Jdiag can only be determined with
ladders provide the optimum geometry for a strong mag- the help of effective Hamiltonians.12 The resulting value
netic coupling. Both for CASSCF and IDDCI, the cou- for Jdiag is much smaller than those found in La2 CuO4 12
pling along the leg gives indeed the largest magnetic in- and SrCu2 O3 ,42 which again can be ascribed to the buck-
teraction. For CASSCF, we find -24 meV, but the inclu- ling of the ladder planes.
sion of electron correlation effects strongly enhances the The lack of a bridging ligand and the relative orienta-
 4

tion of the CuO4 units make the magnetic coupling along

TABLE II: On-site repulsion energy (in eV) for different cop-
the c-axis very weak. The CuO4 plaquettes involved in
per oxide compounds estimated from -4t2 /J using IDDCI pa-
Jc,1 are stacked in a parallel way, comparable to the in- rameters (Upert ) and calculated with IDDCI (Uvar )
terchain coupling in the above-mentioned 1D spin chains
Sr2 CuO3 and Ca2 CuO3 . In the present case, however, compound t (meV) J (meV) Upert Uvar
the CuO4 units are displaced with respect to each other CaCu2 O3
by approximately 1.8 Å. This makes the overlap between inter -143 24.3 3 5.8
the magnetic orbitals (mainly of Cu-3dx2 −y2 character) rung -244 -11.3 21 6.7
even less favorable and Jc,1 (IDDCI) is less than 0.1 meV c, 1 -3 0.09 3
leg -622 -134.1 12 6.5
(∼ 1 K), which is significantly smaller than the ∼ -1 meV
c, 2 -134 0.75 96
for Sr2 CuO3 and Ca2 CuO3 . The relative orientation of La2 CuO4 a -598 150 9.5 7.3
the CuO4 units is rather different for Jc,2 . Despite the Li2 CuO2 b
larger Cu–Cu interatomic distance, Jc,2 is larger than NN -143 12.2 6.7
Jc,1 . This is in line with the observation that the mag- NNN 120 1.9 26
netic orbitals are no longer parallel and the rung oxygen Sr2 CuO3 c
provides some type of bridge between the two copper in-chain 659 -246 7.1
atoms involved in the magnetic coupling parametrized inter 30 -0.44 8.3
by Jc,2 . a Values taken from Ref. 30,45,48
b Values refer to in-chain nearest neighbor (NN) and next-nearest
neighbor (NNN) interactions.13
c Values taken from Ref. 10
B. Hopping parameters

The differences in CASSCF and IDDCI hopping pa-

∼125 meV. Finally, we find a hopping parameter of -30
rameters listed in Table I are less pronounced than for
meV along the diagonal of the buckled Cu4 O12 plaque-
the magnetic coupling parameters. The insensitivity of
ttes. This interaction should probably also be included in
the hopping parameter to the exact details of the electron
a simulation of the macroscopic properties of CaCu2 O3 .
correlation treatment has been observed before for other
systems10,44 and implies that this parameter is essentially
a one-electron property. This allows us to compare our
cluster model IDDCI estimates to those obtained from C. Validity of the superexchange relation
the periodic calculations by Kim et al. within the LDA
+U scheme.11 With the ab initio values for t and J at hand, the ques-
The largest hopping is found along the legs of the tion arises to what extent the simple superexchange rela-
ladders. The IDDCI value (-622 meV) is comparable tion U =-4t2 /J can be used to estimate one of the three
to the hopping along similar Cu–O–Cu bonds found in parameters once the other two are known. Several ex-
the two-dimensional antiferromagnets La2 CuO4 45 and amples can be found in the literature where assumptions
related cuprates.30 Whereas Jrung is more than ten times about the relative size of different J’s have been made
smaller than Jleg , the corresponding hopping parameter solely based on the magnitude of the hopping parame-
is only smaller by a factor of 2.5. The IDDCI estimate ters. Beside for CaCu2 O3 , this strategy has also been
of -244 meV is in remarkable good agreement with the applied for Li2 CuO2 and Li2 VOSiO4 .46,47 Table II re-
LDA+U value of ∼250 meV proposed for the rung. compiles the estimates of U based on the superexchange
The interplane hopping parameters can only be ob- formula (Upert ) using J and t obtained with IDDCI. We
tained from the mapping of the IDDCI wave functions add the values for La2 CuO4 , Li2 CuO2 and Sr2 CuO3 . The
onto an effective Hamiltonian. The lack of an inversion on-site repulsion parameter can also be determined vari-
center in the Cu2 O8 clusters (two CuO4 plaquettes in ationally from quantum chemical calculations by means
different ladder planes) makes that the energy eigenval- of the projection of the IDDCI wave functions onto an ef-
ues of the two lowest doublet states are not sufficient fective Hamiltonian as outlined in Ref. 48. We use Uvar
to calculate t. The difference observed for the magnetic for the values obtained by this more accurate procedure.
interactions along the pathways c,1 and c,2, is more pro- It is readily seen that the applicability of the formula
nounced for the hopping parameters: tc,1 is very small, is not universal. For nearest neighbor interactions along
only -3 meV, while tc,2 is an order of magnitude larger (almost) lineal Cu–O–Cu bonds, Upert is at most of the
and similar to the in-plane interladder hopping tinter . right order of magnitude. For other type of interactions,
These results suggest that the band dispersion in the c- the Upert estimates show a large dispersion and unphys-
direction observed by Kim et al. is due to the hopping ical values as large as 96 eV are obtained for the inter-
of electrons (or holes) between CuO4 plaquettes with a action in CaCu2 O3 along c, 2. Uvar is related (but not
relative orientation as shown on the right in Fig. 1, i.e. equal) to the energy difference between the neutral state
the c, 2 interaction path. Our interplane hopping (tc,2 =- (with mainly Cu-3d9 – O-2p6 – Cu-3d9 contributions),
134 meV) is again very close to the LDA+U value of and the ionic state (mainly Cu-3d10 – O-2p6 – Cu-3d8 ).
 5

The variational determined on-site repulsion parameters very reasonable results for the DDCI parameters.10 The
for CaCu2 O3 obtained with different clusters are more application of the same equation for CaCu2 O3 leads to an
consistent than those calculated with the superexchange overestimation of TN by at least an order of magnitude.
relation. We obtain Uvar =5.8, 6.7 and 6.5 eV for the Obviously, other interactions than Jrung and Jc,2 also
in-plane interladder, the rung and the leg clusters, re- play a role in the magnetic structure. The role of the frus-
spectively. trated in-plane interladder interaction is expected to be
The failure of the perturbative expression to relate t small. The Quantum Monte Carlo simulations of John-
with J shows that the reduction of the magnetic coupling ston et al. for the planar spin ladder SrCu2 O3 suggest
constant to its kinetic exchange compound is too crude that this coupling does not influence the magnetic sus-
an approximation. In the first place, it should be noted ceptibility up to Jinter /Jleg =0.2.50 The four-body cyclic
that within the Anderson model, the magnetic coupling exchange (estimated with ab initio calculations42 to be 4
J is the sum of a ferromagnetic term J F , generally as- meV) and Jdiag may play an important role in the mag-
cribed to direct exchange, and an antiferromagnetic term netic structure.
J AF due to the kinetic exchange. Hence, strictly speak- The close resemblance between the theoretical esti-
ing the superexchange formula only relates t with the mates of t from a periodic modelization of the crystal
antiferromagnetic component and not to the magnetic (the LDA+U calculations of Kim et al.11 ) and from a
coupling itself. Secondly, the hopping parameter t as de- cluster model approach (the present IDDCI estimates)
termined from doped clusters does not have exactly the gives additional evidence of the appropriateness of the
same meaning as the t in the superexchange relation. In local model to extract this type of parameters.
the latter relation, t is related to the Hamilton matrix An interesting observation from the calculations is that
element between the neutral and the ionic states,48 while the magnetic structure cannot be directly extracted from
it parametrizes the mobility of the holes in the cluster the size of the hopping parameters along the different
calculations. The same holds for the values derived from interaction paths. The magnetic interaction along the c-
the LDA band structure calculations. axis is much smaller than can be expected at first sight
from the size of the hopping parameter. This behavior is
not unique for CaCu2 O3 but has also been observed for
IV. CONCLUSIONS other cuprates, e.g. Li2 CuO2 . We conclude that only for
similar interaction paths, the magnitude of t serves as a
The IDDCI magnetic interactions parameters suggest guide for the relative size of the corresponding magnetic
that CaCu2 O3 is best described as a quasi 1D spin chain. couplings.
The chains are coupled in the a direction through the dis-
torted rungs and non-zero interplane interactions along
the c-axis, especially Jc,2 . These interactions can be at Acknowledgments
the origin of the long range ordering below the Néel tem-
perature. The magnetic structure of the compound is Financial support has been provided by the Spanish
however not as simple as the related Ca2 CuO3 . In that Ministry of Science and Technology under Project No.
case the mean-field expression49 that relates the magnetic BQU2002-04029-C02-02, and the DURSI of the General-
in-chain and interchain coupling parameters to TN gives itat de Catalunya (grant SGR01-00315).

∗ 10
Electronic address: [email protected] C. de Graaf and F. Illas, Phys. Rev. B 63, 014404 (2001).
1 11
Z. Hiroi, M. Azuma, M. Takano, and Y. Bando, J. Sol. St. T. K. Kim, H. Rosner, S.-L. Drechsler, Z. Hu, C. Sekar,
Chem. 95, 230 (1991). G. Krabbes, J. Málek, M. Knupfer, J. Fink, and H. Eschrig,
2
T. M. Rice, Z. Phys. B 103, 165 (1997). Phys. Rev. B 67, 024516 (2003).
3 12
E. Dagotto, Rep. Prog. Res. 62, 1525 (1999). C. J. Calzado and J.-P. Malrieu, Phys. Rev. B 63, 214520
4
T. Siegriest, S. M. Zahurak, D. W. Murphy, and R. S. (2001).
13
Roth, Nature 334, 231 (1988). C. de Graaf, I. de P. R. Moreira, F. Illas, O. Iglesias, and
5
E. M. McCarron III, M. A. Subramanian, J. C. Calabrese, A. Labarta, Phys. Rev. B 66, 014448 (2002).
14
and R. L. Harlow, Mat. Res. Bull. 23, 1355 (1988). F. Illas, J. Casanovas, M. A. Garcia-Bach, R. Caballol, and
6
Z. Hiroi and M. Takano, Nature 377, 41 (1995). O. Castell, Phys. Rev. Lett. 71, 3549 (1993).
7 15
S. Matsumoto, Y. Kitaoka, K. Ishida, K. Asayama, Z. Hi- A. B. van Oosten, R. Broer, and W. C. Nieuwpoort, Chem.
roi, N. Kobayashi, and M. Takano, Phys. Rev. B 53, 11942 Phys. Lett. 257, 207 (1996).
16
(1996). C. J. Calzado, J. F. Sanz, J.-P. Malrieu, and F. Illas, Chem.
8
V. Kiryukhin, Y. J. Kim, K. J. Thomas, F. C. Chou, R. W. Phys. Lett. 307, 102 (1999).
17
Erwin, Q. Huang, M. A. Kastner, and R. J. Birgeneau, I. de P. R. Moreira, F. Illas, C. J. Calzado, J. F. Sanz, J.-P.
Phys. Rev. B 63, 5122 (2001). Malrieu, N. Ben Amor, and D. Maynau, Phys. Rev. B 59,
9
H. Rosner, H. Eschrig, R. Hayn, S.-L. Drechsler, and 6593 (1999).
18
J. Málek, Phys. Rev. B 56, 3402 (1997). C. de Graaf, I. de P. R. Moreira, I. F., and R. L. Martin,
 6

Phys. Rev. B 60, 3457 (1999). J. Chem. Phys. 116, 2728 (2002).
19 36
D. Muñoz, F. Illas, and I. de P. R. Moreira, Phys. Rev. J. Miralles, J. P. Daudey, and R. Caballol, Chem. Phys.
Lett. 84, 1579 (2000). Lett. 198, 555 (1992).
20 37
N. Suaud and M.-B. Lepetit, Phys. Rev. Lett. 88, 056405 V. M. Garcı́a, O. Castell, R. Caballol, and J.-P. Malrieu,
(2002). Chem. Phys. Lett. 238, 222 (1995).
21 38
L. Hozoi, A. H. de Vries, A. B. van Oosten, R. Broer, P.-O. Widmark, P.-Å. Malmqvist, and B. O. Roos, Theor.
J. Cabrero, and C. de Graaf, Phys. Rev. Lett. 89, 076407 Chim. Acta 77, 291 (1990).
39
(2002). R. Pou-Amérigo, M. Merchán, I. Nebot-Gil, P.-O. Wid-
22
C. de Graaf, L. Hozoi, and R. Broer, J. Chem. Phys. 120, mark, and B. O. Roos, Theor. Chim. Acta 92, 149 (1995).
40
961 (2004). K. Andersson, M. Barysz, A. Bernhardsson, M. R. A.
23
C. Sousa, J. Casanovas, J. Rubio, and F. Illas, J. Comp. Blomberg, D. L. Cooper, T. Fleig, M. P. Fülscher,
Chem. 14, 680 (1993). C. de Graaf, B. A. Hess, G. Karlström, et al., Molcas ver-
24
C. de Graaf, R. Broer, and W. C. Nieuwpoort, Chem. sion 5.4, Department of Theoretical Chemistry, University
Phys. Lett. 271, 372 (1997). of Lund (2002).
25 41
E. Ruiz, J. Cano, S. Alvarez, P. Alemany, and M. Verda- D. Maynau and N. Ben Amor, casdi suite of programs,
guer, Phys. Rev. B 61, 54 (2000). Université Paul Sabatier, Toulouse (1997).
26 42
T. Klüner, N. Govind, Y. A. Wang, and E. A. Carter, J. C. J. Calzado, C. de Graaf, E. Bordas, R. Caballol, and
Chem. Phys. 116, 42 (2002). J.-P. Malrieu, Phys. Rev. B 67, 132409 (2003).
27 43
N. W. Winter, R. M. Pitzer, and D. K. Temple, J. Chem. J. Cabrero, C. de Graaf, E. Bordas, R. Caballol, and J.-P.
Phys. 86, 3549 (1987). Malrieu, Chem. Eur. J. 9, 2307 (2003).
28 44
P. Durand and J. C. Barthelat, Theor. Chim. Acta 38, 283 C. J. Calzado and J.-P. Malrieu, Chem. Phys. Lett. 317,
(1975). 404 (2000).
29 45
F. Illas, J. Rubio, and J. C. Barthelat, Chem. Phys. Lett. C. J. Calzado, J. F. Sanz, and J.-P. Malrieu, J. Chem.
119, 397 (1985). Phys. 112, 5158 (2000).
30 46
D. Muñoz, I. de P. R. Moreira, and F. Illas, Phys. Rev. B R. Weht and W. E. Pickett, Phys. Rev. Lett. 81, 2502
65, 224521 (2002). (1998).
31 47
J. M. Ricart, R. Dovesi, C. Roetti, and V. R. Saunders, H. Rosner, R. R. P. Singh, W. H. Zheng, J. Oitmaa, and
Phys. Rev. B 52, 2381 (1995). S.-L. Drechsler, Phys. Rev. Lett. 88, 186405 (2002).
32 48
I. de P. R. Moreira and F. Illas, Phys. Rev. B 55, 4129 C. J. Calzado, J. Cabrero, J.-P. Malrieu, and R. Caballol,
(1997). J. Chem. Phys. 116, 3985 (2002).
33 49
Y.-S. Su, T. A. Kaplan, S. D. Mahanti, and J. F. Harrison, H. J. Schulz, Phys. Rev. Lett. 77, 2790 (1996).
50
Phys. Rev. B 59, 10521 (1999). D. C. Johnston, M. Troyer, S. Miyahara, D. Lidsky,
34
F. Illas, I. de P. R. Moreira, C. de Graaf, O. Castell, and K. Ueda, M. Azuma, Z. Hiroi, M. Takano, M. Isobe,
J. Casanovas, Phys. Rev. B 56, 5069 (1997). Y. Ueda, et al., unpublished, Cond-Mat/0001147 (2000).
35
C. J. Calzado, J. Cabrero, J.-P. Malrieu, and R. Caballol,