Is ferromagnetism an intrinsic property of the CuII/GdIII couple? 2. Structures and magnetic properties of novel trinuclear complexes with μ-phenolato-μ-oximato (Cu-Ln-Cu) cores (Ln = La, Ce, Gd)

Article abstract of DOI:10.1021/ic000666u

The present paper is devoted to the study of original trinuclear (Cu^II, Ln^III, Cu^II) complexes (L = La, Ce, Gd). They derive from the polydentate ligands H₂Lⁱ (i = 1, 3, 4) represented in Figure 1. The crystal and molecular structures of two complexes have been determined at room temperature. The (Cu, Gd, Cu) complex of H₂L¹ 1Gd and the (Cu, Ce, Cu) complex of H₂L³ 3Ce crystallize in the triclinic space group P1 (no. 2) with the following cell parameters: a = 14.005(2) A, b = 14.7581(13) A, c = 11.3549(13) A, α = 96.273(9)°, β = 97.648(11)°, γ = 72.946(9)°, V = 2217.7(4) A³, and Z = 2 for 1Gd and a = 11.226(2) A, b = 16.927(3) A, c = 11.010(2) A, α = 108.67(2)°, β = 110.48(1)°, γ = 92.35(2)°, V = 1828.7(5) A³, and Z = 2 for 3Ce. Regarding possible supports for magnetic interactions, it may be noted that, in both complexes, each of the main bridging pathways between the equatorial positions of a copper(II) ion and the related lanthanide ion is double and not symmetrical. It involves a phenolato oxygen atom and an oximato nitrogen-oxygen pair of atoms. The resulting Cu(O,N-O)Gd networks are not planar, but 3Ce displays much larger deviations than does 1Gd. Determination of the thermal dependence of χ_M (molar susceptibility) and the field variations of M (magnetization) show that in 3Gd and 4Gd the Cu-Gd interactions are antiferromagnetic while more usual ferromagnetic interactions are observed for 1Gd. The possibility of a relationship between structural and magnetic parameters is considered.

Full text of DOI:10.1021/ic000666u

5994
Inorg. Chem. 2000, 39, 5994-6000
Is Ferromagnetism an Intrinsic Property of the CuII/GdIII Couple? 2. Structures and
Magnetic Properties of Novel Trinuclear Complexes with µ-Phenolato-µ-oximato
(Cu-Ln-Cu) Cores (Ln ) La, Ce, Gd)
Jean-Pierre Costes,* Franc¸oise Dahan, and Arnaud Dupuis
Laboratoire de Chimie de Coordination du CNRS, UPR 8241, lie´e par conventions
a` l’Universite´ Paul Sabatier et a` l’Institut National Polytechnique de Toulouse,
205 route de Narbonne, 31077 Toulouse Cedex, France
ReceiVed June 20, 2000
The present paper is devoted to the study of original trinuclear (CuII, LnIII, CuII) complexes (Ln ) La, Ce, Gd).
They derive from the polydentate ligands H2Li (i ) 1, 3, 4) represented in Figure 1. The crystal and molecular
structures of two complexes have been determined at room temperature. The (Cu, Gd, Cu) complex of H2L1 1Gd
and the (Cu, Ce, Cu) complex of H2L3 3Ce crystallize in the triclinic space group P1h (no. 2) with the following
cell parameters: a ) 14.005(2) Å, b ) 14.7581(13) Å, c ) 11.3549(13) Å, R ) 96.273(9)°, â ) 97.648(11)°, γ
) 72.946(9)°, V ) 2217.7(4) Å3, and Z ) 2 for 1Gd and a ) 11.226(2) Å, b ) 16.927(3) Å, c ) 11.010(2) Å,
R ) 108.67(2)°, â ) 110.48(1)°, γ ) 92.35(2)°, V ) 1828.7(5) Å3, and Z ) 2 for 3Ce. Regarding possible
supports for magnetic interactions, it may be noted that, in both complexes, each of the main bridging pathways
between the equatorial positions of a copper(II) ion and the related lanthanide ion is double and not symmetrical.
It involves a phenolato oxygen atom and an oximato nitrogen-oxygen pair of atoms. The resulting Cu(O,N-
O)Gd networks are not planar, but 3Ce displays much larger deviations than does 1Gd. Determination of the
thermal dependence of øM (molar susceptibility) and the field variations of M (magnetization) show that in 3Gd
and 4Gd the Cu-Gd interactions are antiferromagnetic while more “usual” ferromagnetic interactions are observed
for 1Gd. The possibility of a relationship between structural and magnetic parameters is considered.
Introduction
difference between 1′Gd and 2′Gd concerns the Cu(O,N-O)-
Gd bridging network which is almost planar in the former
complex and bent in the latter one. These results prompt us to
enlarge our study to the trinuclear (Cu, Ln, Cu) complexes
prepared from the polydentate ligands H2Li (i ) 1,3, and 4, cf.
Figure 1). The experimental data support further the view that
the Cu-Gd interaction mediated by a (O, N-O) double bridge
may be either ferromagnetic or antiferromagnetic, depending
on the degree of planarity of the bridging core. Relevant to the
present study we may note two very recent papers.18,19 They
report on the occurrence of antiferromagnetic interactions in
gadolinium-organic radical derivatives.
In a recent paper,1 we have described two binuclear (Cu, Gd)
complexes (1′Gd and 2′Gd in Figure 1), which despite their
formal resemblance exhibit significantly different magnetic
properties. The Cu-Gd interaction which, in both cases, is
mediated by a double (O, N-O) bridge is ferromagnetic in 1′Gd
but antiferromagnetic in 2′Gd. The latter behavior is unprec-
edented since all the previously reported complexes involving
CuO2Gd 2-15 and Cu(O,N-O)Gd cores 1,16,17 are ferromagnetic.
Scrutinizing the structural data shows that the most pertinent
* Corresponding author. E-mail: costes@lcc-toulouse.fr.
(1) Costes, J. P.; Dahan, F.; Dupuis, A.; Laurent, J. P. Inorg. Chem. 2000,
39, 169.
(2) Bencini, A.; Benelli, C.; Caneschi, A.; Carlin, R. L.; Dei, A.; Gatteschi,
D. J. Am. Chem. Soc. 1985, 107, 8128.
(3) Bencini, A.; Benelli, C.; Caneschi, A.; Dei, A.; Gatteschi, D. Inorg.
Chem. 1986, 25, 572.
(4) Bencini, A.; Benelli, C.; Caneschi, A.; Gatteschi, D.; Pardi, L. Inorg.
Chem. 1990, 29, 1750.
Experimental Section
Materials and Methods. All starting materials were purchased from
Aldrich and were used without further purification. Cu(SalOMe)2, L1-
Cu, L4Cu, and L4Ni complexes and 1-(2,4,4-trimethyl-2-imidazolidinyl)-
1-ethanone oxime ligand were obtained as previously described.1,14,20,21
New complexes are described hereafter.
(5) Matsumoto, N.; Sakamoto, M.; Tamaki, H.; Okawa, H.; Kida, S. Chem.
Lett. 1990, 853.
(6) Sakamoto, M.; Hashimura, M.; Matsuki, K.; Matsumoto, N.; Inoue,
K.; Okawa, H. Bull. Chem. Soc. Jpn. 1991, 64, 3639.
(7) Guillou, O.; Bergerat, P.; Kahn, O.; Bakalbassis, E.; Boubekeur, K.;
Batail, P.; Guillot, M. Inorg. Chem. 1992, 31, 110.
(8) Guillou, O.; Oushoorn, R. L.; Kahn, O.; Boubekeur, K.; Batail, P.
Angew. Chem., Int. Ed. Engl. 1992, 31, 626.
(13) Ramade, I.; Kahn, O.; Jeannin, Y.; Robert, F. Inorg. Chem. 1997, 36,
930.
(14) Costes, J. P.; Dahan, F.; Dupuis, A.; Laurent, J. P. Inorg. Chem. 1997,
36, 3429.
(15) Costes, J. P.; Dahan, F.; Dupuis, A.; Laurent, J. P. New J. Chem. 1998,
1525.
(16) Benelli, C.; Blake, A. J.; Milne, P. E. Y.; Rawson, J. M.; Winpenny,
R. E. P. Chem. Eur. J. 1995, 1, 614.
(9) Andruh, M.; Ramade, I.; Codjovi, E.; Guillou, O.; Kahn, O.; Trombe,
J. C. J. Am. Chem. Soc. 1993, 115, 1822.
(10) Blake, A. J.; Milne, P. E. Y.; Thornton, P.; Winpenny, R. E. P. Angew.
Chem., Int. Ed. Engl. 1991, 30, 1139.
(11) Bouayad, A.; Brouca-Cabarrecq, C.; Trombe, J. C.; Gleizes, A. Inorg.
Chim. Acta 1992, 195, 193.
(17) Stemmler, A. J.; Kampf, J. W.; Kirk, M. L.; Atasi, B. H.; Pecoraro,
V. L. Inorg. Chem. 1998, 37, 2807.
(18) Lescop, C.; Luneau, D.; Belorisky, E.; Fries, P.; Guillot, M.; Rey, P.
Inorg. Chem. 1999, 38, 5472.
(12) Costes, J. P.; Dahan, F.; Dupuis, A.; Laurent, J. P. Inorg. Chem. 1996,
35, 2400.
(19) Caneschi, A.; Dei, A.; Gatteschi, D.; Sorace, L.; Vostrikova, K. Angew.
Chem., Int. Ed. Engl. 2000, 39, 246.
10.1021/ic000666u CCC: $19.00 © 2000 American Chemical Society
Published on Web 11/30/2000

Ferromagnetism in the CuII/GdIII Couple
Inorganic Chemistry, Vol. 39, No. 26, 2000 5995
Table 1. Crystallographic Data for 1Gd and 3Ce
1Gd
3Ce
chem. formula
fw
space group
a, Å
b, Å
c, Å
R, deg
â, deg
γ, deg
V, Å3
C32H52Cu2GdN9O20
1167.16
C30H38CeCu2N9O13
999.89
P1h (no. 2)
14.005(2)
14.7581(13)
11.3549(13)
96.273(9)
97.648(11)
72.946(9)
2217.7(4)
2
1.748
0.71073
293
25.15
P1h (no. 2)
11.226(2)
16.927(3)
11.010(2)
108.67(2)
110.48(1)
92.35(2)
1828.7(5)
2
1.816
0.71073
293
24.53
Z
F
calcd, g cm-3
λ, Å
T, K
µ(Mo KR) cm-1
Ra
0.0253, 0.0326
0.0643, 0.0701
0.0332, 0.0442
0.0862, 0.0957
Rwb
2
2
a R ) ∑||Fo| - |Fc||/∑|Fo|. b Rw ) [∑[w(|Fo| - |Fc| )2]/
∑w|Fo2|2]1/2
.
(L1Cu)2Gd(NO3)3‚3H2O. 1Gd. This complex was directly obtained
as crystals from the solution mixture by slow evaporation. Anal. Calcd
for C32H48Cu2GdN9O18: C, 35.4; H, 4.0; N, 12.4. Found: C, 34.9; H,
4.1; N, 12.3. Mass spectrum (FAB+, 3-nitrobenzyl alcohol matrix): m/z
) 1016, [(L1Cu)2Gd(NO3)2]+.
Methods. Elemental analyses were carried out by the Service de
Microanalyse du Laboratoire de Chimie de Coordination, Toulouse (C,
H, N). Magnetic susceptibility data were collected on powdered samples
of the different compounds with use of a SQUID-based sample
magnetometer on a QUANTUM Design Model MPMS instrument. All
data were corrected for diamagnetism of the ligands estimated from
Pascal’s constants.22 Positive FAB mass spectra were recorded in DMF
as a solvent and 3-nitrobenzyl alcohol matrix with a Nermag R10-10
spectrometer.
X-ray Crystallographic Procedures. Crystal data for 1Gd and 3Ce
are presented in Table 1. Data were measured on an Enraf-Nonius
CAD4 diffractometer with Mo KR (λ ) 0.710 73 Å) radiation and ω
- 2θ scans. The temperature of measurement was 293 K. The
reflections were corrected for Lorentz-polarization effects with the
MolEN package.23 Semiempirical absorption corrections24 based on ψ
scans were applied. The structures were solved using a Patterson
procedure with the SHELXS-97 program25 and refined against all Fo2
(SHELXL-97)26 with a weighting scheme w-1 ) σ2(Fo2) + (aP)2 +
bP, where 3P ) (Fo2 + 2Fc2) and a and b are constants adjusted by
the program. All non-hydrogen atoms were refined anisotropically.
Hydrogen atoms were included using a riding model with U equal to
1.1 times Ueq of atom of attachment, except those bonded to the water
molecules in 1Gd, which were allowed to vary isotropically. Atomic
scattering factors were taken from a standard source.27 Structures were
drawn with the ZORTEP28 program. Selected fractional coordinates
are given in Tables 2 (1Gd) and 3 (3Ce).
Figure 1. Schematic representation of the ligands used in this work
(Ln ) La, Ce, or Gd).
L3Cu. A mixture of Cu(SalOMe)2 (1.83 g, 5 × 10-3 mol) and of
1-(2,4,4-trimethyl-2-imidazolidinyl)-1-ethanone oxime (0.85 g, 5 × 10-3
mol) in acetone (20 mL) was heated for 10 min and then left to cool
with stirring. The precipitate that appeared was filtered off and washed
with acetone and diethyl ether. Yield: 1.6 g (95%). Anal. Calcd for
C15H19CuN3O2: C, 53.5; H, 5.7; N, 12.5. Found: C, 53.4; H, 5.6; N,
12.4. Mass spectrum (FAB+, 3-nitrobenzyl alcohol matrix): m/z ) 337,
[L3Cu + 1]+.
Trinuclear Complexes. As the experimental procedures are quite
similar, we will only describe the detailed preparation of one of them
while analytical results will be reported for the entire set of complexes.
(L3Cu)2Gd(NO3)3. 3Gd. A mixture of L3Cu (0.67 g, 2 × 10-3 mol)
and Gd(NO3)3‚5H2O (0.43 g, 1 × 10-3 mol) in acetone (10 mL) was
heated for 10 min and then left to cool with stirring. The precipitate
that appeared was filtered off and washed with acetone and diethyl
ether. Yield: 0.81 g (80%). Anal. Calcd for C30H38Cu2GdN9O13: C,
35.4; H, 3.8; N, 12.4. Found: C, 35.1; H, 3.6; N, 12.1. Mass spectrum
(FAB+, 3-nitrobenzyl alcohol matrix): m/z ) 956, [(L3Cu)2Gd(NO3)2]+.
(L3Cu)2Ce(NO3)3. 3Ce. Anal. Calcd for C30H38CeCu2N9O13: C, 36.0;
H, 3.8; N, 12.6. Found: C, 35.6; H, 3.4; N, 12.3. Mass spectrum (FAB+,
3-nitrobenzyl alcohol matrix): m/z ) 938, [(L3Cu)2Ce(NO3)2]+. Crystals
were obtained by slow evaporation of the mother solution.
(L3Cu)2La(NO3)3‚2H2O. 3La. Anal. Calcd for C30H38Cu2LaN9O13‚
2H2O: C, 35.4; H, 4.0; N, 12.4. Found: C, 34.9; H, 4.1; N, 12.3. Mass
spectrum (FAB+, 3-nitrobenzyl alcohol matrix): m/z ) 935, [(L3Cu)2La-
(NO3)2]+.
Results and Discussion
The original complexes (1Ln, 3Ln, and 4Ln with Ln ) Gd,
La, Ce) described in the present work are trinuclear with a
(22) Pascal, P. Ann. Chim. Phys. 1910, 19, 5.
(23) Fair, C. K. MolEN. Structure Solution Procedures; Enraf-Nonius: Delft,
Holland, 1990.
(24) North, A. C. T.; Phillips, D. C.; Mathews, F. S. Acta Crystallogr.,
Sect. A 1968, A24, 351.
(25) Sheldrick, G. M. SHELXS-97. Program for Crystal Structure Solution;
University of Go¨ttingen: Go¨ttingen, Germany, 1990.
(26) Sheldrick, G. M. SHELXL-97. Program for the refinement of crystal
structures from diffraction data; University of Go¨ttingen: Go¨ttingen,
Germany, 1997.
(L4Cu)2Gd(NO3)3. 4Gd. Anal. Calcd for C22H34Cu2GdN9O13: C,
28.8; H, 3.7; N, 13.7. Found: C, 28.8; H, 3.5; N, 13.6. Mass spectrum
(FAB+, 3-nitrobenzyl alcohol matrix): m/z ) 856, [(L4Cu)2Gd(NO3)2]+.
(L4Cu)2La(NO3)3. 4La. Anal. Calcd for C22H34Cu2LaN9O13‚
C3H6O: C, 31.4; H, 4.2; N, 13.2. Found: C, 31.5; H, 4.1; N, 13.2.
Mass spectrum (FAB+, 3-nitrobenzyl alcohol matrix): m/z ) 836, [(L4-
Cu)2La(NO3)2]+.
(27) International Tables for Crystallography; Kluwer Academic Publish-
ers: Dordrecht, The Netherlands, 1992; Vol. C.
(28) Zsolnai, L.; Pritzkow, H.; Huttner, G. ZORTEP. Ortep for PC, Program
for Molecular Graphics; University of Heidelberg, Heidelberg,
Germany, 1996.
(20) Costes, J. P.; Dahan, F.; Dupuis, A.; Laurent, J. P. J. Chem. Soc.,
Dalton Trans. 1998, 1307.
(21) Costes, J. P.; Dahan, F.; Dupuis, A.; Laurent, J. P. New J. Chem. 1997,
21, 1211.

5996 Inorganic Chemistry, Vol. 39, No. 26, 2000
Costes et al.
Table 2. Selected Atomic Coordinates and Equivalent Isotropic
Displacement Parameters (Å2 × 100) for
[{L1Cu(H2O)}2Gd(H2O)(NO3)](NO3)2 (H2O)2, 1Gd
a
atom
x/a
y/b
z/c
Ueq
Gd
0.34209(1)
0.26063(2)
0.26549(2)
0.2218(1)
0.1541(1)
0.4256(1)
0.2967(1)
0.4076(1)
0.3423(1)
0.4311(2)
0.2808(2)
0.3774(2)
0.5088(1)
0.1450(2)
0.4277(2)
0.1714(2)
0.3081(2)
0.3881(2)
0.1762(2)
0.2268(2)
0.3033(1)
0.3626(2)
0.44952(1)
0.27275(2)
0.70079(2)
0.4017(1)
0.5392(1)
0.2869(1)
0.5759(1)
0.4161(1)
0.5774(1)
0.3831(1)
0.3737(2)
0.2866(3)
0.4690(2)
0.2365(2)
0.7067(3)
0.2930(2)
0.1446(2)
0.2324(1)
0.7619(1)
0.8250(1)
0.6659(1)
0.3473(2)
0.11235(1)
0.24048(3)
0.20899(2)
0.1998(1)
0.0629(2)
0.1142(2)
0.2608(1)
0.3349(1)
-0.0016(1)
-0.0748(2)
-0.0763(2)
-0.2046(3)
0.1570(2)
0.0862(2)
0.2921(3)
0.3644(2)
0.2878(2)
0.1686(2)
0.3305(2)
0.1509(2)
0.0443(2)
-0.1216(2)
2.470(4)
3.328(7)
3.283(7)
2.99(3)
3.46(4)
3.65(4)
3.00(3)
3.68(4)
3.15(3)
4.98(5)
6.96(7)
9.8(1)
4.03(4)
6.39(6)
7.57(8)
3.41(4)
4.16(5)
3.28(4)
3.35(4)
4.22(5)
2.91(4)
6.27(8)
Cu(1)
Cu(2)
O(1)
O(2)
O(3)
O(4)
O(5)
O(6)
O(7)
O(8)
O(9)
O(10)
O(11)
O(12)
N(1)
N(2)
N(3)
N(4)
N(5)
N(6)
N(7)
Figure 2. Zortep plot for 1Gd with ellipsoids drawn at the 50%
probability level.
Cu(O,N-O)Ln(O,N-O)Cu core while the previously reported
complexes1 (1′Gd and 2′Gd) are binuclear. It may be noted
that the H2L1 ligand may lead to a binuclear 1′Gd or a trinuclear
1Gd complex depending on the ratio of reactants. By contrast,
H2L3 and H2L4 only yield trinuclear species while a binuclear
complex is solely obtained from H2L2. Relevant to the present
study are the works quoted in refs 16 and 17. They concern
polynuclear CunLn complexes involving 5-chloropyridone as
ligand and metallacrown complexes, respectively. Both types
of compounds include bridging networks of the Cu(O,N-O)-
Ln sort.
a Ueq ) one-third of the trace of the orthogonalized Uij tensor.
Table 3. Selected Atomic Coordinates and Equivalent Isotropic
Displacement Parameters (Å2 × 100) for [(L3Cu)2Ce(NO3)3], 3Ce
a
atom
x/a
y/b
z/c
Ueq
Ce
0.22987(2)
0.01351(3)
0.35131(4)
0.0194(2)
0.4226(2)
0.2870(2)
0.2099(2)
0.1039(2)
0.3094(3)
0.1917(4)
0.2385(2)
0.4250(2)
0.4124(3)
0.0189(3)
0.1890(2)
0.0022(3)
-0.1691(2)
0.0140(3)
0.2004(2)
0.4430(3)
0.2916(3)
0.2339(3)
0.2024(3)
0.3597(3)
0.0667(3)
0.25464(1)
0.10845(2)
0.46028(2)
0.1432(1)
0.3786(1)
0.1354(2)
0.2966(2)
0.2657(2)
0.3044(2)
0.2909(2)
0.1288(2)
0.1968(2)
0.0929(2)
0.3243(2)
0.4094(2)
0.4481(2)
0.1033(2)
0.0730(2)
0.1133(2)
0.5602(2)
0.5393(2)
0.3801(2)
0.2879(2)
0.1381(2)
0.3952(2)
0.51767(2)
0.55419(4)
0.53756(4)
0.4082(2)
0.6166(2)
0.6003(3)
0.3228(3)
0.6871(3)
0.7958(3)
0.9088(3)
0.3044(3)
0.4573(3)
0.2747(3)
0.4591(4)
0.6192(3)
0.5820(4)
0.5096(3)
0.7041(3)
0.6445(3)
0.6967(3)
0.4451(3)
0.3558(3)
0.8007(3)
0.3433(3)
0.5542(4)
3.07(1)
2.97(1)
3.56(1)
3.36(5)
3.34(5)
3.79(5)
4.34(6)
4.63(6)
4.96(6)
7.4(1)
4.15(5)
4.27(5)
5.63(7)
6.15(8)
4.64(6)
8.2(1)
3.21(5)
3.63(6)
3.25(6)
3.71(6)
4.37(7)
3.89(6)
4.72(7)
3.51(6)
5.13(8)
From the data of chemical analysis and mass (FAB+)
spectroscopy, the complexes under study may be represented
Cu(1)
Cu(2)
O(1)
O(2)
O(3)
O(4)
O(5)
O(6)
O(7)
O(8)
O(9)
O(10)
O(11)
O(12)
O(13)
N(1)
N(2)
N(3)
N(4)
N(5)
N(6)
N(7)
N(8)
N(9)
by the formulas L1 Cu2Gd(H2O)3(NO3)3 and Li2Cu2Ln(NO3)3
2
(i ) 3 and 4; Ln ) Gd, La, Ce). Almost identical FAB+ spectra
were obtained for the homologous complexes of H2L3 and H2L4
with signals attributable to the [(LiCu)2Ln(NO3)2]+ (100%) and
[LiCuLn(NO3)2]+ (15-20%) species in both series (i ) 3 and
4). Crystal and molecular structures were determined by X-ray
crystallographic analysis in the case of 1Gd and 3Ce for which
appropriate crystals have been obtained.
Description of the 1Gd Structure. One distinctive feature
of the 1Gd structure is that the trinuclear network is not neutral
but dicationic. It is well represented by the formula [(L1Cu-
(H2O))2Gd(H2O)(NO3)]2+. In addition to two trinuclear entities,
the unit cell contains four uncoordinated nitrato ions and four
unbound water molecules. A view of the bimetallic cation is
represented in Figure 2 while relevant bond lengths and angles
of the Cu(II) and Gd(III) environments are quoted in Table 4.
The cation has no crystallographic symmetry; the bond lengths
and angles of the two (L1Cu)Gd/2 moieties differ little.
The central gadolinium ion is linked to each copper ion by a
double (O, N-O) bridge (O standing for the phenolato oxygen
atom and N-O for the oximato group of the L1 ligand). The
five atoms involved in each Cu(O,N-O)Gd bridge are not
exactly coplanar. The dihedral angles (R) between the (OCuN)
and (OGdO) planes are equal to 7.3(3)° and 8.8(2)° in the
Cu(1) and Cu(2) moieties, respectively. The related Cu‚‚‚Gd
separations are equal to 3.6388(4) and 3.6328(3) Å. A sort of
third bridge results from the fact that each OMe sidearm which
intrinsically belongs to the mononuclear (L1Cu) entity is
coordinated to the Gd ion. The polyatomic pathway which
connects the two metal ions is too extended to support a
significant magnetic interaction but, it can contribute to the
stability of the trinuclear species.
a Ueq ) one-third of the trace of the orthogonalized Uij tensor.
ion achieves its environment with three oxygen atoms coming
from a bidentate (η2-coordination) nitrato ion and a water
molecule. As previously noted, the Gd-O bond lengths depend
on the nature of the oxygen atoms; they vary from 2.371(2) to
2.620(2) Å with a mean value of 2.460 Å. The bonds issued
from the oximato and phenolato oxygens are shorter than those
from the water molecules, the largest ones being related to the
methoxy sidearms and the nitrato ions.
Each copper ion has a classical square-pyramidal environ-
ment. The four basal donors (N3O) are afforded by a L1 ligand,
while a water molecule occupies the apical position. As usual,
the apical Cu-O bonds with a mean value of 2.357 Å are longer
than the equatorial ones (mean value 1.908 Å).
The gadolinium ion is nine-coordinated. In addition to the
six oxygen atoms afforded by two L1 ligands, the rare earth

Ferromagnetism in the CuII/GdIII Couple
Inorganic Chemistry, Vol. 39, No. 26, 2000 5997
Table 4. Selected Bond Lengths (Å), Distances (Å), and Angles
(deg) for [{L1Cu(H2O)}2Gd(H2O)(NO3)](NO3)2(H2O)2, 1Gd, and for
[(L3Cu)2Ce(NO3)3], 3Cea
1Gd
3Ce
Cu-Ophenolato
Cu-N(1,4)
Cu-N(2,5)
Cu-N(3,6)
Cu-Owater
Cu-Onitrato
Ln-Onitrato
Ln-Ophenolato
Ln-Ooximato
Ln-Omethoxy
Ln-Owater
1.906(2)-1.911(2)
1.946(2)-1.952(2)
1.923(2)-1.917(2)
1.966(2)-1.985(2)
2.351(2)-2.363(3)
1.901(2)-1.919(2)
1.924(3)-1.921(3)
1.927(3)-1.934(3)
1.968(3)-1.962(3)
2.526(2)-2.550(3)
2.627(3)-2.694(3)
2.603(2)-2.604(2)
2.486(2)-2.410(2)
2.476(2)-2.562(2)
2.371(2)-2.372(2)
2.342(2)-2.403(2)
2.594(2)-2.620(2)
2.416(2)
7.3(3)-8.8(2)
3.6388(4)
Rb
46.5(1)-45.3(1)
3.6070(5)
3.5967(4)
Ln‚‚‚Cu(1)
Ln‚‚‚Cu(2)
3.6328(3)
Figure 3. Zortep plot for 3Ce with ellipsoids drawn at the 50%
probability level.
a Ln ) Gd for 1Gd and Ce for 3Ce. b Dihedral angle between
O-Ln-O and N-Cu-O planes.
lined that, in contrast to the main bridging network Cu(O,N-
O)Ce, the NO3 bridge in 3Ce is not able to mediate any
significant intermetallic interaction. Indeed, it concerns an axial
position of the copper(II) ion which possesses a vanishingly
small spin density. The five atoms of each Cu(O,N-O)Ce
framework are far from being coplanar. The dihedral angle R
between the (OCuN) and (OCeO) planes takes the values
46.5(1)° (Cu(1)) and 45.3(1)° (Cu(2)) which are not very
different from the value (39.1(1)°) observed in 2′Gd1 but much
larger than those found in 1′Gd (6.1(3)°)1 and 1′Ce.30
The intramolecular Cu(1)‚‚‚Cu(2) separation of 7.0149(6) Å
is within the normal range of values for polynuclear (Cu-Ln)
complexes. By contrast, abnormally small values of 3.4417(6)
and 3.989(1) Å are observed for the intermolecular Cu(1)‚‚‚
Cu(1)i and Cu(2)‚‚‚Cu(2)j distances respectively, i and j standing
here for the (-x, -y, 1 - z) and (1 - x, 1 - y, 1 - z) symmetry
operations. Regarding the possibilities of magnetic interaction,
one may note that in both cases the Cu‚‚‚Cui,j vector is almost
perpendicular to the related equatorial coordination planes and
therefore does not mediate any significant magnetic interaction.
Furthermore the axial sites of the Cu(II) ions are known to
support vanishingly small spin densities.
Both copper(II) ions are five-coordinated. As expected, the
axial Cu-O bonds involved in the nitrato bridges are longer
(mean value 2.538 Å) than the equatorial Cu-O (mean value
1.910 Å) and Cu-N (mean value 1.939 Å) bonds. The
lanthanide ion is 10-coordinated. In addition to the four oxygen
atoms afforded by two monometallic units (L3Cu), Ce(III)
completes its surrounding with six oxygens from three bidentate
NO3 anions. Out of these six oxygen atoms, two are also linked
to a Cu(II) ion (see above). The shortest lanthanide-oxygen
bonds (2.410(2) and 2.486(2) Å) are issued from the oximato
groups. The lengths of the other Ce-O bonds vary from
2.603(3) to 2.694(3) Å, the largest values being related to the
bonds involving the nitrato anions.
Bound and unbound nitrato ions and water molecules are
involved in an extensive network of hydrogen bonds possibly
leading to a small shortening of the intermolecular Gd‚‚‚Gd (1
- x, 1 - y, -z) distance (5.9996(3) Å). Other intermolecular
Gd‚‚‚Cu, Gd‚‚‚Gd, and Cu‚‚‚Cu separations are larger than ca.
6.4 Å, while the Cu(1)...Cu(2) distance within the trinuclear
entity is equal to 6.3867(4) Å.
Finally, the two halves (L1Cu)Gd/2 of 1Gd offer many
similarities to each other and to their binuclear analogue 1′Gd.1
The related bond lengths and angles are not fundamentally
different. We note that the dihedral angle R characterizing the
bridging network scarcely varies from 7.3(3)° and 8.8(2)° in
1Gd to 6.1(3)° in 1′Gd while the related Cu‚‚‚Gd distance
decreases from 3.6388(4) and 3.6328(3) Å to 3.6210(3) Å. The
major differences between the trinuclear and binuclear com-
plexes originate in the nature of the actual polynuclear species
(cationic vs neutral) and the coordination number of the
gadolinium ion (nine vs ten).
Description of the 3Ce Structure. We have failed to obtain
well-shaped crystals of the gadolinium complex 3Gd, but we
have succeeded in doing so for the isomorphous cerium complex
(L3Cu)2Ce(NO3)3 3Ce.29 The unit cell contains four trinuclear
entities without any neutral molecule (water or organic solvent).
Each entity is neutral and has no crystallographic symmetry.
One molecular unit is represented in Figure 3, while relevant
bond distances and angles are reported in Table 4.
The two (L3Cu)Ce/2 halves which form the molecular unit
are not identical, but the bond length differences are small. The
largest one (0.076 Å) affects the bond between the cerium ion
and the oximato oxygen atom. The other differences in the Ce-
(III) and Cu(II) environments are at the best equal to 0.030 Å.
An unexpected feature of the structure is that the cerium ion is
triply connected with each copper ion. In addition to the usual
(O, N-O) bridging pathways, a third bridge is afforded by a
bidentate NO3 ion. In each (Cu-Ce) pair, an axial position of
Cu(II) is occupied by a nitrato oxygen atom (O(5) or O(12))
which is already linked to Ce(III). This third bridge is likely
responsible for the shortening of the intramolecular Cu‚‚‚Ce
distances (3.6070(5) and 3.5967(4) Å) compared to the value
(3.6753(6) Å) observed for the related binuclear complex
L1Cu(H2O)Ce(H2O)(NO3)3 1′Ce.30 However, it may be under-
In the series of complexes formed by L4Cu and rare earth
ions, the lack of suitable crystals prevents any structural
determination to be made. The data from chemical analysis and
mass (FAB+) spectroscopy suggest that the structure of these
complexes does not comprise any solvent (organic or water
molecules) linked to the metal ions. They probably have
structural features similar to those of (L3Cu)2Ce(NO3)3, each
pair of metal ions (Cu, Ln) involving a triple bridge and large
values of the dihedral angle between the two halves of the main
bridging pathway Cu(O,N-O)Ln.
(29) Powder diffraction spectra establish an isomorphism relationship for
3Ce and 3Gd.
(30) The structural data characterizing 1′Ce will be published elsewhere.
However, it is interesting to note that 1′Ce and 1′Gd are isomorphous.
Magnetic Study. The magnetic study concerns the (Cu, La,
Cu) and (Cu, Gd, Cu) triads to which a spin-only formalism

5998 Inorganic Chemistry, Vol. 39, No. 26, 2000
Costes et al.
Table 5. Parameters Deduced from the Magnetic Susceptibility
Data of the Different Complexes
JCuGd (cm-1
)
J′CuCu (cm-1
)
gCu
gGd
Ra
1Gd
3Gd
2.5
-1.85
-1.0
3.5
0
0
0
2.10
2.10
2.10
2.05
2.11
1.99
2.01
2.02
1.99
1.99
5 × 10-5
4 × 10-5
7 × 10-5
1 × 10-4
5 × 10-5
4Gd
1′Gdb
2′Gdb
-0.5
a R ) Σ[(øMT)obs - (øMT)calc]2/Σ[(øMT)obs]2. b Cf. ref 1.
complexes are essentially dependent on interactions between
the GdIII and CuII ions. These interactions are ferromagnetic in
1Gd and antiferromagnetic in 3Gd and 4Gd.
Figure 4. Thermal dependence of øMT for 1Gd at 0.1 T. The full line
corresponds to the best data fit.
From the energies E(S,S′) reported above and the equations
relating local and molecular g values,33 it is straightforward to
express the magnetic susceptibility as a function of T and four
parameters,2,9 JCuGd, JCuCu, gCu, and gGd
:
øMT ) [Nâ2/4kT][T/(T - θ)][A/B]
with
A ) [165g(9/2,1)2 +84g(7/2,1)2 exp(9J/2kT) +
84g(7/2,0)2 exp((7J + 2J′)/2kT) + 35g(5/2,1)
Figure 5. Thermal dependence of øMT for 3Gd at 0.1 T. The full line
corresponds to the best data fit.
2
exp(8J/kT)] and B ) [5 + 4 exp(9J/2kT) +
4 exp((7J + 2J′)/2kT) + 3 exp(8J/kT)]
may be applied. Indeed LaIII with a 1S0 ground state is
8
diamagnetic while GdIII with a S7/2 ground state is devoid of
first-order angular momentum.
θ gauges eventual second-order effects (intermolecular cou-
pling, zero field splitting, ...). Least-squares fit of the experi-
mental data with the above relation leads to the values quoted
in Table 5. For comparison, we have reported the parameters
previously obtained for the binuclear complexes 1′Gd and 2′Gd.
We can see that these results are very coherent, particularly for
1Gd and 1′Gd.
Two (Cu, La, Cu) complexes are available, 3La and 4La.
The related complex 1La cannot be prepared. For 3La, the
thermal dependence of øMT is characteristic of a small antifer-
romagnetic interaction since no maximum appears in the øM vs
T curve until 2 K. Least-squares fit of the experimental data
with the equation valid for a copper(II) pair31 leads to JCuCu
)
-1.8 cm-1. The interaction is also antiferromagnetic in 4La,
It may be noted that in the trinuclear complexes 1Gd, 3Gd,
and 4Gd the best fitted value for J′CuCu is 0. The magnetic study
of the (Cu, La, Cu) analogues confirms this result in the case
but its magnitude is much smaller |JCuCu| e 0.1 cm-1
.
In keeping with the occurrence of two types of intramolecular
interactions, the analysis of the magnetic properties of (Cu, Gd,
of 4La but leads to a different evaluation (J′CuCu ) -1.8 cm-1
)
Cu) triads is based on the Hamiltonian H ) -2JCuGdSCuSGd
-
in the case of 3La. If the 3Gd data fit is restricted to three
variables (JCuGd, gCu, and gGd), the fourth parameter (J′CuCu
being held equal to -1.8 cm-1, we obtain JCuGd ) -1.75 cm-1
J′CuCuSCuSCu. It is assumed that the terminal copper centers are
equivalent. The energies E(S,S′) of low-lying spin states can be
expressed as2,9,32
)
,
gCu ) 2.10, and gGd ) 2.01 with a R factor equal to ) 4 ×
10-5. Owing to the experimental uncertainties, these values are
not basically different from those reported in Table 5. The low
sensitivity of øM on the magnitude of the Cu, Cu interaction
may probably be related to the fact that all of the E(S,S′) energies
but one, E(7/2,0), do not depend on this interaction. From the
values quoted in Table 5 we can deduce the energy spectra for
the low-lying spin states of the (Cu, Gd, Cu) complexes. They
are shown in Figure 6 which also contains the 3Gd spectrum
obtained with the second set of parameters. In addition to the
reversal of energy levels on going from 1Gd to 3Gd and 4Gd,
E(9/2,1) ) 0; E(7/2,1) ) 9JGdCu/2;
E(7/2,0) ) 7JGdCu/2 + J′CuCu
;
E(5/2,1) ) 8JGdCu
(7/2,1) and (7/2,0) refer to the states resulting from the coupling
of SGd with S′ ) 1 and 0, respectively, S′ being the intermediate
spin obtained by coupling the two SCu.
The temperature dependence of the experimental values of
the product øMT (Figures 4 and 5) clearly differentiates 1Gd
from 3Gd and 4Gd. For the three complexes øMT is constant
from 300 to ca. 100 K and corresponds to the value (8.6 cm3 K
mol-1) anticipated for three uncoupled ions. As the temperature
is lowered further, øMT for 1Gd increases regularly to reach a
value consistent with a 9/2 spin state at 2 K while, for 3Gd and
4Gd, øMT decreases. The values observed at 2 K are slightly
it appears that the spectrum width is smaller for 4Gd (8.0 cm-1
)
than for 3Gd (14.5 cm-1) and 1Gd (20.0 cm-1) so that, in the
former complex, the populations of the (7/2,0) and (7/2,1) excited
spin states are not negligible. At 2 K they are equal to 7.2%
and 3.5%, respectively, while that of the (5/2,1) ground state is
89%.
5
larger than expected for a /2 spin state, indicating that higher
Additional information may be gained from an analysis of
the field (H) dependence of the magnetization (M). In the case
spin state(s) may be operative in addition to the antiferromag-
netic ground state. This point will be considered later, but it
does not invalidate the conclusion that the behaviors of the three
(32) Griffith, J. S. Struct. Bonding (Berlin) 1972, 10, 87.
(33) Bencini, A.; Gatteschi, D. In Magneto-Structural Correlations in
Exchange Coupled Systems; Willet, R. D., Gatteschi, D., Kahn, O.,
Eds.; D. Reidel: Dordrecht, The Netherlands, 1985.
(31) Bleaney, B.; Bowers, K. D. Proc. R. Soc. London, Ser. A 1952, 214,
451.

Ferromagnetism in the CuII/GdIII Couple
Inorganic Chemistry, Vol. 39, No. 26, 2000 5999
(SGd and two SCu), respectively, at 2 K. The magnetization
increases less than expected for uncorrelated ions, confirming
the antiferromagnetic nature of the interaction. However, the
experimental points are located above the B.f.(5/2) plot. When
the field is raised, the difference between the two sets of values
increases and the magnetization approaches the value anticipated
for three uncoupled spins. This behavior is presumably due to
two factors which converge on enlarging the participation of
5
spin states higher than /2. These factors are the presence of
excited spin states close in energy to the ground state (cf. Figure
6) and the decoupling effect of the magnetic field. Thus the
magnetization at 2 K and H ) 0.5 × 104 G would be
approximated by the sum of the contributions supplied by
respectively the (Cu, Gd, Cu) triads in the ST ) 5/2 state (relative
population of ca. 80%), the triads with ST ) 7/2 (ca. 11%), and
the triads comprising noninteracting spins (ca. 9%). The resulting
value of 2.2 µB compares reasonably with the measured value
of 2.4 µB. Similar comments can be made for 3Gd. In this
complex the difference between the observed magnetization and
the values calculated for a ST ) 5/2 state are smaller than in the
case of 4Gd in accordance with a larger gap between the ground
and excited states (Figure 6) and a larger magnitude of the
antiferromagnetic interaction (1.8 vs 1.0 cm-1).
Figure 6. Energy states for the low-lying spin states of 1Gd, 3Gd,
and 4Gd at 2 K. E(9/2,1) is arbitrarily taken equal to zero in the three
cases. (a) Parameter values as indicated in Table 5. (b) JCuGd ) -1.75
cm-1; JCuCu ) -1.8 cm-1; gCu ) 2.10; gGd ) 2.01.
A point we wish to emphasize concerns the possible relation-
ship between the magnetic properties of the (Cu, Gd) pairs and
the bending of the bridging network. If we consider the values
of the dihedral angle R between the (OCuN) and (OGdO) planes,
it appears that the ferromagnetism of the Cu-Gd entities in
1Gd and 1′Gd (J ) 2.5 and 3.5 cm-1, respectively) tallies with
small R values (7.3(3)°, 8.8(2)°, and 6.1(3)°, respectively).
Conversely, large R values (from ca. 40° to 46°) characterize
2′Gd and 3Gd (judging by the isomorphous 3Ce complex),
which display antiferromagnetic Cu-Gd interactions (J from
-0.5 to -1.8 cm-1).
Figure 7. Field dependence of the magnetization for 1Gd. The full
line corresponds to the Brillouin function for a S ) /2 spin state.
The ferromagnetic behavior of the (Cu-Gd) pair has been
9
attributed9,13,35 to coupling between the electronic (3dCu-4fGd
)
ground configuration and the excited configuration arising from
a 3dCu-5dGd electron transfer. In any case this mechanism
stabilizes the resulting S ) 4 spin state with respect to the S )
3 one. However, the experimental data obtained for complexes
with a (CuO2Gd) core show that, as the dihedral angle R
becomes larger, the magnitude of the interaction decreases and
tends to vanish when R approaches a value of the order of 40°.34
It is not unrealistic to assume that for 2′Gd, 3Gd, and 4Gd the
ferromagnetic contribution is reduced to such an extent that an
underlying weak antiferromagnetism becomes perceptible. To
account for this antiferromagnetism, two mechanisms are
available: the Heitler-London type interaction9,36,37 and/or the
Anderson mechanism.9,38,39 In the former approach the antifer-
romagnetism of a (Cu-Gd) pair would arise in the ground
configuration (3dCu-4fGd) from overlap between natural orbitals.
The latter model is based on the interaction between the ground
configuration and the excited configurations corresponding to
either the 3d f 4f or the 4f f 3d metal-metal charge transfer,
both leading to an S ) 3 excited-pair state. The Heitler-London
mechanism and the Anderson one have been assumed9 to be
Figure 8. Field dependence of the magnetization for 4Gd. The full
line corresponds to the Brillouin function for a S ) /2 spin state and
the dotted line to the Brillouin function for three isolated spins.
5
of 1Gd the variation of M vs H at 2 K (Figure 7) is very similar
to those previously reported for ferromagnetically coupled (Cu,
Gd) pairs.1,12,34 It nicely follows the Brillouin function
9
(abbreviated as B.f.(9/2) in the following) for a ST ) /2 spin
state and thus confirms the ferromagnetic nature of the Cu, Gd
interaction in 1Gd.
For 3Gd and 4Gd, the M vs H plots are linear in the low-
field regime (H e 0.8 × 104 G). They are in agreement with
the zero-field susceptibility values determined independently.
The experimental values of the magnetization obtained for 4Gd
in the field range ((0-5) × 104 G) are represented in Figure 8
together with the theoretical curves corresponding to the
Brillouin functions for a S ) 5/2 state and three uncoupled spins
(35) Goodenough, J. B. In Magnetism and the Chemical Bond; Inter-
science: New York, 1963.
(36) Kahn, O. In Magneto-Structural Correlations in Exchange Coupled
Systems; Willet, R. D., Gatteschi, D.; Kahn, O., Eds.; D. Reidel:
Dordrecht, The Netherlands, 1985.
(37) Girerd, J. J.; Charlot, M. F.; Kahn, O. Chem. Phys. Lett. 1981, 82,
534.
(38) Anderson, P. W. Phys. ReV. 1956, 115, 2.
(39) Anderson, P. W. In Magnetism; Rado, G. T., Suhl, H., Eds.; Academic
Press: NewYork, 1963; Vol. 1, Chapter 2.
(34) Costes, J. P.; Dahan, F.; Dupuis, A.; Laurent, J. P. Inorg. Chem. 2000,
39, 165.

6000 Inorganic Chemistry, Vol. 39, No. 26, 2000
Costes et al.
inoperative in the complexes built around a CuO2Gd core. It
has been stated that, due to the contraction of the 4f orbital
around the gadolinium nucleus and their shielding by the 6s
and 5p orbitals, all the integrals involving a 4f-3d density would
vanish. In the case of complexes 2′Gd, 3Gd, and 4Gd the
observed antiferromagnetism demands that the Heitler-London
interaction and/or the Anderson one become operative, implying
that the total 4f-3d overlap density takes a finite value.
Commun de Spectroscopie de Masse) for her contribution to
the mass spectra data. The authors are greatly indebted to Dr.
J. P. Laurent for fruitful discussion and comments.
Supporting Information Available: X-ray crystallographic files
including the structural data for [{L1Cu(H2O)}2Gd(H2O)(NO3)](NO3)2-
(H2O)2 1Gd and for [(L3Cu)2Ce(NO3)3] 3Ce in CIF format. This
material is available free of charge via the Internet at http://pubs.acs.org.
Acknowledgment. We thank Dr. A. Mari for his contribu-
tion to the magnetic measurements and Dr. S. Richelme (Service
IC000666U