Exchange Coupling in an Isostructural Series of Face-Sharing Bioctahedral Complexes [LMII(μ-X)3MIIL]BPh4 M = Mn, Fe, Co, Ni, Zn; X = Cl, Br; L = 1,4,7-Trimethyl-1,4,7-triazacyclononane

Article abstract of DOI:10.1021/ic970119h

The reaction of the divalent metal halides ZnCl₂, ZnBr₂, MnCl₂·CH₃CN, MnBr₂, FeCl·4CH₃CN, CoCl₂4CH₃CN, CoBr₂, NiCl₂·6H₂O, and NiBr₂, respectively, with the macrocycle 1,4,7-trimethyl-1,4,7-triazacyclononane (L) (1:1) in anhydrous acetonitrile, acetone, chloroform, or ethanol affords upon additon of NaBPh₄ the isomorphous series of complexes [LM^II(μ-X)₃M^IIL]BPh₄: 1, M = Zn, X = Cl; 2, Zn, Br; 3, Mn, Cl; 4, Mn, Br; 5, Fe, Cl: 6, Co, Cl; 7, Co, Br; 8, Ni, Cl; 9, Ni, Br. Six of these complexes have been structurally characterized by single-Crystal X-ray crystallography; they crystallize in the triclinic space group P1? (No. 2) with Z = 4. Crystal data are as follows 1, a = 16.654(1), b = 17.042(1), c = 17.684(1) A?, α = 97.30(1). β= 93.58(1), γ = 117.46(1)°; 3, a = 16.632(8), b = 17.012(8), c = 17.855(5) A?, α = 97.16(3), β= 93.37(3), γ = 117.24(3)°; 5, a = 16.658(3), b = 17.064(3), c = 17.741(4) A?, α = 97.32(3), β= 93.47(3), γ = 117.36(3)°; 6, a = 16.640(3), b = 17.040(3), c = 17.686(4) A?, α = 97.39(3), β= 93.58(3), γ = 117.39(3)°; 8, a = 16.608(3). b = 16.995(3), c = 17.555(3) A?, α = 97.36(1), β= 93.52(1), γ = 117.52(1)°; 9, a = 16.680(3), b = 17.016(2), c = 17.715(3)A?, α = 96.99(1). β= 93.70(1), γ = 117.42(1)°, All complexes consist of a dinuclear, face-sharing bioctahedral monocation with three μ₂-Cl or μ₂-Br bridging ligands and two LM fragments and well-separated tetraphenylborate anions (1:1). The cations cocrystallize in two different forms: an enantiomeric form with (λλλ) (or (δδδ)) conformation at both LM fragments and a meso form with an (λλλ) conformation at one LM fragment and (δδδ) at the other (ratio 1:1). From temperature-dependent magnetic susceptibility measurements (2-293 K) it was established that the spins of the unpaired electrons in 3 (d⁵d⁵ high spin), 4 (d⁵d⁵), 6 (d⁷d⁷ high spin), 7 (d⁷d⁷), 8 (d⁸d⁸), and 9 (d⁸d⁸) are intramolecularly, weakly antiferromagnetically coupled in each case, Surprisingly, the spins order ferromagnetically in 5 (d⁶d⁶ high spin). This is in contrast to the previously reported complex [(thf)₃Fe^II(μ-Cl)₃Fe^II(thf) ₃][SnCl₅(thf)] (thf = tetrahydrofuran)⁵ for which a new analysis of the temperature-dependence of the magnetic susceptibility and of field-dependent Mo?ssbauer spectra establish a weak intramolecular antiferromagnetic coupling. The origin of this difference is analyzed.

Full text of DOI:10.1021/ic970119h

2834
Inorg. Chem. 1997, 36, 2834-2843
Exchange Coupling in an Isostructural Series of Face-Sharing Bioctahedral Complexes
[LMII(µ-X)3MIIL]BPh4 (M ) Mn, Fe, Co, Ni, Zn; X ) Cl, Br;
L ) 1,4,7-Trimethyl-1,4,7-triazacyclononane)
Ursula Bossek,1a Daniela Nu1hlen,1a Eckhard Bill,1a Thorsten Glaser,1a Carsten Krebs,1a
Thomas Weyhermu1ller,1a Karl Wieghardt,*,1a Marek Lengen,1b and A. X. Trautwein1b
Max-Planck-Institut fu¨r Strahlenchemie, Stiftstrasse 34-36, D-45470 Mu¨lheim an der Ruhr, Germany,
and Institut fu¨r Physik, Medizinische Universita¨t, D-23538 Lu¨beck, Germany
ReceiVed January 31, 1997X
The reaction of the divalent metal halides ZnCl2, ZnBr2, MnCl2‚4CH3CN, MnBr2, FeCl2‚4CH3CN, CoCl2‚4CH3-
CN, CoBr2, NiCl2‚6H2O, and NiBr2, respectively, with the macrocycle 1,4,7-trimethyl-1,4,7-triazacyclononane
(L) (1:1) in anhydrous acetonitrile, acetone, chloroform, or ethanol affords upon additon of NaBPh4 the isomorphous
series of complexes [LMII(µ-X)3MIIL]BPh4: 1, M ) Zn, X ) Cl; 2, Zn, Br; 3, Mn, Cl; 4, Mn, Br; 5, Fe, Cl; 6,
Co, Cl; 7, Co, Br; 8, Ni, Cl; 9, Ni, Br. Six of these complexes have been structurally characterized by single-
crystal X-ray crystallography; they crystallize in the triclinic space group P1h (No. 2) with Z ) 4. Crystal data
are as follows 1, a ) 16.654(1), b ) 17.042(1), c ) 17.684(1) Å, R ) 97.30(1), â ) 93.58(1), γ ) 117.46(1)°;
3, a ) 16.632(8), b ) 17.012(8), c ) 17.855(5) Å, R ) 97.16(3), â ) 93.37(3), γ ) 117.24(3)°; 5, a )
16.658(3), b ) 17.064(3), c ) 17.741(4) Å, R ) 97.32(3), â ) 93.47(3), γ ) 117.36(3)°; 6, a ) 16.640(3), b )
17.040(3), c ) 17.686(4) Å, R ) 97.39(3), â ) 93.58(3), γ ) 117.39(3)°; 8, a ) 16.608(3), b ) 16.995(3), c )
17.555(3) Å, R ) 97.36(1), â ) 93.52(1), γ ) 117.52(1)°; 9, a ) 16.680(3), b ) 17.016(2), c ) 17.715(3)Å, R
) 96.99(1), â ) 93.70(1), γ ) 117.42(1)°. All complexes consist of a dinuclear, face-sharing bioctahedral
monocation with three µ2-Cl or µ2-Br bridging ligands and two LM fragments and well-separated tetraphenylborate
anions (1:1). The cations cocrystallize in two different forms: an enantiomeric form with (λλλ) (or (δδδ))
conformation at both LM fragments and a meso form with an (λλλ) conformation at one LM fragment and (δδδ)
at the other (ratio 1:1). From temperature-dependent magnetic susceptibility measurements (2-293 K) it was
established that the spins of the unpaired electrons in 3 (d5d5 high spin), 4 (d5d5), 6 (d7d7 high spin), 7 (d7d7), 8
(d8d8), and 9 (d8d8) are intramolecularly, weakly antiferromagnetically coupled in each case. Surprisingly, the
spins order ferromagnetically in 5 (d6d6 high spin). This is in contrast to the previously reported complex [(thf)3-
FeII(µ-Cl)3FeII(thf)3][SnCl5(thf)] (thf ) tetrahydrofuran)5 for which a new analysis of the temperature-dependence
of the magnetic susceptibility and of field-dependent Mo¨ssbauer spectra establish a weak intramolecular
antiferromagnetic coupling. The origin of this difference is analyzed.
Introduction
complexes containing metal ions with dn (n g 4) electronic
configuration are rather scarce. Sobota et al. have in recent
years synthesized three isostructural complexes of the type
[(thf)3MII(µ-Cl)3MII(thf)3]+, where thf is tetrahydrofuran and MII
represents Fe, Co, and Ni with high-spin d6, d7, and d8
configurations,5,6 respectively, and reported their structures and
magnetic properties. Intramolecular, very weak antiferromag-
netic coupling between the two paramagnetic metal ions has
been established in each case by temperature-dependent mag-
netic susceptibility measurements; the results are summarized
in Table 1.
We report here on an analogous but more comprehensive
series of face-sharing bioctahedral complexes of the type [LMII-
(µ-X)3MIIL]BPh4, where L is the facially capping ligand 1,4,7-
trimethyl-1,4,7-triazacyclononane, MII represents the divalent
metal ions Mn (high-spin d5), Fe (high-spin d6), Co (high-spin
d7), Ni (d8), and Zn (d10), and X is a single atom bridging ligand,
chloride or bromide.
Dinuclear, face-sharing bioctahedral complexes of the type
[X3MIII(µ-X)3MIIIX3]3-, where X represents the halides Cl-,
Br-, and I- and MIII is a trivalent first-, second- or third-row
paramagnetic transition metal ion with dn electronic configu-
ration (n ) 1-3), have played a key role in our understandig
of molecular magnetic properties such as intramolecular ex-
change coupling phenomena and metal-metal bond formation.2
Compounds of this type are well suited for such studies in the
solid state because intramolecular spin exchange is in general
significantly stronger than intermolecular interactions. In the
first-row transition metal series these studies are focused on early
transition metal ions TiIII, VIII,3 and CrIII with d1, d2, and d3
configurations, respectively. It has been proposed that due to
the close proximity of the two metal ions at 2.5-4.0 Å the
dominant magnetic superexchange pathway within such a
dinuclear cation is direct through-space which has also been
described as incipient metal-metal bond formation.4
Chart 1 gives a list of these compounds and their labels. They
crystallize in the triclinic space group P1h with two crystallo-
graphically independent formula units (Z ) 4). Due to the
presence of three N-bound methyl groups at each macrocycle
In contrast, reports on the magnetic properties of dinuclear
X Abstract published in AdVance ACS Abstracts, June 1, 1997.
(1) (a) Max-Planck-Institut fu¨r Strahlenchemie. (b) Medizinische Univer-
sita¨t Lu¨beck.
(2) Cotton, F. A.; Ucko, D. A. Inorg. Chim. Acta 1972, 6, 161 and
references therein.
(3) Bouma, R. J.; Teuben, J. H.; Beukema, W. R.; Bansemer, R. L.;
Huffman, J. C.; Caulton, K. G. Inorg. Chem. 1984, 23, 2715.
(4) Niemann, A.; Bossek, U.; Wieghardt, K.; Butzlaff, C.; Trautwein, A.
X.; Nuber, B. Angew. Chem., Int. Ed. Engl. 1992, 31, 311.
(5) Janas, Z.; Sobota, P.; Lis, T. J. Chem. Soc., Dalton Trans. 1991, 2429.
(6) Janas, Z.; Lis, T.; Sobota, P. Polyhedron 1992, 11, 3019.
S0020-1669(97)00119-5 CCC: $14.00 © 1997 American Chemical Society

Bioctahedral Complexes [LMII(µ-X)3MIIL]BPh4
Inorganic Chemistry, Vol. 36, No. 13, 1997 2835
Table 1. Comparison of Structural and Magnetic Properties of Dinuclear Tris(µ-chloro)-Bridged Complexes
complexa
M1‚‚‚M2, Å
M-Clb, Å
Cl-M-Cl, deg
M1-Cl-M2, deg
J, cm-1 b
ref
[VII2Cl3(thf)6]2[Zn2Cl6]
2.973(1)
3.086(2)
n.d.
2.993(2)
2.921(2)
2.477
2.488(9)
n.d.
2.403
2.416
87.3-88.0
85.0-86.2
n.d.
84.9-86.4
86.2-88.8
73.8
76.7
n.d.
77.0
74.4
-37.5
-3
-1.7
-6.0
n.d.
3, 4
5
6
6
12
[FeII2Cl3(thf)6][SnCl5(thf)]
[CoII2Cl3(thf)6][SnCl5(thf)]
[NiII2Cl3(thf)6][SnCl5(thf)]
[NiII2Cl3L′2]BF4‚CH3CN
a Abbreviations: n.d. ) not determined; L′ ) 1,4,7-trithiacyclononane; thf ) tetrahydrofuran. b H ) -2JBS1BS2.
[L2Mn2(µ-Cl)3]BPh4 (3). A solution of the ligand L (0.25 g, 1.46
mmol) and MnCl2‚4CH3CN (0.40 g, 1.41 mmol) in acetonitrile (30
mL) was stirred at 20 °C for 15 min. Addition of NaBPh4 (0.5 g, 1.46
mmol) initiated the precipitation of colorless crystals. Yield: 0.35 g
(34%). Anal. Calcd for C42H62N6Mn2Cl3B (878): C, 57.45; H, 7.11;
N, 9.57; Cl 12.1. Found: C, 58.0; H, 7.3; N, 9.4; Cl, 11.5.
[L2Mn2(µ-Br)3]BPh4 (4). This complex was prepared as described
for 3 by using MnBr2 as starting material.9 Upon cooling the reaction
mixture to 0 °C for 12 h colorless crystals were obtained. Yield: 0.30
g (60%). Anal. Calcd for C42H62N6Mn2Br3B (1011.4): C, 49.88; H,
6.18; N, 8.31. Found: C, 49.6; H, 6.3; N, 8.3.
Chart 1. Synthesized Complexes and Labels
[L2Fe2(µ-Cl)3]BPh4 (5). This complex was prepared as described
for 3 by using FeCl2‚4CH3CN as starting material. Pale yellowish
crystals were obtained. Yield: 0.35 g (34%). Anal. Calcd for
C42H62N6Fe2Cl3B (879.9): C, 57.33; H, 7.10; N, 9.55; Cl, 12.1.
Found: C, 57.2; H, 7.2; N, 9.9; Cl, 12.2.
[L2Co2(µ-Cl)3]BPh4 (6). To a solution of CoCl2‚6CH3CN (0.39 g;
1.0 mmol) and the ligand L (0.17 g; 1.0 mmol) in dry acetone (25 mL)
was added N2H4‚2HCl (0.10 g). After the mixture was stirred for 2 h
at 20 °C, a colorless precipitate was filtered off and NaBPh4 (0.17 g)
was added to the resulting blue solution: Within 3 d at 0 °C pink
microcrystals precipitated. Yield: 0.19 g (43%). Anal. Calcd for
C42H62N6Co2Cl3B (886.0): C, 56.94; H, 7.26; N, 9.49; Cl, 12.0.
Found: C, 56.6; H, 7.1; N, 9.3; Cl, 11.8.
[L2Co2(µ-Br)3]BPh4 (7). To a solution of CoBr2 (0.22 g; 1.0 mmol)
and the ligand L (0.17 g; 1.0 mmol) in chloroform (40 mL) was added
a solution of NaBPh4 (0.17 g) in acetone (15 mL). The stirred two-
phasic solution was heated to reflux for 2.5 h after which time the
reaction volume was reduced by evaporation to one-half. A violet
microcrystalline solid formed. Yield: 0.46 g (90%). Anal. Calcd for
C42H62N6Co2Br3B (1019.5): C, 49.49; H, 6.13; N, 8.24. Found: C,
50.3; H, 6.3; N, 8.3.
[L2Ni2(µ-Cl)3]BPh4 (8). To a hot solution of NiCl2‚6H2O (0.43 g;
1.81 mmol) in an acetone/acetonitrile mixture (15 mL; 2:1 vol) were
added 20 drops of ethyl orthoformate to remove the water. After
addition of the ligand L (0.31 g; 1.81 mmol) the solution was refluxed
until a clear green-blue solution was obtained. Addition of NaBPh4
(0.31 g) initiated the precipitation of blue-green crystals. Yield: 0.60
g (75%). Anal. Calcd for C42H62N6Ni2Cl3B (885.6): C, 56.97; H, 7.06;
N, 9.49; Cl, 12.0. Found: C, 56.8; H, 7.2; N, 9.4; Cl, 11.7.
[L2Ni2(µ-Br)3]BPh4 (9). To a clear solution of NiBr2 (0.22 g; 1.0
mmol) in dry ethanol (40 mL) was added the ligand L (0.17 g; 1.0
mmol) dissolved in ethanol (10 mL). The solution was refluxed for
1.5 h and NaBPh4 (0.34 g) dissolved in ethanol (10 mL) was added. A
green precipitate formed. Yield: 0.46 g (90%). Anal. Calcd for
C42H62N6Ni2Br3B (1018.9): C, 49.50; H, 6.13; N, 8.24. Found: C,
49.1; H, 6.4; N, 8.0.
and of the sterically very demanding diamagnetic tetraphe-
nylborate anions, the individual cations are well separated in
the solid state. They are magnetically dilute. Consequently,
the electronic structure and the magnetic properties of these
cations are readily studied in the solid state without interference
of intermolecular electronic or magnetic interactions. Each
metal ion is in a pseudo-octahedral environment comprising
three facially coordinated tertiary amine nitrogens and the three
bridging halide ions (Cl- or Br-). We were primarily interested
in studying the intramolecular exchange coupling as a function
of the metal-metal electronic configuration of the individual
metal ion. More specifically, we intend with the present series
of complexes to study the effect of a filled t62g set where a
direct through-space interaction between the unpaired electrons
is not possible, versus an incompletely filled t2g subshell where
such direct through-space interactions are possible.
An unexpected result of this investigation is the observation
that in [LFeII(µ-Cl)3FeIIL]BPh4 (5) an intramolecular ferromag-
netic coupling prevails. This is in contrast to the complex
[(thf)6FeII (µ-Cl)3][SnCl5(thf)] reported by Sobota et al.5 in
2
which the two high-spin iron(II) ions are weakly antiferromag-
netically coupled. Here we present a detailed magnetochemical
and Mo¨ssbauer spectroscopic analysis of these two diferrous
complexes.
Experimental Section
Preparation of Complexes. The following water-free starting
complexes have been prepared according to published procedures:7
CoCl2‚4CH3CN, FeCl2‚4CH3CN, MnCl2‚4CH3CN, MnBr2‚4CH3CN.
The ligand 1,4,7-trimethyl-1,4,7-triazacyclononane (L) has been syn-
thesized as described previously.8 The preparation of complexes has
been performed in water-free solvents and under an argon blanketing
atmosphere. Sobota’s complex has been synthesized according to the
procedure given in ref 5.
[L2Zn2(µ-Cl)3]BPh4 (1). A solution of ZnCl2 (0.27 g, 2.0 mmol)
in acetonitrile (60 mL) to which the ligand L (0.34 g, 2.0 mmol) was
added was stirred at ambient temperature for 1.5 h. Addition of tetra-
n-butylammonium tetraphenylborate (0.61 g, 1.0 mmol) initiated the
slow precipitation (5 d) of colorless crystals which were collected by
filtration. Yield: 0.72 g (40%). Anal. Calcd for C42H62N6Zn2Cl3B
(898.95): C, 57.08; H, 8.09; N, 9.51. Found: C, 56.9; H, 7.5; N, 9.3.
[L2Zn2(µ-Br)3]BPh4 (2). This complex was prepared as described
for 1 by using ZnBr2. Yield: 0.51 g (50%). Anal. Calcd for C42H62N6-
Zn2Br3B (1032): C, 48.88; H, 6.05; N, 8.81. Found: C, 49.3; H, 6.4;
N, 8.5.
X-ray Crystallography. Crystal, data collection, and refinement
parameters are given in Table 2. A colorless cubic crystal of 1 was
placed on an Enraf-Nonius CAD4 diffractometer; a pale yellow crystal
of 3, a pale yellow crystal of 5, a violet parallelepipede of 6, and green
parallelepipedes of 8 and 9 were placed on a Siemens P4 diffractometer.
Graphite-monochromated Mo KR radiation (λ ) 0.710 73 Å) was used
throughout. Intensity data were collected at 293 ( 2 K for 1, 5, 6, 8,
and 9 and at 193 ( 2 K for 3; they were corrected for Lorentz and
polarization effects in the usual manner; correction for absorption effects
was carried out for 1, 5 and 9 by ψ-scans; for 6 and 8 no absorption
correction was deemed necessary. The structures were solved by direct
(7) Hathaway, B. J.; Holah, D. G. J. Chem. Soc. 1964, 2400.
(8) Wieghardt, K.; Chaudhuri, P.; Nuber, B.; Weiss, J. Inorg. Chem. 1982,
21, 3086.
(9) In this preparation extreme care must be taken that the solvent, the
ligand L, and MnBr2 are free of adventitious water. Otherwise varying
amounts of [LMn(µ-Br)2(µ-OH)MnL]BPh4 cocrystallize.

2836 Inorganic Chemistry, Vol. 36, No. 13, 1997
Bossek et al.
a
Table 2. Crystallographic Data for Complexes [L2M2(µ-X)3]BPh4
complex
1
3
5
6
8
9
empirical formulab
fw
AZn2Cl3
898.9
AMn2Cl3
878.0
AFe2Cl3
879.8
ACo2Cl3
886.0
ANi2Cl3
885.6
ANi2Br3
1018.9
a, Å
b, Å
c, Å
R, deg
â, deg
γ, deg
V, Å3
16.654(3)
17.042(3)
17.684(4)
97.30(3)
93.58(3)
117.46(3)
4376(2)
13.2
16.632(8)
17.012(8)
17.855(5)
97.16(3)
93.37(3)
117.24(3)
4418(3)
7.89
16.658(3)
17.064(3)
17.741(4)
97.32(3)
93.47(3)
117.36(3)
4401.4(15)
8.78
16.640(3)
17.040(3)
17.686(4)
97.39(3)
93.58(3)
117.39(3)
4373.3(15)
9.79
16.608(3)
16.995(3)
17.555(3)
97.36(1)
93.52(1)
117.52(1)
4316.9(13)
10.9
16.680(3)
17.016(2)
17.715(3)
96.99(1)
93.70(1)
117.42(1)
4388.27(15)
36.3
µ, cm-1
d(calcd), g cm-3
T, °C
1.33
20
0.069
1.32
-80
0.051
1.33
20
0.045
1.35
20
0.041
1.36
20
0.040
1.54
20
0.043
Rc
a Features common for all complexes: triclinic crystal system; space group P1h; Z ) 4; radiation Mo KR, λ ) 0.7101 Å. b A ) C42H62N6B. c R
) ∑||Fo| - |Fc||/∑|Fo|.
as racemate of two enantiomeric forms.11 Consequently, the
and difference Fourier methods by using the SHELXTL-PLUS program
package.10 The function minimized during full-matrix least-squares
formation of the above dinuclear complexes leads, in principle,
refinement was ∑w(|Fo| - |Fc|)2. The hydrogen atoms were placed at
to two enantiomeric forms each of which contains two LM
calculated positions with isotropic thermal parameters; the methyl
groups were treated as rigid bodies. All non-hydrogen atoms were
MII(λλλ)]+ and [(δδδ)MII(µ-X)3MII(δδδ)]+ and, in addition, the
refined with anisotropic thermal parameters. It has not been possible
to resolve the disorder of the methylene carbon atoms of the respective
second monocation (involving metal ions M3 and M4) in all structures
fragments of identical conformation, namely [(λλλ)MII(µ-X)3-
meso form [(δδδ)MII(µ-X)3MII(λλλ)]+, which is the achiral
diastereomer. As we will show below by X-ray crystallography,
the enantiomers and the meso form cocrystallize in the ratio
1:1.
by a split atom model.
Physical Measurements. Electronic absortion spectra of complexes
were measured in the range 350-2000 nm on a Perkin-Elmer Lambda
19 spectrophotometer in dry acetonitrile solution. The magnetic
susceptibilities of powered samples of complexes were measured in
the temperature range 2-293 K on a SQUID susceptometer (MPMS
Quantum Design) at 1, 4, and 7 T. The Mo¨ssbauer spectra were
recorded on alternating constant-acceleration spectrometer. The mini-
mal experimental line width was 0.24 mm s-1 full-width at half-height.
The sample temperature was maintained constant either in an Oxford
Variox or an Oxford Mo¨ssbauer-Spectromag cryostat. The latter is a
split-pair superconducting magnet system for applied fields up to 8 T
where the temperature of the samples can be varied in the range 1.5-
250 K. The field at the sample is oriented perpendicular to the γ-beam.
The 57Co/Rh source (1.8 GBq) was positioned at room temperature
inside the gap of magnet system at a zero-field position. Isomer shifts
are referenced relative to iron metal at 295 K.
Description of the Crystal Structures. The crystal struc-
tures of the isomorphous series of complexes 1, 3, 5, 6, 8, and
9 have been determined by single-crystal X-ray crystallography.
They crystallize in the triclinic space group P1h with very similar
unit cell parameters which are given in Table 2. There are four
formula units [L2M2(µ-X)3]BPh4, in the unit cell. Therefore,
two such units are located on two different, crystallographically
independent positions in each case. Table 3 summarizes
selected bond distances, and Table 4 gives averaged bond angles.
Each structure consists of well-separated monocations [L2M2-
-
(µ-X)3]+ and BPh4 anions. Both crystallographically inde-
pendent monocations do not lie on any crystallographic sym-
metry element, and consequently, the dimensions of four
independent octahedral LMX3 units have been determined for
each structure.
Results
Before we discuss some pertinent structural features of
individual monocations, we describe an interesting disorder
phenomenon found in all structures under consideration here.
We exemplify this by considering the structure 3 only, but the
same observations apply for all structures. Figure 1 shows the
structure of the two monocations [L2Mn2(µ-Cl)3]+ with the
individual atoms displayed as thermal ellipsoids drawn at the
40% probability level. A comparison of the two cations reveals
that the ellipsoids of the methylene carbon atoms of the
coordinated amines at Mn1 and Mn2 are significantly smaller
than those of the coordinated macrocycle at Mn3 (and Mn4).
More quantitatively, this observation is corroborated by the
numerical values of the anisotropic (and isotropic Ueq) thermal
parameters for these methylene carbon atoms (see Supporting
Information). As a consequence, all C-C and C-N bond
lengths of the 1,4,7-triazacyclononane backbone at Mn1 and
Mn2 are normal single bonds and are observed in the usual
range 1.51 ( 0.01 and 1.48 ( 0.01 Å, respectively. In contrast,
due to apparently larger thermal motions of the methylene
carbon atomssat least at Mn3sthe C-C bonds are unrealisti-
Syntheses. The crystalline salts [L2M2(µ-X)3]BPh4 where
L is the macrocycle 1,4,7-trimethyl-1,4,7-triazacyclononane, M
is a divalent transition metal ion (Mn, Fe, Co, Ni, Zn), and X
represents the bridging ligand chloride or bromide, were
prepared from acetonitrile, acetone, chloroform or ethanol
solutions of water-free halides MX2 or their solvates MX2‚4CH3-
CN by addition of 1 equiv of the macrocycle L and Na[BPh4],
respectively. Upon evaporation of the solvents and standing
of the solution under an argon blanketing atmosphere crystals
of colorless zinc(II) (1, 2), manganese(II) (3, 4) and iron(II)
(5), pink cobalt(II) (6, 7), and blue-green nickel(II) compounds
(8, 9) were obtained in reasonable yields (50-70%). See Chart
1 for complexes and labels.
It is well established by single-crystal X-ray crystallography
that in pseudo-octahedral mononuclear complexes LMX3 the
coordinated macrocycle L always adopts a conformation where
the three five-membered chelate rings, M-N-C-C-N, have
either (λλλ) or (δδδ) configuration, i.e. the LM fragment exists
(10) Full-matrix least-squares structure refinement program package Si-
emens SHELXTL-PLUS: Sheldrick, G. M., (University of Go¨ttingen,
Go¨ttingen, Germany.
(11) (a) Bhula, R.; Osvath, P.; Weatherburn, D. C. Coord. Chem. ReV. 1988,
91, 89. (b) Chaudhuri, P.; Wieghardt, K. Prog. Inorg. Chem. 1987,
35, 329.

Bioctahedral Complexes [LMII(µ-X)3MIIL]BPh4
Inorganic Chemistry, Vol. 36, No. 13, 1997 2837
Table 3. Selected Bond Distances (Å) of the Two Independent
Cations in 1, 3, 5, 6, 8, and 9
1
3
5
6
8
9
M1‚‚‚M2 3.058(1) 3.077(2) 3.014(2) 3.048(2) 3.044(1) 3.234(1)
M1-X1 2.467(1) 2.536(2) 2.490(2) 2.464(2) 2.432(1) 2.591(1)
M1-X2 2.584(1) 2.557(2) 2.508(2) 2.489(2) 2.461(1) 2.622(1)
M1-X3 2.501(1) 2.503(2) 2.456(2) 2.443(2) 2.413(1) 2.566(1)
M2-X1 2.485(2) 2.526(2) 2.484(2) 2.463(2) 2.440(1) 2.595(1)
M2-X2 2.515(1) 2.587(2) 2.544(2) 2.525(2) 2.492(1) 2.655(1)
M2-X3 2.449(1) 2.556(2) 2.501(1) 2.477(1) 2.443(1) 2.589(1)
M1-N1 2.171(4) 2.284(4) 2.223(4) 2.180(4) 2.122(3) 2.133(6)
M1-N2 2.169(4) 2.274(5) 2.215(5) 2.161(5) 2.110(3) 2.101(7)
M1-N3 2.159(4) 2.281(4) 2.206(5) 2.170(5) 2.118(3) 2.132(6)
M2-N4 2.191(4) 2.254(4) 2.189(5) 2.142(4) 2.094(3) 2.106(6)
M2-N5 2.181(4) 2.269(4) 2.194(5) 2.153(5) 2.103(3) 2.107(6)
M2-N6 2.189(4) 2.266(3) 2.193(4) 2.149(3) 2.106(2) 2.107(4)
M3‚‚‚M4 3.068(1) 3.078(2) 3.026(2) 3.062(2) 3.054(1) 3.243(1)
M3-X4 2.473(1) 2.531(2) 2.486(2) 2.461(2) 2.431(1) 2.588(1)
M3-X5 2.551(2) 2.522(2) 2.472(2) 2.460(2) 2.434(1) 2.574(1)
M3-X6 2.465(1) 2.570(2) 2.530(2) 2.508(2) 2.469(1) 2.629(1)
M3-N7 2.175(4) 2.263(4) 2.199(5) 2.154(4) 2.109(3) 2.103(5)
M3-N8 2.171(4) 2.270(4) 2.209(5) 2.157(5) 2.109(3) 2.121(6)
M3-N9 2.169(4) 2.258(5) 2.193(6) 2.160(5) 2.102(3) 2.103(7)
M4-X4 2.463(2) 2.526(2) 2.480(2) 2.458(1) 2.430(1) 2.585(1)
M4-X5 2.486(1) 2.558(2) 2.513(2) 2.492(2) 2.459(1) 2.601(2)
M4-X6 2.532(1) 2.536(2) 2.492(2) 2.470(2) 2.442(1) 2.603(1)
M4-N10 2.177(4) 2.268(4) 2.199(6) 2.159(5) 2.105(3) 2.110(6)
M4-N11 2.178(4) 2.262(5) 2.208(7) 2.165(6) 2.111(4) 2.119(8)
M4-N12 2.169(5) 2.273(3) 2.200(4) 2.157(4) 2.108(2) 2.096(5)
Table 4. Selected Averaged Bond Angles (deg) of One
Independenta Cation in Crystals of 1, 3, 5, 6, 8, and 9
Figure 1. Structure of the two crystallographically independent
monocations in crystals of 3. The atoms are drawn at the 40%
probability level.
1
3
5
6
8
9
N-M-N 82.7(2) 79.1(2) 81.2(1) 82.7(1) 84.3(1) 84.2(2)
N-M-X 95.4(1) 96.8(1) 95.7(1) 95.6(1) 95.2(1) 95.2(2)
X-M-X 86.5(1) 87.2(1) 87.3(1) 86.1(1) 85.4(1) 85.5(1)
M-X-M 75.4(1) 74.4(1) 74.2(1) 76.0(1) 76.9(1) 76.8(2)
Scheme 1
a Values for the ordered enantiomeric form are given only.
cally short within the coordinated macrocycle (1.41 ( 0.01 Å).
A conformational analysis of two sets of three five-membered
of the three Fe1-Cl distances is 7.454 Å, and that of Fe2-Cl
is larger at 7.529 Å (∆ ) 0.075 Å). The same holds for the
sums of Fe1-N and Fe2-N distances where the latter is now
smaller (∑, 6.644 Å, 6.576 Å; ∆ ) 0.068 Å). Note that the
overall sum of Fe1-N and Fe1-Cl distances is 14.098 Å
whereas that of Fe2-N and Fe2-Cl is 14.105 Å (∆ ) 0.007
Å). In the meso form the two LFeCl3 octahedra at Fe3 and
Fe4 are more similar: The sum of the three Fe-Cl distances at
Fe3 is 7.488 Å and at Fe4 7.485 Å and, similarly, the sum of
the Fe-N distances (6.601 Å at Fe3 and 6.607 Å at Fe4; ∆ )
0.006 Å). The N3Fe(µ-Cl)3FeN3 core of the meso form is more
symmetric than that in the corresponding enantiomers. Interest-
ingly, in the crystal structure of [{(thf)3FeII}2(µ-Cl)3][SnCl5(thf)]5
the O3Fe(µ-Cl)3FeO3 core also shows some degree of distortion
but the Fe-Cl distances of both O3FeCl3 octahedra are within
experimental error identical (average Fe-Cl: 2.488(9) Å). Note
that the monocation in this structure also does not possess
crystallographically imposed symmetry.
chelate rings Mn-N-C-C-N at Mn1 and Mn2 reveals that
both sets adopt the same conformation namely (λλλ)Mn1‚‚‚
Mn2(λλλ) and its isomer (δδδ)Mn1‚‚‚Mn2(δδδ) which are
related by a crystallographic inversion center in the space group
P1h. This monocation is fully ordered. In contrast, the other
monocation (Mn3, Mn4) appears to be disordered. It is not
possible to determine the conformation of the three Mn3-N-
C-C-N chelate rings. This can be understood if a superposi-
tion of a (λλλ)Mn and a (δδδ)Mn fragment prevails at this site
in the solid state as is schematically shown in Scheme 1 for
one such ring. Note that the thermal parameters of the atoms
of the N3Mn(µ-Cl)3MnN3 core in both cations are normal and
it is meaningful to compare the metrical details of the two cores
(see below). We propose that the monocation comprising the
metal ions Mn3 and Mn4 actually consist of the achiral meso
form [(λλλ)Mn‚‚‚Mn(δδδ)]+ which is superimposed by its
“turned around” form [(δδδ)Mn‚‚‚Mn(λλλ)]+ in the solid state.
Thus crystals of [{LMII}2(µ-X)3][BPh4] cocrystallize as a
mixture of their enantiomers and the meso form (1:1).
In the following we briefly discuss the difference of the
metrical details in the N3M(µ-Cl)3MN3 cores of the enantiomeric
and the meso forms in the diferrous complex 5. It is significant
that the three Fe-Cl bond lengths within each LFeCl3 octahe-
dron are not equidistant; i.e. neither the (λλλ)Fe(µ-Cl)3Fe(λλλ)
(or its enantiomer) nor the meso form (λλλ)Fe(µ-Cl)3Fe(δδδ)
possess C3 (or higher) symmetry. The octahedra around Fe1
and Fe2 of the enantiomeric monocation are different: The sum
It should be pointed out that the observed different distortions
of the enantiomeric and meso forms, of the N3M(µ-X)3MN3
cores are steric in nature rather than electronic. The relative
orientations of the six methyl groups of the two macrocycles
differ slightly in both forms, and more importantly, they are in
close vicinity to the µ-chloro or µ-bromo bridges. This is nicely
borne out by a comparison of the Ni-Cl bond distances in 8
and a similar complex [L′′2Ni2(µ-Cl)3][BF4]‚CH3CN,12 where
L′′ represents the macrocycle 1,4,7-trithiacyclononane. Thus

2838 Inorganic Chemistry, Vol. 36, No. 13, 1997
Bossek et al.
Table 5. Electronic Spectra of Complexes in Acetonitrile Solution
(350-2000 nm) at 298 K
complexa
λ )
max, nm (ꢀ, L mol-1 cm-1
6
7
8
9
1200 (6), 553 (sh), 523 (26)
1142 (33), 500 (80)
1053 (32), 633 (16), 387 (32)
1092 (34), 650 (30), 329 (70)
a Complexes 1-5 did not show absorption maxima with ꢀ > 5 L
mol-1cm-1 in the measured range.
the N-CH3 groups in 8 are replaced by sterically less demand-
ing thioether sulfur donors. The three Ni-Cl distances in 8 at
Ni1 differ at 2.4313(1), 2.461(1), and 2.432(1) Å and similarly
at Ni2 at 2.440(1), 2.492(1), and 2.443(1) Å. Even in the meso
form with Ni-Cl distances at Ni3 of 2.431(1), 2.434(1), and
2.469(1) Å and at Ni4 of 2.430(1), 2.459(1), and 2.442(1) Å
these differences prevail albeit to a lesser degree. In contrast,
in [L′′2Ni2(µ-Cl)3]+ four Ni-Cl distances are observed in the
very narrow range of 2.419(3) to 2.411(3) Å and only one is
slightly longer at 2.440(3) Å and one is shorter at 2.396(3) Å.
Similarly, in [(thf)3Ni(µ-Cl)3Ni(thf)3][SnCl5(thf)]6 the six Ni-
Cl distances are within experimental error equidistant (average
2.403(6) Å).
Electronic Spectra and Magnetic Properties of Complexes.
The electronic spectra of complexes have been recorded in the
range 350-2000 nm in acetonitrile solution at ambient tem-
perature; the results are given in Table 5. Compounds 1-5
are colorless and display no absorption in this range with ꢀ >
5 L mol-1 cm-1. The pink complexes 6 and 7 exhibit typical
weak d-d transitions of octahedral high-spin cobalt(II) (d7) in
the visible. Similarly, the spectra of 8 and 9 are typical of
octahedral nickel(II) (d8).
The temperature-dependence of the molar magnetic suscep-
tibility of powdered solid samples of complexes 3-9 has been
measured on a SQUID magnetometer in the range 2-298 K in
an applied magnetic field of 1.0 T. The data were corrected
for underlying diamagnetism by use of tabulated Pascal’s
constants; no correction for temperature-independent paramag-
netism was applied.
Figure 2. Temperature dependence of the magnetic moment, µeff, per
dinuclear unit of powdered samples of complexes 3-5 measured at 1
T. Bottom: Magnetization data of 5 measured at various fields. The
solid lines represent best fits to the spin-Hamiltonian eq 1 with
parameters listed in Table 6.
Complex 5 exhibits weak ferromagnetic coupling between
the two high-spin ferrous ions of the monocation. Figure 2
(bottom) shows the magnetization M in NgeµB units as a function
of µBB/kT measured at three different applied magnetic field
strengths (1, 4, and 7 T). The solid lines represent best fits
obtained by using the parameter set in Table 6 for 5. The St )
4 ground state of 5 is confirmed by this fit. As a consequence
of the zero-field splitting the magnetization data cannot be fit
to a simple Brillouin function. From the fit shown in Figure 2
(bottom) the sign and the magnitude of the zero-field parameter
D was established to be +5.5 cm-1. A calculated error contour
map of J vs D showed two minima with identical J values. The
global minimum was found at D ) +5.5 cm-1 whereas a local
minimum had D ) -2.4 cm-1 but with a much larger error. It
is not possible to fit the magnetization data satisfactorily with
The data were fitted to the isotropic Heisenberg, Dirac, van
Vleck (HDvV) model by using the spin Hamiltonian in eq 1,
2
Si(Si + 1)
2
H ) -2JBS ‚BS +
Di Siz
-
+ µBgiBSi‚BB
∑
1
2
{
}
[
]
3
i)1
(1)
which consists of a HDvV exchange, a zero-field splitting, and
Zeeman term. All symbols in eq 1 have their usual meaning.
Since the dinulear complexes have approximately C3 symmetry,
the rhombicity, E/D, is considered to be zero. The parameters
were optimized to fit optionally either øM or øMT as a function
of temperature because the fits of øM are very sensitive to traces
of paramagnetic (mononuclear) impurities (PI). Their contribu-
tions are included in the best fits shown in Figures 2-4;
numerical results are summarized in Table 6. Figure 2 shows
the temperature-dependence of the magnetic moment, µeff, per
dinuclear unit for complexes 3-5. Moderately strong intra-
molecular antiferromagnetic coupling is observed in the man-
ganese(II) complexes 3 and 4. The g value obtained from the
above fitting procedure for 4 is not reliable. This unrealistic
value is probably due to the presence of a dinuclear impurity
such as [LMn(µ-OH)(µ-Br)2MnL]BPh4 which we have isolated
from acetonitrile/water mixtures and which we will report
separately.9
D ) -2.4 cm-1
.
Compounds 6 and 7 contain octahedral high-spin cobalt(II)
ions with a T ground state (in Oh symmetry), which is split in
C3V local symmetry into A and E. This splitting affects a
quenching of the first-order orbital momentum. If this splitting
is large the orbital momentum contributes by several tens of
wavenumbers to D and induces a strong anisotropy of the g
tensor.13 It is therefore justified to use eq 1 to fit the
susceptibility data of 6 and 7. This is shown in Figure 3. For
both compounds very satisfactory fits were obtained. It is noted
that the value of J is not very sensitive to variation of |D|
(12) Blake, A. J.; Halcrow, M. A.; Schro¨der, M. Acta Crystallogr. 1992,
C48, 1844.
(13) Kahn, O. Molecular Magnetism; VCH Publishers: New York, 1993;
p 208.

Bioctahedral Complexes [LMII(µ-X)3MIIL]BPh4
Inorganic Chemistry, Vol. 36, No. 13, 1997 2839
Octahedral nickel(II) in 8 and 9 exhibits, in general, non-
negligible zero-field splitting. As described above the fit of
the temperature-dependent susceptibility data is not very sensi-
tive to variations of |D| in the range 5-25 cm-1. Therefore,
we have fixed the value for D at 10 cm-1. The resulting fits
are excellent. Both complexes show intramolecular antiferro-
magnetic coupling.
In order to compare the magnetic properties of 5 with those
reported by Sobota et al.5 for [(thf)6FeII (µ-Cl)3][SnCl5(thf)], we
2
have prepared a fresh sample of this compound and measured
its temperature-dependent susceptibility. The data are shown
in Figure 4. Clearly, the effective magnetic moment decreases
with decreasing temperature indicating a diamagnetic ground
state (St ) 0). A fit of the data using the simple spin-
Hamiltonian H ) -2J‚BS1BS1 (S1 ) S2 ) 2) with g and J being
the only variables is possible and has been reported by Sobota
et al.5 (They do not show their data.) These authors arrive at
J ) -3 cm-1 and g ) 2.4. The fit obtained by using these
parameters to our data is shown as dotted line in Figure 4sa
reasonable but not excellent fit. By using the spin-Hamiltonian
eq 1 including zero-field splitting the fit was improved consider-
ably. We obtain g and J values of 2.34 and -2.3 cm-1
,
respectively, with |D| values in the range 1-4 cm-1
.
By using these parameters with a positiVe D only the correct
trend in the field-dependent magnetization data measured at 1,
4, and 7 T (not shown) can be reproduced. At low temperatures
(2-10 K) the fits do not agree well. Inclusion of a paramagnetic
impurity of ∼11% [LFeIIICl3] did improve the fit, but this is
not corroborated by the zero-field Mo¨ssbauer spectra (see below)
which rule out such large amount of FeIII in the sample.
Figure 3. Temperature dependence of the magnetic moment, µeff, per
dinuclear unit of powdered samples of complexes 6-9. The solid lines
represent best fits obtained as described in Figure 2.
The relatively large g value of 2.34 points to significant zero-
field splitting. From error contour maps of J over D steep local
minima of the error were obtained for small (+2 cm-1) and
large (-60 cm-1) values of D. Excluding the minima at large
zero-field splittingsin agreement with the field-dependent
Mo¨ssbauer spectrasit appears that in Sobota’s complex zero-
field splitting and spin coupling are of the same order of
magnitude (D ) +1 to +4 cm-1; J ) -1 to -3 cm-1). Using
these parameters and a g value at 2.3, an excellent fit of the
temperature-dependent susceptibility data is obtained, but as
pointed out, the magnitization data are not reproduced in every
detail.
Figure 4. Temperature dependence of the magnetic moment, µeff, per
dinuclear unit of a powdered sample of [(thf)6FeII2(µ-Cl)3][SnCl5(thf)].
The solid line represents the best fit to eq 1 with parameters listed in
Table 6. The dashed line was calculated by using the spin-Hamiltonian
H ) -2JBS1BS2 (S1 ) S2 ) 2) and Sobota’s published parameters J )
-3 cm-1 and g ) 2.4.
In summary, the above analysis allows two important
conclusions to be drawn: In Sobota’s complex the two ferrous
ions are weakly antiferromagnetically coupled and the local
zero-field splitting parameter D is positive and of the same order
of magnitude as the coupling constant J.
Mo1ssbauer Spectroscopy. Zero-field Mo¨ssbauer spectra of
5 and Sobota’s complex are displayed in Figure 5; the
corresponding Mo¨ssbauer parameters are given in Table 7.
Table 6. Magnetic Properties of Complexes
complex
J, cm-1 a
gb
PI, %c
D, cm-1 d
3
4
5
6
7
8
9
-11.6
-8.5
2.07
1.84
2.08
2.86
2.80
2.26
2.25
1.8 (S ) 5/2)
2.7 (S ) 5/2)
0
0 (fixed)
0 (fixed)
+5.5
(40 (fixed)
(40 (fixed)
(10 (fixed)
(10 (fixed)
The zero-field spectrum of crystalline 5 at 4.2 K consists of
a broad quadrupole doublet with distinct shoulders which
suggest the presence of essentially three different subspectra
(Figure 5A,B). Numerical fits with three Lorentzian doublets
yielded intensity ratios of 1:2:1. The “nested” arrangement of
these subspectra with similar isomer shifts but different quad-
rupole splittings is in agreement with the differing bonding
situation at each LFeCl3 octahedron of the two monocations in
the solid state of 5; they are typical of octahedral FeII (3d6 high
spin). At a given temperature the quadrupole splittings, ∆EQ,
scatter in the range 0.98-2.08 mm s-1. In addition, for two of
the subspectra ∆EQ is significantly temperature dependent which
indicates either thermal population of close lying excited spin-
orbital states or temperature-dependent structure variations. The
zero-field Mo¨ssbauer spectrum of Sobota’s complex is remark-
+5.8
-13.1
-13.3
-10.5
-12.0
0
1.4 (S ) 3/2)
4.5 (S ) 1)
4.1 (S ) 1)
a Coupling constant (H ) -2JBS1BS2). b g value (fit parameter).
c Percentage of a monomeric paramagnetic impurity with its assumed
spin in parentheses. d Zero-field-splitting parameter.
between 35 and 45 cm-1; it changes by ( 0.2 cm-1 in this
range of |D|. Therefore, we have fixed D at 40 cm-1 in the
fitting procedure; the sign of D cannot be determined from these
data. Both compounds 6 and 7 are intramolecularly antiferro-
magnetically coupled.

2840 Inorganic Chemistry, Vol. 36, No. 13, 1997
Bossek et al.
Table 8. Expectation Values of the Valence Contributions to the
Electric Field Gradient Tensor (EFG)val for 3d Orbitals in C3V
Symmetry with a1 Ground Statea
a The orbital functions are expressed as linear combinations of the
usual |ij d functions in octahedral symmetry.23 The tensor elements
for the basic orbitals |ij are taken from ref 24. In order to convert the
elements of (EFG)val to quadrupole splitting multiply by 4.2 mm s-1
/
(4/7)e r-3 . Note that the molecular z-axis is taken collinearly to the
3-fold axis in C3V.
For solid 5 and Sobota’s complex the sign of the main
component, Vzz, of the electric field gradient tensor (EFG) and
asymmetry parameter, η,14 were unambiguously determined
from high-field Mo¨ssbauer measurements at elevated temper-
atures (120 and 180 K; 3.5 and 7 T applied field). These spectra
are not shown. For both compounds Vzz is negative and η is
zero with an experimental error of less than 0.1. The same
results should apply for 5 in solution. With the reasonable
assumption that the EFG of the FeII sites is dominated by the
nonspherical valence charge distributions14 this result indicates
2
a dz -type ground-state orbital for the sixth d electron which
Figure 5. Zero-field Mo¨ssbauer spectra of 5 (A, B, powder sample at
295 K and 4.2 K, respectively; C, acetonitrile solution at 80 K) and
[(thf)6FeII2(µ-Cl)3][SnCl5(thf)] (D, powder sample at 4.2 K). The
experimental data in (C) are corrected for precipitated microcrystals
of 5 (21%); the subspectra of the solid were subtracted by using the
77 K parameters given in Table 7.
exceeds the “spherical” half-filled d-shell. From corresponding
(EFG)val tensor elements listed in Table 8 it is then obvious
that out of the possible ground state functions a1 and 1e in
(approximate) C3V symmetry only a1 yields a large negative
expectation value for Vzz (as is observed). Taking Vzz
)
-4/7e r-3 , r-3 ) 5a0-3, and Q ) 0.15b for a hypothetical pure
valence contribution from a1 and neglecting covalency and
lattice contributions to the EFG, one would arrive at a
Table 7. Mo¨ssbauer Parameters of 5 (Crystalline and in CH3CN
Solution) and of Crystalline [(thf)6FeII2(µ-Cl)3][SnCl5(thf)]
temp, K
quadrupole splitting of ∆EQ ) -4.2 mm s-1 15
Since the ∆EQ
.
parama 4.2
77
160 295 rel intensity
values of 5 in solution and even more of Sobota’s complex in
the solid state approach this theoretical value, we postulate for
both complexes a relatiVely isolated a1 ground state for the FeII
sites. On the other hand, the reduced quadrupole splittings of
solid 5 then owe their origin to distortions of local C3 symmetry
which perturb the splitting of the d orbitals and induce spin-
orbit mixing of a1 and 1e states and thereby add positive
contributions to Vzz. An estimate based on the (EFG)val
components given in Table 8 shows that mixing of only 15%
1e contributions into the a1 ground state is sufficient to reduce
∆EQ to -1.5 mm s-1 (with η ) 0.06). This is close to the
experimental results for solid 5. Thus it is quite plausible that
the distortions obserVed in the molecular structure of solid 5
induce perturbations of the electronic structure.
It is now possible to assign the different Mo¨ssbauer subspectra
of solid 5 to individual FeII sites (Fe1‚‚‚Fe4) in the crystal of 5
by crude quantitation of the local deviations (i) from ideal C3
(C3v) symmetry and (ii) from the the Fe-Cl bond length of
2.488 Å in [Fe2(thf)6Cl3][SnCl5(thf)]. To this end we have
calculated (i) a distortion parameter ∆1 for each FeII site by
summing up the squares of deviations of the individual Fe-Cl
subspectrum 1c
of 5
δ
1.10 1.08 1.05 0.96
2.08 1.86 1.57 1.31
1.10 1.09 1.06 0.98
1.52 1.33 1.23 1.03
1.13 1.10 1.07 0.97
0.96 0.92 1.03 0.97
25%
50%
25%
∆EQ
δ
subspectrum 2c
of 5
∆EQ
δ
∆EQ
subspectrum 3c
of 5
CH3CN soln of 5
δ
∆EQ
1.16b
∼100%
2.79b
[(thf)6Fe2Cl3]
[SnCl5(thf)]
δ
∆EQ
1.30 1.29b
3.26 3.22b
a Isomer shift, δ (referenced vs R-Fe at 295 K), and quadrupole
splitting, ∆EQ, in mm s-1, respectively. b Measured at 80 K. c See Figure
5.
ably different (Figure 5D). It shows only one doublet with δ
) 1.3 mm s-1 and a large quadrupole splitting ∆EQ ) 3.27
mm s-1 at 4.2 K although it is noted that the spectrum shows
a slight asymmetry and a relatively large line width of Γ )
0.36 mm s-1 which indicates some degree of microheteroge-
neity. The enormous differences in ∆EQ suggest the presence
of different electronic ground states in 5 and Sobota’s complex.
Interestingly, the Mo¨ssbauer spectrum of 5 measured in CH3-
CN solution also displays a large quadrupole splitting of 2.8
mm s-1 at 80 K as shown in Figure 5C. Furthermore, the FeII
sites of the meso form and of the enantiomers become
indistinguishable as only one subspectrum is detected.
(14) Gu¨tlich, P. In Mo¨ssbauer Spectroscopy, Gonser, U., Ed.; Springer-
Verlag: Berlin, Heidelberg, New York, 1975; p 53.
(15) (a) Lauer, S.; Marathe, V. R.; Trautwein, A. X. Phys. ReV. 1979, A19,
1852. (b) Bominaar, E. L.; Guillin, J.; Sawaryn, A.; Trautwein, A. X.
Phys. ReV. 1989, B39, 72.

Bioctahedral Complexes [LMII(µ-X)3MIIL]BPh4
Inorganic Chemistry, Vol. 36, No. 13, 1997 2841
Table 9. Assignment of the Mo¨ssbauer Subspectra to Fe(II) Sites
in Crystalline 5 at 295 K
FeII
∆1,
∆2,
∆EQ, mm s-1 f
(at 4.2 K /295 K)
sitea
10-3 Å2 b,c
10-2 Åd
subspectrume
Fe1
Fe2
Fe3
Fe4
1.39
1.91
1.83
0.56
-0.4
+2.1
+0.8
+0.7
2 (25%)
3 (25%)
2 (25%)
1 (25%)
1.52/1.03
0.96/0.97
1.52/2.03
2.08/1.31
a Labels as in Table 3 and Figure 1 (replace Mn for Fe). b ∆1 )
3
∑
i)1(dFe-Cl(i) - dFe-Cl )2; dFe-Cl ) 1/3∑i3)1dFe-Cl(i). The bond lengths
dFe-Cl(i) are taken from Table 3. c Note that the calculated ∆1 values
for the two iron sites in [(thf)6Fe2(µ-Cl)3][SnCl5(thf)] are smaller: 0.05
× 10-3 and 0.33 × 10-3 Å-2
Figure 5. f Quadrupole splittings from Table 7.
.
d ∆2 ) dFe-Cl -2.488 Å, with 2.488 Å
being the average Fe-Cl distance in [Fe2(thf)6Cl3][SnCl5(thf)]. e See
bond length, dFe-Cl(i), from the average Fe-Cl bond length,
dFe-Cl , and a corresponding parameter ∆2 for each FeII site
by taking the difference between dFe-Cl and 2.488 Å. The
results are shown in Table 9. The smaller the deviation
parameters ∆1 and ∆2 the larger |∆EQ| should be according to
the arguments presented above. At 4.2 and 295 K one of the
Mo¨ssbauer subspectra in Figure 5A which accounts for one iron
site (subspectrum 1 with 25% relative intensity) has a signifi-
cantly larger ∆EQ than the other two subspectra which accout
for three other sites (50% + 25% relative intensity). The
deviation parameters ∆1 and ∆2 are significantly larger for Fe4
than those of Fe1, Fe2, and Fe3. Therefore, we assign
subspectrum 1 (Figure 5A) to Fe4 of the meso form of one of
the monocations in 5. For Fe2 this reasoning is opposite: Large
deviation parameters result in stronger spin-orbit mixing and,
therefore, in strongly reduced ∆EQ. Hence, we assign subspec-
trum 3 to Fe2. The remaining subspectrum 2 then represents
the practically unresolved overlaps of signals from sites Fe1
and Fe3.
The magnetic Mo¨ssbauer spectra of Sobota’s complex
provided further insight into the zero-field and exchange
interactions of this compound which proved to be helpful
resolving the ambiguities of the analysis of the magnetic
susceptibility data (see above). We have measured these spectra
in a wide temperature range (1.7-120 K). In weak or
moderately strong fields (1-4 T) the spectra showed only weak
magnetic splittings due to partial cancellation of internal and
applied fields. Resolved magnetic hyperfine lines were only
observed in a strong field of 7 T. This observation is fully
consistent with a weak intramolecular antiferromagnetic interac-
tion of the iron(II) spins and a positive zero-field splitting
yielding locally almost nonmagnetic “ms,i ) 0” ground levels.
Spin-Hamiltonian simulations for the strong field case revealed
a pronounced sensitivity on the parameters D and J which is
indicative of delicate mixing of magnetic substates by coupling
zero-field and exchange interactions. The “best results” of these
spin-Hamitonian simulations, obtained for the series of measured
Mo¨ssbauer spectra by adopting identical iron sites with the
constraint E/D ) 0, are shown in Figure 6. Despite the fact
that some details of the fits are poor, the essential features of
the spectra, i.e. the magnetic splittings for a wide temperature
range, are satisfactorily reproduced. This lends confidence to
the derived mixed nature of the spin ground level and the
sequence of substates; the D and J values obtained should,
therefore, be essentially correct.
Figure 6. Magnetic Mo¨ssbauer spectra of solid [(thf)6FeII2(µ-Cl)3]-
[SnCl5(thf)]. The solid lines represent spin-Hamiltonian simulations with
Si ) 2, η ) 0; J ) -0.8 cm-1, Di ) +3.5 cm-1, g ) 2.0; Ahi/gNµN
)
h
(-1.0,-10.2,-19.7)T, ; ∆EQ ) -3.27 mm s-1 (-3.19 mm s-1 at 120
K), and δ ) 1.3 mm s-1 (1.27 mm s-1 at 120 K).
Due to the increasing ambiguities we refrained from extending
the parameter set.
The spin-Hamiltonian parameters obtained in this fashion
from magnetic Mo¨ssbauer spectroscopy do not perfectly match
the set obtained from the analysis of the magnetic susceptibility.
We attribute this to the differing sensitivity of both methods
toward sample inhomogeneities. However, the Mo¨ssbauer
simulations do allow us to limit the parameter set to the range
0 < D < +5 cm-1 and -0.5 > J > -2 cm-1
.
Discussion. For bioctahedral face-sharing complexes con-
taining two 3d3 paramagnetic ions the observed exchange
coupling is invariably antiferromagnetic. An empiral correlation
between the M‚‚‚M distance and the magnitude of the coupling
constant J shows that the latter is not greatly affected by the
nature of the three bridging ligands (O2-, HO-, EtO-, Cl-, Br-,
J-) or the metal ion (VII, CrIII, MnIV).4 It was concluded that
in face-sharing complexes with d3-d3 configurations no magnetic
superexchange mediated by bridging ligands occurs but rather
a direct metal-metal interaction (overlap of a1 orbitals)
dominates. This is in excellent agreement with extended Hu¨ckel
MO calculations by Leuenberger and Gu¨del16 for [CrIII2X9]3-
complexes (X ) Cl, Br, I) and, more recently, ab initio
calculations by Ceulemans et al.17 on [TiIII2Cl9]3- (a d1-d1
system). These theoretical papers show that the dominant
contribution to the exchange interaction in these systems arises
from the direct interaction of the a1 orbitals (t2g in octahedral
symmetry).
Keeping these results in mind, we now attempt to rationalize
the exchange pathways in bioctahedral face-sharing complexes
containing two first-row transition metal ions with metal-metal
(n ) 5-8) configuration. For the manganese(II) complexes
(d5 high spin) 3 and 4 additional superexchange paths via the
The misfits of the spectral pattern are related to the high
sensitivity of the hyperfine splittings toward marginal variations
of D, E/D, and J. Such variations are easily induced by strain
on the FeII sites in the powder crystallites. For instance, missing
absorptions in the simulations could be corrected by allowing
for inhomogeneous spectral contributions with 0 < E/D < 0.1.
(16) Leuenberger, B.; Gu¨del, H. U. Inorg. Chem. 1986, 25, 181.
(17) Ceulemans, A.; Heylen, G. A.; Chibotaru, L. F.; Maes, T. L.; Pierloot,
K.; Ribbing, C.; Vanquickenborne, L. G. Inorg Chim. Acta 1996, 251,
15.

2842 Inorganic Chemistry, Vol. 36, No. 13, 1997
Bossek et al.
Figure 7. Splitting caused by a trigonal distortion (C3V) superposed
on an octahedral field.23
half-occupied eg orbitals must be considered. In contrast, for
the nickel(II) complexes 8 and 9 a direct metal-metal interaction
is not possible because the metal ion t2g orbitals are filled. In
the cobalt(II) complexes 6 and 7 both possibilities are available.
Notwithstanding the fine details in all of these complexes the
exchange coupling is always antiferromagnetic. Thus the
ferromagnetically coupled iron(II) complex 5 in conjunction with
Sobota’s antiferromagnetically coupled complex represent test
cases for any consistent interpretation of the exchange coupling
in complexes containing this structural motif.
The tris(µ-chloro)-bridged complex 3 as well as its tris(µ-
bromo)-bridged analog 4 display a weak antiferromagnetic
coupling of similar magnitude. The slightly diminished coupling
in 4 as compared to 3 might be explained by the expected
increase of the metal‚‚‚metal distance on substituting the chloro
for bromo bridges (see the structural changes in 8 and 9) and
the resulting diminished overlap integral of the a1 orbitals.
Interestingly, the antiferromagnetic exchange coupling in 9
is stronger than in 8 despite the fact that the Ni‚‚Ni distance in
9 at 3.234 Å is longer than in 8 at 3.044 Å. As pointed out
above the direct metal-metal interaction via a1 orbitals can be
ruled out because the only available magnetic orbitals of the
octahedral NiII ions are eg orbitals. In 8 and 9 a σ-superex-
change via bridging halide ions prevails and overlap of the e
orbitals with a filled p orbital of the bridge is slightly stronger
(more covalent) for the bromo complex 9 than for the chloro
species 8.18 The Ni-Br-Ni and Ni-Cl-Ni angles at 76.9 and
76.8°, respectively, are similar and much smaller than 90°. It is
noted that Elerman et al.19 have recently reported the magnetism
of a dinuclear triply phenoxo bridged dinickel(II) complex which
is of the face-sharing bioctahedral type. The Ni-O-Ni angles
range from 86.2 to 88.3° which is close to 90°, and the Ni‚‚‚Ni
distance is at 2.884 Å. A ferromagnetic exchange coupling
(J ) +11.5 cm-1) has been reported and interpreted in terms
of a superexchange pathway.
Figure 8. Generation of possible excited states 3-5 by one-electron
transfer steps in the 1e orbitals assuming (a) parallel 1 or (b) antiparallel
2 spin alignment in the ground state of tris(µ-chloro)diferrous com-
plexes. (c) Relative energies of the ground and excited states without
interaction between the metal ions (â ) 0) and with such interaction
(â * 0).
Exchange coupling in the cobalt(II) complexes 6 and 7 with
local high-spin d7 configuration at the cobalt(II) ions is
antiferromagnetic and within experimental error of identical
magnitude in both species. It is conceivable that the expected
diminished direct exchange going from 6 to 7 (due to the
increasing Co‚‚‚Co distance) is compensated by a more effective
σ-superexchange pathway in the tris(µ-bromo) complex. A
more detailed analysis is not possible at this point because the
order of the a1 and e orbitals in C3 symmetry (Figure 7) is not
known.
(Figure 7). Mo¨ssbauer spectroscopy of Sobota’s complex
clearly shows that in this species a nearly unperturbed a1 ground
2
state prevails giving rise to an a1 1e22e2 electronic configuration
at the FeII ions. It is then immediately obvious that direct
exchange via a1 orbitals does not provide a suitable pathway
for exchange because these are filled. The half-filled 1e and
2e magnetic orbitals yield according to the Goodenough-
Kanamori rules20 an antiferromagnetic interaction. Remarkably,
the observed antiferromagnetic coupling is rather weak (J )
In the following we analyze the electronic structures of 5
and Sobota’s complex which both contain the tris(µ-chloro)-
diferrous moiety but different terminal ligands. As pointed out,
in trigonal symmetry the t2g metal orbital split into a1 and 1e
(20) (a) Anderson, P. W. Phys. ReV. 1950, 79, 350. (b) Goodenough, J. B.
Phys. ReV. 1955, 100, 564. (c) Goodenough, J. B. J. Phys. Chem.
Solids 1958, 6, 287. (d) Kanamori, J. J. Phys. Chem. Solids 1959, 10,
87. (e) Anderson, P. W. In Magnetism; Rado, G. T., Suhl, H., Eds.;
Academic Press: New York, 1963; Vol. 1, Chapter 2. (f) Ginsberg,
A. P. Inorg. Chim. Acta 1971, 5, 45.
(18) Hatfield, W. E. In Theory and Applications of Molecular Paramagnet-
ism; Boudreaux, E. A., Mulay, L. N., Eds.; John Wiley & Sons: New
York, 1976; p 352.
(19) Elerman, Y.; Kabak, M.; Svoboda, I.; Fuess, H.; Griesar,K.; Haase,
W. Z. Naturforsch. 1996, 51b, 1132.

Bioctahedral Complexes [LMII(µ-X)3MIIL]BPh4
Inorganic Chemistry, Vol. 36, No. 13, 1997 2843
-2 cm-1) as compared to 3, 4, and 6-9. A reason for this
observation may be the differing terminal ligands (six tetrahy-
drofuran ligands in Sobota’s complex vs two 1,4,7-trimethyl-
1,4,7-triazacyclononane macrocycles in our complexes). This
is corroborated by the data in Table 1. For [(thf)6Co2(µ-Cl)3]-
[SnCl5(thf)] an antiferromagnetic coupling of J ) -1.7 cm-1
has been reported whereas for 6 J is -13.1 cm-1. Similarly, J
) -6 cm-1 for [(thf)6Ni2(µ-Cl)3][SnCl5(thf)] and -10.5 cm-1
for 8. Similar observations have been reported by Roman et
al.21 for a series of (µ-oxalato)dinickel(II) complexes with
differing terminal ligands.
coupling between the two ferrous ions. Configurations 1 and
2 are degenerate states neglecting the necessary resonance
integral âii for coupling, but the excited states 3-5, which
correspond to metal-metal charge transfer configurations, are
energetically split due to intraatomic exchange, K, and double
orbital occupation (Figure 8c). The energy of the excited state
5 is larger than that of 4 because one orbital of the FeIII site in
5 is occupied by two electrons in contrast to 4 where each orbital
is only half-filled. The Coulomb repulsion, U′, of two electrons
in one orbital exceeds that of two electrons in two orbitals U
(U′ > U). The interaction parameter âii mixes then the ground
subspace with the excited one in a fashion that lifts the
degeneracy of the ground space.22 The magnitude of stabiliza-
tion of 1, or 2 is then inversely proportional to the energy
difference to the excited states. Excited state 3 with S ) 4 is
lowest in energy, and consequently, the S ) 4 ground state of
5 is more stabilized than the corresponding S ) 0 state.
In the foregoing discussion we have ignored the possible
influence of the potential exchange which exhibits always a
smaller ferromagnetic contribution than kinetic exchange.20 We
assume that in 5 and Sobota’s complex the potential exchange
contributions are very similar due to their similar core structures.
At this point it is not at all straightforward to find an
additional ferromagnetic exchange pathway between two ferrous
ions with a1 ground state as is experimentally established for
complex 5.
From the Mo¨ssbauer spectra we had concluded that the local
ground state of the FeII ions in 5 and Sobota’s complex differ.
The local ground state of an FeII ion in 5 is a mixture of a1
(∼85%) and 1e (∼15%). We identify two possible origins for
this orbital admixture in crystalline 5: structural and electronic.
(i) Due to packing effects in crystalline 5 the local symmetry
at each iron(II) site is less than trigonal. This has been verified
by the crystal structure determination of 5. In Sobota’s complex,
on the other hand, the symmetry is much closer to C3V. A pure
a1 ground state can only exist in C3 symmetry, and lowering
the symmetry facilitates mixing of wave functions a1 and 1e.
(ii) It should also be considered that the donor atoms of the
terminal ligands in 5 are six pure σ-donors (tertiary nitrogen
atoms) whereas there are six oxygen donor atoms in Sobota’s
complex (thf) which may exert some π-donor capability in
addition to being good σ-donors. π-Donation destabilizes the
1e orbitals in Sobota’s complex and increases the energy
difference between the a1 and 1e orbitals. Thereby, mixing of
these orbitals becomes less effective as compared to the situation
in 5.
Acknowledgment. We thank the Fonds der Chemischen
Industrie and the Deutsche Forschungsgemeinschaft (Schwer-
punktprogramm “Bioanorganische Chemie”) for financial sup-
port.
Supporting Information Available: Listings of crystallographic
data, bond lengths and angles, anisotropic thermal parameters, and
calculated positions of hydrogen atoms for complexes 1, 3, 5, 6, 8,
and 9 (66 pages). Ordering information is given on any current
masthead page.
IC970119H
(22) Hotzelmann, R.; Wieghardt, K.; Flo¨rke, U.; Haupt, H.-J.; Weatherburn,
D. C.; Bonvoisin, J.; Blondin, G.; Girerd, J.-J. J. Am. Chem. Soc. 1992,
114, 1681.
(23) Casey, A. T.; Mitra, S. In Theory and Applications of Molecular
Paramagnetism; Boudreaux, E. A., Mulay, L. N., Eds.; John Wiley
& Sons: New York, 1976; pp 109, 110.
Now we focus on exchange pathways in a pure 1e configu-
ration. Figure 8 gives the energetically lowest excited states,
which can be generated by an intramolecular one-electron
transfer,20a taking into account parallel or antiparallel spin
(24) Gu¨tlich, P.; Link, R.; Trautwein, A. X. In Mo¨ssbauer Spectroscopy
and Transition Metal Chemistry; Springer-Verlag: Berlin, Heidelberg,
New York, 1978.
(21) Roman, P.; Guzman-Miralles, C.; Luque, A.; Beitia, J. I.; Cano, J.;
Lloret, F.; Julve, M.; Alvarez, S. Inorg. Chem. 1996, 35, 3741.