Slow magnetic relaxation in trigonal-planar mononuclear Fe(II) and Co(II) bis(trimethylsilyl)amido complexes - A comparative study

Article abstract of DOI:10.1021/ic401677j

Alternating current magnetic investigations on the trigonal-planar high-spin Co²⁺ complexes [Li(15-crown-5)] [Co{N(SiMe ₃)₂}₃], [Co{N(SiMe₃) ₂}₂(THF)] (THF = tetrahydrofuran), and [Co{N(SiMe ₃)₂}₂(PCy₃)] (Cy = -C ₆H₁₃ = cyclohexyl) reveal that all three complexes display slow magnetic relaxation at temperatures below 8 K under applied dc (direct current) fields. The parameters characteristic for their respective relaxation processes such as effective energy barriers U_eff (16.1(2), 17.1(3), and 19.1(7) cm^-1) and relaxation times τ₀ (3.5(3) × 10^-7, 9.3(8) × 10^-8, and 3.0(8) × 10^-7 s) are almost the same, despite distinct differences in the ligand properties. In contrast, the isostructural high-spin Fe²⁺ complexes [Li(15-crown-5)] [Fe{N(SiMe₃)₂}₃] and [Fe{N(SiMe₃)₂}₂(THF)] do not show slow relaxation of the magnetization under similar conditions, whereas the phosphine complex [Fe{N(SiMe₃)₂}₂(PCy₃)] does, as recently reported by Lin et al. (Lin, P.-H.; Smythe, N. C.; Gorelsky, S. I.; Maguire, S.; Henson, N. J.; Korobkov, I.; Scott, B. L.; Gordon, J. C.; Baker, R. T.; Murugesu, M. J. Am. Chem. Soc. 2011, 135, 15806.) Distinctly differing axial anisotropy D parameters were obtained from fits of the dc magnetic data for both sets of complexes. According to density functional theory (DFT) calculations, all complexes possess spatially nondegenerate ground states. Thus distinct spin-orbit coupling effects, as a main source of magnetic anisotropy, can only be generated by mixing with excited states. This is in line with significant contributions of excited determinants for some of the compounds in complete active space self-consistent field (CASSCF) calculations done for model complexes. Furthermore, the calculated energetic sequence of d orbitals for the cobalt compounds as well as for [Fe{N(SiMe₃)₂} ₂(PCy₃)] differs significantly from the prediction by crystal field theory. Experimental and calculated (time-dependent DFT) optical spectra display characteristic d-d transitions in the visible to near-infrared region. Energies for lowest transitions range from 0.19 to 0.35 eV; whereas, for [Li(15-crown-5)][Fe{N(SiMe₃)₂}₃] a higher value is found (0.66 eV). Zero-field ⁵⁷Fe Moessbauer spectra of the three high-spin iron complexes exhibit a doublet at 3 K with small and similar values of the isomer shifts (δ), ranging between 0.57 and 0.59 mm/s, as well as an unusual small quadrupole splitting (ΔE_Q = 0.60 mm/s) in [Li(15-crown-5)][Fe{N(SiMe₃)₂}₃].

Full text of DOI:10.1021/ic401677j

Article
pubs.acs.org/IC
Slow Magnetic Relaxation in Trigonal-Planar Mononuclear Fe(II) and
Co(II) Bis(trimethylsilyl)amido ComplexesA Comparative Study
Andreas Eichhofer,*,† Yanhua Lan,‡ Valeriu Mereacre,‡ Tilmann Bodenstein,† and Florian Weigend†
̈
†Institut fur Nanotechnologie, Karlsruher Institut fur Technologie (KIT), Campus Nord, Hermann-von-Helmholtz-Platz 1, 76344,
̈
̈
Eggenstein-Leopoldshafen, Germany
‡Institut fur Anorganische Chemie, Karlsruher Institut fur Technologie (KIT), Campus Sud, Engesserstr. 15, 76131, Karlsruhe,
̈
̈
̈
Germany
S
* Supporting Information
ABSTRACT: Alternating current magnetic investigations on the trigonal-
planar high-spin Co2+ complexes [Li(15-crown-5)] [Co{N(SiMe3)2}3],
[Co{N(SiMe3)2}2(THF)] (THF = tetrahydrofuran), and [Co{N-
(SiMe3)2}2(PCy3)] (Cy = −C6H13 = cyclohexyl) reveal that all three
complexes display slow magnetic relaxation at temperatures below 8 K under
applied dc (direct current) fields. The parameters characteristic for their
respective relaxation processes such as effective energy barriers Ueff (16.1(2),
17.1(3), and 19.1(7) cm−1) and relaxation times τ0 (3.5(3) × 10−7, 9.3(8) ×
10−8, and 3.0(8) × 10−7 s) are almost the same, despite distinct differences
in the ligand properties. In contrast, the isostructural high-spin Fe2+
complexes [Li(15-crown-5)] [Fe{N(SiMe3)2}3] and [Fe{N-
(SiMe3)2}2(THF)] do not show slow relaxation of the magnetization under similar conditions, whereas the phosphine complex
[Fe{N(SiMe3)2}2(PCy3)] does, as recently reported by Lin et al. (Lin, P.-H.; Smythe, N. C.; Gorelsky, S. I.; Maguire, S.; Henson,
N. J.; Korobkov, I.; Scott, B. L.; Gordon, J. C.; Baker, R. T.; Murugesu, M. J. Am. Chem. Soc. 2011, 135, 15806.) Distinctly
differing axial anisotropy D parameters were obtained from fits of the dc magnetic data for both sets of complexes. According to
density functional theory (DFT) calculations, all complexes possess spatially nondegenerate ground states. Thus distinct spin−
orbit coupling effects, as a main source of magnetic anisotropy, can only be generated by mixing with excited states. This is in line
with significant contributions of excited determinants for some of the compounds in complete active space self-consistent field
(CASSCF) calculations done for model complexes. Furthermore, the calculated energetic sequence of d orbitals for the cobalt
compounds as well as for [Fe{N(SiMe3)2}2(PCy3)] differs significantly from the prediction by crystal field theory. Experimental
and calculated (time-dependent DFT) optical spectra display characteristic d−d transitions in the visible to near-infrared region.
Energies for lowest transitions range from 0.19 to 0.35 eV; whereas, for [Li(15-crown-5)][Fe{N(SiMe3)2}3] a higher value is
found (0.66 eV). Zero-field 57Fe Moßbauer spectra of the three high-spin iron complexes exhibit a doublet at 3 K with small and
̈
similar values of the isomer shifts (δ), ranging between 0.57 and 0.59 mm/s, as well as an unusual small quadrupole splitting
(ΔEQ = 0.60 mm/s) in [Li(15-crown-5)][Fe{N(SiMe3)2}3].
single-ion molecule magnet behavior, which was first observed
in lanthanide-containing complexes,8,9 was also reported for
several first-row transition metal complexes of iron10−12 and
cobalt13−15 for different coordination geometries including the
trigonal-planar complex [Fe{N(SiMe3)2}2 (PCy3)] (Cy =
cyclohexyl).16 Slow magnetic relaxation at a zero dc (direct
current) field was for the first time observed in the PPh4+ salt of
the tetrahedral cobalt thiolate complex [Co(SPh)4]2−.14
INTRODUCTION
■
Coordinatively unsaturated compounds of iron are of
increasing interest because of their magnetic properties.1
Three-coordinated complexes with [LFeIIX] chromophores
(L = β-diketiminate; X = -Cl, -CH3, -NHTol, -NHt-Bu)2 as well
as two-coordinated complexes like Fe[C(SiMe3)3]2,3 Fe[N(t-
Bu)2]2,4 [Fe{N(H)Ar}2] (Ar = C6H3-2,6-(C6H2-2,4,6-i-Pr3)2 or
C6H3-2,6-(C6H2-2,4,6-Me3)2),5 and [Fe(C(SiMe3)3)2]− display
remarkably large orbital contributions to their ground-state
Mononuclear three-coordinate complexes of Fe2+ and Co2+
are for example accessible as Lewis base adducts of well-known
homoleptic bis(trimethylsilyl)amide complexes [M{N-
(SiMe3)2}2] (M = Co17,18 and Fe19,20) like [Fe{N(SiMe3)2}2
(PCy3)],16 [Fe{N(SiMe3)2}2 (THF)] (THF = tetrahydrofur-
an),19,20 [Co{N(SiMe3)2}2 (PPh3)]21, and [M{N(SiMe3)2}2
magnetic behavior. Furthermore, zero-field Moßbauer spectra
̈
of Fe[C(SiMe3)3]2 and Fe[N(t-Bu)2]2 at 4.2 K exhibit internal
fields (Hint) of 152 and 105 T respectively, which are far larger
than those for any other iron complex. Consequently, all of
these compounds show unusual slow relaxation of the
magnetization at low temperatures, with numbers for the
energy barriers of the spin reversal up to 186 cm−1 in
[Fe{N(SiMe3) (Dipp)}2] (Dipp = C6H3-2,6-i-Pr).6,7 Such
Received: July 3, 2013
Published: January 27, 2014
© 2014 American Chemical Society
1962
dx.doi.org/10.1021/ic401677j | Inorg. Chem. 2014, 53, 1962−1974

Inorganic Chemistry
Article
(py)] (M = Fe, Co; py = pyridine),22 or as anions in [Na(12-
crown-4)2] [M{N(SiMe3)2}3] (M = Fe, Co).23 First-row
transition-metal silylamides have been shown to be generally
useful precursors for a wide variety of derivatives of these
metals.24−27 Investigation of transition-metal amides in our
laboratory originally stems from the interest to utilize them as
precursor complexes for the synthesis of mixed metal
chalcogenide cluster complexes28 as well as the synthesis of
polymeric metal chalcogenolato complexes.29
(Figure 1). The symmetry of the complexes is slightly distorted
from an idealized D3 point group symmetry (for bond distances
Herein we report on the synthesis and structural as well as
physical characterization of three trigonal-planar Co2+ bis-
(trimethylsilyl)amido complexes, in comparison to their Fe2+
analogues, with a focus on their magnetic properties.30
RESULTS AND DISCUSSION
■
Synthesis and Structure. The complexes [Li(15-crown-
5)] [M{N (SiMe3)2}3] (M = Fe(1), Co(2)) were synthesized
in good yields by the reaction of [M{N(SiMe3)2}2]2 (M = Fe,19
Co17) with lithium bis(trimethylsily)amide in toluene in the
presence of 15-crown-5 as reported for the amido complexes
[Li(12-crown-4)2] [M{N(SiMe3)2}3] (M = Mn, Fe, Co)23 (see
Scheme 1).
Figure 1. Molecular structure of the anion [Co(N(SiMe3)2)3]− (2a) in
the crystal of 2 (30% ellipsoids, H atoms omitted for clarity). Selected
bond lengths [pm] and angles [deg]: Co(1)−N(1) 196.9(2), Co(1)−
N(2) 197.4(2), Co(1)−N(3) 197.5(2), N(1)−Co(1)−N(2)
119.63(9), N(1)−Co(1)−N(3) 120.78(9), N(2)−Co(1)−N(3)
119.59(9). Selected bond lengths [pm] and angles [deg] for
isostructural [Fe(N(SiMe3)2)3]− (1a): Fe(1)−N(1) 198.4(2),
Fe(1)−N(2) 198.9(2), Fe(1)−N(3) 198.9(2); N(1)−Fe(1)−N(2)
119.75(9), N(1)−Fe(1)−N(3) 120.66(9), N(2)−Fe(1)−N(3)
119.58(9).
Scheme 1
Intensely green [Co{N(SiMe3)2}2 (THF)] (4)30 and green
[Co{N(SiMe3)2}2 (PCy3)] (6) were synthesized by the
recrystallization of [Co{N(SiMe3)2}2]2 in THF and from
toluene in the presence of 2 equiv of tricyclohexylphosphine,
respectively (see Scheme 2).
Scheme 2
Figure 2. Molecular structure of [Co(N(SiMe3)2)2 (THF)] (4) in the
crystal along a (30% ellipsoids, H atoms are omitted for clarity).
Symmetry transformation for generation of equivalent atoms: 1−x, y,
−z+3/2. Selected bond lengths [pm] and angles [deg]: Co(1)−N(1)
189.8(2), Co(1)−N(1)′ 189.8(2), Co(1)−O(1) 203.6(3), N(1)−
Co(1)−N(1)′ 141.78(12), N(1)−Co(1)−O(1) 109.11(6), N(1)′−
Co(1)−O(1) 109.11(6).
and angles see figure caption). With respect to the structures of
the isostructural M(III) neutral molecules [M{N(SiMe3)2}3]
(M = Fe,31 Co17) one observes extended M−N bond distances
in 1 and 2 (M = Fe, ∼7 pm; M = Co, ∼10 pm). This was also
observed before for isostructural [Li(12-crown-4)2] [M{N-
(SiMe3)2}3] and related to the larger ionic radii of the metal
ions in 1 and 2 as well as a weakening of the M−N bond by the
excess charge.23
Complex 4 crystallizes in the orthorhombic space group Pbcn
(Table 1); the molecular structure of 4 is shown in Figure 2. A
crystallographically imposed 2-fold axis is running through the
Co−O bond. The cobalt atom is coordinated by two
bis(trimethylsilyl)amido and one THF ligand in a distorted
trigonal-planar geometry. The Co−N distances (189.8(2) pm)
are shorter than the Co−O bond length (203.6(3) pm), and
the N−Co−N angle (141.78(12)°) is much larger than the two
To do a comparative study, we also reproduced the reported
isostructural iron compounds [Fe{N(SiMe3)2}2(L)] (L = THF
(3)19,20 and L = PCy3 (5)16). The detailed synthetic procedures
can be found in the Experimental Section.
The ionic complexes 1 and 2 crystallize both in the
monoclinic space group P21/n (Table 1), with one CH2Cl2
molecule in the asymmetric unit. The cations are built by
dimerization of two [Li(15-crown-5)]+ units, a structural motif
that has not been observed before (Supporting Information,
Figure S1). In the monomeric anions [M{N(SiMe3)2)3]− (M =
Fe (1a), Co(2a)) the metal ions are planary coordinated by the
three nitrogen atoms of the bis(trimethylsilyl)amido groups
1963
dx.doi.org/10.1021/ic401677j | Inorg. Chem. 2014, 53, 1962−1974

Inorganic Chemistry
Article
O−Co−N angles (109.11(6)°). Complex 4 is isostructural to
the corresponding iron compound 3, which diplays only slightly
different structural parameters (Fe−N: 191.6(5), Fe−O
207.1(6) pm; N−Fe−N: 144.0(3), O−Fe−N 108.0(1)°).19
Complex 6 crystallizes in the monoclinic space group P21/n
(Table 1). Two bis(trimethylsilyl)amido ligands and the
tricyclohexyl phosphine ligand form a distorted trigonal-planar
geometry around the cobalt atom (Figure 3). The Co−N
slightly different structural parameters (Fe−N: 195, 195; Fe−P
252 pm; N−Fe−N: 128.5, P−Fe−N 115.2, 116.3°).16
In comparison, structural differences between the cobalt and
the iron analogues are restricted to the slightly smaller ionic
radius of Co2+ in comparison to Fe2+ (Fe2+, high-spin, CN 4, 77
pm; Co2+, high-spin, CN 4, 72 pm).32 Upon going from 1 to 3
to 5 and from 2 to 4 to 6, the molecular point group symmetry
of the complexes is reduced from D3 over C2 to C1. Deflexions
of the metal atoms from the trigonal plane formed by the
coordinating ligand atoms are small (1: 0.72(12), 2: 0.66(13),
3: 0,20 4: 0, 5: 3.0,16 6: 2.6(1) pm). The shortest interatomic
distances between the metal atoms in the crystal lattices were
found to range from 874 to 924 pm (1: 894.0, 2: 896.0, 3:
873.7, 4: 880.2, 5: 923.7, 6: 916.8 pm).
The measured powder patterns of 1−6 show a good
agreement with the calculated ones based on the single crystal
data (Figures S2−S7 in the Supporting Information), which
proves the crystalline purity of the compounds.
Magnetic Behavior. For all complexes 1−6, dc and ac
(alternating current) magnetic properties have been studied on
crystalline powders. Taking into account the extreme air and
moisture sensitivity of the compounds, we prepared the dc
measurements of each compound on several samples of
different batches till we obtained at least two reproducible
measurements. For those we proceeded to carry out the ac and
temperature-dependent magnetization measurements.
Figure 3. Molecular structure of [Co(N(SiMe3)2)2(PCy3)] (6) in the
crystal (30% ellipsoids, H atoms omitted for clarity). Selected bond
lengths [pm] and angles [deg]: Co(1)−N(1) 192.9(2), Co(1)−N(2)
192.3(2), Co(1)−P(1) 251.8(1), N(1)−Co(1)−N(2) 125.44(8),
N(1)−Co(1)−P(1) 117.49(9), N(2)−Co(1)−P(1) 117.03(6).
Static magnetic behavior of complexes 1−6 were studied
between 1.8 and 300 K in a field of 0.1 T and by magnetization
measurements from 0 to 5 T at 2, 3, 4, 5, and 6 K. The μeff
versus T and M versus H curves at different temperatures were
simultaneously fitted using the PHI program33 by means of an
anisotropic spin Hamiltonian (SH) (with gx = gy):
distances (192.3(2), 192.9 pm) are distinctly shorter than the
Co−P bond length (251.8(1) pm), and the N−Co−N angle
(125.44(8)°) is larger than the two P−Co−N angles (117.03,
117.49(6)°). The cobalt complex 6 is isostructural to the
recently published iron compound 5, which only displays
Table 1. Crystallographic Data for [Li(15-crown-5)] [M{N(SiMe3)2}3] (M = Fe (1), Co(2)), [Co{N(SiMe3)2}2 (THF)] (4),30
and [Co{N(SiMe3)2}2 (PCy3)] (6)
1·CH2Cl2
2·CH2Cl2
4
6
sum formula
fw [g/mol]
crystal system
space group
cell a [pm]
b
C29H76Cl2FeLiN3O5Si6
849.16
C29H76Cl2CoLiN3O5Si6
852.24
C16H44CoN2OSi4
451.82
C30H69CoN2Si4P
660.14
monoclinic
P21/n
monoclinic
P21/n
monoclinic
P21/n
orthorhombic
Pbcn
1461.4(3)
2052.3(4)
1671.0(3)
107.77(3)
4772.6(18)
4
1461.4(3)
2049.4(4)
1672.7(3)
108.03(3)
4763.9(16)
4
1360.9(3)
1116.5(2)
1781.9(4)
1725.4(4)
1178.1(2)
2081.0(4)
111.33(3)
3940.2(14)
4
c
β [°]
V [106pm3]
Z
2707.6(9)
4
T [K]
180(2)
1.182
180(2)
1.188
180(2)
1.108
0.817
980
180(2)
1.113
dc [g cm−3
]
μ(λ) [mm−1
]
0.613
0.657
0.618
F[000]
1832
1836
1444
2θmax [deg]
meas reflns
52
54
52
51
22 197
35 936
15 163
2554
0.0446
1867
146
26 244
7394
unique reflns
Rint
8890
10 070
0.0350
0.0811
0.0992
5950
reflns with I > 2σ(I)
refined params
6878
7071
421
439
499
a
R1(I > 2σ(I))
0.0461
0.0470
0.0376
0.1086
0.0421
0.1183
b
wR2(all data)
0.1279
0.1304
a
b
2
2
R1 = ∑||Fo|−|Fc||/∑|Fo|. wR2 = [∑[w(Fo − Fc2)2]/∑[w(Fo )2]]1/2
.
1964
dx.doi.org/10.1021/ic401677j | Inorg. Chem. 2014, 53, 1962−1974

Inorganic Chemistry
Article
2
2
2
̂
̂
̂
̂
̂
H = D(Sz − 1/3S) + E(Sx − Sy )
̂
̂
̂
+ μB(SxgxBx + SygyBy + SzgzBz)
(1)
̂
where D = axial ZFS parameter, E = rhombic ZFS parameter, S
= spin operator, B = magnetic induction, g = Lande
μB = Bohr magneton.
́
factor, and
It is known that the reliability of such a model depends on
the electronic structure of the system under study, particularly
in the presence of (near-)degeneracies.34 For the complexes 1−
6, density functional theory (DFT) calculations reveal that the
ground state of each complex exhibits an orbital quantum
number L = 0 (see section Theory). Thus nonvanishing
contributions to L can only be generated by mixing with excited
states.35 To estimate the reliability of the SH parameters, we
refer to the complete active space self-consistent field
(CASSCF) and spin−orbit configuration interaction (SOCI)
calculations (see section Theory). The amount of mixing and
thus the estimated reliability of eq 1 at this level of theory is
quantified in the third column of Table 4: large contributions of
excited determinants indicate that strong spin−orbit effects
may occur, and hence the (perturbative) SH in eq 1 might be
only a crude approximation in these cases. Dynamic magnetic
properties were probed in ac measurements performed in the
1.8−10 K range using a 3.0 Oe oscillating ac field at frequencies
between 1 and 1500 Hz. From the frequency dependence of
the out-of-phase signal under an applied dc field at different
temperatures, the relaxation time τ was extracted by fits to eq 2
in the Supporting Information (Tables S1, S4, and S6). The
energy barriers Ueff for reversal of magnetization direction were
obtained according to Glauber’s theory,36 where the thermal
variation of τ is described by an Arrhenius expression (eq 4 in
the Supporting Information) with τ0 being a pre-exponential.
Below, we will first discuss the dc and ac magnetic properties of
the cobalt complexes 2, 4, and 6, followed by those of their iron
analogues 1, 3, and 5.
The effective magnetic moments (μeff) of 2, 4, and 6
continuously decrease with decreasing temperature (Figure 4a).
This deviation from the ideal Curie behavior is in the absence
of any close Co···Co contacts likely attributable to magnetic
anisotropy, which is indicative of a significant ZFS and g-factor
anisotropy resulting from the pseudotrigonal crystal field.
Magnetic anisotropy is also indicated by the room temperature
values of μeff, which are much larger than the theoretical spin-
only value of 3.87 μB mol−1 for one Co2+ ion (high-spin, S = 3/
2), which in turn results in the large g values derived from the
μeff at room temperature. Comparable large values for μeff at
room temperature have been reported for [Co(N(SiMe3)2)2
(THF)] (5.88 μB mol−1)30 and other three-coordinated Co2+
complexes [Co(N(SiMe3)2)2 (PMe3)] (4.71 μB mol−1),30
[Co(N(SiMe3)2)2 (PPh3)] (4.84 μB mol−1),21 [Co(N-
(SiMe3)2)2(py)] (4.6922, 5.2730 μB mol−1), and [Na(12-
crown-4)]2 [Co(N(SiMe3)2)3] (5.75 μB mol−1).30 The field
dependence of the magnetization (M) of 2, 4, and 6 at 2 K
displays similar curves (Supporting Information, Figure S8),
which suggests similar ground states of the molecules as also
supported by the similar values of μeff at 2 K. The lack of
saturation (MS (S = 3/2) = 3 N μB) is also indicative of
magnetic anisotropy in all three complexes.
Figure 4. Temperature dependence of the magnetic moment (μeff) of
(a) [Li(15-crown-5)] [Co(N(SiMe3)2)3] (2), [Co(N(SiMe3)2)2
(THF)] (4), [Co(N(SiMe3)2)2 (PCy3)] (6), and (b) [Li(15−
crown−5)] [Fe{N(SiMe3)2}3] (1), [Fe{N(SiMe3)2}2 (THF)] (3),
[Fe{N(SiMe3)2}2 (PCy3)] (5). Solid lines represent the results of
the simultaneous fittings with the temperature-dependent magnet-
ization (Table 2, Figures S8 and S22 in the Supporting Information),
according to a spin Hamiltonian (eq 1) by the PHI program.33
the Supporting Information. All three cobalt complexes display
anisotropic g parameters and large negative D values. The g
values for 4 and 6 are anisotropic, with gx = gy < gz, whereas
similar fits of the g factor of 2 resulted in gx = gy > gz. Quite
similar anisotropic g values like those in 2 have been reported
recently for [Na(14-crown-4)]2 [Co(N(SiMe3)2)3] (gx = gy =
2.97, gz = 2.75).30 However, this is in view of the large negative
D value (and a negative spin−orbit coupling parameter λ for
Co2+ (d7)) not in line with the “consistency criterion” derived
from perturbation theory which requires D = 0.5λ(gz − gx).37
For S = 3/2 ions, χT curves are known to be insensitive to the
sign of D;37,69 whereas, they are not when coupled by an E
parameter and simultaneously fit with temperature-dependent
magnetization data. Reasonable fits (R = 7.72 × 10−4) with a
positive D = 46.3 cm−1 resulted in gx = gy = 2.73 < gz = 3.13
(also not consistent with the above formula) and in significantly
increased E values (E = 17.4 cm−1), which gives rise to an E/
D > 1/3. Therefore the data of 2 were fitted with an isotropic g
factor and a negative D = −57 cm−1, which slightly deteriorates
the goodness of fit but results in a “proper”38 E/D < 1/3 (not
the case for positive D and isotropic g). Comparable large D
and E values have been found for trigonal-planar as well as for
To probe this magnetic anisotropy further, we performed
least-squares fits to eq 1 as described above. The fittings are in
general quite good. The best sets of parameters are listed in
Table 2 and shown as solid lines in Figure 4a and Figure S9 in
1965
dx.doi.org/10.1021/ic401677j | Inorg. Chem. 2014, 53, 1962−1974

Inorganic Chemistry
Article
a
Table 2. Results of the Fittings of the DC Magnetic Data by the PHI Program33 and of the Evaluation of the AC Magnetic
b
c
Data of 1−6
gx = gy
gz
D [cm−1
]
E [cm−1
]
R [10−3
]
Hdc [Oe]
Ueff [cm−1
]
τ0 [s]
d
[Li(15-crown-5)] [Co{N(SiMe3)2}3] (2)
[Co{N(SiMe3)2}2 (THF)] (4)
2.79
2.68
2.58
2.18
2.07
2.14
2.79
2.90
2.90
1.91
2.28
2.61
−57
−72
−82
+9.9
−20
−33
12.7
13.5
0
3.32
0.003
0.85
0.006
0.25
3.15
800
16.1(2)
18.1(3)
19.1(7)
3.5(3) × 10−7
9.3(8) × 10−8
3.0(8) × 10−7
600
[Co{N(SiMe3)2}2 (PCy3)] (6)
750
e
[Li(15-crown-5)] [Fe{N(SiMe3)2}3] (1)
0.0
4.0
3.4
0−1000
0−1000
600
e
[Fe{N(SiMe3)2}2 (THF)] (3)
[Fe{N(SiMe3)2}2 (PCy3)] (5)
16.0(3)
1.6(2) × 10−6
a
Simultaneous treatment of χT vs T and M vs H plots at different temperatures (Figure 4 and Figures S9 and S23 in the Supporting Information).
b
c
Figure 5 and S27 in the Supporting Information. Parameters: g, uniaxial magnetic anisotropy D, transversal magnetic anisotropy E, goodness-of-fit
d
factor R (least-square approach), external magnetic field Hdc, effective energy barrier Ueff and relaxation time τ0. A TIP of 0.0005 cm3 mol−1 has
e
been considered in the simulation of the μeff vs T data of 2; Lande
́
factor g refined isotropic (see text). A TIP of 0.0005 and 0.0007 cm3 mol−1 has
been considered in the simulation of the μeff vs T data of 1 and 3 respectively.
pentacoordinated Co2+ complexes: [Co(N(SiMe3)2)2 (THF)]
(D = −73 cm−1, E = 14.6 cm−1), [Co(N(SiMe3)2)2 (PMe3)]
(D = −74 cm−1, E = 9.6 cm−1), [Co(N(SiMe3)2)2 (py)] (D =
−82 cm−1, E = 21.0 cm−1), [Na(12-crown-4)]2 [Co(N-
(SiMe3)2)3] (D = −62 cm−1, E = 10.0 cm−1),30 [Co(bzimpy)-
Cl2] (bzipym = 2,6-bis{benzimidazol-2′-yl}-pyridine) (D = 73.4
cm−1, E = 3.28 cm−1)39, and octahedral [Co(iphos)2 (H2O)2]
(iphos = imino-bis{methylphophosphonate}) (D = 52 cm−1, E
=
17 cm−1).40 The large g-factor variations and D values
found for the cobalt complexes are indicative of the presence of
strong spin−orbit coupling contributions as recently out-
lined.35,41,42 It was already mentioned above that these are not
considered by the Hamiltonian (eq 1) used in the fittings.
Therefore fittings according to eq 1 might incorporate spin−
orbit interactions (including first-order terms) into D on the
one hand and into the principal values of gz and gx of the g
matrix on the other hand, which depends on the degree of
admixture of excited determinants (see Table 4).
The dynamic magnetic behavior of 2, 4, and 6 in terms of the
ac measurements is shown in Figure 5a and in the Supporting
Information, Figures S10−S21. In the absence of an external dc
field, the out-of-phase component of the ac susceptibility (χ″)
of 2, 4, and 6 has a much lower intensity than the in-phase
component (χ′), which indicates fast zero-field tunneling of the
magnetization.43,44 With the application of a static dc field the
intensity of χ″ is significantly enhanced for all three cobalt
complexes 2, 4, and 6 and reaches a similar order to that of χ′.
This effect is ascribed to a removal of the state degeneracy by
the dc field, which helps to suppress or slow down the
relaxation process through quantum tunneling.45,46 It is
worthwhile to mention that for compound 2 there is an
indication of a second relaxation process occurring at higher
frequencies (Supporting Information, Figure S10). This might
be attributed to a partial desolvation during the sample
preparation which leads to multiple species present in the probe
with slightly different relaxation barriers. The relaxation time of
2, 4, and 6 versus the inverse temperature, plotted in Figure 5b,
shows a curvature feature down to low temperatures suggesting
a crossover from thermally activated relaxation processes to
those which are associated with quantum tunneling of the
magnetization. The data at higher temperatures which
correspond to the thermally activated process follows an
Arrhenius law, leading to almost the same sets of parameters
(Table 2). These values are comparable to those of recently
reported examples such as the distorted square pyramidal
complexes [{ArNCMe}2 (NPh) ]Co(NCS)2 (Ueff = 11.1
cm−1, τ0 = 3.6 × 10−6 s) and [{ArNCPh}2 (NPh) ]Co-
Figure 5. (a) Temperature dependence of the out-of-phase χ″
component of the ac magnetic susceptibility at 750 Oe at different
frequencies for [Co{N(SiMe3)2}2 (PCy3)] (6) (solid lines are a guide
for eyes). (b) Relaxation time (τ) versus the inverse temperature (T−1)
for [Li(15-crown-5)] [Co(N(SiMe3)2)3] (2), [Co(N(SiMe3)2)2
(THF)] (4), and [Co(N(SiMe3)2)2 (PCy3)] (6). Standard deviations
of the relaxation times were determined from a nonlinear least-squares
analysis using the program Origin R (Version 6.1, OriginLab
Cooperation, 1991−2000); the error of ln τ was estimated from that
of τ by eq 6 in the Supporting Information. Error bars are omitted as
they are within the radius of the symbols. Linear fits represent
Arrhenius laws (eq 4 in the Supporting Information) and are shown as
solid lines (for corresponding parameters, see Table 3).
(NCS)2 (Ueff = 16.7 cm−1, τ0 = 5.1 × 10−7 s),13 the tetrahedral
complex [Co(SPh)4]2− (Ueff = 21 cm−1, τ0 = 1.0 × 10−7 s),14
and the pseudotetrahedral complex [(3G)CoCl]− (1; 3G =
1,1,1-tris-[2N-(1,1,3,3-tetramethylguanidino)methyl]ethane)
(Ueff = 24 cm−1, τ0 = 1.9 × 10−10 s).15 The relaxation times in
the order of 10−7 s are typical for an Orbach process. From
1966
dx.doi.org/10.1021/ic401677j | Inorg. Chem. 2014, 53, 1962−1974

Inorganic Chemistry
Article
least-squares fittings of the plots of χ″ versus χ′ (known as
Cole−Cole plots, Supporting Information, Figures S13, S17,
and S21) onto a distribution of single relaxation processes
(Debye model, eq 3 in the Supporting Information) the
distribution coefficients α could be derived (Tables S3, S5, and
S7). For all complexes, values of α < 0.2 were found which is
indicative for small distributions of single relaxation processes.
The room temperature μeff (5.4, 5.3, and 5.7 μB mol−1) of all
three iron complexes 1, 3, and 5 are found to be larger than the
theoretical spin-only value of 4.9 μB mol−1 for one Fe2+ ion
(high-spin, S = 2) (Figure 4b). The g values derived from the
μeff at room temperature for the S = 2 complexes amount to
2.16 (1), 2.20 (3) and 2.33 (5). However the enhancement is
not pronounced as strong as for the cobalt complexes which
might be understood by use of the equation geff = ge − α × λ/
10Dq, which quantifies the “g-factor variation” assuming
isotropic behavior and no first-order orbital angular momen-
the influence of strong spin−orbit interaction while those of 1
(almost no mixing with excited determinants) are thought to be
reliable.
In agreement with the findings from the static properties, the
dynamic magnetic behavior of the iron compounds 1, 3, and 5
also reveals pronounced differences between the different types
of molecules. Ac measurements performed under similar
conditions as for the cobalt analogues display no out-of-phase
signals χ″ for 1 and 3 even if an external dc field is applied up to
0.1 T. The absence of slow relaxation of magnetization for
complex 1 is in line with the positive D value. However, it
should be mentioned that typical single-molecule magnet
behavior has also been observed under an applied field for
mononuclear cobalt complexes that have a positive ZFS.15,49 In
contrast, a field-induced slow relaxation of magnetization was
observed in complex 5 similar to that reported.5 The
characteristic parameters of the relaxation process are
tum.53 Both ground terms of the free ions (Co2+, F9/2; Fe2+,
4
determined as Ueff = 16.0(3) cm−1 and τ0 = 1.6(2) × 10−6
s
5D4) split in a trigonal-planar ligand field of D3h point group
symmetry to give an A ground term, for which α = 4.10Dq is
expected to be almost the same for each pair of similar ligated
complexes, so that only the difference between the spin−orbit
coupling constants of the free ions of Co2+ (−177 cm−1) and
Fe2+ (−102 cm−1) remains.47 The experimental μeff of 1 and 3
are comparable to that of the reported three-coordinate
complex [Fe(N(SiMe3)2)2(py)] (5.3 μB mol−1),22 whereas 5
displays a higher value, although not as high as recently
reported (6.4 μB mol−1).16 When the compounds are cooled
from 300 to 100 K, the μeff of 1 and 3 slightly decrease, while
that of 5 slightly increases (Figure 4b). Below 50 K, 3 and 5
display a rapid downturn of the magnetic moment with further
decreasing temperature, which is, in the absence of any close
M···M contacts, indicative for the presence of zero-field
splitting (ZFS) of the spin states in these complexes.16,10 In
contrast, 1 does not display such an early deviation from ideal
Curie behavior. In agreement with distinctly different values of
μeff at 2 K (2.9, 4.1, and 5.0 μB mol−1), the curves of the field
dependence of the magnetization (M) of 1, 3, and 5 at this
temperature are also different (Supporting Information, Figure
S22). This behavior suggests either different ground states for
these complexes or distinct different magnetic anisotropies or
both. The lack of saturation (MS (S = 2) = 4 N μB) at high field
is quite pronounced for 1 which in principle indicates less
influence of magnetic anisotropy. The best sets of parameters of
simultaneous least-squares fittings of the μeff versus T and M
versus H curves at different temperatures using eq 1 are listed
in Table 2 and shown as solid lines in Figure 4 and Figure S23
in the Supporting Information. The g and D values of 1, 3, and
5 obtained in this way differ significantly. Fits of the anisotropic
g-factor of 1 resulted in gx = gy > gz which is in view of the
positive D = +9.9(1) cm−1 in agreement with the consistency
criterion mentioned before.37 The value of gz = 1.91 however is
unusual for a high-spin Fe2+ compound.37 In contrast the D
value of 3 is found to be negative (D = −20(1) cm−1) similar to
values reported for the three-coordinated complexes [(IPr)Fe-
{N (SiMe3)2}2] (IPr = 1,3-bis(diisopropylphenyl)imidiazol-
2ylidene) (g = 2.27, D = −18.2 cm−1) and [[(IMes)Fe
{N(SiMe3)2}2] (IMes =1,3-bis(2,4,6-trimethylphenyl)-
imidiazol-2ylidene) (g = 2.24, D = −23.3 cm−1).48 5 displays
distinctly larger values of gz = 2.61 and D = −33.0 cm−1
together with a moderate E = 3.4 cm−1. Again strong mixing
with excited determinants in the case of 3 and 5 (Table 4)
suggest that the ZFS parameters of theses complexes include
from our experimental data (Supporting Information, Figures
S24−S28), which differ from those in reference 16 (Ueff = 29.2
cm−1, τ0 = 6.0 × 10−7 s). At the present time, we have no sound
explanation for this discrepancy. The absence of slow relaxation
of magnetization for 3 is remarkable in view of the negative D
values derived from the fittings of the dc magnetic data and
from CASSCF/SOCI calculations.
The general difference between the ac magnetic behavior of
the cobalt and iron compounds might, apart from the difference
of the values of the spin−orbit coupling constant λ of the free
ions (larger g-factor variation of the Co2+ complexes), also be
explained by the fact that Co2+ is a noninteger spin system, a
so-called Kramers ion.14,50 Quantum tunneling by E is among
others thought to be one of the reasons for lowering the
theoretical spin reversal barrier. For every S > 1/2 the rhombic
anisotropy parameter E mixes only the MS 2 states. Hence,
the MS components of the Kramers doublets can not be
mixed by E. Although the other components can mix, for
example, for an S = 3/2 ion the MS = +3/2 state can mix with
MS = −1/2 and vice versa, these states are energetically
separated by D (in terms of eq 1) which makes mixing of MS
states and thus quantum tunneling in Kramers ions less
probable compared to non-Kramers ions. From Table 2 there
seems to be a correlation in 1−6 between the lowering of the
symmetry which is accompanied by the exchange of one strong
σ-donor/π-donor −N(SiMe3)2 ligand against a σ-donor/weak
π-donor THF ligand and a σ-donor/π-acceptor PCy3 ligand on
the one side, and the increase of the magnetic anisotropy
expressed by the anisotropy of the g-factors and the D values on
the other side. In addition, at a first glance one could also
conclude that differences in magnetic anisotropy are mirrored
by different sizes of the energy barrier for the relaxation of
magnetization (although less pronounced for the cobalt
complexes). However one has to consider that the observed
energy barriers are by far smaller than the calculated ones (U =
S2|D| for integer spin and U = (S2 − 0.25)|D| for half integer
spin) which is a commonly observed phenomenon and usually
assigned to the presence of quantum tunnelling effects.10,14 It
was also already pointed out that the ZFS parameters derived
from fittings to eq 1 and from CASSCF/SOCI calculations on
model complexes partially include contributions from strong
spin−orbit coupling effects. Furthermore, as already indicated
by the strong curvature of the ln τ versus 1/T graph it might be
possible that the Arrhenius model cannot be reasonably applied
here to fully describe the relaxation behavior of 1−6. Other
1967
dx.doi.org/10.1021/ic401677j | Inorg. Chem. 2014, 53, 1962−1974

Inorganic Chemistry
Article
mechanisms of spin interaction and spin relaxation might be
also effective like recently shown.1,6,7
For 1a and 3 it matches the qualitative orbital schemes derived
from crystal field theory for an idealized trigonal-planar
geometry (d1 case) and the removal of degeneracy of the e-
symmetry orbitals in the distorted structures (Figure S35 in the
We rule out the presence of dominant intermolecular
interactions because the shortest M···M distances are long
and not very different (1: 894.0, 2: 896.0, 3: 873.7, 4: 880.2, 5:
923.7, 6: 916.8 pm). In this respect it was found for
[Co(SPh)4]2− that mitigating intermolecular interaction by
diluting the sample has only an influence on the quantum
tunneling pathways (at low temperatures) whereas the value of
the energy barrier Ueff remains unchanged.14
Theory. The interpretation of the magnetic data of 1 to 6
leaves some questions especially concerning the distinctly
different behavior of 3 and 5 in the ac magnetic measurements.
In addition we were primarily interested in the question
whether the energy of the first transition may be used as rough
estimate when estimating the size of the ZFS in 1−6 as outlined
before.16,35,51
Electronic structures of compounds 1a, 2a, and 3−6 were
calculated with density functional theory (DFT), and are shown
in Figure 6 and Figures S29−S34 in the Supporting
Information. Data presented below were obtained with the
BP86 functional (and def2-TZVP bases), and the results are
qualitatively the same for the B3LYP functional. We start with
the discussion of the energetic sequence of the calculated
minority spin (beta spin) frontier metal d orbitals (Figure 6).
Supporting Information).2,52 For 1a and 3 the dz orbital is
2
lowest in energy and thus occupied. In 5 the dxz orbital instead
2
of the dz is lowest in energy and occupied, being stabilized by
the π-acceptor abilities of the PCy3 ligand; this was also
observed in reference 16. This change of electron density
distribution from the axial position toward the trigonal plane
may influence the magnetic anisotropy, which would be in line
with the different ac magnetic properties of 5 in comparison
with 1 and 3.
For all three cobalt complexes 2a, 4, and 6 the dxz and dyz
orbitals are lowest in energy and thus occupied. Highest in
2
2
2
energy are dxy and dx −y , dz is in between. Therefore the lowest
orbitals are either energetic degenerate and equally populated
(2a) or nondegenerate (4, 6). Again, this order is contrary to
the qualitative orbital schemes derived from crystal field theory
which neglects electron−electron interactions (Figure S35 in
the Supporting Information).2,52 Because of these schemes, the
degenerate dxz and dxy orbitals would be unequally populated in
2a resulting in a Jahn−Teller distortion which is not evidenced
by the calculated and measured structural parameters. Thus, in
contrast to the iron complexes, the orbital sequence is the same
for all cobalt complexes irrespective of the kind of ligands.
In all complexes 1a, 2a, and 3−6, minority spin d orbitals are
nondegenerate, leading to L = 0 for the ground states. Thus
only mixing of excited states into the ground state by spin−
orbit coupling can act as a main source of magnetic anisotropy
expressed by the size of the ZFS. Within a perturbative
approach, the magnitude of the coupling is proportional to the
energetic accessibility of excited states.37,53 Calculations of the
electronic d−d transitions by time-dependent density func-
tional theory (TDDFT) were performed on 1a, 2a, and 3−6
(Table 3). As expected, for compounds 1a, 3, and 5 and for
compounds 2a, 4, and 6, four and six d−d transitions,
respectively, are obtained, which are partly degenerate in 1a
and 2a. With the exception of compound 5 the positions of the
peaks are quite well-reproduced in the experimental spectra as
far as they are in the window of the experiment (see section
Electronic Spectra and Figures S36−S40 and Tables S9−S14 in
the Supporting Information).
The values for the lowest energy transition ΔE range from
0.19 to 0.35 eV (1533−2824 cm−1) in 2a and 3−6, the value
for 1a is significantly higher, 0.66 eV (5324 cm−1) (Table 3).
The energy difference between the lowest energy transitions of
trigonal-planar 1 and 5 amounts to 0.41 eV (3307 cm−1) and is
therefore comparable to those reported for 5 and the
tetrahedral complex [Fe{N(SiMe3) }2(dppe)] which display
values of 0.39 eV (3146 cm−1) or 0.44 (3549) from PBE or
B3LYP calculations respectively.16 Note that the excitation
energies obtained by TDDFT, i.e. the poles of the response on
an alternating field, are quite different from the energy
differences of occupied and virtual ground state orbitals. The
latter serve as a starting point, but large changes may happen
within the TDDFT procedure, in particular, if exchange
interactions are relevant. In view of the latter investigations
which reveal that 5 behaves as a single-ion magnet whereas
[Fe{N(SiMe3) }2(dppe)] does not, these numbers may help to
rationalize the distinctly smaller influences of ZFS (no out-of-
phase ac signal) for 1 compared to 5, but not the different
magnetic behavior of 3 and 5, for which the energy of the first
Figure 6. Energies and occupations of the five minority spin frontier
orbitals of (a) [Fe{N(SiMe3)2}3]− (1a), [Fe{N(SiMe3)2}2(THF)] (3),
[Fe{N(SiMe3)2}2(PCy3)] (5) and (b) [Co(N(SiMe3)2)3]− (2a),
[Co(N(SiMe3)2)2(THF)] (4), [Co(N(SiMe3)2)2(PCy3)] (6) and
the orientation of the metal d orbital contribution. The z axis is
perpendicular to the trigonal plane, and the x axis points to the THF
or PCy3 ligand.
1968
dx.doi.org/10.1021/ic401677j | Inorg. Chem. 2014, 53, 1962−1974

Inorganic Chemistry
Article
a
Table 3. Calculated (TDDFT) and Experimentally Observed d−d Transition Energies ΔE in 1−6
[Fe{N(SiMe3)2}3]− (1a)
theory experiment
ΔE
1.31
[Fe{N(SiMe3)2}2(THF)] (3)
theory experiment
ΔE
[Fe{N(SiMe3)2}2(PCy3)] (5)
theory experiment
ΔE
1.43
1.43
0.66
0.66
f
ε
ΔE
1.33
1.28
0.81
0.29
f
ε
ΔE
1.71
1.29
0.68
0.25
f
ΔE
1.33
1.03
ε
1.23
1.23
0.05
0.05
100
1.11
1.04
0.04
0.02
1.25
17
27
0.69
0.05
0.31
0.02
25
32
1.07
[Co{N(SiMe3)2}3]− (2a)
[Co{N(SiMe3)2}2(THF)] (4)
[Co{N(SiMe3)2}2(PCy3)] (6)
theory experiment
theory experiment
theory experiment
ΔE
f
ΔE
1.78
ε
ΔE
f
ΔE
1.83
ε
ΔE
f
ΔE
1.89
1.7
ε
1.69
1.69
1.15
1.04
0.19
0.19
2.9
155
1.59
1.46
1.06
0.93
0.52
0.35
3.4
105
1.87
1.69
1.14
0.99
0.58
0.19
1.11
0.32
0.07
0.43
0.03
0.05
61
90
19
28
2.9
0.006
0.004
0.16
0.007
0.13
0
0.92
20
0.81
10
0.98
0.85
0.34
0.0003
0.0003
a
Parameters: transition energy ΔE [eV], corresponding oscillator strength f (multiplied by 103) and extinction coefficient(ε [l mol−1 cm−1]). For the
spectra, see Figures S30−S35 in the Supporting Information, and for an assignment of the d−d transitions see Tables S9−S14 in the Supporting
Information.
transition is very similar. Obviously the above estimation, which
is considering only the energy denominator of the first term in
the perturbative expansion for the spin−orbit coupling, is not
sufficient here, probably, as the electronic situation (orientation
of occupied d-orbitals) in 5 is very different from that in 3.
Improvement may be achieved by regarding also subsequent
terms arising from higher transitions as well as finding suitable
weights for each term.
For the three cobalt complexes, the calculated lowest d−d
excitation energies are quite similar among each other (unlike
for the iron compounds). They amount to 0.19 eV (1533
cm−1) for 2a, to 0.35 eV (2823 cm−1) for 4, and to 0.19 eV
(1533 cm−1) for 6. This similarity is in line with similar g
parameters, relaxation times and energy barriers (Table 2) for
compounds 2, 4, and 6 and a moderate variance of the
anisotropy parameters D and E derived from the experimental
data (see Table 2).
Table 4. Calculated (CASSCF) Lowest Transition Energies
ΔE, Contributions of the CASSCF 2S + 1A Ground State (Co:
S = 3/2, Fe: S = 2) to the Lowest SOCI Eigenvector, ZFS
Parameters D, and E/D of the Model Complexes 1−6
a
b
ΔE [cm−1/eV]
contrib. [%] D [cm−1
]
E/D
c
c
c
[Co(NH2)3]−
504.3/0.06
1196.8/0.15
1489.3/0.19
2539.4/0.32
80.8/0.01
69.6
83.5
87.3
97.5
55.4
75.2
97.5
77.1
71.8
0.25
0.32
0.28
0.01
0.01
0.01
[Co(NH2)2(OH2)]
[Co(NH2)2(PH3)]
[Fe(NH2)3]−
+12.3
−75.9
−53.2
[Fe(NH2)2(OH2)]
[Fe(NH2)2(PH3)]
477.9/0.06
a
Lowest transition energies obtained from CASSCF naturally differ
from those derived by TDDFT (Table 3). However, these transitions
b
could not be detected due to experimental limitations. The meaning
of the D parameter in terms of eq 1 becomes unreliable in the case of
c
strong contributions of excited determinants. The sign of calculated D
becomes ambiguous in the limit of extreme rhombicity (E/D → 1/3).
DFT calculations of D and E in the way proposed by Neese
and co-workers54,55 yield similar values for all compounds of
the respective metal. For the iron complexes, calculated D
values ranges from −3.7 to −2.7 cm−1, whereas for the cobalt
compounds from +10.6 to +11.3 cm−1; calculated values of E/D
ranges from 0.21 to 0.23 for the iron compounds and from 0.05
to 0.07 for the cobalt compounds. This discrepancy to the
experimental data is not unexpected. A systematic computa-
tional study of high-spin Co(II)S4-containing complexes
already revealed that ab initio multireference wave function
methods are superior to DFT computational methods when
applied to calculate D values in a wider range than −10 < D <
+10 cm−1.56
We therefore performed complete active space self-consistent
field (CASSCF) and spin−orbit configuration interaction
(SOCI) calculations on model complexes of 1a, 2a, and 3−6
(see Experimental Section). The spin−orbit coupling con-
tribution to the ZFS tensor is calculated by means of quasi-
degenerate perturbation theory and effective Hamiltonian
theory.57 The results are summarized in Table 4 together
with the first CASSCF excitation energies, and contributions of
the CASSCF ground state to the lowest SOCI eigenvector. For
a comparison of the experimentally determined and calculated
ZFS parameters (Tables 2 and 4) one has to consider that in
the case of distinct admixtures of excited states, the calculated
ZFS parameters will loose their original physical meaning. In
addition, it is known that in the case of large rhombicity (E/D
→ 1/3), as obtained for all three cobalt complexes, the sign of
the calculated D is vague.58 It was also already mentioned above
that strong spin−orbit coupling contributions, whenever
present, are not considered by the Hamiltonian (eq 1) used
in the fittings. In agreement with this the discrepancies between
calculated and experimentally derived D values are found to be
pronounced in 2/2a, 3, and 5 with strong admixtures of excited
determinants (>20%) and lowest energy transitions below 500
cm−1. In contrast calculated D values of the model complexes of
1a (+12.3 cm−1), 4 ( 77.1 cm−1), and 6 ( 71.8 cm−1) are in
good agreement with the experimental ones (1: +9.9 cm−1, 4:
−72 cm−1, 6: −82 cm−1), in line with admixtures of excited
determinants less than 20% and lowest energy transitions above
1000 cm−1.
Electronic Spectra. Electronic spectra of compounds 1−6
were measured in a region from 5.8 to 0.56 eV (46 800−4520
cm−1), with the compounds in THF solution and as
polycrystalline powders between two quartz plates, and they
1969
dx.doi.org/10.1021/ic401677j | Inorg. Chem. 2014, 53, 1962−1974

Inorganic Chemistry
Article
were compared with the transitions calculated by TDDFT
(Table 3, Figures S36−S40, Tables S9−S14 in the Supporting
Information). Because of the occurrence of sharp bands, which
we assign to vibration overtones, the spectra were not recorded
further into the infrared (IR) region. In view of the extinction
coefficients of the bands in solution (ε > 2000 and ε < 150 l
mol−1 cm−1) all spectra can be roughly divided into a region
above and below 2.5 eV (20 165 cm−1), the former ones
assigned to charge transfer bands and the latter ones belonging
to d−d transitions. The position and appearance of the charge-
transfer bands differ for the same compound measured in
solution and in the solid state, which could be attributed to
solvatochromic effects and the nonvalidity of Lambert−Beers
law for the crystal mulls. However, the positions of these bands
correspond to the region observed for the d0 metal-ion complex
are too low in intensity, according to small calculated values of
the oscillator strength.
Information about d−d transitions in the absorption spectra
of comparable bis(trimethylsilyl)amido compounds of iron and
cobalt, especially in the formal oxidation state 2+ in the
literature, are scarce. For the trivalent complexes [M{N-
(SiMe3)2}3] (M = Fe, Co), transitions have been reported at
2.0 eV (16 132 cm−1) and 2.48 eV (20 004 cm−1) for the iron
compound,59 while the cobalt analogue was found to show two
absorptions at 2.73 eV (22 020 cm−1) and 2.07 eV (16 697
cm−1).60 Intensities were found to be quite strong in the both
cases ([Fe{N(SiMe3)2}3]: ε ≈ 400−450 l mol−1 cm−1;
[Co{N(SiMe3)2}3]: ε ≈ 2900−3600 l mol−1 cm−1). For the
four- and three-coordinated Co2+ complexes [Co{N(SiMe3)2}2
(py)2] and [Co{N(SiMe3)2}2 (py)], two narrow bands around
1.91 eV (15 406 cm−1) and 1.78 eV (14 358 cm−1) with ε ≈
300−350 1 mol−1 cm−1 were observed quite similar to the
transitions we have found.22 Absorption bands further in the
near-IR have so far not been reported.30 However, it is known
that in four-coordinate tetrahedral and pseudotetrahedral
cobalt(II) and iron(II) complexes the 4T1(F)←4A2 and
5T2←5E transitions, respectively, appear as broad absorptions
in the near-IR.61
[Sc{N(SiMe )2}3] (5.06 eV (40 814 cm−1) and 3.86 eV (31
3
135 cm−1))59 and the trivalent cobalt and iron analogues
[M{N(SiMe3)2}3] (M = Co: 4.63 eV (37 346 cm−1) and 3.90
eV (31 457 cm−1),60 M = Fe 3.68 eV (29 683 cm−1), 3.14 eV
(25 324 cm−1))59.
5
For a free Fe2+ ion the ground state is D4, which in a
trigonal-planar crystal field of D3h point-group symmetry splits
5
5
5
4
into A′ + E′ + E″, whereas the F9/2 ground state of a free
Co2+ ion would split in the same field into 4A′ + (4A1″, 4A2″) +
Moßbauer Spectra. 57Fe Moßbauer spectra of all three
̈
̈
59
4E′ + E″. In the ideal symmetric complexes one would
therefore expect two d−d transitions for the iron and three d−d
transitions for the cobalt complexes. Lowering of the symmetry
should affect a splitting of the degenerated states, leading to
four transitions in the iron case and six for the cobalt
complexes.
4
compounds 1, 3, and 5 at 3 K and at higher temperatures
exhibit a doublet (Figures S42−S45 in the Supporting
Information). This behavior is characteristic of an integer
spin paramagnet. Small values of the isomer shifts (δ) of 0.59
(1), 0.57 (3), and 0.59 mm/s (5) at 3 K are comparable with
other values found for high-spin three-coordinated iron(II) ions
in a weak ligand field (Supporting Information, Table
S15).2,62−64 Similar δ values for all three compounds indicate
a negligible influence of the different bonding situations in 1, 3,
and 5 on the s electron density at the iron nuclei. The observed
decrease in the isomer shifts of all compounds with increasing
temperature (Figures S42−S44 in the Supporting Information)
results from a second-order Doppler-shift contribution.
Contrary to the isomer shifts, the quadrupole splitting values
of 0.60 (1), 1.86 (3), and 1.25 mm/s (5) at 3 K are very
different. The formal oxidation state +II and the spin state
(high-spin) is assumed to be similar for all three complexes, as
indicated by the results from single crystal XRD, magnetic
measurements, and quantum-chemical calculations. Therefore
the differences in the electric field gradient should mainly arise
from different valence-electron contributions due to the
different ligands and different degrees of distortion of the
ligand field environment. The unusual small ΔEQ value of
compound 1, which is in principle comparable to those of high-
spin Fe(III)-S/SR complexes, is similar to that found in
trigonal-planar [Fe(SC6H3-2,4,6-t-Bu3)3]− (ΔEQ = 0.81 mm/
s).62 Upon lowering the symmetry in 3 and 5 compared to 1
the symmetric electron configuration (A′)2(E″)2(E′)2, which is
thought to be consistent with a relatively small quadrupole
splitting,60 changes, and ΔEQ increases. In agreement with this,
similar large values of ΔEQ (1.74 −1.61 mm/s) were also
observed before for the two unsymmetrically coordinated
trigonal-planar complexes [(β-ketiminate) Fe(R)] (R = −CH3,
Cl−).2
In the region from 2.5 to 0.56 eV (20 165−4520 cm−1) we
observed for the iron complexes 1, 3, and 5 in principle one
band around 1.15 eV (9276 cm−1), which is split upon lowering
the symmetry. As the solution spectra of 5 in THF are due to
quick ligand exchange almost the same as that of 3 (not
shown), we remeasured 5 in toluene.
For the cobalt complexes 2, 4, and 6, we observe in the same
spectral region two bands around 1.8 eV (14 520 cm−1) and
0.85 eV (6856 cm−1). Splitting of the bands is only observed in
the case of the solid-state spectrum of 6. The bright-green color
of the cobalt complexes can therefore be attributed to the d−d
transitions around 1.8 eV (14520 cm−1), whereas the slight
differences in the pale colors of the iron compounds are
determined by the different onsets of the charge transfer bands.
A comparison of the experimental spectra with the calculated
d−d transitions (Table 3) has to take into account that the
differences in the position of the bands could originate from the
difference in temperature between measurements and calcu-
lations. In addition the experimental intensities and thus the
possibility of observing the transitions could also depend on
temperature as well as the sample environment (crystal lattice
or solution). For the cobalt complexes 2, 4, and 6 the
agreement between theory and experiment is in general quite
good, suggesting that the lowest-energy transitions are located
further down in the IR region, which makes them difficult to
detect due to the appearance of vibration modes of the ligands.
For the iron complexes we found a good match for the spectra
of 1 and 3, while that of 5 does not agree with the predicted
one. This might indicate that, upon cooling, the electron
distribution and orbital sequence of 5 distinctly changes.
However, also in the case of the iron complexes, the lowest-
energy transitions are to be found below 0.5 eV (4033 cm−1) or
As is revealed in Figures S43−S45 in the Supporting
Information, at temperatures between 144 and 3 K the
quadrupole splittings are slightly dependent on temperature.
For compound 1 the change is of 0.07 mm/s from 120 to 3 K;
for 2, it is 0.04 mm/s from 100 to 3 K, and for 3 it is −0.06
1970
dx.doi.org/10.1021/ic401677j | Inorg. Chem. 2014, 53, 1962−1974

Inorganic Chemistry
Article
mm/s from 144 to 3 K. Changes in ΔEQ with temperature are
not unexpected in complexes of the type studied here, as a
result of rearrangement in the thermal population of the large
number of spin−orbit states.
line with modest variances in the anisotropy parameters and
lowest energy differences. However, one has to remember that,
for the determination of the effective energy barriers, probably
more influential parameters like spin-reversal relaxation path-
ways other than an Orbach process also have to be taken into
consideration, as was recently outlined for example by
Atanosov et al.7 In addition, to obtain a more refined picture
of the magnetic properties of the complexes under inves-
tigation, one would also have to confirm the obtained
anisotropic g values and the sign and size of the D values by
high-field electron spin resonance and magnetic measurements
on single crystals.
CONCLUSION
■
The comparative experimental and theoretical study of the
physical properties of the two sets of trigonal-planar M = Fe2+,
Co2+ high-spin complexes [Li(15-crown-5)] [M{N(SiMe3)2}3],
[M{N(SiMe3)2 }2(THF)], and [M{N(SiMe3)2 }2(PCy3)] leads
to the following conclusions.
Single-crystal X-ray diffraction analysis suggests only small
deviations (<3.0 pm) from trigonal planarity for the metal
atoms in all six complexes. In agreement with considerable
magnetic anisotropy derived from the dc magnetic data, the
Co2+ complexes display slow relaxation of the magnetization at
low temperatures; whereas, in the case of the iron compounds,
only [Fe{N(SiMe3)2}2 (PCy3)] comprises sufficiently high
anisotropy to show a similar behavior in the ac magnetic
measurements. The energetic ordering of the minority spin
frontier orbitals reveals that L = 0 for the ground states of all
complexes. Thus only mixing of excited states into the ground
state by spin−orbit coupling can act as a main source of
magnetic anisotropy, as is indicated to be present in all
compounds except [Li(15-crown-5)] [Fe{N(SiMe3)2}3] by
CASSCF/SOCI calculations on model complexes. To relate
the different behavior of the complexes in the ac measurements
to the ZFS parameters derived either from fits to a spin
Hamiltonian or from CASSCF/SOCI calculations on model
complexes, one has to be careful for two reasons. The absolute
value of D derived from CASSCF/SOCI calculations loses its
original meaning in the case of strong contributions of excited
determinants. In addition strong mixing with excited states
suggests the presence of orbital momentum contributions,
which are not considered by the Hamiltonian used in the
fittings of the dc magnetic data. In general the different
behavior of the two sets of complexes in the dc and ac magnetic
measurements is in agreement with a stronger spin−orbit
coupling constant λ of the free Co2+ ion in comparison with
Fe2+. In addition Co2+ is a so-called Kramers ion in which
mixing of the ground MS levels by transverse ZFS E is
forbidden, which leads to a minimization of the probability of
quantum tunneling.
EXPERIMENTAL SECTION
■
Synthesis. Standard Schlenk techniques were employed through-
out the syntheses using a double manifold vacuum line with high-
purity dry nitrogen (99.9994%) and an MBraun Glovebox with high-
purity dry argon (99.9990%). The solvent THF (tetrahydofuran) and
toluene were dried over sodium benzophenone, n-heptane over
LiAlH4, and distilled under nitrogen. Anhydrous dimethylenechloride
(CH2Cl2) (H2O < 0.005%) obtained from Aldrich was degassed,
freshly distilled, and stored over molecular sieves under nitrogen.
LiN(SiMe3)2, PCy3 (Cy = cyclohexyl, C6H11), and anhydrous CoCl2
were purchased from Aldrich. LiN(SiMe3)2 was distilled prior to use.
19
17
[Fe{N(SiMe3)2}2]2 and [Co{N(SiMe3)2}2]2 were synthesized
according to literature procedures.
[Li(15-crown-5)][Fe{N(SiMe3)2}3] (1). [Fe{N(SiMe3)2}2]2 (0.25 g,
0.332 mmol) and Li N(SiMe3)2 (0.111 g, 0.664 mmol) were dissolved
in 10 mL of toluene. Addition of 15-crown-5 (0.13 mL, 0.664 mmol)
resulted in the formation of a heavier green phase and a lighter
colorless phase. When it is cooled to approximately −70 °C by a dry
ice/methanol bath, the green phase becomes highly viscous. The
supernatant solution was replaced by 10 mL of toluene, warmed up,
stirred for a few minutes, and cooled again. After removal of the lighter
phase the rest was warmed to room temperature again and dried in
vacuum to give a dirty-green residue, which was redissolved in 5 mL of
CH2Cl2. Addition of 4.5 mL of n-heptane led to the formation of
crystals upon standing in the refrigerator (−42 °C). Another 5 mL of
n-heptane were added to complete the crystallization, and then the
whole solvent was removed at low temperatures before the crystals
were washed two times with heptane at room temperature to give a
total yield of 0.36 g (71%) of 1.
Anal. Calcd for C28H74O5FeLiN3Si6 (764.21): C 44.0, H 9.8, N 5.5.
Found: C 43.6, H 9.8, N 5.6%.
When it is heated, 1 decomposes visibly above 190 °C.
[Li(15-crown-5)] [Co{N(SiMe3)2}3] (2) was prepared in a similar
procedure to that of compound 1 from 0.25 g (0.332 mmol) of
[Co{N(SiMe3)2}2]2 and equimolar amounts of LiN(SiMe3)2 and 15-
crown-5 to yield 0.33 g (65%) of 2.
Anal. Calcd for C28H74O5CoLiN3Si6 (767.30): C 43.8, H 9.7, N 5.5.
Found: C 44.9, H 9.8, N 5.3%. A slightly enhanced value for C and
simultaneous decrease of N could be explained by traces of CH2Cl2 in
the crystalline powders.
The special magnetic behavior of [Fe{N(SiMe3)2}2 (PCy3)]
among the three iron complexes correlates with a change in the
local magnetic anisotropy. DFT calculations reveal for this
complex a change of the energetic order and occupation of the
2
dz and the dxz/dyz minority spin frontier orbitals upon
coordination of the σ-donor/π-acceptor PCy3 ligand. This
2
differs from the prediction by crystal-field theory, for which dz
is lowest in energy as observed in the other two iron complexes.
In contrast, the absence of a spin-reversal barrier in [Li(15-
crown-5)] [Fe{N(SiMe3)2}3] is in line with a positive D
(confirmed by CASSCF/SOCI calculations and also indicated
When it is heated, 2 decomposes visibly above 170 °C.
[Fe{N(SiMe3)2}2(THF)] (3). Addition of 1 mL of THF to [Fe{N-
(SiMe3)2}2]2 (2 g, 2.66 mmol) resulted, after gentle heating, in the
formation of a dark-green solution. Slow removal of excess THF under
reduced pressure afforded pale-green crystals of 3 with a total yield of
2.3 g (97%).
by Moßbauer spectroscopy) and the highest calculated energy
̈
for the lowest transition by TDDFT and CASSCF/SOCI.
In case of the cobalt compounds, for all three complexes, the
dxz/dyz minority spin frontier orbitals are occupied. Exchanging
the strong σ-donor/π-donor −N(SiMe3)2 ligand against one σ-
donor/weak π-donor THF ligand or a σ-donor/π-acceptor
PCy3 ligand has only a moderate influence on the orbital
energies and does not change their sequence. The energy
barriers of the cobalt complexes differ only slightly, which is in
Anal. Calcd for C16H44OFeN2Si4 (448.73): C 42.8, H 9.9, N 6.2.
Found: C 42.5, H 10.1, N 6.2%.
Crystals of 3 melt at around 45 °C and are extremely air- and
moisture-sensitive. Visible decomposition results in a quick color
change and formation of a black oil.
[Co{N(SiMe3)2}2(THF)] (4). Addition of 1 mL of THF to [Co{N-
(SiMe3)2}2]2 (2 g, 2.63 mmol) resulted, after gentle heating, in the
formation of a dark-green solution. Slow removal of excess THF under
1971
dx.doi.org/10.1021/ic401677j | Inorg. Chem. 2014, 53, 1962−1974

Inorganic Chemistry
Article
reduced pressure afforded intensely green crystals of 4, with a total
yield of 2.3 g (96%).
Anal. Calcd for C16H44OCoN2Si4 (451.81): C 42.5, H 9.8, N 6.2.
Found: C 41.9, H 9.8, N 6.1%.
Crystals of 4 melt at around 75 °C and are extremely air- and
moisture-sensitive. Visible decomposition results in a quick color
change from original intense green to black.
[Fe{N(SiMe3)2}2(PCy3)] (5). [Fe{N(SiMe3)2}2]2 (0.5 g, 0.66 mmol)
and PCy3 (0.37 g, 1.34 mmol) were dissolved in 5 mL of heptane;
when the mixture was gently heated, a dark brown solution resulted.
Evaporation of half of the solvent afforded the formation of a pale
powder that dissolves upon gentle heating to give 5 as huge pale-green
crystals after it stands in the freezer (−40 °C) for 3 d. These crystals
were washed two times with cold (−60 °C) pentane and dried under
vacuum to give a total yield of 0.53g (61%).
Zero-field-cooled temperature-dependent susceptibilities were re-
corded for 1−6 in dc mode using a MPMS-5S (Quantum Design)
SQUID magnetometer over a temperature range from 2 to 300 K in a
homogeneous 0.1 T external magnetic field. The ac susceptibility
measurements were performed with an oscillating ac field of 3 Oe and
ac frequencies ranging from 1 to 1500 Hz. The magnetization curves
were measured on the same instrument up to a dc field of 5 T. The
samples were contained in gelatin capsules filled in a glovebox under
argon atmosphere, owing to the high degree of moisture and oxygen
sensitivity of the compounds. The samples were transferred in sealed
Schlenk tubes from the glovebox to the magnetometer and then
rapidly transferred to the helium-purged sample space of the
magnetometer. The data were corrected for the sample holder
including the gelatin capsule and for diamagnetism using Pascal’s
constants.69−71 Details about the simulations are given in the
Supporting Information.
Anal. Calcd for C30H69FeN3PSi6 (657.05): C 54.8, H 10.6, N 4.3.
Found: C 54.9, H 10.6, N 4.3%.
The Moßbauer spectra were acquired using a conventional
̈
spectrometer in the constant-acceleration mode equipped with a
57Co source (3.7 GBq) in rhodium matrix. Isomer shifts are given
relative to α-Fe at room temperature. The samples were pressed in a
homemade holder between two 0.075 mm thick polyimide foils
(covered on the sample side with a 300 nm Al layer) separated by a
Viton O-ring inside a glovebox and then inserted inside an Oxford
[Co{N(SiMe3)2}2(PCy3)] (6). [Co{N(SiMe3)2}2]2 (1 g, 1.33 mmol)
and PCy3 (0.74 g, 2.66 mmol) were dissolved in 10 mL of heptane;
when the mixture was gently heated, a dark green solution resulted.
Evaporation of half of the solvent afforded the formation of a pale
powder that dissolves upon gentle heating to give 6 as green crystals
after it stands in the freezer (−40 °C) for 3 d. These crystals were
washed two times with cold (−60 °C) pentane and dried under
vacuum to give a total yield of 1.2 g (68%).
Instruments Moßbauer-Spectromag 4000 Cryostat. The sample
̈
temperature can be varied between 3.0 and 300 K. Three Kelvin
temperature could be achieved by pumping the sample space.
Quantumchemical Calculations. Density functional calculations
for complexes 1−6 were carried out with the program system
TURBOMOLE72 for the ground state73,74 as well as for the excited
states,75 using functionals BP8676,77 and B3LYP,78 def2-TZVP79 basis
sets, and corresponding RI-J auxiliary basis sets.80
Anal. Calcd for C30H69CoN3PSi6 (660.14): C 54.6, H 10.5, N 4.2.
Found: C 54.6, H 10.6, N 4.1%.
Crystallography. Because of the extreme air and moisture
sensitivity of the compounds, crystals suitable for single-crystal X-ray
diffraction were selected in perfluoroalkylether oil in a glovebox and
transferred rapidly under argon atmosphere to the diffractometer
equipped with an Oxford Cryosystem. Single-crystal X-ray diffraction
data of 1, 2, 4, and 6 were collected using graphite-monochromatised
Mo Kα radiation (λ = 0.710 73 Å) on a STOE IPDS II (imaging plate
diffraction system). Raw intensity data were collected and treated with
the STOE X-Area software Version 1.39. Data for all compounds were
corrected for Lorentz and polarization effects.
On the basis of a crystal description, a numerical absorption
correction was applied for 4.65 The structures were solved with the
direct methods program SHELXS of the SHELXTL PC suite of
programs66 and were refined with the use of the full-matrix least-
squares program SHELXL. Molecular diagrams were prepared using
Diamond.67
In 1, 2, 3, and 6 all Fe, Co, Si, N, and C atoms were refined with
anisotropic displacement parameters, while H atoms were placed in
fixed positions. In 1 and 2, solvent CH2Cl2 molecules and parts of the
15-crown-5 ether molecules were refined with a split model of site
disorder. In 4 the C atoms (C(1), C(3), C(4), and C(6)) were refined
with a split model of site disorder.
CCDC-873291 (1), 873292 (2), 863806 (4), and 884241 (6)
contain the supplementary crystallographic data for this Paper. These
data can be obtained free of charge at www.ccdc.cam.ac.uk/conts/
retrieving.html (or from the Cambridge Crystallographic Data Centre,
12 Union Road, Cambridge CB2 1EZ, UK; fax: (internat.) +44−1223/
336−033; e-mail: deposit@ccdc.cam.ac.uk).
X-ray powder diffraction patterns (XRD) for 1 and 2 (suspension of
crystals in CH2Cl2) and 3−6 (powder of crystals) were measured on a
STOE STADI P diffractometer (Cu Kα1 radiation, Germanium
monochromator, Debye−Scherrer geometry, Mythen 1K detector) in
sealed glass capillaries. The theoretical powder diffraction patterns
were calculated on the basis of the atom coordinates obtained from
single-crystal X-ray analysis by using the program package STOE
WinXPOW.68
Physical Measurements. C, H, and N elemental analyses were
performed on an Elementar vario Micro cube instrument.
The ultraviolet−visible absorption spectra in THF (1−4) and
toluene (5, 6) were measured on a Perkin-Elmer Lambda 900
spectrophotometer in quartz cuvettes. Solid-state spectra were
measured in transmission as micrometer-sized crystalline powders
between quartz plates in front of a Labsphere integrating sphere.
The CASSCF and SOCI calculations on the model complexes of 1
and 2 were performed using the ORCA program package in version
3.0.0.55 The X-ray geometries were truncated to [M(NH2)3]−,
[M(NH2)2(OH2)], and [M(NH2)2(PH3)] (M = Fe, Co) (Tables
S16 and S17 in the Supporting Information). After replacing the
functional groups with their simplified derivatives, only the coordinates
of the hydrogen atoms were optimized at the DFT level of theory
using the TURBOMOLE72 program system employing the B3LYP78
functional. In all calculations def2-TZVPP79 basis sets were used.
Molecular orbitals were determined for the average of the 10 and 5
high-spin roots at the CASSCF level of theory, choosing the five 3d-
based orbitals of Fe and Co as active, giving rise to CAS(6,5) and
CAS(7,5). The spin−orbit coupling matrix was constructed within the
active space and diagonalized to give the ZFS parameters, using
effective Hamiltonian theory.57 In test calculations, the spin−spin
coupling contributions to the ZFS tensor are below 1 cm−1.
ASSOCIATED CONTENT
■
S
* Supporting Information
Equations 1−6 for the simulation of the dc and ac magnetic
data. Tables S1−S17 for the relaxation fitting parameters from
least-squares fitting and Cole−Cole plots for χ″ versus χ′,
calculated and experimentally observed d−d transitions,
Moßbauer parameters for three-coordinate Fe2+ complexes,
̈
Cartesian coordinates of model complexes from the CASSCF
calculations. Figures S1−S45 for the molecular structure of the
cation [Li2(15-crown-5)2]2+ (1b), measured and simulated X−
ray powder patterns, field dependence of the magnetization M
at 2 K, plots of magnetization M versus H measured between 2
and 6 K and the corresponding fittings, frequency dependence
of χ′ and χ″ at 2 K under different dc fields, temperature
dependence χ′ and χ″ under a given dc field at different
frequencies, frequency dependence of χ′ and χ″ under a given
dc field at different temperatures, Cole−Cole plots for χ″ versus
χ′, contour plots of the calculated minority spin d orbitals
ordered by their orbital energies, schematic 3d orbital level
1972
dx.doi.org/10.1021/ic401677j | Inorg. Chem. 2014, 53, 1962−1974

Inorganic Chemistry
Article
(17) Burger, H.; Wannagat, U. Monatsh. Chem. 1963, 94, 1007.
̈
schemes, comparison of the experimental electronic spectra as a
powder of crystals between quartz plates and in THF with the
(18) Murray, B. D.; Power, P. P. Inorg. Chem. 1984, 23, 4584.
(19) Andersen, R. A.; Faegri, K.; Green, J. C.; Haaland, A.; Lappert,
M. F.; Leung, A.-P.; Rypdal, K. Inorg. Chem. 1988, 27, 1782.
(20) Olmstead, M. M.; Power, P. P.; Shoner, S. C. Inorg. Chem. 1991,
30, 2547.
(21) Bradley, D. C.; Hursthouse, M. B.; Smallwood, R. J.; Welch, A. J.
J. Chem. Soc., Chem. Commun. 1972, 872.
d−d transitions calculated by TDDFT, zero-field Moßbauer
̈
spectra at different temperatures. This material is available free
of charge via the Internet at http://pubs.acs.org. CCDC-
873291 (1), 873292 (2), 863806 (4), and 884241 (6) contain
the supplementary crystallographic data for this Paper. These
data can be obtained free of charge at www.ccdc.cam.ac.uk/
conts/retrieving.html (or from the Cambridge Crystallographic
Data Centre, 12 Union Road, Cambridge CB2 1EZ, UK; fax:
(internat.) +44−1223/336−033; e-mail: deposit@ccdc.cam.a-
c.uk).
(22) Panda, A.; Stender, M.; Olmstead, M. M.; Klavins, P.; Power, P.
P. Polyhedron 2003, 22, 67.
(23) Putzer, M. A.; Neumuller, B.; Dehnicke, K.; Magull, J. Ber.
̈
Bunsen-Ges. 1996, 129, 715.
(24) Power, P. P. Chem. Rev. 2012, 112, 3482.
(25) Power, P. P. Chemtracts 1994, 6, 181.
(26) Lappert, M. F.; Power, P. P.; Sanger, A. R.; Srivastava, R. C. In
Metal and Metalloid Amides; Horwood, E., Ed.; Chichester/Halsted
Press: New York, 1980.
(27) Nguyen, T.; Panda, A.; Olmstead, M. M.; Richards, A. F.;
Stender, M.; Brynda, M.; Power, P. P. J. Am. Chem. Soc. 2005, 127,
8545.
AUTHOR INFORMATION
Corresponding Author
*E-mail: andreas.eichhoefer@kit.edu. Phone: 49-(0)721-608-
26371. Fax: 49-(0)721-608-26368.
■
Notes
The authors declare no competing financial interest.
(28) Eichhofer, A.; Hampe, O.; Lebedkin, S.; Weigend, F. Inorg.
̈
Chem. 2010, 49, 7331.
ACKNOWLEDGMENTS
This work was supported by the Karlsruhe Institut fur
(29) Eichhofer, A.; Buth, G.; Dolci, F.; Fink, K.; Mole, R. A.; Wood,
̈
■
P. T. Dalton Trans. 2011, 40, 7022.
̈
(30) A paper which also contains investigations on the synthesis
structure as well as electronic and dc magnetic properties of
[Co{N(SiMe3)2}2(THF)] (4) and related trigonal-planar cobalt(II)
complexes was published during the reviewing process of this Paper by
Bryan, A. M.; Long, G. J.; Grandjean, F.; Power, P. P. Inorg. Chem.
2013, 52, 12152.
Technologie (KIT, Campus Nord). The authors wish to
thank A. K. Powell for generous support, K. Fink for helpful
discussions concerning the calculation of ZFS parameters, and
S. Stahl for the performance of the elemental analysis. T.
Bodenstein acknowledges funding from Deutsche Forschungs-
gemeinschaft via SFB/TRR 88 “3MET”.
(31) Hursthouse, M. B.; Rodesiler, P. F. J. Chem. Soc., Dalton Trans.
1972, 2100.
REFERENCES
■
(32) Shannon, R. D. Acta Crystallogr. 1976, A32, 751.
(33) Chilton, N. F.; Anderson, R. P.; Turner, L. D.; Soncini, A.;
Murray, K. S. J. Comput. Chem. 2013, 34, 1164.
(34) McGarvey, B. R.; Telser, J. Inorg. Chem. 2012, 51, 6000.
(35) Gomez-Coca, S.; Cremades, E.; Aliaga-Alcalde, N.; Ruiz, E. J.
Am. Chem. Soc. 2013, 135, 7010.
(1) Zadrozny, J. M.; Xiao, D. J.; Atanasov, M.; Long, G. J.; Grandjean,
F.; Neese, F.; Long, J. R. Nat. Chem. 2013, 577.
(2) Andres, H.; Bominaar, E. L.; Smith, J. M.; Eckert, N. A.; Holland,
P. L.; Munck, E. J. Am. Chem. Soc. 2002, 124, 3012.
̈
(3) Reiff, W. M.; LaPointe, A. M.; Witten, E. H. J. Am. Chem. Soc.
2004, 126, 10206.
(36) Glauber, R. J. J. Math. Phys. 1963, 4, 294.
(4) Reiff, W. M.; Schulz, C. E.; Whangbo, M.-H.; Seo, J. I.; Lee, Y. S.;
Potratz, G. R.; Spicer, C. W.; Girolami, G. S. J. Am. Chem. Soc. 2009,
131, 404.
(5) Merill, W. A.; Stich, T. A.; Brynda, M.; Yeagle, G. J.; Fettinger, J.
C.; de Hont, R.; Reiff, W. M.; Schulz, C. E.; Britt, R. D.; Power, P. P. J.
Am. Chem. Soc. 2009, 131, 12693.
(6) Zadrozny, J. M.; Atanasov, M.; Bryan, A. M.; Lin, C.-Y.; Rekken,
B. D.; Power, P. P.; Neese, F.; Long, J. R. Chem. Sci 2013, 4, 125.
(7) Atanasov, M.; Zadrozny; Long, J. R.; Neese, F. Chem. Sci 2013, 4,
139.
(8) Ishikawa, N.; Sugita, M.; Ishikawa, T.; Koshihara, S.; Kaizu, Y. J.
Am. Chem. Soc. 2003, 125, 8694.
(9) Woodruff, D. N.; Winpenny, R. E. P.; Layfield, R. A. Chem. Rev.
2013, 113, 5110.
(10) Freedman, D. E.; Harman, W. H.; Harris, T. D.; Long, G. J.;
Chang, C. J.; Long, J. R. J. Am. Chem. Soc. 2010, 132, 1224.
(11) Harman, W. H.; Harris, T. D.; Freedman, D. E.; Fong, H.;
Chang, A.; Rinehart, J. D.; Ozarowski, A.; Sougrati, M. T.; Grandjean,
F.; Long, G. J.; Long, J. R.; Chang, C. J. J. Am. Chem. Soc. 2010, 132,
18115.
̌
(37) Boca, R. Coord. Chem. Rev. 2004, 248, 757.
(38) Harriman, J. E. Theoretical Foundations of Electron Spin
Resonance; Academic Press: New York, 1978.
(39) Boca, R.; Dlhan, L.; Linert, W.; Ehrenberg, H.; Fuess, H.; Haase,
̌ ́ ̌
W. Chem. Phys. Lett. 1999, 307, 359.
(40) Jankovics, H.; Daskalakis, M.; Raptopoulou, C. P.; Terzis, A.;
Tangoulis, V.; Giapintzakis, J.; Kiss, T.; Saligoglou, A. Inorg. Chem.
2002, 41, 3366.
(41) Atanasov, M.; Ganyushin, D.; Pantazis, D. A.; Sivalingam, K.;
Neese, F. Inorg. Chem. 2011, 50, 7460.
(42) Palii, A. V.; Clemente-Juan, J. M.; Coronado, E.; Klokishner, S.;
Ostrovsky, S.; Reu, O. S. Inorg. Chem. 2010, 49, 8073.
(43) Pointillart, F.; Bernot, K.; Sessoli, R.; Gatteschi, D. Chem.Eur.
J. 2007, 13, 1602.
(44) Ferbinteanu, M.; Kajiwara, T.; Choi, K.-Y.; Nojiri, H.;
Nakamoto, A.; Kojima, N.; Cimpoesu, F.; Fujimura, Y.; Takaishi, S.;
Yamashita, M. J. Am. Chem. Soc. 2006, 128, 9008.
(45) Gatteschi, D.; Sessoli, R.; Villain, J. Molecular Nanomagnets;
Oxford University Press: Oxford, 2006.
(46) Koizumi, S.; Nihei, M.; Shiga, T.; Nakano, M.; Nojiri, H.;
Bircher, R.; Waldmann, O.; Ochsenbein, S. T.; Gudel, H. U.;
(12) Weismann, D.; Sun, Y.; Lan, Y.; Wolmershauser, G.; Powell, A.
̈
̈
K.; Sitzmann, H. Chem.Eur. J. 2011, 17, 4700.
Fernandez-Alonso, F.; Oshio, H. Chem.Eur. J. 2007, 13, 8445.
(13) Jurca, T.; Farghal, A.; Lin, P.-H.; Korobkov, I.; Murugesu, M.;
Richeson, D. S. J. Am. Chem. Soc. 2011, 133, 15814.
(47) Bendix, J.; Brorson, M.; Schaffer, C. E. Inorg. Chem. 1993, 32,
̈
2838.
(14) Zadrozny, J. M.; Long, J. R. J. Am. Chem. Soc. 2011, 138, 20732.
(15) Zadrozny, J. M.; Liu, J.; Piro, N. A.; Chang, C. J.; Hill, S.; Long,
J. R. Chem. Commun. 2012, 48, 3897.
(16) Lin, P.-H.; Smythe, N. C.; Gorelsky, S. I.; Maguire, S.; Henson,
N. J.; Korobkov, I.; Scott, B. L.; Gordon, J. C.; Baker, R. T.; Murugesu,
M. J. Am. Chem. Soc. 2011, 135, 15806.
(48) Layfield, R. A.; McDouall, J. J. W.; Scheer, M.; Schwarzmaier,
C.; Tuna, F. Chem. Commun. 2011, 47, 10623.
(49) Reiff, W. M.; Witten, E. H. Polyhedron 1984, 3, 443.
(50) Kramers, H. A. Proc. R. Acad. Sci. Amsterdam 1930, 33, 959.
(51) Cremades, E.; Ruiz, E. Inorg. Chem. 2011, 50, 4016.
1973
dx.doi.org/10.1021/ic401677j | Inorg. Chem. 2014, 53, 1962−1974

Inorganic Chemistry
Article
(52) Companion, A. L.; Komorynskyl, M. A. J. Chem. Educ. 1964, 41,
257.
(53) Ribas Gispert, J. Coordination Chemistry; Wiley-VCH Verlag
GmbH & Co. KGaA: Weinheim, 2008, p 265ff.
(54) Neese, F. J. Chem. Phys. 2007, 127, 164112.
(55) Neese, F. Wiley Interdiscip. Rev.: Comput. Mol. Sci. 2012, 2, 73.
(56) Maganas, D.; Sottini, S.; Kyritsis, P.; Groenen, E. J. J.; Neese, F.
Inorg. Chem. 2011, 50, 8741.
(57) Neese, F. J. Am. Chem. Soc. 2006, 128, 10213−10222.
(58) Zein, S.; Duboc, C.; Lubitz, W.; Neese, F. Inorg. Chem. 2007, 47,
134.
(59) Alyea, E. C.; Bradley, D. C.; Copperthwaite, R. G.; Sales, K. D. J.
Chem. Soc., Dalton Trans. 1973, 185.
(60) Ellison, J.; Power, P. P.; Shoner, S. C. J. Am. Chem. Soc. 1989,
111, 8044.
(61) Lever, A. B. P. Inorganic Electronic Spectroscopy, 2nd ed., third
impression; Elsevier: Amsterdam, 1984, p 448.
(62) MacDonnell, F. M.; Ruhland-Senge, K.; Ellison, J. J.; Holm, R.
H.; Power, P. P. Inorg. Chem. 1995, 34, 1815.
(63) Evans, D. J.; Hughes, D. L.; Silver, J. Inorg. Chem. 1997, 36, 747.
(64) Day, B. M.; Pugh, T.; Hendriks, D.; Guerra, C. F.; Evans, D. J.
F.; Bickelhaupt, M.; Layfield, R. A. J. Am. Chem. Soc. 2013, 135, 13338.
(65) X-RED32 1.01, Data Reduction Program; Stoe & Cie GmbH:
Darmstadt, Germany, 2001.
(66) Sheldrick, G. M. SHELXTL PC version 5.1 An Integrated System
for Solving, Refining, and Displaying Crystal Structures from Diffraction
Data; Bruker Analytical X-ray Systems: Karlsruhe, 2000.
(67) Diamond Version 2.1d, K.; Crystal Impact GbR: Brandenburg,
Germany, 1996−2000.
(68) STOE, WinXPOW; STOE & Cie GmbH: Darmstadt, Germany,
2000.
(69) Kahn, O. Molecular Magnetism; Wiley-VCH: Weinheim,
Germany, 1993.
(70) Lueken, H. Magnetochemie; B. G. Teubner: Stuttgart, Leibzig,
1999 p 426.
(71) Haberditzl, W. Angew. Chem. 1966, 78, 277−288; Angew. Chem.,
Int. Ed. Engl. 1966, 5, 288−298.
(72) TURBOMOLE V6.4; TURBOMOLE GmbH: Karlsruhe, 2012,
http://www.turbomole.de. TURBOMOLE is a development of
University of Karlsruhe and Forschungszentrum Karlsruhe 1989−
2007, TURBOMOLE GmbH since 2007.
(73) Treutler, O.; Ahlrichs, R. J. Chem. Phys. 1995, 102, 346.
(74) Eichkorn, K.; Treutler, O.; Oehm, H.; Haser, M.; Ahlrichs, R.
̈
Chem. Phys. Lett. 1995, 240, 283.
(75) Bauernschmitt, R.; Ahlrichs, R. Chem. Phys. Lett. 1996, 256, 454.
(76) Becke, A. D. J. Chem. Phys. 1993, 98, 5648.
(77) Perdew, J. P. Phys. Rev. B 1986, 33, 8822.
(78) Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. B 1988, 37, 785.
(79) Weigend, F.; Ahlrichs, R. Phys. Chem. Chem. Phys. 2005, 7, 3297.
(80) Weigend, F. Phys. Chem. Chem. Phys. 2006, 8, 1057.
1974
dx.doi.org/10.1021/ic401677j | Inorg. Chem. 2014, 53, 1962−1974