Planar three-coordinate high-spin FeII complexes with large orbital angular momentum: Moessbauer, electron paramagnetic resonance, and electronic structure studies

Article abstract of DOI:10.1021/ja012327l

Moessbauer spectra of [LFe^IIX]⁰ (L = β-diketiminate; X = Cl^-, CH₃^-, NHTol^-, NHtBu^-), 1.X, were recorded between 4.2 and 200 K in applied magnetic fields up to 8.0 T. A spin Hamiltonian analysis of these data revealed a spin S = 2 system with uniaxial magnetization properties, arising from a quasidegenerate M_s = ±2 doublet that is separated from the next magnetic sublevels by very large zero-field splittings (3|D| > 150 cm^-1). The ground levels give rise to positive magnetic hyperfine fields of unprecedented magnitudes, B_int = +82, +78, +72, and +62 T for 1.CH₃, 1.NHTol, 1.NHtBu, and 1.Cl, respectively. Parallel-mode EPR measurements at X-band gave effective g values that are considerably larger than the spin-only value 8, namely g_eff = 10.9 (1.Cl) and 11.4 (1.CH₃), suggesting the presence of unquenched orbital angular momenta. A qualitative crystal field analysis of g_eff shows that these momenta originate from spin-orbit coupling between energetically closely spaced yz and z² 3d-orbital states at iron and that the spin of the M_s = ±2 doublet is quantized along x, where x is along the Fe-X vector and z is normal to the molecular plane. A quantitative analysis of g_eff provides the magnitude of the crystal field splitting of the lowest two orbitals, |ε_yz - ε_z² = 452 (1.Cl) and 135 cm^-1 (1.CH₃). A determination of the sign of the crystal field splitting was attempted by analyzing the electric field gradient (EFG) at the ⁵⁷Fe nuclei, taking into account explicitly the influence of spin-orbit coupling on the valence term and ligand contributions. This analysis, however, led to ambiguous results for the sign of ε_yz - ε_z². The ambiguity was resolved by analyzing the splitting Δ of the Ms = ±2 doublet; Δ = 0.3 cm^-1 for 1.Cl and Δ = 0.03 cm^-1 for 1.CH₃. This approach showed that z² is the ground state in both complexes and that ε_xz - ε_z² ≈ 3500 cm^-1 for 1.Cl and 6000 cm^-1 for 1.CH₃. The crystal field states and energies were compared with the results obtained from time-dependent density functional theory (TD-DFT). The isomer shifts and electric field gradients in 1.X exhibit a remarkably strong dependence on ligand X. The ligand contributions to the EFG, denoted W, were expressed by assigning ligand-specific parameters: W_X to ligands X and W_N to the diketiminate nitrogens. The additivity and transferability hypotheses underlying this model were confirmed by DFT calculations. The analysis of the EFG data for 1.X yields the ordering W_{N(diketiminate)} < W_Cl < W_N′HR, W_CH3 and indicates that the diketiminate nitrogens perturb the iron wave function to a considerably lesser extent than the monodentate nitrogen donors do. Finally, our study of these synthetic model complexes suggests an explanation for the unusual values for the electric hyperfine parameters of the iron sites in the Fe-Mo cofactor of nitrogenase in the M^N state.

Full text of DOI:10.1021/ja012327l

Published on Web 02/27/2002
Planar Three-Coordinate High-Spin FeII Complexes with Large
Orbital Angular Momentum: Mo1ssbauer, Electron
Paramagnetic Resonance, and Electronic Structure Studies
Hanspeter Andres,† Emile L. Bominaar,† Jeremy M. Smith,‡ Nathan A. Eckert,‡
Patrick L. Holland,*,‡ and Eckard M¨unck*,†
Contribution from the Departments of Chemistry, Carnegie Mellon UniVersity,
4400 Fifth AVenue, Pittsburgh, PennsylVania 15213, and UniVersity of Rochester,
Rochester, New York 14627
Received October 10, 2001
Abstract: Mo¨ssbauer spectra of [LFeIIX]0 (L ) â-diketiminate; X ) Cl-, CH3-, NHTol-, NHtBu-), 1.X, were
recorded between 4.2 and 200 K in applied magnetic fields up to 8.0 T. A spin Hamiltonian analysis of
these data revealed a spin S ) 2 system with uniaxial magnetization properties, arising from a quasi-
degenerate MS ) (2 doublet that is separated from the next magnetic sublevels by very large zero-field
splittings (3|D| > 150 cm-1). The ground levels give rise to positive magnetic hyperfine fields of
unprecedented magnitudes, Bint ) +82, +78, +72, and +62 T for 1.CH3, 1.NHTol, 1.NHtBu, and 1.Cl,
respectively. Parallel-mode EPR measurements at X-band gave effective g values that are considerably
larger than the spin-only value 8, namely geff ) 10.9 (1.Cl) and 11.4 (1.CH3), suggesting the presence of
unquenched orbital angular momenta. A qualitative crystal field analysis of geff shows that these momenta
originate from spin-orbit coupling between energetically closely spaced yz and z2 3d-orbital states at iron
and that the spin of the MS ) (2 doublet is quantized along x, where x is along the Fe-X vector and z is
normal to the molecular plane. A quantitative analysis of geff provides the magnitude of the crystal field
-1
2
splitting of the lowest two orbitals, |ꢀyz - ꢀz | ) 452 (1.Cl) and 135 cm (1.CH3). A determination of the
sign of the crystal field splitting was attempted by analyzing the electric field gradient (EFG) at the 57Fe
nuclei, taking into account explicitly the influence of spin-orbit coupling on the valence term and ligand
2
contributions. This analysis, however, led to ambiguous results for the sign of ꢀyz - ꢀz . The ambiguity was
resolved by analyzing the splitting ∆ of the MS ) (2 doublet; ∆ ) 0.3 cm-1 for 1.Cl and ∆ ) 0.03 cm-1 for
-1
1.CH3. This approach showed that z2 is the ground state in both complexes and that ꢀxz - ꢀz ≈ 3500 cm
2
for 1.Cl and 6000 cm-1 for 1.CH3. The crystal field states and energies were compared with the results
obtained from time-dependent density functional theory (TD-DFT). The isomer shifts and electric field
gradients in 1.X exhibit a remarkably strong dependence on ligand X. The ligand contributions to the EFG,
denoted W, were expressed by assigning ligand-specific parameters: WX to ligands X and WN to the
diketiminate nitrogens. The additivity and transferability hypotheses underlying this model were confirmed
by DFT calculations. The analysis of the EFG data for 1.X yields the ordering WN(diketiminate) < WCl < WN′HR
,
WCH and indicates that the diketiminate nitrogens perturb the iron wave function to a considerably lesser
3
extent than the monodentate nitrogen donors do. Finally, our study of these synthetic model complexes
suggests an explanation for the unusual values for the electric hyperfine parameters of the iron sites in the
Fe-Mo cofactor of nitrogenase in the MN state.
Introduction
nitrogenase. The N2 reduction site of nitrogenase, the iron-
molybdenum cofactor (FeMoco), has been investigated with
Research aimed at understanding the chemical and biochemi-
cal processes that make up the global nitrogen cycle is
intensively pursued in many laboratories. The reduction of
dinitrogen to ammonia especially has been scrutinized for many
decades because it is often the limiting step in nitrogen
assimilation by plants and animals.1 Biological N2 reduction is
brought about by microbes that use the enzyme system
many chemical, biochemical, and spectroscopic techniques, and
the X-ray structure of the cofactor has been reported for the
proteins from three organisms.2 Particularly surprising was the
discovery that six of the seven iron sites that constitute the
FeMoco have a trigonal sulfido coordination. Three-coordinate
(2) (a) Chan, M. K.; Kim, J.; Rees, D. C. Science 1993, 260, 792-794. (b)
Kim, J.; Rees, D. C. Science 1992, 257, 1677-1682. (c) Kim, J.; Rees, D.
C. Nature 1992, 360, 553-560. (d) Peters, J. W.; Stowell, M. H. B.; Soltis,
S. M.; Finnegan, M. G.; Johnson, M. K.; Rees, D. C. Biochemistry 1997,
36, 1181-1187. (e) Kim, J.; Woo, D.; Rees, D. C. Biochemistry 1993, 32,
7104-7115. (f) Mayer, S. M.; Lawson, D. M.; Gormal, C. A.; Roe, S. M.;
Smith, B. E. J. Mol. Biol. 1999, 292, 871-891.
* To whom correspondence should be addressed. E-mail: (P.H.)
holland@chem.rochester.edu; (E.M.) em40@andrew.cmu.edu.
† Carnegie Mellon University.
‡ University of Rochester.
(1) Postgate, J. Nitrogen Fixation, 2nd ed.; Edward Arnold: Baltimore, 1987.
9
3012 VOL. 124, NO. 12, 2002 J. AM. CHEM. SOC.
10.1021/ja012327l CCC: $22.00 © 2002 American Chemical Society

II
Planar Three-Coordinate High-Spin Fe Complexes
A R T I C L E S
understand zero-field splittings, g-tensors, isomer shifts, mag-
netic hyperfine interactions, and electric field gradients (EFGs)
of the individual iron atoms. Thus, we have so far focused on
understanding 1.X, the monomeric complexes of similar iron-
ligand structure.
In this article, we describe detailed Mo¨ssbauer and EPR
studies of 1.Cl and the new compound 1.CH3. Our studies
revealed magnetic hyperfine fields at the 57Fe nuclei of
considerable magnitude. Thus, 1.CH3 exhibits an internal field
of +82 T, which is 20 T larger than the “extraordinarily large”
+62.4 T hyperfine field reported for R-iron(II) octaethyltetra-
azaporphyrin.7 Analysis of our data in the framework of a crystal
field model revealed that the unusually large hyperfine field of
1.CH3 results from mixing of closely spaced z2 and yz d-orbitals
by spin-orbit coupling, creating an electronic ground state with
nearly the maximum attainable orbital angular momentum. Our
analysis also revealed a substantial ligand contribution to the
electric field gradient tensor, and moreover, this contribution
can be partitioned and assigned to the diketiminate nitrogens
and ligand X. The large ligand contribution found here seems
to be a feature of planar structures, providing insight into the
nature of the rather small quadrupole splittings of the trigonal
irons of the nitrogenase cofactor.8 Finally, we report the results
of DFT and TD-DFT quantum chemical calculations which
substantiate the major features of the electronic level stucture
deduced from the crystal field analysis. However, as the
quantum chemical calculations reported in this paper do not
incorporate spin-orbit coupling, the unusual magnetic properties
of the ground states in 1.X could only be indirectly assessed
by these techniques.
Figure 1. View of the structure of 1.X along the axis perpendicular to the
FeNN plane. The iron and nitrogen atoms are shown as black and grey
circles, respectively. The hydrogen atoms are omitted for clarity.
iron is rare in synthetic chemistry,3 so efforts to expand the
database of such compounds are expected to be useful in refining
our understanding of the spectroscopic and biochemical signa-
tures of the FeMoco.
We recently reported a new three-coordinate iron(II) com-
pound that is supported by a bulky â-diketiminate ligand.4 The
terminal chloride ligand can be substituted with other ligands,
giving compounds 1.X (Figure 1) that differ only in the nature
of X. Although such compounds are highly sensitive to air and
water, their thermal stability enables detailed spectroscopic
study, and below are described the first steps toward the
elucidation of ligand contributions to the electronic structure
of three-coordinate Fe(II).
In addition, we have reported that 1.Cl reacts with N2 and
reductant to give interesting dimers, [LFeNNFeL]n- (n ) 0,
2), that contain a bridging N2 with a substantially weakened
N-N bond.5 With the combination of formal N-N bond
reduction and low-coordinate iron, these complexes promise
substantial progress in understanding the role of low coordina-
tion at iron in N2 activation. Moreover, the diketiminate
nitrogens have a delocalized negative charge, are electron rich,
soft, and polarizable,6 and thus have properties associated with
the biological thiolate/sulfide ligands. These similarities suggest
that results obtained from the diketiminate model complexes
will provide insight into iron-sulfur clusters such as FeMoco.
Preliminary Mo¨ssbauer and EPR studies of [LFeNNFeL]n- (n
) 0, 2) indicate that these complexes have novel electronic
properties. However, to assess the meaning of the spectral
features of these exchange-coupled dimers, it is necessary to
2. Materials and Methods
2.1. Synthesis and Characterization. All compounds were syn-
thesized and handled in an MBraun Labmaster glovebox under an
atmosphere of N2 with <1 ppm of O2 and H2O. Solvents were
deoxygenated and purified using columns of alumina and deoxygenizer
(Glass Contour Co.; pentane, diethyl ether, tetrahydrofuran), distillation
from Na/benzophenone (hexamethyldisiloxane, C6D6), or activated 4
Å molecular sieves (light mineral oil). Celite was dried under vacuum
overnight at 250 °C. Lithium amides were prepared by treating a
pentane solution of the corresponding amine with butyllithium and
collecting the solid precipitate. The 1H NMR spectra of the new
compounds have not been fully assigned; they consisted solely of broad
singlets. 1H NMR spectroscopic measurements of magnetic susceptibil-
ity used the method of Evans9 and were corrected for diamagnetism
using standard values.10 Elemental analyses were performed by Desert
Analytics (Tucson, AZ).
LFeCH3 (1.CH3). Methylmagnesium chloride (3.0 M in tetrahy-
drofuran, 0.04 mL, 0.1 mmol) was added to a solution of 1.Cl (46 mg,
78 µmol) in diethyl ether (10 mL). The solution turned from red to
pale orange. Volatile materials were removed under vacuum, and the
residue was extracted with pentane (5 mL) and filtered through Celite.
Addition of hexamethyldisiloxane (2 mL) and cooling to -35 °C gave
(3) (a) Bradley, D. C.; Hursthouse, M. B.; Rodesiler, P. F. Chem. Commun.
1969, 14-15. (b) Alyea, E. C.; Bradley, D. C.; Copperthwaite, R. G.; Sales,
K. D.; Fitzsimmons, B. W.; Johnson, C. E. J. Chem. Soc. D 1970, 1715-
1716. (c) Seidel, W.; Lattermann, K. J. Z. Anorg. Allg. Chem. 1982, 488,
69-74. (d) Roddick, D. M.; Tilley, T. D.; Rheingold, A. L.; Geib, S. J. J.
Am. Chem. Soc. 1987, 109, 945-946. (e) Power, P. P.; Shoner, S. C. Angew.
Chem. 1991, 103, 308-309. (f) Olmstead, M. M.; Power, P. P.; Shoner, S.
C. Inorg. Chem. 1991, 30, 2547-2551. (g) Bartlett, R. A.; Ellison, J. J.;
Power, P. P.; Shoner, S. C. Inorg. Chem. 1991, 30, 2888-2894. (h)
MacDonnell, F. M.; Ruhlandt-Senge, K.; Ellison, J. J.; Holm, R. H.; Power,
P. P. Inorg. Chem. 1995, 34, 1815-1822. (i) Stokes, S. L.; Davis, W. M.;
Odom, A. L.; Cummins, C. C. Organometallics 1996, 15, 4521-4530. (j)
Putzer, M. A.; Neumueller, B.; Behnicke, K.; Magull, J. Chem. Ber. 1996,
129, 715-719. (k) O’Donoghue, M. B.; Zanetti, N. C.; Davis, W. M.;
Schrock, R. R. J. Am. Chem. Soc. 1997, 119, 2753-2754. (l) O’Donoghue,
M. B.; Davis, W. M.; Schrock, R. R.; Reiff, W. M. Inorg. Chem. 1999,
38, 243-252. (m) Siemeling, U.; Vorfeld, U.; Neumann, B.; Stammler,
H.-G. Inorg. Chem. 2000, 39, 5159-5160.
(7) Reiff, W. M.; Frommen, C. M.; Yee, G. T.; Sellers, S. P. Inorg. Chem.
2000, 39, 2076-2079.
(8) (a) Zimmermann, R.; Mu¨nck, E.; Brill, W. J.; Shah, V. K.; Henzl, M. T.;
Rawlings, J.; Orme-Johnson, W. H. Biochim. Biophys. Acta 1978, 537,
185-207. (b) Huynh, B. H.; Mu¨nck, E.; Orme-Johnson, W. H. Biochim.
Biophys. Acta 1979, 527, 192-203. (c) Huynh, B. H.; Henzl, M. T.;
Christner, J. A.; Zimmermann, R.; Orme-Johnson, W. H.; Mu¨nck, E.
Biochim. Biophys. Acta 1980, 623, 124-138. (d) Yoo, S. J.; Angove, H.
C.; Papaefthymiou, V.; Burgess, B. K.; Mu¨nck, E. J. Am. Chem. Soc. 2000,
122, 4926-4936.
(4) Smith, J. M.; Lachicotte, R. J.; Holland, P. L. Chem. Commun. 2001, 1542-
1543.
(5) Smith, J. M.; Lachicotte, R. J.; Pittard, K. A.; Cundari, T. R.; Lukat-Rodgers,
G.; Rodgers, K. R.; Holland, P. L. J. Am. Chem. Soc. 2001, 123, 9222-
9223.
(6) Randall, D. W.; DeBeer George, S.; Holland, P. L.; Hedman, B.; Hodgson,
K. O.; Tolman, W. B.; Solomon, E. I. J. Am. Chem. Soc. 2000, 122, 11632-
11648.
(9) (a) Evans, D. F. J. Chem. Soc. 1959, 2003. (b) Schubert, E. M. J. Chem.
Educ. 1992, 69, 62.
(10) Kahn, O. Molecular Magnetism; Wiley: New York, 1993.
9
J. AM. CHEM. SOC. VOL. 124, NO. 12, 2002 3013

A R T I C L E S
Andres et al.
orange crystals of 1.CH3. An additional crop of crystals was collected
for a total yield of 36 mg (80% yield). 1H NMR (C6D6, 400 MHz): δ
132(1H), 44(18H), -4(4H), -28(12H), -111(2H), -122(4H), -134-
(12H). Evans µeff (C6D6, 298 K) ) 5.5 ( 0.3 â. FTIR (mineral oil):
3055(w), 1621(w), 1505(s), 1360(s), 1321(s), 1253(m), 1209(s), 1149-
(m), 1133(s), 1096(s), 1055(m), 1030(m), 933(m), 889(m), 824(m), 775-
(s), 751(s), 669(w) cm-1. UV/vis (Et2O): 380, 430(sh), 520 nm. Anal.
Calcd for C36H56N2Fe: C, 75.50; H, 9.86; N, 4.89. Found: C, 75.42;
H, 9.96; N, 4.85.
LFeNHR (R ) p-tolyl, 1.NHTol; R ) tert-butyl, 1.NHtBu). A
solution of LiNHR (1.8 mmol) in diethyl ether (5 mL) was added to a
suspension of 1.Cl (1.7 mmol) in diethyl ether (15 mL). The reaction
mixture rapidly became dark red-brown. The reaction was stirred for 2
h, filtered through Celite, concentrated, and cooled to -35 °C to give
red crystals. The connectivity of the compounds was verified via X-ray
crystallography.
1
Data for 1.NHTol. Yield: 68%. H NMR (C6D6, 400 MHz): δ
Figure 2. 4.2 K Mo¨ssbauer spectra of polycrystalline samples of 1.CH3
(A) and 1.Cl (B) recorded at 4.2 K in zero field. The solid lines are the
result of least-squares fitting doublets to the data.
122, 96, 94, 37.9, 4.0, 0.1, -25.7, -100, -113, -114. Evans µeff (C6D6,
298 K) ) 5.3 ( 0.5 â. FTIR (mineral oil): 3313 cm-1(νN-H). UV/vis
(Et2O): 436(sh), 480(sh) nm.
1
controller. For the analysis of the spectra, we have used software written
by Dr. M. P. Hendrich at Carnegie Mellon University.
Data for 1.NHtBu. Yield: 62%. H NMR (C6D6, 400 MHz): δ
119, 43, 38, -1.7, -28.3, -29.1, -98.5, -103.8, -122.8. Evans µeff
(C6D6, 298 K) ) 4.8 ( 0.3 â. FTIR (mineral oil): 3282 cm-1 (νN-H).
UV/vis (Et2O): 512, 561 nm.
3. Results
3.1. Synthesis and Structure. The synthesis and structure
of 1.Cl were reported recently.4 It was possible to replace the
chloride ligand with other monoanionic ligands through simple
metathesis reactions, giving the relatives 1.CH3, 1.NH(p-tolyl),
and 1.NH(tert-butyl). Proton NMR spectroscopy and magnetic
measurements are consistent with a high-spin iron(II) config-
uration at room temperature. Each of these complexes has been
analyzed by X-ray crystallography, in each case revealing a
planar three-coordinate geometry of the type shown in Figure
1. The structures of the amidoiron complexes will be described
fully at a later time.
Complex 1.CH3 crystallized in the same space group as 1.Cl,4
and the two structures are isomorphous, having a crystal-
lographic C2 axis coincident with the Fe-X bond. This
symmetry gives a rigorously planar geometry at the iron atom
in each case. The Fe-N distances are slightly longer in 1.CH3
(1.973(1) Å vs 1.948(2) Å), decreasing the bite angle of the
â-diketiminate ligand (94.85(8)° vs 96.35(1)°) relative to 1.Cl.
The Fe-C bond is notable because 1.CH3 is the first crystal-
lographically characterized three-coordinate iron-alkyl com-
plex.13 This bond is extremely short (2.009(3) Å), as one would
expect from the low coordination number. The strong Fe-C
interaction foreshadows the especially large effect of this ligand
on the EFG of 1.CH3 shown below.
3.2. Mo1ssbauer Studies. Figure 2A shows a 4.2 K Mo¨ssbauer
spectrum of a polycrystalline sample of 1.CH3 recorded in the
absence of an applied magnetic field. The spectrum consists of
a quadrupole doublet with ∆EQ ) 1.74 mm/s and δ ) 0.48
mm/s. The quadrupole splitting was found to be temperature-
dependent (at 200 K we observed ∆EQ ) 1.57 mm/s),
suggesting the presence of a low-lying excited state with orbital
composition different from that of the ground state. The spectra
in Figures 3 and 4 were obtained in magnetic fields B up to 8.0
T applied parallel to the observed γ-radiation. It can be seen
that the applied field induces magnetic hyperfine interactions
of considerable magnitude. To our knowledge, the internal field,
X-ray Structural Determination of 1.CH3. An orange crystal (0.4
× 0.3 × 0.3 mm) was mounted under Paratone-8277 on a glass fiber
and placed in a cold nitrogen stream at -80 °C on the X-ray
diffractometer. X-ray intensity data were collected on a standard
Siemens SMART CCD area detector system equipped with a normal
focus Mo-target X-ray tube operated at 2.0 kW (50 kV, 40 mA). A
total of 1321 frames of data (1.3 hemispheres) were collected using a
narrow frame method with scan widths of 0.3° in ω and exposure times
of 10 s/frame using a detector-to-crystal distance of 5.09 cm (maximum
2θ angle of 56.6°). Frames were integrated with the Siemens SAINT
program to yield a total of 7348 reflections, of which 2501 were
independent (Rint ) 1.66%). Laue symmetry revealed a monoclinic
crystal system, and the final unit cell parameters were determined from
the least-squares refinement of three-dimensional centroids of 4546
reflections. The data were corrected for absorption with the SADABS11
program.
Systematic absences allowed the space group to be assigned as C2/
c, using the XPREP program (Siemens, SHELXTL,12 version 5.04).
The structure was solved using direct methods and refined employing
full-matrix least-squares on F2 (SHELXTL). All non-hydrogen atoms
were refined with anisotropic thermal parameters. Hydrogen atoms were
included in idealized positions with riding thermal parameters. The
structure refined to a goodness of fit (GOF) of 1.034 and final residuals
of R1 ) 2.82% and wR2 ) 7.31% (I > 2σ(I)).
2.2. Sample Preparation. Mo¨ssbauer samples (polycrystalline
samples suspended in mineral oil) were transported and analyzed in
Delrin cups with tight-fitting lids to avoid exposure of the samples to
oxygen. When such lids were not used, sample decomposition was
evident, even at low temperatures. Quartz EPR tubes containing
polycrystalline samples suspended in mineral oil were flame-sealed
under <1 atm N2.
2.3. Spectroscopic Instrumentation. Mo¨ssbauer spectra were
collected on constant acceleration spectrometers, which allowed record-
ing spectra between 1.5 and 200 K in fields up to 8.0 T. Isomer shifts
are quoted relative to Fe metal at room temperature. The spectra were
analyzed using the WMOSS software (WEB Research Co., Edina, MN).
EPR spectra were recorded on a BrukerESP 300 equipped with an ESR
910 helium continuous flow cryostat and an Oxford temperature
(11) Blessing, R. H. Acta Crystallogr. A 1995, 51, 33.
(12) SHELXTL: Structure Analysis Program, Version 5.04; Siemens Industrial
Automation Inc.: Madison, WI, 1995.
(13) Cambridge Structural Database (October, 2000). Allen, R. H.; Kennard,
O. Chem. Des. Autom. News 1993, 8, 31-37.
9
3014 J. AM. CHEM. SOC. VOL. 124, NO. 12, 2002

II
Planar Three-Coordinate High-Spin Fe Complexes
A R T I C L E S
Figure 5. Energy level diagram for an S ) 2 system for a large negative
D as a function of the rhombicity E/D. The diagram is not strictly pertinent
to the systems discussed but serves well for illustrating the principal features
of the Mo¨ssbauer and EPR spectra.
Figure 3. Mo¨ssbauer spectra of polycrystalline 1.CH3 recorded at 4.2 K
in parallel applied fields of 8.0 T (A), 3.0 T (B), and 0.5 T (C). The magnetic
splitting of the 0.5 T spectrum reflects an internal field of +82 T. The
solid lines are spectral simulations based on eq 1 using the parameters listed
in Table 1. The ∆m ) 0 transitions (lines 2 and 5) of the experimental data
have slightly larger intensities than anticipated for a sample containing
randomly oriented molecules, suggesting texture. See text for the doublet
in (C) marked by the bracket. The simulations yield too small amplitudes
for the middle two lines of the six-line magnetic spectrum, especially in
(C). This mismatch is caused by the fact that Bint is somewhat distributed
yielding a broadening that is larger for the outer lines; the simulation
program uses a velocity-independent linewidth, which is too large for the
inner two lines.
For 1.CH3, a quadrupole doublet is observed in zero field,
and magnetic hyperfine interactions arise only after an external
magnetic field is applied. Such spectral behavior is characteristic
of an integer spin paramagnet.14 As can be seen from inspection
of Figure 3, a fraction of molecules already exhibits fully
developed magnetic splittings in applied fields of moderate
strength, indicating an electronic system with a quasi-degenerate
ground doublet having a small energy splitting ∆.15 The spectral
simulations described below suggest ∆ ≈ 0.03 cm-1. The
observation of sizable magnetic splittings at low applied field
also demonstrates that the intradoublet relaxation rate is slow
at 4.2 K.
High-spin (S ) 2) ferrous compounds are generally well
described by the spin Hamiltonian,16
E
D
2
Hˆ ) D Sˆz2 - 2 + (Sˆx2 - Sˆy ) + âSˆ‚g‚Bˆ + Sˆ‚A‚Iˆ -
{
}
gnânBˆ‚Iˆ + Hˆ Q (1a)
and
1
12
2
Hˆ Q
)
eQVz′z′[3ˆIz′2 - I(I + 1) + η(ˆIx′2 - ˆIy′ )] (1b)
Figure 4. 4.2 K Mo¨ssbauer spectra of 1.CH3 recorded in parallel applied
fields of 0.25 T (A), 0.15 T (B), and 0.05 T (C). The solid line in Figure
4A is a spectral simulation based on eq 1 using the parameters of Table 1
and assuming isotropic exchange coupling between two ferrous sites (J )
0.01 cm-1). See text for the doublet in (A) marked by the bracket.
In eq 1, all quantities have their conventional meanings. The
quadrupole interaction is expressed in the principal axes system
(x′,y′,z′) of the EFG tensor, where Vz′z′ is the largest component
and 0 e η e 1/3. For |D| . âB and D < 0, the spin Hamiltonian
has a ground doublet (see Figure 5) split by ∆ ) 3D(E/D)2.
The parameters D and E/D can be used to produce a ground
doublet with the desired ∆ value that is well isolated from the
excited sublevels of the S ) 2 manifold. The ground doublet
has a magnetization axis (z) along which the expectation values
of the electronic spin saturate to Sz ≈ (2 in fields above 0.1
T, while the components perpendicular to z, Sx and Sy , remain
near zero (see, for instance, ref 14). This behavior gives rise to
Bint ) 82.0 T, observed for 1.CH3 exceeds those reported for
other compounds by a considerable margin.7 The maximum Bint
is already observed in applied fields below 0.5 T. With
increasing applied field, the effective field at the nucleus, Beff
) Bint + B, also increases, showing that Bint is positive for the
ground level. The observation of a large and positive Bint
indicates that orbital contributions to the magnetic hyperfine
field dominate. Given the large magnitude of the orbital
contribution, it is anticipated that 1.CH3 has a very unusual
electronic structure.
In general, high-spin FeII complexes are readily identified
by their characteristic isomer shifts. However, three-coordinate
iron complexes are rare, and thus an appropriate isomer shift
database is not available. The X-ray structures of 1.CH3 and
1.Cl show that the iron sites are formally FeII.4 Because three
ligands will produce a weak ligand field, a high-spin (S ) 2)
configuration is anticipated. In the following we will develop
arguments that support such an assignment.
(14) (a) Mu¨nck, E. Methods Enzymol. 1978, 54, 346-379. (b) Mu¨nck, E.
Physical Methods in Inorganic and Bioinorganic Chemistry; University
Science Books: Sausolito, CA, 2000; Chapter 6, pp 287-319. (c) Mu¨nck,
E. The Porphyrins; Academic Press: New York, 1979; Vol. IV, Chapter
8.
(15) Surerus, K. K.; Hendrich, M. P.; Christie, P.; Rottgardt, D.; Orme-Johnson,
W. H.; Mu¨nck, E. J. Am. Chem. Soc. 1992, 114, 8579-8590.
(16) (a) Greenwood, N. N.; Gibb, T. C. Mo¨ssbauer Spectroscopy; Chapman
and Hall: London, 1971. (b) Gu¨tlich, P.; Link, R.; Trautwein, A. Mo¨ssbauer
Spectroscopy and Transition Metal Chemistry; Springer-Verlag: Berlin,
1979. (c) Goldanski, V. I., Herber, R., Eds. Chemical Applications of
Mo¨ssbauer Spectroscopy; Academic Press: New York, 1968.
9
J. AM. CHEM. SOC. VOL. 124, NO. 12, 2002 3015

A R T I C L E S
Andres et al.
Table 1. Fine Structure and Hyperfine Parameters of 1.X
δ,a
mm/s
∆EQ,b
mm/s
B ,c
T
∆,d
cm-1
int
ligand X
Cl-
ηb
geff
µeff/âf
e
0.74
0.71
-1.61 0.5
+1.42 0.6
+1.11 0.2
+1.74 0.2
+62
0.3
10.9(5)
5.4
5.3
4.8
5.5
N′HTol-
+78 <10-3
N′HtBu- 0.63
+72
+82
0.2
0.03 11.4(1)
-
CH3
0.48
a Isomer shift relative to metallic iron at 298 K. b ∆EQ and asymmetry
parameter η are quoted in the x′,y′,z′ system of eq 1b; Vz′z′, the largest
component of EFG(-), is along the x-axis of the zero-field splitting tensor
of eq 1a. See also footnote 21. c The internal magnetic field at 57Fe nucleus
is along the z-axis of the spin Hamiltonian eq 1a. For the crystal field
analysis and the DFT calculations, the coordinate system in Figure 10 was
used; in this system Bint is along the x-axis. d Zero-field splitting of MS )
(2 states. e Effective g value for MS ) (2 states obtained from EPR.
f Effective magnetic moment at room temperature expressed in Bohr
magnetons obtained from solution magnetic susceptibility measurements
(precision (0.5â).
Figure 6. Mo¨ssbauer spectra of complex 1.CH3 (A) and 1.Cl (B) recorded
at 200 K in an 8.0 T applied field. The solid lines are theoretical curves
based on eq 1 for the parameters listed in Table 1 assuming fast relaxation
of the electronic spin. To separate the quasi-degenerate ground doublet from
some complications that render a precise analysis of the spectra
quite difficult.
excited states we have used D ) -100 cm-1
.
Because the sample was suspended in Nujol to prevent
orientation of the crystallites by the applied field, the spectra
in Figure 3 were simulated with the assumption that the
molecules are randomly distributed. This assumption, however,
is not strictly justified. The spectra of Figures 3 and 4 exhibit
texture effects that give rise to slightly increased intensities for
the nuclear ∆m ) 0 transitions. This observation suggests that
the crystallites have a tendency to settle in the Nujol host such
that the magnetization axes of the molecules slightly favor a
direction perpendicular to the γ beam. The presence of texture
is also apparent in the spectrum of Figure 2A. Thus, least-
squares fitting of the spectrum with two independent Lorentzian
lines revealed that the area of the low-energy line is 2% larger
than that of the high-energy line. The larger width of the low-
energy line can be explained by noticing that the Earth’s
magnetic field (ca. 0.06 mT in Pittsburgh18) induces a hyperfine
field of 0.8 T along the electronic z-axis in the presence of a
dominant quadrupole interaction of positive sign along x.
a uniaxial internal magnetic field Bint ) - S ‚A/gnân.14 Although
the application of eq 1 to the complexes 1.X is not strictly
justified from a theoretical perspective (see section 4.1), we can
use its features to simulate the Mo¨ssbauer spectra of the ground
doublet. (In effect we are using eq 1 to emulate a fictitious S )
1/2 spin Hamiltonian with gx ) gy ) 0 and an off-diagonal
element ∆/2). The z-component of the orbital contribution to
the A-tensor for a high-spin ferrous ion with an isolated orbital
ground state is given by Az(orb) ) (gz - 2)P, where P is a
scaling factor (P ≈ 40-60 T) that depends on the radial
extension of the 3d-orbitals, and gz is the z-component of the
g-tensor.17 The saturation field for 1.CH3 which is attained when
the MS ) -2 level is exclusively populated, Bsat ) 2(gz - 2)P
≈ 82 T, suggests gz ≈ 2.8, a value well in excess of the spin-
only value, 2.
The solid lines drawn through the spectra of Figure 3 are
spectral simulations based on eq 1 for S ) 2, D ) -100 cm-1
,
and E/D ) 0.01. This choice of parameters produces an isolated
ground doublet with ∆ ) 0.03 cm-1. The 0.5 T spectrum of
Figure 3 is characteristic of a uniaxial system, i.e., of a system
for which Bint is directed along the electronic z-axis in essentially
all molecules of the polycrystalline sample. Bint has Euler angles
RE and âE relative to the principal axes system of the EFG
tensor; âE is the polar angle between z and z′, the latter being
the principal axis belonging to the largest component of the
EFG tensor. Quite frequently such situations lead to a manifold
of equivalent solutions, referred to as the ambiguity problem.8a
Our simulations, however, show that the largest component of
the EFG tensor must be either perpendicular to the magnetization
axis of the electronic system or tilted by 45° relative to this
axis, i.e., âE ) 90° or 45°. The molecular symmetry of the
complex and the 200 K spectrum of Figure 6 (see below) do
not support a solution for which âE ) 45°, so we are left with
âE ) 90°. Since the component of the EFG tensor along Bint is
negative, it follows that ∆EQ > 0. RΕ is undetermined because
the electronic ground doublet is uniaxial. We have arbitrarily
chosen RE ) 0, which directs Vz′z′ along x; this choice is
supported by the molecular geometry (see section 4.1.1). The
simulations shown in Figure 3 were obtained from eq 1 by using
the parameters listed in Table 1. In the following we discuss
Previous Mo¨ssbauer and EPR studies of integer spin para-
magnets in our laboratory have revealed that ∆ is generally
distributed about some mean value.15,19 Such distributions result
in a broadening of integer-spin EPR signals (see section 3.3).
The hyperfine field of molecules with larger ∆ values saturates
at higher fields than those of molecules with a smaller ∆. A
distribution in ∆ explains some of the spectral features in Figures
4B, such as the line broadening and the presence of a broad
absorption feature that extends over the entire velocity range.
However, the persistence of a doublet (7% of the total
absorption) in fields as strong as 0.5 T requires amending the
spin Hamiltonian with terms describing spin-dipolar and,
perhaps, exchange interactions between neighboring complexes.
The X-ray structure of 1.CH3 reveals pairs of molecules with
Fe-Fe distances of 9.5 and 13 Å.4 At these distances, neighbor-
ing molecules have dipolar coupling parameters of ca. 0.01
(18) Barton, C. E. J. Geomagn. Geoelectr. 1997, 49, 123-148.
(19) (a) Hendrich, M. P.; Mu¨nck, E.; Fox, B. G.; Lipscomb, J. D. J. Am. Chem.
Soc. 1989, 112, 5861-5865. (b) Juarez-Garcia, C.; Hendrich, M. P.;
Holman, T. R.; Que, L.; Mu¨nck, E. J. Am. Chem. Soc. 1991, 114, 518-
525. (c) Fox, B. G.; Hendrich, M. P.; Surerus, K. K.; Andersson, K. K.;
Froland, W. A.; Lipscomb, J. D.; Mu¨nck, E. J. Am. Chem. Soc. 1993, 115,
3688-3701. (d) Mu¨nck, E.; Surerus, K. K.; Hendrich, M. P. Methods
Enzymol. 1993, 227, 463-479. (e) Yoo, S. J.; Angove, H. C.; Burgess, B.
K.; Hendrich, M. P.; Mu¨nck, E. J. Am. Chem. Soc. 1999, 121, 2534-
2545.
(17) Oosterhuis, W. T.; Lang, G. Phys. ReV 1969, 178, 439-456.
9
3016 J. AM. CHEM. SOC. VOL. 124, NO. 12, 2002

II
Planar Three-Coordinate High-Spin Fe Complexes
A R T I C L E S
cm-1 20
.
Our Mo¨ssbauer simulation programs do not contain
terms describing spin-dipolar interactions, but the effects of
weak coupling are readily illustrated by assuming isotropic
exchange coupling between two ferrous sites, Hexch ) JS1‚S2.
The spectral simulation shown in Figure 4A was obtained by
using the parameters of Table 1 and J ) 0.01 cm-1. It can be
seen that the simulation produces the doublet feature (bracket)
in the center of the spectrum; this feature is absent for J ) 0.
The reader may keep in mind that a system consisting of two
interacting complexes has four closely spaced levels that are
nearly equally populated at 4.2 K; the doublet results from
insufficiently decoupled spins, essentially the up/down and
down/up combinations of an interacting pair. Texture, distribu-
tion of ∆, and dipolar couplings all affect the low-field spectra,
and for these reasons precise simulations of these spectra are
very difficult to achieve.
Figure 7. Mo¨ssbauer spectra of polycrystalline 1.Cl recorded at 4.2 K in
parallel applied fields of 8.0 T (A), 2.0 T (B), and 1.0 T (C). The magnetic
splitting of the 1.0 T spectrum reflects an internal field of +62 T. See text
for comments on the doublet in (C) marked by the bracket.
Because of the magnetic anisotropy of the electronic ground
doublet, only the z-component of the magnetic hyperfine field
can determined from the low-temperature spectra. Even for an
8.0 T applied field, the spectra do not indicate any Zeeman
mixing between the ground doublet and excited states. For most
high-spin ferrous compounds, the electronic spin relaxes rapidly
above 100 K, so the internal magnetic field is proportional to
the thermal expectation value S , Bint ) - S ‚A/gnân.14 Since
field splitting between the ground doublet and the next group
of excited states is estimated to be ca. 150 cm-1 21
.
Figures 2B and 7 show 4.2 K spectra of polycrystalline 1.Cl.
The spectra of 1.Cl have many similarities with those of 1.CH3,
with some notable exceptions. The zero-field spectrum in Figure
2B consists of a quadrupole doublet with ∆EQ ) -1.61 mm/s
and δ ) 0.74 mm/s (∆EQ ) -1.32 mm/s at 200 K). As
described for 1.CH3 the electronic ground state of 1.Cl comprises
a quasi-degenerate doublet that is well isolated from the excited
states of the S ) 2 manifold, and consequently the applied field
spectra of 1.Cl (Figure 7) have the same general features as
those of 1.CH3. Again, the internal magnetic field is positive,
but its magnitude, Bint ) 62.0 T, is smaller than the value Bint
) 82.0 T observed for 1.CH3. The best simulations to the field
th
th
the D values are generally smaller than 10 cm-1
,
S
is
th
essentially independent of D at 200 K. Under these circum-
stances, the x- and y-components of the magnetic hyperfine
tensor can be determined as well. Figure 6A shows a Mo¨ssbauer
spectrum of 1.CH3 recorded at 200 K in a parallel field of 8.0
T. This spectrum was simulated in the limit of fast fluctuations
of the electronic spin, using the Hamiltonian of eq 1 and the
parameters determined from the 4.2 K spectra. The simulations
readily established that the 200 K spectrum reflects essentially
the spectral features of the quasi-degenerate ground doublet,
with only minor contributions of higher lying states. In the
language of eq 1, the next group of electronic states has to be
higher than 150 cm-1 (|D| > 50 cm-1) to provide the correct
magnetic splittings (see section 4.1). Smaller values of D would
result in a significant population of the MS ) (1 levels resulting
in reduced Sz th values. Since Sx th ) Sy th ) 0 for the ground
doublet, our data cannot provide estimates for the x- and
y-components of the magnetic hyperfine tensor.
Interestingly, simulations with D values larger than 150 cm-1
produce Sz th values at 200 K that are slightly too large,
suggesting some population of excited states. This observation
is in agreement with the zero-field spectra which show that ∆EQ
has declined by 10% relative to its value at 4.2 K. Our
simulations of the 200 K spectrum rule out the possibility that
the EFG is tilted by âE ) 45° relative to the z-axis of the
electronic system. Finally, the recoilless fraction at 200 K has
declined by 80% relative to its value at 4.2 K, resulting in a
high noise level at this temperature, despite accumulating 107
counts per velocity channel.
dependence of the spectra were obtained for ∆ ) 0.35 cm-1
.
As for 1.CH3, the largest component of the EFG tensor was
found to be perpendicular to the direction of Bint. Significantly,
however, ∆EQ of 1.Cl is negative, suggesting that the two
complexes have different orbital ground states (see, however,
section 4.1.3). These conclusions are supported by the satisfying
spectral simulation of the 200 K spectrum shown in Figure 6B.
In contrast to the sample of 1.CH3, polycrystalline 1.Cl did
not exhibit texture effects. This is most readily confirmed by
noting that the two lines of the zero-field spectrum in Figure
2B have equal areas within the uncertainties of the fits. Note
also that the low-energy line is not broadened as observed for
1.CH3. The lack of broadening reflects the fact that 1.Cl has a
significantly larger ∆ value. From the parameters quoted in
Table 1, a Bint ) 60 mT is predicted for B(Earth) ) 0.06 mT.18
3.3. EPR Studies. Figure 8, parts A and B, shows parallel-
mode X-band EPR spectra of polycrystalline 1.CH3 and 1.Cl
recorded at 2 K. Both low-temperature spectra exhibit a broad
asymmetric EPR transition with maxima at 59.4 and 35.6 mT
for 1.CH3 and 1.Cl, respectively. These transitions are observed
up to 50 K, exhibiting a temperature dependence of the
integrated intensity according to the Curie law.22 We assign these
Although the spin Hamiltonian of eq 1 provides an adequate
framework for simulating the spectra of complex 1.CH3, we do
not wish to suggest that it properly describes the entire spin
quintet. It describes well the low-temperature spectra and
provides a large value of D, on the basis of which the zero-
(21) In eq 1 we have used the standard form of a spin Hamiltonian, and for the
chosen values of D and E/D the internal magnetic field, Bint, will be along
the electronic z-axis. The theoretical analysis presented in section 4.1 shows
that the internal field is along the Fe-X bond. For the sake of the crystal
field analysis, we adopt in section 4.1 the commonly used coordinate system
for which the z-axis is perpendicular to the plane defined by the two
diketiminate nitrogens and ligand X.
(22) Wertz, E. W.; Bolton, J. R. Electron Spin Resonance; McGraw-Hill: New
York, 1972.
(20) Carlin, R. L. Magnetochemistry; Springer-Verlag: Berlin, 1986.
9
J. AM. CHEM. SOC. VOL. 124, NO. 12, 2002 3017

A R T I C L E S
Andres et al.
the iron sites are high-spin in these complexes. This spin state
is suggested by the low coordination number for Fe and is
further corroborated by the large effective g values and effective
magnetic momenta (see Table 1).23 The lowest two sublevels
of the spin quintet are combinations of MS ) (2 states that are
closely spaced in energy (see Table 1) and separated from the
next sublevels of the same spin manifold by at least 150 cm-1
.
The spin quantization axis, denoted ê and defined by Sê|(2 )
(2|(2 , was found to be perpendicular to the principal axis of
the EFG tensor, that corresponds to the eigenvalue with the
largest magnitude. (N.B.: Direction ê corresponds to the
z-direction in the spin Hamiltonian (eq 1a), written in a “proper”
coordinate frame for which 0 e E/D e 1/3.) Bint is positive along
ê, in contrast to the internal fields in virtually all high-spin
ferrous complexes. This observation, the large zero-field split-
tings of the S ) 2 multiplet, and the geff and µeff values in excess
of the spin-only values geff ) 8 and µeff ) 4.9 â suggest the
presence of substantial unquenched orbital momenta. Bint
contains three contributions, viz. the Fermi contact term, arising
from spin polarization of the s electrons by exchange interactions
with the unpaired 3d electrons, an orbital term representing the
magnetic field induced by the orbital motion of the 3d electrons,
and the spin dipolar term associated with a nonspherical spin
density in the 3d shell:16
Figure 8. X-band EPR spectra of polycrystalline samples of 1.CH3 (A)
and 1.Cl (B) recorded in parallel mode at T ) 2.0 K. The spectra were
collected at a microwave frequency of 9.27 GHz, 20 µW microwave power,
0.98 mT modulation amplitude, and 100 kHz modulation frequency. The
arrow in (A) denotes the shoulder discussed in the text.
resonances to a quasi-degenerate (∆ < 0.3 cm-1) MS ) (2
ground doublet of an S ) 2 multiplet. For this doublet, the
resonance condition can be written as15
∆E ) hν ) ∆2 + (4g âB )2
(2a)
x
z
z
For ∆ , 4gzâBz, as observed for 1.CH3, this expression can be
expanded as
Bint ) Bfc + BL + Bdp ) 2â r-3 (κ Sˆ - Lˆ ) + Bdp (3)
2
2
∆
∆
∆E = 4gzâBz +
) geffâBz 1 +
(2b)
2
(
)
8gzâBz
2(geffâBz)
where κ is a dimensionless constant expressing the s-orbital
contribution. Orbital angular momentum in common high-spin
ferrous complexes is almost quenched under the influence of
the crystal field.24 Some orbital momentum usually persists as
a result of crystal field state mixing by spin-orbit coupling.
The size of the resulting orbital momentum is then approxi-
mately proportional to the ratio of the spin-orbit coupling
constant λ and the crystal field splitting ∆ꢀcrys. As the crystal
field splittings are typically an order of magnitude larger than
λ, the resulting orbital terms are small. The idealized C2V
symmetry of 1.X gives rise to crystal field states which transform
according to one-dimensional irreducible representations. These
states have quenched orbital angular momenta. However, the
momenta observed in 1.X are large and suggest that there is a
substantial state mixing by spin-orbit coupling. For this
“unquenching” mechanism to be effective, the orbital ground
state must be accidentally degenerate, or nearly so, with a low-
lying excited orbital level. All pairs of 3d-orbitals whose energy
separations could possibly be small have been listed in Table
2. For geffâB gkBT, the expectation values in eq 3 can be
evaluated for the lowest level of the Zeeman-split spin quintet.
In the absence of an applied magnetic field, the internal field is
quenched by the zero-field splitting ∆ (both - Sê and - Lê ).
For 1.CH3, the second term in the parentheses of eq 2b will be
about 0.01, and thus a value of ∆ ) 0.03 cm-1 gives geff
)
11.4. The spectrum of Figure 8A exhibits a shoulder at g )
16.7, which may reflect the presence of spin-dipolar interactions
in the solid or, alternatively, may represent a distribution of ∆
values. For small ∆ values the intensity of the resonance is
proportional to ∆2/(geffâBz)2.15 Spectral simulations based on
this expression suggest that the g ) 16.7 shoulder represents
less than 1% of the molecules. EPR spectra of frozen toluene
solutions of 1.CH3 (not shown) lack this shoulder but still exhibit
the sharp feature at geff ) 11.4. Since the Mo¨ssbauer spectra of
1.CH3 are obtained for polycrystalline samples, we have
displayed a representative EPR spectrum of a solid sample in
Figure 8A. Taking into account the uncertainties in ∆, we
estimate geff ) 11.30-11.46, where the upper bound is the
theoretical maximum for geff (see section 4.1.1).
The Mo¨ssbauer data for 1.Cl suggest that ∆ ) 0.35 cm-1
,
and therefore the general expression (eq 2a) has to be used to
describe the EPR resonance. Since the terms in eq 2a have
comparable magnitudes, geff cannot be determined from a
spectrum taken at a single frequency. However, given that geff
is theoretically confined to 8.0 < geff < 11.46, ∆ can be
constrained from the spectrum of Figure 8B to 0.255-0.284
cm-1. We have also taken an X-band spectrum in perpendicular
mode (not shown); in this mode the cavity is tuned at a slightly
higher frequency, which allowed us to limit geff to the range
10.4-11.46.
(23) The experimental effective magnetic moments at room temperature for
1.CH3 and 1.Cl are clearly above the spin-only value of 4.9 µB (S ) 2).
Our crystal field model predicts for 1.CH3 and 1.Cl powder-averaged
magnetic moments of 5.47 and 5.38 µB at 300 K, respectively. We have
performed SQUID susceptibility studies on polycrystalline samples between
4 and 300 K, but we encountered problems in reproducing the temperature
dependence of the effective magnetic moments, which might be attributable
to alignment of the crystallites by the applied magnetic field. Such alignment
is not surprising in view of the extreme magnetic anisotropy of the ground
doublet. For the Mo¨ssbauer studies we have supressed alignment by
suspending the polycrystalline samples in Nujol. Further SQUID suscep-
tibility studies are in preparation.
4. Discussion
4.1. Crystal Field Model for 1.X. 4.1.1. Fine Structure and
Orbital Angular Momentum in 1.X. The spectroscopic results
for 1.X (X ) Cl-, NHTol-, NHtBu-, CH3-) demonstrate that
(24) Pavel, E. G.; Kitajima, N.; Solomon, E. I. J. Am. Chem. Soc. 1998, 120,
3949-3962.
9
3018 J. AM. CHEM. SOC. VOL. 124, NO. 12, 2002

II
Planar Three-Coordinate High-Spin Fe Complexes
A R T I C L E S
Table 2. Effective g Values, Angular Momenta, and Spin
Quantization Axes of High-Spin Ferrous Ion with a Doubly
Degenerate Orbital Ground State
degenerate orbitalsa
geff
Rb
spin quant. axisc
z2 x2-y2
8
8
10
10
10
10
10
11.5
11.5
12
0
0
1
1
1
1
1
x3
x3
2
z2 xy
xz xy
x
x
y
y
z
x
y
z
yz x2-y2
yz xy
xz x2-y2
xz yz
yz z2
xz z2
xy x2-y2
a Orbitals labeled d and d′ in Supporting Information (eqs S.1). z2 stands
for 3z2-r2. b Expectation value - Lê in lowest magnetic sublevel. c Denoted
ê in eq 4.
Figure 9. Plot of geff values as a function of (ꢀd - ꢀd′)/λ for R ) 0 (dashed
x
line), 1 (dotted line), 3 (full line), and 2 (dash-dotted line) in eq 5a.
However, since ∆ is small (Table 1), a large Bint can easily be
generated by applying a magnetic field with a nonvanishing
component along the spin quantization axis. The values for
- Lê to be used in eq 3 are listed in Table 2. These values are
the maximum orbital momenta attainable for the orbital pairs
and pertain to the limit in which the crystal field splitting
between these orbitals vanishes. If the smallest term in eq 3,
Bdp, is ignored, - Lê can be expressed as
B
fc,ê - Bint,ê
Bfc,ê
- Lˆê ) 2κ
(4)
The values for - Lê in Table 2 can be compared with the
experimental values obtained from eq 4 by substituting Bint
≈
70 T (Table 1), Bfc ≈ -50 T (based on the estimate of 12.7 T
per unpaired electron; see ref 16a), and 0.2 < κ < 0.4 into this
expression. The substitution yields 0.96 < - Lê < 1.92, which
includes seven of the cases listed in Table 2.
Fortunately, geff provides a more discriminatory criterion for
selecting the lowest orbital pair. The effective g value arising
from spin-orbit coupling between crystal field split 3d-orbitals
d and d′ is given by
Figure 10. Central structural core in complexes 1.X with an idealized
trigonal geometry (left) and a realistic distorted geometry (right) together
with schematic 3d-orbital level schemes. The z-axis of the Cartesian
coordinate frame is parallel to the trigonal axis and points toward the reader.
The symmetry labels refer to C3h (left) and C2V symmetry and ignore the
differences between the ligands.
geff ) 2geS + 2R cos[2Θ]
(5a)
not on its sign. The maximum of geff is obtained at ꢀd ) ꢀd′ for
each orbital pair. These maxima, which can be evaluated by
substituting ꢀd - ꢀd′ ) 0 in eq 5c, are listed in the second column
of Table 2. The experimental geff values can be compared with
the predicted maxima. The first seven cases yield maxima for
geff that are lower than the geff values observed in the series
1.X and can therefore be excluded. Two of the remaining three
cases, listed in the last lines of the table, can be eliminated,
based on the following qualitative crystal field considerations.
Figure 10 shows the core of 1.X with idealized planar trigonal
symmetry (θ ) 120°) and a schematic representation of the
actual distorted geometry of these complexes (θ ≈ 132°). Also
shown are the corresponding qualitative orbital schemes which
illustrate the removal of the degeneracies from the e-symmetry
orbitals by the crystal field in the distorted structure.26 The xy
orbital is raised in energy relative to x2-y2 because the
diketiminate nitrogens form nearly a 90° angle at the iron, which
places them favorably in the lobes of the xy orbital. In the
unlikely case that the splitting between xy and x2-y2 were large
enough to stabilize x2-y2 as the ground state, the large splitting
where Θ describes mixing of the two states under the influence
of crystal field and spin-orbit coupling,
Θ ) arctan
-
2 + 1
(5b)
4λR
ꢀd - ꢀd′
4λR
|
(
|
)
(
x
)
ꢀ - ꢀ
d
d′
λ ) -107 cm-1 is the spin-orbit coupling constant25 in λLˆ‚Sˆ,
S ) 2, ge ) 2.0023, ꢀd - ꢀd′ is the crystal field splitting, and R
is a coefficient of which the value is listed in Table 2 for each
orbital pair. In the limits of small and large crystal field
splittings, eq 5a simplifies to:
2
ꢀd - ꢀd′
4λ
ꢀd - ꢀd′
1
R
geff ≈ 8 + 2R -
,
, 1 (5c)
(
)
|
|
λ
λ
λ
geff ≈ 8 + 8R2
,
, 1
(5d)
|
| |
|
ꢀd - ꢀd′
ꢀd - ꢀd′
Figure 9 shows the geff values as a function of (ꢀd - ꢀd′)/λ. It
can be seen that geff depends on the magnitude of ꢀd - ꢀd′ but
(26) Campanion, A. L.; Komarynsky, M. A. J. Chem. Educ. 1964, 41, 257-
(25) Bendix, J.; Broson M.; Scha¨ffer, C. E. Inorg Chem. 1993, 32, 2838-2849.
262.
9
J. AM. CHEM. SOC. VOL. 124, NO. 12, 2002 3019

A R T I C L E S
Andres et al.
(>104 cm-1) would essentially quench the orbital momentum
arising from this orbital pair. Hence, the last case in Table 2
can be excluded as well. To decide between the remaining two
cases, (yz, z2) and (xz, z2), we note that yz is located in a plane
perpendicular to the Fe-X vector, whereas xz is in the plane
bisecting the N-N vector. Increasing θ increases (decreases)
the spatial overlap between xz (yz) and the symmetric (anti-
symmetric) combination of the nitrogen pz-orbitals, splitting xz
and yz as indicated in Figure 10. Thus, we conclude that the
large orbital momenta in 1.X originate from spin-orbit coupling
between the z2 and yz levels. Substitution of the maximum of
- Lê for the (yz, z2) pair into eq 3 yields κ ≈ 0.33, which is in
the middle of the range reported for this quantity.17
The last column in Table 2 indicates the quantization axis
(ê) of the spin, using the Cartesian coordinates defined in Figure
10 as a reference frame. Thus, the present analysis specifies
the spatial relationship between the principal axes of the fine-
structure and hyperfine tensors in the spin Hamiltonian with
respect to the molecular frame. For each orbital pair in Table
2, there is only one Cartesian component of the orbital angular
momentum operator that yields a nonzero matrix element, e.g.,
z2|Lx|yz ) ix3 . Thus, the spin-orbit coupling operator in
the space spanned by basis {z2, yz} simplifies to
the magnetic quantum number. The latter type of interaction is
commonly parametrized by the DSê term in the spin Hamil-
2
tonian formalism.27 However, as the second-order perturbation
terms are typically smaller by a factor λ/∆ꢀcrys than the first-
order terms, the energies and states given in eq 9 should provide
good approximations to those obtained from a complete
2
treatment. The inability of the DSê operator to represent the
entire energy spectrum of the spin quintet does not invalidate
the spectroscopic data analysis based on the quadratic spin
Hamiltonian eq 1a, because a precise description of the level
spacings is irrelevant at the employed cryogenic temperatures,
where the excited zero-field levels are virtually depopulated.
The zero-field splitting attains the maximum value |λ|x3 )
120-170 cm-1 in the degenerate case.
The smallness of the crystal field splitting between the z2
and yz orbitals introduces the concern of state mixing by low-
symmetry components of the crystal field,
|d ) cos ϑ|z2 + sin ϑ|yz
|d′ ) - sin ϑ|z2 + cos ϑ|yz
(10a)
where, contrary to the case of spin-orbit mixing, the coefficients
are real numbers. The only nonzero matrix element of the
angular momentum operator in the representation {d,d′} is an
invariant of the transformation 10a,
λLˆ‚Sˆ f λLˆxSˆx
(6)
For ꢀyz ) ꢀz2, the Hamiltonian can be diagonalized by using the
orbital states,
d|Lx|d′ ) z2|Lx|yz
(10b)
|Ψ( ) (|z2 ( i|yz )/x2
(7)
and, therefore, the effective g value is not affected by any such
crystal field effect.
The magnitude of the crystal field splitting |ꢀyz - ꢀz2| between
the lowest two 3d-orbitals in 1.X can be evaluated from eq 5a,
using the value for geff and an estimate for λ. The evaluation
results in the splittings 452 cm-1 for 1.Cl and 135 cm-1 for
1.CH3, where we have adopted the geff values listed in Table 1
and the rounded free ion value λ ) -100 cm-1. (N.B.: Equation
5c gives 395 and 133 cm-1, respectively, and illustrates its
accuracy, particularly in the second case.) Thus, the crystal field
splitting in 1.Cl is larger than that in 1.CH3, quenching the
orbital angular momentum to a greater extent than in 1.CH3,
and reducing the net internal field (eq 3) by an amount close to
that observed (Table 1). The last statement follows from the
expression for the change in the magnetic hyperfine orbital term,
∆AL ) 1/2 BL(CH3)[geff(CH3) - geff(Cl)]/[geff(CH3) - 8], from
which we predict a difference of 20-40 T in Bint for the two
complexes. However, to determine whether z2 or yz is the ground
state, we have to resort to an analysis that includes zero-field
splitting ∆ and the EFG tensor.
4.1.2. Evaluation of the Zero-Field Splitting ∆. The MS )
(2 ground state of the truncated spin-orbit coupling operator
that exclusively acts between the z2 and yz states in the 3d6 S )
2 manifold is strictly degenerate (see eqs 6-9). The actual
levels, however, were found to be split by energy ∆ in zero
applied field (see Table 1). These splittings originate from spin-
orbit couplings between the sets {z2,yz} and {xz,xy,x2-y2},
which generate orbital angular momenta perpendicular to the
spin quantization axis, x. It follows from Table 2 that the
and spin states that are quantized along x, i.e., Sx|MS ) MS|MS .
The energy eigenvalues can now be expressed as
ꢀM ) - |λ| Ψ(|Lˆx|Ψ( MS|Sˆx|MS
(8a)
S
in which the matrix elements are given by
x
Ψ(|Lˆx|Ψ( ) - 3
(8b)
(8c)
MS|Sˆx|MS ) MS
The states, |Ψ(, MS ) A|Ψ(>|MS> (A is an antisymme-
trizer), and energies arising from spin-orbit coupling between
the two degenerate crystal field levels are given by
+
x
|Ψ+,2 ,|Ψ-,-2 , ꢀ(2 ) 2 3|λ|
+
x
3|λ|
|Ψ+,1 ,|Ψ-,-1 , ꢀ(1
|z2,0 ,|yz,0 , ꢀ0 ) 0
)
(9)
-
x
|Ψ-,1 ,|Ψ+,-1 , ꢀ(1 ) - 3|λ|
-
x
|Ψ-,2 , |Ψ+,- 2 , ꢀ(2 ) - 2 3|λ|
The energies in eq 9 depend linearly on the coupling constant
λ, and accordingly, the level energies increase linearly with the
magnetic quantum number. Spin-orbit coupling between non-
degenerate orbital levels, such as those between the designated
orbital pair and the remaining three d-orbital states in 1.X, give
rise to a scheme with splittings depending quadratically on λ
in which the term energies progress as a quadratic function of
(27) (a) Abragam, A.; Bleaney, B. Electron Paramagnetic Resonance of
Transition Ions; Clarendon Press: Oxford, 1970. (b) Griffith, J. S. The
Theory of Transition-Metal Ions; Cambridge University Press: Cambridge,
1971.
9
3020 J. AM. CHEM. SOC. VOL. 124, NO. 12, 2002

II
Planar Three-Coordinate High-Spin Fe Complexes
A R T I C L E S
Thus, unlike the effective g value, ∆ is a nonsymmetric function
of ꢀyz - ꢀz2, having large and small values for positive and
negative arguments, respectively, and is therefore a sensitive
marker of the sign of the crystal field splitting. Based on the
experimental value for ∆ in 1.Cl (Table 1), ꢀxz is estimated to
be 3600 cm-1 for ꢀyz - ꢀz2 ) 400 cm-1 and 100 cm-1 for ꢀyz
-
ꢀz2 ) -400 cm-1. (N.B.: The magnitudes used here for the
crystal field splittings were estimated from geff (see section
4.1.1); the energies of the remaining 3d-orbitals were obtained
from the TD-DFT calculations described in section 4.2.2.) These
estimates concern the denominator of eq 11c and reflect the
change in the numerator associated with the sign flip of the
crystal field splitting between z2 and yz at fixed ∆. The
corresponding values for 1.CH3 (i.e., ꢀxz - ꢀyz ) 6000 and 1500
cm-1 for ꢀyz - ꢀz2 ≈ (130 cm-1, respectively) show the same
trend as that found in 1.Cl but are overall larger because ∆ for
1.CH3 is smaller than that in the chloride complex (see Table
1). The splitting ꢀxz - ꢀyz ≈ 100 cm-1 obtained for 1.Cl by
assuming that yz is the ground state of the crystal field seems
to be unrealistically small for a metal ion in a severely distorted
trigonal coordination sphere. Indeed, TD-DFT suggests that the
energy separation for this orbital pair is at least 1 order of
magnitude larger (section 4.2.2). The splitting ꢀxz - ꢀyz ≈ 3600
cm-1 estimated in the presence of a z2 ground state is on the
right order of magnitude, and we therefore conclude that z2 is
the ground state of the crystal field in 1.Cl.
2
Figure 11. ∆ calculated as a function of crystal-field splitting (ꢀyz - ꢀz )
for ꢀxz ) 3500 cm-1. The ordinate is on a logarithmic scale with an axis
break between 0 and 0.0001.
interactions in the pairs (z2,xz), (yz,xz), and (yz,xy) have the
required property. Among these interactions, the one involving
xy is expected to have the least effect on ∆, because xy is the
highest 3d-orbital level with an estimated excitation energy of
∼12 000 cm-1 (see section 4.2). The zero-field splitting ∆ was
calculated by numerical diagonalization of
H ) HCF + λLˆ‚Sˆ
(11a)
where the first term represents the crystal field which splits the
3d-obital energies into a pattern such as the one shown in Figure
10 on the right. The calculation gives the eigenstates
Given that splitting ꢀyz - ꢀz2 in 1.CH3 has a magnitude of
only 135 cm-1, a determination of its sign may seem to be a
rather academic exercise. One could argue that by replacing a
(n)
MS,ML
|Ψn
)
c
|MS,ML
(11b)
∑
σ-π donor (Cl-) with a σ donor (CH3-), the energy gap ꢀxz
-
MS,ML
ꢀyz (which depends on the π interaction of xz with X) will drop.
This prediction seems to support a yz ground state in 1.CH3,
because ꢀxz - ꢀyz ≈ 1500 cm-1 obtained for 1.CH3 with a yz
ground state is smaller than the value 3600 cm-1 for 1.Cl.
However, we consider the size of the reduction thus obtained
(2100 cm-1) to be unrealistically large. This judgment is based
on TD-DFT calculations for 1.Cl and 1.CH3 (section 4.2.2) that
result in nearly equal ꢀxz - ꢀyz values (≈4650 cm-1) for the
two species and indicate that the splitting between xz and yz is
predominantly determined by the metal-diketiminate interac-
tions. The alternative possibility of a z2 ground state, for which
ꢀxz - ꢀyz (6000 cm-1) is found to be greater than the value for
1.Cl (3600 cm-1), is not satisfactory either. It should be noted,
however, that while these estimates are based on a common
(free-ion) value for |λ|,25 the value for ꢀxz - ꢀyz obtained from
eq 11c for a given value of ∆ diminishes if one adopts a smaller
value for |λ|. In particular, a lowering of |λ| in 1.CH3 by 30%
reduces ꢀxz - ꢀyz from 6000 cm-1 to about 3600 cm-1. Taking
into consideration that a 30% reduction in |λ| is not unprec-
edented and that the computational results presented in section
4.2 support a z2 ground state, we tentatively conclude that z2 is
the ground state of the crystal field in 1.CH3. Finally, given
that z2 is lowest in 1.Cl, the decrease of ꢀyz - ꢀz2 in passing
from 1.Cl (435 cm-1) to 1.CH3 (135 cm-1) can be rationalized
by noting that ꢀz2 increases (due to an enhanced σ interaction
between z2 and X), while ꢀyz is constant (due to a lack of σ and
π interactions between yz and X).
and energy eigenvalues, ꢀn, from which ∆ is evaluated by taking
the difference of the lowest two values, ꢀ1 - ꢀ0. These
calculations were performed for different sets of orbital energies
and revealed a large effect of the order of the crystal field
energies for z2 and yz on ∆. The behavior can be understood
by considering the approximate expression
2
3
x
3 - 1
|λ|
x
3
∆ ≈
(11c)
(
)
2
x
2
x
(ꢀxz + 3|λ|)
which was derived by a perturbational treatment of the effect
of spin-orbit coupling involving the |xz,MS states with crystal
field excitation energy ꢀxz on the energies of the degenerate MS
) (2 ground levels for ꢀyz ) ꢀz2) 0, given in eq 9. The terms
x3 and -1 in the quadratic factor in eq 11c originate from the
couplings of xz with the z2 and yz components of the orbital
ground state given in eq 7, respectively. The two contributions
have opposite signs and tend to cancel. In the presence of a
crystal field splitting between z2 and yz, among other things,
x
the factor has to be replaced by the more general form 3 sin
γ - cos γ, where γ is defined as in eq S.1a of the Supporting
Information. The weight of the negative cosine term increases
if the yz orbital is lowered by the crystal field relative to z2 in
energy. As a result, ∆ decreases until perfect cancellation is
1
1
attained at γ ) /6 π, for which Θ ) γ - /4 π ) - 1/12 π.
Substituting Θ into eq 5b and solving for the crystal field
splitting yields the condition ꢀyz - ꢀz2 ) -4|λ| for ∆ ) 0. The
value of ∆ increases again for ꢀyz - ꢀz2 < -4|λ| but does not
become anywhere as large as for ꢀyz - ꢀz2 > 0 (see Figure 11).
In summary, the analysis of the ∆ values supports the idea
that the z2 orbital is the lowest 3d level in the crystal field of
the chloride complex, and most likely so in the methyl complex.
9
J. AM. CHEM. SOC. VOL. 124, NO. 12, 2002 3021

A R T I C L E S
Andres et al.
Table 3. Calculated Principal Components of EFG(-) in Modelsa
4.1.3. Evaluation of the Electric Field Gradient Tensor.
The sign of the EFG28 is determined by the electronic charge
distribution about the Fe nucleus. In particular, an oblate charge
distribution, such as that found for an electron in a yz orbital,
gives rise to ∆EQ > 0, whereas a prolate distribution, as in the
case of an occupied z2 orbital, yields ∆EQ < 0.16 Table 1 shows
that ∆EQ is negative in 1.Cl and positive in 1.CH3, suggesting
that the ground orbitals in these systems are z2 and yz,
respectively. However, this suggestion is incorrect because the
Vmax components28 in the two complexes are perpendicular to
the spin quantization axes (see section 3.2). The Vmax values
associated with electrons in the z2 and yz orbitals are along the
z- and x-axes, respectively. As the spin is predicted to be
quantized along x, irrespective of the order of the two orbitals,
we would indeed obtain the observed perpendicular orientation
for a z2 ground state but a parallel orientation (along x, i.e., the
Fe-X vector) for a yz ground state. Thus, the observed sign
flip in ∆EQ cannot be attributed to a reversal in the order of the
two orbitals and requires a more thorough analysis.
So far we have only considered the valence contributions to
the EFG. To explain the rather volatile behavior of the EFG in
the 1.X series, we have extended our analysis by taking into
consideration two additional contributions, namely: (i) the effect
of spin-orbit coupling on the valence part of the electric field
gradient and (ii) ligand contributions to the electric field gradient,
arising from the unusual coordination geometry at the metal.
The S ) 2 state of 1.X results from a half-filled shell with five
spin-parallel (R) electrons, a sixth electron with minority spin
(â) aligned antiparallel to the majority spin, and a large number
of spin-paired electrons. The charge distribution in the molecule
can be decomposed into a distribution for the sixth electron and
one for the other electrons in the system. The two distributions
yield EFG tensors at the iron nucleus which we generically refer
to as the “metal contribution” and the “ligand contribution” to
the EFG.29 In general, the ligands influence the metal contribu-
tion by splitting the degenerate 3d manifold into a nondegenerate
set of crystal field states and by admixing ligand orbitals into
the 3d-orbitals by which they are delocalized toward the ligands.
We show in section 4.2 by applying TD-DFT that the delocal-
ization of the 3d-orbitals in 1.X is much smaller than expected
on the basis of molecular orbital calculations. This circumstance
greatly facilitates the evaluation of the influence of the spin-
orbit coupling between the ground state and the nearby excited
states on the EFG by allowing a crystal field treatment. A
detailed analysis of the metal contribution is presented in section
S.1 of the Supporting Information. The ligand contributions to
the EFG originate from the anisotropy in the charge distribution
around the iron nucleus, arising from all charges in the complex
other than the sixth 3d electron. The relevant charges are located
in the essentially non-overlapping metal-ligand areas and
furnish additive contributions to the EFG with respect to the
ligands. These contributions are parametrized by the ligand-
e
f
V ,c
V ,c
V ,c
W ,d
V
,
W
,
x
y
z
X
z,intp
X,mb
modelb
FeIIICl3
mm/s
mm/s
mm/s
mm/s
mm/s
mm/s
g
g
2.90a 0.97
3.79a 1.26
3.56
4.73
5.50
6.93
0.3-1.3
h
FeIII(N′H2)3
g
g
g
g
1.19
1.58
1.83
2.31
0.9-1.9
1.2-2.0
FeIII(CH3)3
FeIII(N′H2)2Clh
FeIII(N′H2)2CH3
FeIIIL′Cli
-0.82 -2.44 3.26
-1.12 -3.13 4.25
-2.23 -1.97 4.20
-2.57 -2.84 5.42
0.76 -2.83 2.07
-1.44 -0.16 1.60
-0.29 -2.15 2.45
3.34
4.41
4.21
5.46
h
0.55
0.17
0.87
0.74
0.2-0.5
i
FeIIIL′CH3
-3.99
0.21 3.79
a Density functional, B3LYP; basis sets, 6-31G (first line) and 6-311G
(second line). Cartesian coordinates are defined in Figure 10. b In first five
complexes, all non-hydrogen atoms are in the x-y plane, and bond angle
θ (Figure 10) is 120°. Crystallographic structures of cores in 1.X for X )
Cl and CH3 were adopted for complexes listed in last two lines. The
5
electronic spin S is /2. c Converted from EFGs calculated in atomic units
by multiplying with factor -1.719 (mm/s)/(A.U.) based on 57Fe nuclear
quadrupole moment Q ) 0.17 barn. The quadrupole splitting is given by
∆EQ ) Vz(1 + η2/3)1/2 d WX values obtained by WX ) 1/3 Vz for [FeIIIX3]0
.
(eq S.2); values in the last four lines are for the diketiminate nitrogens, WN
) 1/2(Vz - WX), obtained by using the WX values for [FeIIIX3]0. e Interpolated
values Vz,intp ) 2WN′H + WX, for X ) Cl and CH3, respectively, based on
2
eq S.2. f WX ranges deduced from quadrupole interactions in series 1.X
(section 4.1.3). g Vx ≈ Vy ) -1/2 Vz. h Prime is used to distinguish N from
diketiminate nitrogen. i Results for structure L′ in which side chains of
phenyls were replaced by hydrogens and tert-butyls by methyls.
specific quantities WX, where X refers to a ligand, which are
assumed to be transferable between the complexes. Departures
from transferability are expected when, upon replacing a ligand
in the coordination shell of the three-coordinate iron site, the
charge distributions in the two unaltered bonds are modified.
However, the DFT calculations presented in section 4.2 indicate
that these nonadditivity effects are small compared to the
magnitude of the WX parameters. The details of the analysis of
the ligand contributions to the EFG in the series 1.X are
presented in section S.2 of the Supporting Information. The
effects of spin-orbital coupling and covalent reduction on the
metal contribution to the EFG have been parametrized by the
admixing parameter V and the reduction factor f, respectively,
which are defined in section S.1. The values for V and f can be
confined to narrow ranges, based on the g values (Figure S.1,
Supporting Information) and DFT calculations. These constraints
have been combined with the electric hyperfine data to construct
the allowed areas in a graphical representation of the WX
parameters (Figures S.2-4). The solutions lie in areas in which
either yz or z2 is the ground state. As the latter state is the ground
state according to the analysis of ∆ given in section 4.1.2, the
domains where z2 is lowest must contain the solution. The
limiting values for the WX parameters obtained from this analysis
have been listed in the last column of Table 3. The values for
WX are comparable in magnitude to the metal contributions to
the EFG. For example, a single methyl ligand may yield a
quadrupole splitting of as much as -2 mm/s. The ordering of
the WX parameters for the ligands in the series 1.X is
(28) For ∆EQ > 0, the MI ) (3/2 sublevels of the nuclear excited state are
higher in energy than the MI ) (1/2 levels. Note that the sign of ∆EQ is
the reverse of the sign of the electric field gradient component with the
largest magnitude. Quantum chemical codes usually provide the EFG rather
than ∆EQ and η, which may lead to confusion. To remind the reader of the
sign change, we label the gradient used by Mo¨ssbauer spectroscopists as
EFG(-). The operator for EFG(-) is denoted Vˆ, and the largest component
in magnitude of EFG(-) is referred to as Vmax. Our quantities Vi correspond
to eQVii/2 in the commonly used Hamiltonian, eq 1b, where Q is the nuclear
quadrupole moment of 57Fe and e is the positive unit charge.
WN(diketiminate) < WCl < WN′HR, WCH and indicates a dramatic
3
difference in the capabilities of the two nitrogen-based ligands
of perturbing the charge distribution about the nucleus of the
coordinated iron site. The small values for ∆EQ in 1.X, as
compared to the majority of high-spin ferrous complexes,
(29) The ligand contribution includes “covalency” contributions arising from
the overlap charges in the metal-ligand bonds.
9
3022 J. AM. CHEM. SOC. VOL. 124, NO. 12, 2002

II
Planar Three-Coordinate High-Spin Fe Complexes
A R T I C L E S
indicates an extensive cancellation of positive ligand contribu-
tions and negative metal contributions to EFG(-) along the
normal of the molecular plane.
numbers were obtained as small differences of larger numbers
and therefore are prone to any source of error. Moreover, the
W values show a significant dependence on the basis set. The
WX values deduced on the basis of additivity and transferability
from the computed (fifth column) and empirical (seventh
column) EFG(-)’s are in fair agreement. The computed W values
corroborate the conclusion of section 4.1.3 in that the diketimi-
nate nitrogens differ from the nitrogens in ligands of the type
N′HR in their ability to perturb the charge distribution around
the nucleus of the coordinated metal site. Moreover, the W
parameters of N′H2 and CH3, for which we inferred strongly
overlapping domains in the previous section, appear clearly in
the order WN′H < WCH , according to the calculations.
4.2. Density Functional Calculations. 4.2.1 Electric Field
Gradients. In the previous section we have introduced a scheme
for parametrizing the ligand contributions to the EFG(-) at the
iron nuclei in 1.X which relied on the premises that (a) the EFG
tensors for the ligands are additive quantities and (b) the values
for the WX parameters are transferable between the complexes.
These hypotheses were tested by performing quantum chemical
calculations in a number of model systems, listed in Table 3.
Among the computational methodologies available to us, density
functional theory (DFT) appeared to us the most appropriate
technique for this purpose, as it focuses on the determination
of the quantity from which the EFG is calculated, namely the
charge distribution.30,31 The calculations were carried out with
the Gaussian 98 package (release A.9)32 for charge-neutral
models in the MS ) 5/2 state and converged to solutions in which
the unpaired electrons are in metal 3d-orbitals. In the absence
of a sixth 3d electron, the metal contribution to the EFG(-)
vanishes in the Fe3+ state. Table 3 lists the eigenvalues of the
EFG(-) tensors computed for five trigonal models, including
three models with homoleptic coordinations and two with
heteroleptic ones. Assuming additivity, the Vz values for the
homoleptic models were used to evaluate the WX parameters
for the three monodentate ligands (fifth column). These values
were subsequently transferred to the heteroleptic models and
added in order to obtain the Vz components (sixth column). The
latter values and those obtained from the calculations for the
heteroleptic species (fourth column) are in excellent agreement,
independent of the basis set, and lend support to the notions of
both additivity and transferability. Heteroleptic coordination
leads to field gradient tensors with three different principal
components (see Table 3). Interpolation of the in-plane com-
ponents, based on the expressions for the Vx and Vy in eq S.2
of the Supporting Information, gave a somewhat less satisfactory
agreement, possibly because the EFG tensors may not be quite
axial along the atom-metal vectors, as assumed in section 4.1.3.
The last four lines of Table 3 contain the results for models
representing the nontrigonal diketiminate coordinations in 1.Cl
and 1.CH3 together with estimated WN parameters for the
diketiminate nitrogens (fifth column), obtained upon transferring
the WX parameters given in the same column to these mixed-
ligand species. Although transferability requires equal WN values
for the two models, these values are actually calculated to be
different for the two species. We note, however, that these
2
3
The DFT calculations for [FeIICl3]- and [FeII(CH3)3]- (using
the structures of the corresponding ferric species and the spin-
unrestricted B3LYP functional and basis set 6-311G provided
by the Gaussian 98 package32) resulted in d6 ground states with
doubly occupied z2 orbitals and gradients Vz ) -1.09 mm/s
(chloride) and +1.76 mm/s (methide). These values can be used
to estimate the W parameters for the FeII complexes, denoted
W′. Thus, subtraction of the valence contribution VVal(z2) ) -4.9
mm/s (computed for FeII with a z2 ground state, using the same
functional/basis set) and division by 3 yields the values W′Cl
)
1.3 mm/s and W′CH ) 2.2 mm/s. A comparison of these values
3
with the W parameters for the ferric species (Table 3) shows
that W′X ≈ WX and suggests that the ligand contributions are
practically independent of the oxidation state of the metal ion.
4.2.2. Excited Orbital States. Analysis of the fine structure
and hyperfine interactions in the ground states of the complexes
in the series 1.X allowed us to draw detailed conclusions about
the nature of these states and of two low-lying excited orbital
states (section 4.1). The latter states have low excitation energies
and are therefore difficult to detect by means of electronic
absorption spectroscopy or magnetic circular dichroism (MCD)
measurements, due to overlap with vibrational overtones and
the spectral limitations of the available MCD instrumention.
The acquired knowledge about the ground and excited orbital
states in 1.X makes the series a suitable candidate for testing
the predictive power of both DFT30 and recently developed time-
dependent (TD) DFT methodologies.33 We used for this purpose
the commonly used spin-unrestricted B3LYP functional and the
basis set 6-311G provided by the Gaussian 98 package.32 The
calculations were performed on stripped versions of 1.Cl and
1.CH3 in which the diketiminate ligand was simplified by
replacing the side chains by hydrogens for saving computational
time (structure 1). We also did calculations on more complete
structures in which substitution was limited to the side chains
of the phenyl rings and replacement of tert-butyl by methyl
(structure 2) but found only minor changes in the results, using
the EFG at the iron nucleus and excitation energies as indicators.
The calculations were performed for MS ) 2 and converged to
a state in which the four unpaired electrons occupy molecular
orbitals with predominantly 3d character.34 The calculations
placed the two spin-paired 3d electrons of the d6 configuration
(30) (a) Parr, R. C.; Yang, W. Density-Functional Theory of Atoms and
Molecules; Oxford University Press: New York, 1989. (b) Dreizler, R.
M.; Gross, E. K. U. Density Functional Theory, an Approach to the
Quantum Many-Body Problem; Springer-Verlag: Berlin, 1990.
(31) Available DFT programs do not consider spin-orbit coupling, that is, an
interaction shown above to be essential for understanding the electronic
properties of the ground state. This precludes a direct calculation of the
EFGs in 1.X.
(32) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M.
A.; Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J. A., Jr.; Stratmann,
R. E.; Burant, J. C.; Dapprich, S.; Millam, J. M.; Daniels, A. D.; Kudin,
K. N.; Strain, M. C.; Farkas, O.; Tomasi, J.; Barone, V.; Cossi, M.; Cammi,
R.; Mennucci, R.; Pomelli, C.; Adamo, C.; Clifford, S.; Ochterski, J.;
Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.; Malick, D. K.;
Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Cioslowski, J.; Ortiz,
J. V.; Baboul, A. G.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.;
Komaromi, I.; Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham,
M. A.; Peng, C. Y.; Nanayakkara, A.; Challacombe, M.; Gill, P. M. W.;
Johnson, B.; Chen, W.; Wong, M. W.; Andres, J. L.; Gonzalez, C.; Head-
Gordon, M.; Replogle, E. S.; Pople, J. A. Gaussian 98; Gaussian, Inc.:
Pittsburgh, PA, 1998.
(33) (a) Gross, E. K. U.; Dobson, J. F.; Petersilka, M. In Density Functional
Theory; Nalewajski, R. F., Ed.; Springer: Heidelberg, 1996. (b) Stratmann,
R. E.; Scuseria, G. E.; Frisch, M. J. J. Chem. Phys, 1998, 109, 8218-
8224. (c) Bauernschmitt, R.; Ahlrichs, R. Chem Phys Lett.. 1996, 256, 454.
(d) Casida, M. E.; Jamorski, C.; Casida, K. C.; Salahub, D. R. J. Chem.
Phys. 1998, 108, 4439.
(34) Gaussian 98 calculations for LCuIISCPh3, L ) â-diketiminate, by Randall
et al.6 gave solutions in which the unpaired electron occupied a ligand
orbital, at variance with experiment. No such problem arose for LFeIIX.
9
J. AM. CHEM. SOC. VOL. 124, NO. 12, 2002 3023

A R T I C L E S
Andres et al.
Kohn-Sham orbitals35 with major ligand components and the
numbers in parentheses are the energy eigenvalues (in cm-1
)
of the DFT Fock matrix, taken with respect to the eigenvalue
for the â-HOMO. Apart from small components, these orbitals
are either of 3d type, ligand type, or 3d-ligand mixtures in
roughly equal proportions. The TD-DFT output lists the excited
states as linear combinations of configurations that are obtained
from the Kohn-Sham monodeterminantal ground state by
making single substitutions therein: |occupied f virtual . With
the caveats that all orbitals contain ligand components and that
there are substitutions of charge-transfer type |d f ligand with
small coefficients for each excited state, the lowest four excited
levels obtained by TD-DFT are |(z2)â f (yz + L)â + |(z2)â f
(yz - L)â ≈ |(z2)â f (yz)â (2108), |(z2)â f (x2-y2)â (6179),
|(z2)â f (xz)â (6729), |(z2)â f (xy + L)â + |(z2)â f (xy -
L)â ≈ |(z2)â f (xy)â (12 535). In the case of the yz and xy
excitations, the configuration interaction leads to an “unmixing”
of the transitions into d f d′ and charge transfer, and it is
therefore incorrect to identify the excited states with the (highly
covalent) virtual Kohn-Sham orbitals. Such an identification
is also improper from an energetic point of view, as can be
seen by comparing the orbital energies with the TD-DFT
excitation energies (in parentheses), the former exceeding the
latter by as much as a factor of 3. The lowest four excitations
are metal-based and involve the standard Cartesian 3d-orbitals,
thus corroborating the analysis given in section 4.1. The TD-
DFT calculation correctly predicts the ordering ꢀz2 < ꢀyz < ꢀxz
but fails to reproduce the accidental near degeneracy of z2 and
yz; the computed orbital splitting (2108 cm-1) gives a too-small
effective g value (9.1, using eqs 5a,b) that lies far outside the
error margin for this quantity (Table 1). The calculated xz-yz
splitting is 4071 cm-1 and close to the value estimated from
the analysis of zero-field splitting ∆ in section 4.1.2. The TD-
DFT results for 1.Cl are similar to those for 1.CH3, apart from
Figure 12. Contour plot of the HOMO of â-spin in 1.Cl projected on a
truncated structural model used in the DFT calculation (structure 2). â is
the minority spin.
in a z2 type orbital (Figure 12) which, in the case of 1.CH3, is
slightly contaminated by admixture with an x2-y2 component.
The prediction of a ground state with z2 character is in agreement
with the ground state of the crystal field that was inferred from
the fine structure and hyperfine data in 1.X. However, the
comparison between computed and observed hyperfine param-
eters in 1.X is a more problematic matter. These parameters
are strongly affected by spin-orbit coupling (see section 4.1),
and since our DFT calculations ignore magnetic interactions
altogether, a meaningful comparison between theory and experi-
ment is excluded from the beginning. The only possible
exception is the Fermi contact field, which is insensitive to
spin-orbit coupling in cases, like the present one, where the
spin of the ground state is not affected by the coupling. The
contact fields that we computed by DFT for free iron ions were
in good agreement with experiment with respect to both sign
Afc < 0 and magnitude; however, calculations for the same ions
placed in a coordination environment resulted in dramatically
reduced contact fields, occasionally even having positive sign.
This result should not misguide one into thinking that the
positive hyperfine fields observed in 1.X are caused by a sign
flip of the contact field because the same behavior is also found
in DFT calculations for complexes with normal (negative)
contact fields. Moreover, a sign flip of the contact field would
not explain the exceptionally large effective g values in 1.X.
Contact fields calculated with unrestricted Hartree-Fock theory
using the same basis set are in much better agreement with
experiment and show that the erroneous results for this property
obtained by the DFT calculations are not related to the choice
of the basis set but are rather methodological in nature.
the x2-y2 excitation energy, which is lower by about 1700 cm-1
.
The latter result confirms that Cl- is a weaker σ donor than
-
CH3
.
Oldfield and co-workers have shown that DFT is a reliable
tool for predicting the electric hyperfine parameters in a large
number of metalloporphyrins.36 The good agreement between
the calculated and empirical WX parameters for 1.X confirms
the utility of DFT. The complexes 1.X differ from the earlier
studied macrocyclic species in that there is a low-lying excited
state which affects the EFG through admixture by spin-orbit
coupling. This complication makes the ab initio prediction of
the quadrupole splittings in systems such as 1.X virtually
impossible. For the spin-orbit effect on the EFG to be
significant, the energy level of the admixing state should lie
within ∼5|λ| from the ground level (see Figure S.1). As the
relevant energy gaps amount to only a few hundred wavenum-
bers, a reliable prediction of the quadrupole interactions in 1.X
requires the TD-DFT calculation for the excitation energies to
be accurate within 100 cm-1, exceeding the currently available
computational precision by at least 1 order of magnitude.
4.2.3. Isomer Shifts of 1.X. A comparison of the isomer shifts
observed for 1.X (Table 1) with those for complexes with
coordination number greater than 3 indicate that these values
follow the rule that δ decreases with decreasing coordination
In the following discussion the minority spin of the theoretical
MS ) 2 state is labeled â. The calculation for 1.CH3 (structure
2) yields a highest occupied â spinorbital (â-HOMO) with
approximately (z2)â character and a sequence of virtual orbitals
which appear as (yz + L)â (27 396), (x2-y2)â (31 126), (xy +
L)â (34 226), (L)â (36 847), (L)â (36 977), (L)â (38 831), (xz)â
(39 346), (yz - L)â (43 995), (xy - L)â (45 987), where L labels
(35) Kohn, W.; Sham, L. Phys. ReV. A 1965, 140, 1133.
(36) Havlin, R. H.; Godbout, N.; Salzmann, R.; Wojdelski, M.; Arnold, W.;
Schulz, C. E.; Oldfield, E. J. Am. Chem. Soc. 1998, 120, 3144-3151.
9
3024 J. AM. CHEM. SOC. VOL. 124, NO. 12, 2002

II
Planar Three-Coordinate High-Spin Fe Complexes
A R T I C L E S
number.37 Moreover, Table 1 reveals remarkably large changes
in the isomer shift upon replacements at a single ligand position,
X. In comparing the tabulated numbers, it is to be noted that a
smaller δ value is associated with a higher electron density at
the 57Fe nucleus. We tested this prediction by means of DFT
calculations on the model complexes [FeIIL′Cl]0 and [FeIIL′-
(CH3)3]0, where L′ is defined as in Table 3, and indeed found
that this trend is corroborated by the values for the electron
density, F(0), at the iron sites. Thus, F(0) in 1.CH3 is calculated
which a trigonal thiolate coordination is enforced by using
sterically encumbered ligands.3h The analysis described here and
the data cited in the Supporting Information suggest that there
is an extensive cancellation of positive and negative contribu-
tions to the EFG(-) along the normal of the molecular plane,
leading to unusually small quadrupole splittings for these three-
coordinate ferrous species. We appreciate now that three-
coordinate FeII complexes may have small isomer shifts (the
Power complex has δ ) 0.57 mm/s at 4.2 K) and unusually
small quadrupole splittings, resulting from substantial ligand
contributions in these planar environments. Without this rec-
ognition, it was difficult to assign oxidation states to the
“trigonal” iron sites of the cofactor. In fact, until suitable ferric
complexes with trigonal sulfido/thiolato coordination become
available, caution is advised in assigning oxidation states to
three-coordinate sites in multinuclear complexes.
-3
to be greater than in 1.Cl by ∆F(0) ) 0.7ao (ao ) 1 Bohr).
The increase in the density corresponds to a decrease in the
isomer shift, ∆δ ) -0.26 mm s-1 (Table 1), which leads to a
slope R ) ∆δ/∆F(0) ≈ -0.37 mm s-1 ao , which is in
3
accordance with earlier estimates of this quantity.38
4.3. Trigonal Sites in Fe-Mo Cofactor in Nitrogenase.
Finally, we wish to comment on some spectroscopic properties
that have puzzled us for over 20 years. The Mo¨ssbauer spectra
of the MoFe7S9 cluster in the S ) 3/2 state MN (see ref 8) yield
one quadrupole doublet with ∆EQ ≈ 0.7 mm/s and δ ) 0.41
mm/s at 100 K, where fast spin relaxation prevails. Six of the
seven iron sites of the cofactor cluster have a nearly trigonal
sulfido coordination, and thus it is not surprising that they have
the same, or nearly the same, quadrupole splittings. However,
the magnitude of the quadrupole splitting was substantially
smaller than anticipated for a cluster in which at least four8d or
perhaps six39 iron sites are thought to be ferrous (N.B.: The
majority of the S ) 2, FeII complexes has ∆EQ values between
2.5 and 3.3 mm/s).14 Nearly the same quadrupole splitting, ∆EQ
) 0.81 mm/s, has been reported by Power and collaborators
for the high-spin FeII complex [Fe(SC6H2-2,4,6-t-Bu3)3]- in
Acknowledgment. P.L.H. thanks Rene Lachicotte for as-
sistance with crystallography and Kevin Mooney for measuring
SQUID data. We thank Thomas Cundari for sharing the results
of calculations that were not included here. We thank Vladislav
Vrajmasu and Sebastian Stoian for assistance in the project.
P.L.H. acknowledges funding from the University of Rochester.
This research was supported by National Science Foundation
Grant MCD-9416224 (E.M.) and National Institute of Health
Grant GM-22701 (E.M.).
Supporting Information Available: Sections S.1 and S.2,
Table S.1, and Figures S.1-4 describing the analysis of the
electric hyperfine interactions; Tables S.2-7 and Figure S.5
containing the crystallographic data for complex 1.CH3 (PDF).
This material is available free of charge via the Internet at
http://pubs.acs.org.
(37) Kurtz, D. M., Jr. Chem. ReV. 1990, 90, 585.
(38) Bominaar, E. L.; Guillin, J.; Sawaryn, A.; Trautwein, A. X. Phys. ReV. B
1989, 39, 72-79.
(39) True, A. E.; Nelson, M. J.; Venters, R. A.; Orme-Johnson, W. H.; Hoffman,
B. M. J. Am. Chem. Soc. 1998, 110, 1935-1943.
JA012327L
9
J. AM. CHEM. SOC. VOL. 124, NO. 12, 2002 3025