Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
J Am Chem Soc. Author manuscript; available in PMC 2010 September 9.
Published in final edited form as:
PMCID: PMC2768349

Direct Spectroscopic Observation of Large Quenching of First Order Orbital Angular Momentum with Bending in Monomeric, Two-Coordinate Fe(II) Primary Amido Complexes and the Profound Magnetic Effects of the Absence of Jahn- and Renner-Teller Distortions in Rigorously Linear Coordination


The monomeric iron(II) amido derivatives Fe{N(H)Ar*}2 (1), Ar* = C6H3-2,6-(C6H2-2,4,6-Pri3)2, and Fe{N(H)Ar#}2 (2), Ar# = C6H3-2,6-(C6H2-2,4,6-Me3)2, were synthesized and studied in order to determine the effects of geometric changes on their unusual magnetic properties. The compounds, which are the first stable homoleptic primary amides of iron(II), were obtained by the transamination of Fe{N(SiMe3)2}2, with HN(SiMe3)2 elimination, by the primary amines H2NAr* or H2NAr#. X-ray crystallography shows that they have either strictly linear (1) or bent (2, N–Fe–N = 140.9(2)°) iron coordination. Variable temperature magnetization and applied magnetic field Mössbauer spectroscopy studies reveal a very large dependence of the magnetic properties on the metal coordination geometry. At ambient temperature, the linear 1 displayed an effective magnetic moment in the range 7.0 to 7.50 μB, consistent with essentially free ion magnetism. There is a very high internal orbital field component, HL ≈ 170 T which is only exceeded by a HL ≈ 203 T of Fe{C(SiMe3)3}2. In contrast, the strongly bent 2 displays a significantly lower μeff value in the range 5.25 to 5.80 μB at ambient temperature and a much lower orbital field HL value of 116 T. The data for the two amido complexes demonstrate a very large quenching of the orbital magnetic moment upon bending the linear geometry. In addition, a strong correlation of HL with overall formal symmetry is confirmed. ESR spectroscopy supports the existence of large orbital magnetic moments in 1 and 2, and DFT calculations provide good agreement with the physical data.


Although the gas phase electron diffraction structure of the linear d5 species Mn(CH2But)2 was reported over three decades ago,1 the synthesis and X-ray structure of Mn{C(SiMe3)3}2 in 1985 marked the first determination of the structure of a stable, open shell, two-coordinate molecular transition metal species in the solid state.2 Since then the half-filled shell manganese(II) species have been joined by several other two coordinate d5 derivatives and the range of open-shell first row transition metal complexes has been extended to complexes with d4, d6, d7 and d8 electron configurations.3,4 These include derivatives of Cr(I),5 Cr(II),58 Mn(II),1,2,711,1315 Fe(II),7,8,1012,14,21 Co(II),7,8,10,11,15,17 and Ni(II),68 which all possess a high spin electron configuration. Investigations of the magnetic properties of the rigorously linear Fe{C(SiMe3)3}2 have shown that it exhibits essentially free ion magnetic behavior (μeff = 6.8 ± 0.2 μB) and the largest internal hyperfine fields (152 T) observed to date.19,20 In addition, it has been shown that the almost linear Fe(NBut2)2 (N–Fe–N = 179.45(8)°) also possesses a high magnetic moment (μeff = 5.55 μB) and a large internal hyperfine field (105 T). However, there have been no studies on two-coordinate species that feature different geometries with the same ligand class. Such a study provides a rare opportunity to investigate the effects of lowering the symmetry of the two-coordinate iron environment on the orbital contribution to the magnetic moment and internal hyperfine field while maintaining a very similar ligand set.

We report the synthesis of the closely related iron(II) amido complexes Fe{N(H)Ar*}2 (1) (Ar* = C6H3-2,6-(C6H2-2,4,6-Pri2)2) and Fe{N(H)Ar#}2 (2) (Ar# = C6H3-2,6-(C6H2-2,4,6-Me3)2). They were characterized by X-ray crystallography, magnetization, zero and applied field Mössbauer effect and ESR spectroscopy and DFT calculations. The complexes possess local Dh (1), Figure 1, and C2v (2), Figure 2, symmetry at iron. We show that the symmetry change from strict linearity induces dramatic effects on the magnetic properties of 2.

Figure 1
Thermal ellipsoid (30%) drawing of 1. Hydrogen atoms (except N-H) are not shown. Fe---C(7),C(7)A 2.792 Å. Dashed lines indicate possible very weak intramolecular interactions.
Figure 2
Thermal ellipsoid (30%) drawing of 2. Hydrogen atoms (except N-H) are not shown. Fe(1)---C(16) 2.690 Å, Fe(1)---C(31) 2.588 Å. Dashed lines indicate weak intramolecular aryl π --- Fe interactions.

Experimental Section

Synthetic Methods, Physical Measurements, and Calculations

General Procedures

All manipulations were performed with the use of modified Schlenk techniques under argon or in a Vacuum Atmospheres drybox under N2. Solvents were dried and collected using a Grubbs-type22 solvent purification system or distilled from Na/K alloy, and degassed by sparging with dry Ar for 10 min. Ar#NH2 was prepared from reduction of Ar#N3 with LiAlH4 in Et2O,23,24, Ar#N3 having been prepared by treating Ar#Li25 with p-toluenesulfonyl azide26 in Et2O. Ar*NH2 was prepared in an analogous manner. Fe{N(SiMe3)2}2 was synthesized according to a modified literature procedure,10a using anhydrous FeCl2 (FeCl·4H2O, Alfa, dehydrated at ca. 250 °C under reduced pressure for 1 d) instead of the bromide. All physical measurements were obtained under strictly anaerobic conditions. 1H NMR spectra were recorded on a Varian 300 MHz instrument and referenced internally to residual silicone vacuum grease, d 0.29 ppm in C6D6. IR spectra were recorded as Nujol mulls between CsI plates on a Perkin-Elmer 1430 spectrophotometer. UV-vis spectra were recorded as dilute hexane solutions in 3.5 mL quartz cuvettes using a HP 8452 diode array spectrophotometer. Melting points were determined on a Meltemp II apparatus using glass capillaries sealed with vacuum grease, and are uncorrected.

Fe{N(H)Ar*}2 (1)

A Schlenk flask was charged with Ar*NH2 (1.403 g, 3.0 mmol), Fe{N(SiMe3)2}2 (0.678 g, 1.8 mmol), and a stirring bar. The mixture was then shaken as it was heated to ca. 200 °C, melting and co-dissolving as the reaction ensued. Intermittent vacuum was applied over ca. 10 min to remove the HN(SiMe3)2 byproduct as a white vapor. A deep red solid resulted, and, under reduced pressure, the entire flask was heated from bottom to top at ca. 180 °C for 5 min to to complete the reaction, removing all the HN(SiMe3)2 as well as unreacted Fe{N(SiMe3)2}2. The resulting solid was extracted with 40 mL hexane, filtered on a medium frit with Celite, and concentrated to ca. 15 mL. After 2d standing at ca. 25 °C deep red rod-shaped crystals appeared, yield 0.350 g (22%), mp 263 °C. 1H NMR (300 MHz, C6D6, 25 °C) δ 7.09 (br, s, ArH), 1.21 (br, m), 0.84 (br, s), 0.19 (br, m), −13.2 (br, s), −57.2 (br, s). IR, cm−1: ν(N–H) 3350; ν(Fe–N) 400. UV-vis, nm (ε, M−1cm−1): 450 (2840), 424 sh (2360), 320 (12100), 310 sh (11800).

Fe{N(H)Ar#}2 (2)

A Schlenk flask was charged with Ar#NH2 (0.988 g, 3.0 mmol) and [Fe{N(SiMe3)2}2] (0.678 g, 1.8 mmol), and a stirring bar. As the solids began reacting at ca. 25 °C (indicated by a reddening of the crystalline Ar#NH2), toluene (ca. 10 mL) was added and the mixture was stirred and heated to reflux for 5 min to afford a deep red solution. The solvent was removed under reduced pressure and the entire flask heated from bottom to top at ca. 180 °C for 5 min. The resulting red solid was extracted with ca. 100 mL hexane, filtered via cannula, and left to stand at ambient temperature for 2 d to afford lighter red block-like crystals, yield 0.420 g (39 %), mp 248 °C. 1H NMR (300 MHz, C6D6, 25 °C) δ 7.10 (br, s, ArH), 2.14 (br, s), 1.20 (br, s), 0.84 (br, s), 0.27 (br, m), −64.3 (br, s). IR, cm−1: ν(N–H) 3340, 3300; ν(Fe–N) tentatively assigned as 340, 315. UV-vis, nm (ε, M−1cm−1): 434 (3880), 322 (13300).

X-ray Crystallography

Crystals of appropriate quality for X-ray diffraction studies were removed from a Schlenk tube under a stream of nitrogen and immediately covered with a thin layer of hydrocarbon oil (Paratone). A suitable crystal was then selected and attached to a glass fiber and quickly placed in a low-temperature stream of nitrogen (90 K).27 Data for compounds 1 and 2 were obtained on a Bruker SMART 1000 instrument using Mo Kα radiation (λ = 0.71073 Å) in conjunction with a CCD detector. The collected reflections were corrected for Lorentz and polarization effects by using Blessing’s method as incorporated into the program SADABS.28,29 The structures were solved by direct methods and refined with the SHELXTL v.6.1 software package.30 Refinement was by full-matrix least-squares procedures with all carbon-bound hydrogen atoms included in calculated positions and treated as riding atoms. N-bound hydrogens were located directly from the Fourier difference map.

Magnetic measurements

DC magnetization measurements were obtained at Northeastern University using a Quantum Design MPMS SQUID magnetometer in applied fields up to 5 T on ca. 60 mg samples in Quantum Design Delrin holders sealed under nitrogen with Apiezon-N grease.

57Fe Mössbauer Effect Spectroscopy

Zero-field Mössbauer spectra were recorded using a standard constant acceleration spectrometer with an accompanying Janis Vari-Temp Cryostat for temperatures as low as 1.3 K with temperature measurement based on calibrated silicon diodes and vapor pressure thermometry for samples sealed under nitrogen with Devcon Epoxy in cylindrical nylon holders. A 50 mCi (57Co(Rh)) gamma-ray source was used for spectral determinations with calibration based on the hyperfine patterns of α-Fe foil (6.5 μm thick) and polycrystalline α-Fe2O3. Longitudinal applied field Mössbauer spectra (Eγ parallel to H0) were obtained using Janis Super Vari-Temp cryostats for magnetic fields up to 9 T (Knox College) and 8 T (Carnegie Mellon University) for complexes 1 and 2, respectively.

Electron Spin Resonance Spectroscopy

Perpendicular and parallel-polarization continuous-wave electron spin resonance (CW ESR) spectra were recorded using a Bruker ECS106 spectrometer equipped with a dual-mode cavity (ER 4116DM) operating at X-band microwave frequencies. Cryogenic temperatures were achieved and controlled using an Oxford Instruments ESR900 liquid helium cryostat in conjunction with an Oxford Instruments ITC503 temperature and gas flow controller.

DFT Calculations

The electronic structure calculations were performed with the Gaussian 03 program31 using the hybrid BLYP approximation to the exchange-correlation functional within DFT theory, which was previously shown to perform well for the description of the low-valent iron compounds. The BLYP was combined with a double-ζ quality 6–31g* basis set for most of the optimized geometries. Several truncated models were used for these calculations: models M1 and M2 were based on the X-ray data for linear 1 and bent 2 respectively, where the alkyl groups on the flanking aryls were replaced with hydrogens. In the models M1′ and M2′ terphenyl ligands were replaced by phenyl groups, while M1″ is simply the (predicted linear) Fe(NHMe)2.

Results and Discussion

Synthesis and Structural Characterization

The complexes 1 and 2 were synthesized by a transamination approach that involved the treatment of Fe{N(SiMe3)2}2 with two equivalents of the appropriate primary arylamine in accordance with eq. (1).


The compounds 1 and 2, which represent the first examples of stable, homoleptic, primary amides of iron, were isolated as red crystalline solids. Primary amido {–N(H)R} homoleptic Fe(II) derivatives were previously unavailable most likely because the diminished steric protection afforded by a ligand that carries only one organic substituent at nitrogen was insufficient to stabilize such complexes. However, the very large sizes of the Ar# and Ar* substituents enable monomeric species to be readily isolated. IR spectroscopy revealed absorptions at 3350 (1), 3340 (2), and 3300 (2) cm−1, consistent with the presence of –NH moieties in each complex. Despite several attempts using a 1:1 ratio of reactants at elevated temperature, the imide (FeNAr*,#)2 was never isolated (although analogous divalent Group 14 imide dimers have been characterized in our laboratory44).

X-ray crystallography showed that both 1 and 2 were well separated monomers with the closest metal-metal approaches being 11.70 Å for 1 and 9.30 Å for 2. Complex 1, Figure 1, has a linear N–Fe–N array with a crystallographically required inversion center at iron. The strictly linear structure of 1 is unique for an iron(II) amide in the crystalline phase. In contrast, the N–Fe–N geometry in 2 is bent (Figure 2) with an N–Fe–N angle of ca. 141°. There are also apparently weak secondary interactions with the metal center from a flanking ring from each ligand as exemplified by the distances Fe(1)---C(7) = 2.792(1) Å in (1) and Fe(1)---C(16) = 2.690(3) Å and Fe(1)---C(31) = 2.588(3) Å in 2 as illustrated by dashed lines in figures 1 and and2.2. The Fe–N distances in 1 (1.902(1) Å) and 2 (1.911(3) Å, average) may be compared with those previously reported for Fe{N(SiMe2Ph)2}2 (Fe–N = 1.903(7) Å, N–Fe–N = 172.1(1)°),11 Fe{N(SiMePh2)2}2 (Fe–N = 1.917(2) Å, N–Fe–N = 169.0(1)°),11,17 Fe{N(Mes)BMes2}2 (Fe–N = 1.938(2), N-Fe-N = 166.6(1)°)7 and Fe(NBut2)2 (Fe–N = 1.879(2) Å, N-Fe-N = 179.45(8)°).21 Thus, the almost equal Fe–N bond lengths in 1 and 2 lie within the previously known limits. It is also noteworthy that, with the exception of the almost linear Fe(NBut2)2 (note: the silylamide Fe{N(SiMe3)2}2 is linear in the vapor phase10a but is dimerized and bent at iron in the solid10b), the three previously observed solid state structures of two-coordinate iron amides all feature bending of the metal coordination geometry, although we observe that the degree of bending in 2 is ca. 20° greater than that previously reported.4 Thus it seems probable that the strongly bent geometry in 2 is a result of the lower steric requirement of the Ar# substituent vs. Ar*. In the solid state, two-coordinate metal complexes often display deviations from linearity owing to the tendency of the generally electron deficient metal center (8 valence electrons for Fe(II)) to interact with electron rich moieties such as flanking aryl groups via bending or dimerization. Bending can be prevented by using very large substituent groups which require the distance between them to be maximized to avoid steric congestion as in linear 1. With less bulky ligands the substituents can approach each other more closely and bending becomes allowed as in 2. Even so, the secondary interactions between the iron and a carbon in 2 (2.690(3) and 2.588(3) Å) are not much shorter than those in 1 (2.79(2) Å) and their weakness is emphasized by the fact that they exceed a single Fe–C bond length by ca. 0.5–0.6 Å.1419

Magnetism, ESR and Mössbauer Spectroscopic Data

In view of the linear and significantly bent structures of complexes 1 and 2 respectively, one has for the first time a unique opportunity to investigate great extremes of orbital magnetism for an open shell first transition series ion in a common oxidation and spin state, namely high-spin Fe(II), in complexes that carry electronically similar ligands. As 1 is linear, there is no requirement for linear (first order) Jahn-Teller distortion of its coordination environment because there is no vibrational mode that can couple to the electronic level that might afford such a distortion.32 A spontaneous bending distortion (so called Renner or Renner-Teller effect) can arise from higher order non-linear terms in the vibronic coupling32 and lead to quenching of first order orbital angular momentum. However this effect is likely to be somewhat, if not completely, vitiated for the solid state form of two-coordinate complexes with bulky ligands. In any event, these observations and the high degree of coordinative unsaturation lead to the expectation of extraordinary orbital contributions to the effective magnetic moment, isothermal magnetization and internal hyperfine fields of complexes such as 1, in some cases values broaching those of the corresponding free ion.20 On the other hand, the bent structure of complex 2 mimics the essential end result of a large Renner Effect distortion, i.e., a sterically allowed (aryl-π)---Fe interaction-driven distortion leading to bending and concomitant partial quenching of orbital magnetism. The orbital ground state of 1 is likely to be intrinsically or at least accidently degenerate in view of the non-bonding, δ symmetry of the ground dxy, dx2y2 pair (Figure 3) with respect to the primary amide ligands. This leads to a δ3π2σ1 ground configuration which is the basis of the 5Δg orbital ground term33 in local Dh symmetry. This ground state exhibits first order orbital angular momentum owing to the presence of the aforementioned orbital degeneracy and the existence of a rotational symmetry element (C8, 45° rotation) connecting the dxy, dx2y2 pair.34 The magnetization, ESR and Mössbauer spectroscopy results that follow confirm that the latter ground state term does indeed lead to (a) very large (first order) orbital angular momentum and (b) appreciable orbital magnetism in the limit of its spin-orbit splitting, ultimately resulting in a so called quasi-degenerate spin-orbit35,36 electronic ground state. Lastly, we partially rationalize our results in terms of DFT calculations for complexes 1 and 2.

Figure 3
Crystal field splitting diagram (high-spin d6) showing atomic orbital symmetry labels for ideal homoleptic linear two-coordination.

Effective Moments and Isothermal Magnetization

The temperature dependence of the effective moment for the bent complex, 2, in various applied fields, is presented in Figure 4. It is evident that the effective magnetic moments at ambient temperature range from ca. 5.25 to ca. 5.80 μB. These values for μeff are near or greater than those normally observed for the upper range of μeff values for typical high-spin Fe(II).37 (Recall that spin-only behavior for high-spin Fe(II) corresponds μeff = √24 or ca. 4.9 μB). We know of no other detailed magnetization studies of genuine linear and/or bent two coordinate high-spin Fe(II) systems save for those on the linear Fe{C(SiMe3)3}220 and the almost linear (N–Fe–N = 179.45(8)°) diamide Fe(NBut2)2.21 Suffice it to say that the effective moment vs. temperature profiles of both 1 and 2 are rather reminiscent of those predicted and observed for certain pseudo-octahedral complexes of Fe(II), especially in the limit of a Stevens orbital angular momentum reduction factor, k ≈ 0, i.e., little or no electron delocalization due to covalency. This is consistent with the sigma bonding anionic nature of the (first short period element) primary amide ligands.10a Unoccupied, ligand-centered molecular orbitals are not energetically close to the occupied Fe 3d orbitals, minimizing any nephelauxetic delocalization behavior. The gradual rise in μeff as T decreases to ca. 70 K is expected since the greater than half filled shell d6 configuration of high spin Fe(II) requires λ (± ζ3d/2S) to be negative (the free ion single electron spin-orbit constant ζ3d for Fe(II) is ~ 400 cm−1). That is, one is progressively populating larger values of J (= L + S) at lower temperatures. Finally, at lowest temperatures, the effects of single ion zero-field splitting appear to predominate and μeff decreases.

Figure 4
Zero-field cooled DC temperature dependence of the effective magnetic moment of complex 2 in various applied fields.

While complexes 1 and 2 exhibit qualitatively similar magnetization behavior, the magnitude of μeff as a function of temperature for 1 is ca. 1.2 to 2.0 μB greater than that for 2 (cf. Figures 4 and and5).5). This is consistent with a measurably larger orbital contribution to μeff. for the linear complex. Such an observation is not unexpected in view of its linear structure leading to the absence of so-called linear or first order Jahn-Teller distortion. Figures 4 and and55 clearly show DC field alignment effects for complexes 1 and 2. This effect is expected to lead to larger limiting μeff values for 1 owing in part to its intrinsically larger ground state orbital angular momentum. These interpretations have much firmer, direct support from consideration of a comparison of the applied field Mössbauer spectra of 1 and 2, vide infra. In passing, we point out a knee in μeff versus T in the vicinity of 275 K for 2 that is not observed for 1. This probably suggests a structural phase transformation for 2 although we do not have ambient temperature structures or calorimetry (DSC, etc.) data to pursue this possibility further at this time.

Figure 5
Zero-field cooled DC temperature dependence of the effective magnetic moment of complex 1 in various applied fields. The small irregularities at ~94 and ~64 K are believed to be the result of condensation of dioxygen and antiferromagnetic ordering of ...

Figure 6 shows the (field cooled) isothermal magnetization of 1 (top) and 2 (bottom) at 1.8 K. It is apparent that 2 is not saturated at 5 T. This is likely the result of strong local magnetic anisotropy. On the other hand, 1 exhibits a magnetization well in excess of that of the spin-only value for S = 2 (22,340 emu/mol). In fact its value is some 96% of the value (27,913 emu/mol) corresponding to an S = 5/2 spin-only value. This implies a very large orbital contribution for 1. In effect, we have an orbital magnetism contribution that is essentially equivalent to adding one full spin unit to the spin-only behavior of high-spin Fe(II). Nevertheless, both 1 and 2 are simple, rapidly relaxing paramagnets (in zero field) where 1 exhibits large first order orbital contributions to its moment.38 There is no evidence of long range cooperative magnetic effects, namely ferromagnetism or canted anti-ferromagnetism capable of enhancing the measured magnetic moments (the intermolecular Fe---Fe distances in both compounds exceeds 9 Å).

Figure 6
Field cooled DC magnetization versus applied field at 1.8 K for complexes 1 (top) and 2 (bottom). The spin-only maximum for Fe(II) magnetization is ~22,340 emu.

Zero and Applied Magnetic Field Mössbauer Spectroscopy

Complex 1 exhibits a single, quadrupole doublet (Figure 7) in zero field to as low as 4.2 K. This behavior is characteristic of a unique iron environment and rapidly relaxing paramagnetism. At first sight, its small and relatively temperature independent quadrupole splitting values (ΔEQ = 0.68 mm/s at 293 K, 0.74 mm/s at 77.5 K and 0.96 mm/s at 20 K), seem atypical of high spin iron Fe(II), especially when one recalls that the valence shell electron occupation contribution to the z-component of the electric field gradient tensor, 4/7q <r−3>3d, is large (ca. 3 mm/s) and positive for an electron in either orbital of the ground dxy, dx2y2 pair, Figure 3. Apparently, the field gradient tensor arising from covalence anisotropy owing to the axially disposed ligands (which should be negative in this case) largely cancels that induced by the occupation of the in-plane 3d orbitals. The isomer shifts of 1 (δ = 0.38 mm/s at 293 K, 0.46 mm/s at 77.5 K and 0.45 mm/s at 20 K) fit in nicely with the expected decrease in δ with decreasing coordination number for 57Fe of a given spin, oxidation state and coordination environment. Presently, however, there is an insufficient Mössbauer parameter database for two-coordinate iron from which to make quantitative comparisons.

Figure 7
Zero-field Mössbauer spectrum of complex 1 at 293 K.

The application of a magnetic field has a spectacular effect on the Mössbauer spectrum of 1 (Figure 8). The not quite fully resolved nuclear Zeeman splitting apparent in the 0.1 T spectrum (bottom) is diagnostic of slow—but not infinitely slow—electron spin relaxation. The features present at +20 and −19 mm/s indicate that the applied field H0 of only 1000 gauss causes an effective field Heff at the 57Fe nucleus of some 122 T (1.22 megagauss). This results from slow electron spin relaxation, leading to dramatic Zeeman splitting of the nuclear spin levels. The infinitely slow relaxation limit is essentially reached in the 1 T spectrum and all six (two ΔmI = 0, two ΔmI = +1 and two ΔmI = −1) gamma-ray transitions39 of the typical nuclear Zeeman pattern of 57Fe are clearly evident with Heff now ≈ 124 T. At H0 = 5 T, the two ΔmI = 0 transitions have all but disappeared. This signifies that the internal field at the iron atoms is now largely polarized parallel to H0 and Eγ, where Eγ is the electric field component of the incident γ-radiation. At this point, Heff is ca. 126 T. Finally, at 9 T, one has more or less “complete polarization” of the sample moments and Heff ≈ 130 T. Note that the effective internal fields have been obtained neglecting any quadrupolar shift of the inner four nuclear gamma resonances of the Zeeman pattern relative to the outer two (small in the case of complex 1). Furthermore, the magnitude of these nuclear splittings has increased monotonically with increasing magnitude of the applied field. This implies that the internal field is positive and that the dominant contribution to this field is the orbital component HL. The other contributions to the internal hyperfine field are embodied in the equation for the effective field, namely: Heff = H0 + Hinternal = H0 + HF + HD + HL where HF is the Fermi contact contribution to Hint and HD is the dipolar term. Making reasonable assumptions concerning HD and HF as outlined in reference 36, one calculates HL for complex 1 to be ca. 170 T (1.70 megagauss). The values for Hint and HL are among the largest ever observed regardless of spin, oxidation state or coordination environment in all of known iron chemistry to date. To our knowledge, the values of these parameters for 1 are exceeded only by those for the linear two-coordinate complex Fe{C(SiMe3)3}220 for which Hint = 152 T and HL = 203 T with H0 = 0 at 4.2 K. Without a doubt, the ground state of 1 has an extremely large first order orbital angular momentum.

Figure 8
Longitudinal applied field Mössbauer spectra of complex 1 at 4.2 K. Note that the ΔmI = 0 transitions disappear between 1.0 and and 5.0 T.

The zero-field Mössbauer spectra of bent 2 presented in Figure 9 are noteworthy in that the isomer shifts (δ = 0.76 mm/s at 293 K and 0.95 mm/s at 4.2 K) are markedly larger than for any previously characterized two-coordinate Fe(II) complex. However, these larger δ values are consistent with an increase in the effective coordination number due to weak Fe---C interactions with the flanking aryl rings on the amide ligand (Figure 2, dashed lines). The local symmetry of the coordination environment (with or without the proposed flanking interactions) is at most C2v. This is a point group whose irreducible representations are all uni-dimensional and thus non-degenerate. Hence, the bending leads to the expectation of a measureable quenching of first order orbital angular momentum relative to complex 1. Essentially this is confirmed by the applied field Mössbauer spectra of complex 2 (Figure 10) that are now considered. The sequence of Figure 8 is roughly reproduced in Figure 10 except that the field resolution is higher. The asterisks of Figure 10 correspond to the two ΔmI = 0 transitions which have all but disappeared at 1T. We note that there is some impurity background absorption evident in the spectra especially at 8 T. However this does not interfere with the interpretation that follows.

Figure 9
Zero-field Mössbauer spectra of complex 2 at various temperatures.
Figure 10
Longitudinal applied field Mössbauer spectra of complex 2 at 4.2 K. The asterisks mark the ΔmI = 0 nuclear transitions whose intensity diminish beyond 1 T.

It is clear that the internal field is positive again as for complex 1 indicating a dominant orbital contribution. However at H0 = 8 T, the value of Heff is ca. 73 T from which one calculates HL = 116 T. This represents a ~32% decrease in orbital angular momentum (using the HL value as a metric) relative to complex 1, a large but by no means total quenching of L. In fact in the context of orbital angular momentum, the behavior of the bent complex 2 (N–Fe–N angle = 141°) is reminiscent of that of a series of planar three-coordinate (C2v) ferrous complexes of reference 36. Specifically, the CH3-substituted Fe(II) β-diketiminate analogue (N–Fe–N = ~95°) in the series has an internal field of 82 T. This corresponds to HL ≈ 133 T (neglecting HD) where the authors interpret the large internal field in terms of mixing of closely spaced or accidentally degenerate ground state d orbitals via spin-orbit coupling. This leads to a quasi-degenerate electronic ground state with near maximal orbital angular momentum for the particular ground state d orbital pair involved, namely (dz2, dyz) in the three coordinate iron series. Note however, that while this system is somewhat similar to 2 of this work, neither has a first order contribution to L. Referring to Figure 3, the ground state orbital (dxy, dx2y2) pair of complex 1 is degenerate or at least accidentally so and is likely to be similarly accidently degenerate in 2 owing to their non-bonding δ symmetry in the present linear and bent systems. However, there is no longer a very large first order contribution in 2 since in C2v the C8(45°) rotational symmetry element that transforms dxy to dx2y2 or vice versa is lost. One can however invoke the aforementioned spin-orbit coupling “unquenching” mechanism of reference 36 for 2. It is likely that the similarity of the orbital magnetism of 2 to that of the three-coordinate C2v β-diketiminate complexes of reference 36 is not fortuitous.

In summary, strictly speaking, the electronic ground states of linear two coordinate ferrous complexes are better considered either as (i) a pure spin-orbit doublet (mJ = ± 4 arising from the ground J state for the 5D term of Fe(II))20 or (ii) such a doublet mixed to varying degrees (in lower symmetry) with nearby adjacent spin-orbit states where the extent of mixing depends on the prevailing symmetry of the complex. This contention arises, in the context of perturbation theory, from the facts that (a) we have minimal ligation with only two relatively weak field anions, and (b) the spin-orbit coupling perturbation to the Hamiltonian H = λLS is near maximal owing to the approach of L to 2. In any event, (i) should lead to ESR silence and infinitely slow paramagnetic relaxation in H0 = 0. Only an absence of rhombic ZFS will lead to true ESR silence if D is greater than the energy of the microwave quantum. Specifically, the zero-rhombicity Fe{C(SiMe3)3}220 is ESR silent out to 26 T at 4.2 K and exhibits fully resolved magnetic hyperfine splitting of its zero-field Mössbauer spectrum up to ~50 K.41 Its uniaxial magnetism and extraordinary internal hyperfine field have been adequately explained in terms of an isolated pure mJ = ± 4 ground state.42 The ground state described in situation (ii) ensures a degree of “allowedness” to ESR transitions but will require an externally applied magnetic field (albeit usually relatively small) in order to induce slow paramagnetic relaxation and resolved hyperfine splitting of Mössbauer spectra, and where HL is expected to be reduced somewhat owing to the mixing. This is the case for complex 1 of this work and the diamide system Fe(N But2)2.21 In this light, it is instructive to consider Table 3, especially the first three entries wherein it is seen that the value of HL is qualitatively quite sensitive to the overall (formal) symmetry of the complex even for the Dh entries. However, while tantalizing, the unequivocal confirmation of the importance of such secondary (formal) symmetry considerations clearly demands the study of numerous other linear and bent two coordinate complexes of high-spin Fe(II).

Table 3
Comparison of HL values for different symmetries.

ESR Spectroscopy

The 4.2 K parallel-polarization CW ESR spectra (Figure 11, top) of the linear and bent Fe(II) amido complexes are typical of those for high-spin ferrous (S = 2) compounds wherein the zero-field splitting (Δ) of the ms = ± 2 doublet is less than the energy of the incident microwave photon ( = 0.314 cm−1 for 9.42 GHz radiation). The effective g-values (11.22 and 11.60) measured at the point where the spectrum crosses the zero-line for duplicate preparations of 1 and 2, respectively, are significantly greater than the spin-only g-value of 8. This is again diagnostic of large orbital angular momentum in the electronic ground state achieved either by orbital degeneracy of the ground state (as expected for high-symmetry molecules)20 or spin-orbit coupling (usually in the case of heavier element complexes). The ESR studies of 1 and 2 are thus more or less in harmony with the magnetic studies discussed above. That the CW ESR spectrum of the bent Fe(II) amido complex 2 is much broader than that for the linear complex 1 is likely due to more pronounced D-strain, i.e., structural inhomogeneity leading to a distribution of zero-field splitting values. This hypothesis is also supported by the X-ray structural data for 2 which shows greater thermal atomic motion than that in 1 (cf. thermal ellipsoids in Figures 1 and and2).2). As described earlier,43 the field position for the |2, +2right angle bracket ↔ |2, 2right angle bracket transition is a function of Δ which itself can be related to the axial (D) and rhombic (E) terms of the zero-field splitting tensor H^=D[S^z22+ED(S^x2S^y2)]. Small structural changes give rise to different values for D and E, altering the field position (B0) that meets the resonance condition and thus broadening the EPR line.

Figure 11
Parallel (hν = 9.42 GHz, top) and perpendicular (hν = 9.69 GHz, bottom) polarization CW ESR spectra of 0.01 M toluene solutions of linear 1 (left) and bent 2 (right) at 4.2 K. Modulation amplitude = 10 G, frequency = 100 kHz, time constant ...

The shift of the uniaxial g-value, gz from the free-electron value (2.0023) scales linearly with the expectation value of orbital angular momentum operator left angle bracketLzright angle bracket in the electronic ground state. Using literature formulas, the approximations given in reference 36, and the values for HL determined from our Mössbauer studies, we compute left angle bracketLzright angle bracket = 8.8κ for 1 and 6.8κ for complex 2, where κ is dimensionless constant related to the Fe s orbital contribution to the description of the ground state (usually 0.2 < κ < 0.4). If we assume κ for both complexes is the same (a fair assumption based on the similarity of the coordinating ligands) then clearly gz should be larger for complex 1. However, in the parallel-mode EPR experiment, the resonant condition is met when Δ2+(4gzβBz)2=hν, where Δ = 3D(E/D)2. Thus depending on the exact values for the principal elements of the ZFS tensor, the effective g-value for complex 2 could be larger than that for complex 1. Indeed, as it is considerably bent away from axial symmetry, we expect E/D to be relatively large for complex 2.

On the other hand, we note that the effective g-values derived from the 10 mM (and repeated at 1 mM) frozen toluene solution ESR spectra (Figure 11) may indeed not be fully consistent with the magnetization and Mössbauer spectroscopy results for the neat, undiluted polycrystalline solids. Needless to say, both types of data clearly suggest large orbital contributions for 1 and 2. However, the geff-value for the bent complex is larger than that of the linear analogue. In the absence of other data (primarily frozen solution applied field Mössbauer spectra for Fe57 enriched versions of 1 and 2), we speculate that, though unlikely, this result may suggest linearity for 2 in frozen solution owing to different packing effects for the frozen solution versus the neat polycrystalline solid.

DFT Calculations

The gas phase optimized structural parameters for M1 and M2, models based on the X-ray coordinates of 1 and 2 with their alkyl groups replaced by hydrogens, given in Table 4 reproduce most of the experimental bond lengths and bond angles reasonably well, cf. Table 2. The differences of a few degrees in the optimized and experimental N–Fe–N and C–Fe–N angles are most probably due to packing effects (which are always present with the large terphenyl ligands) that are not taken into account in the optimized gas phase structure. The quintet state was determined to be lowest in energy for both complexes by computing various spin states for M1 and M2, and for M1′ and M2′, simpler models with the Ar* and Ar# on nitrogen replaced by Ph. The corresponding energies are listed in Table 5. Calculations were also carried out on the M1/M2 system where the N–Fe–N angle was varied from 145 to 180°. The corresponding curves for triplet and quintet spin states are illustrated in Figure 12. The quintet ground state lies 33.9 kcal/mol lower than the triplet and 64.5 kcal/mol lower than the singlet state for the linear model. For the bent model the corresponding energy differences are 31.0 and 61.4 kcal/mol, respectively.

Figure 12
Relative energies (a.u.) computed for triplet and quintet spin states by varying the N–Fe–N angle in M1/M2 from 145° to 180°.
Table 2
Important interatomic distances (Å) and angles (°) for compounds 1 and 2.
Table 4
Selected optimized geometrical parameters (B3LYP/6–31g(d)) for the M1 and M2 models, which have the naked terphenyl C6H3–2,6-Ph2 as the Ar substituent on each nitrogen atom. Experimental values from X-ray data are given in parentheses. ...
Table 5
Computed energy (a.u.) for the quintet, triplet and singlet spin states of model compounds. Structures were applied from X-ray data to the linear M1 and bent M2 models (C6H3–2,6-Ph2 as the Ar substituent) and to M1′ and M2′ (Ph ...

The electronic structures of 1 and 2 are similar, with the exception of some interactions between the flanking aryl orbitals and metal-centered orbitals that occur mainly in the bent form, and are weaker and more distant in the linear form. The breaking of symmetry from a linear to a bent structure has the effect slightly increasing the HOMO-LUMO gap (68.3 kcal/mol for the linear and 77.6 kcal/mol for the bent structure).

In order to further check whether the quintet ground state of the iron amide obtained from DFT exhibits some multiconfigurational character, we have performed a CASSCF calculation on the linear model M1″ (in which the Ar* or Ar# are replaced by a methyl group) using Alhrichs double-ζ basis set with one polarization function. Linear geometry was predicted to be a potential minimum for this model. A total of 10 electrons were included in 14 active orbitals, resulting in overall 1249248 configurations. The computed occupation numbers illustrate the bonding, in which three unpaired electrons are located in dzy, dxy and dx2y2 metal-centered orbitals and the fourth unpaired electron resides in a nominally bonding orbital composed of iron dxz and two px AO’s of nitrogen (Figure 13). For the active space selected, the quintet ground state is highly monodeterminental, as evidenced by the weight of the leading CAS configuration (0.976) interaction of the Fe dxz orbital with the pz AO’s of nitrogen, an interaction which is symmetry forbidden in the linear form of this compound. Finally, the spin density is very similar in both cases, and is almost exclusively centered at the metal. A small spin delocalization is observed in both cases on the nearby nitrogen atoms, but no delocalization onto the aryl ligands occurs.

Figure 13
CAS orbitals for the model Fe{N(H)Me}2 (M1″), predicted to be linear, with the corresponding occupation numbers. The designation M indicates a metal localized orbital.

It is clear that the DFT calculations are in very good agreement with the experimental data. Most importantly, a ground state quintet is provided to be the most stable in both linear and bent structures. Furthermore, the unpaired electrons are localized for the most part on the iron d-orbitals such that they can provide the very large orbital moment in a linear geometry.


The main finding of the experimental and theoretical results in this paper is that the very large first order orbital angular momentum associated with the sterically fixed linear coordination geometry of the Fe(II) ion in 1 owes to the absence of first order Jahn- and Renner-Teller distortion effects that normally lift ground state electronic degeneracy. Thus the orbital angular momentum is greatly diminished by simple bending to a non-linear form. A direct comparison of this type was previously unavailable and is here permitted for the first time by the steric adaptability of the terphenyl amide ligands.

Table 1
Selected crystallographic data for compounds 1 and 2.

Supplementary Material





We are grateful to the National Science Foundation (Grant CHE-0641020) for financial support. W.M.R. is grateful for partial support by the NSF Division of Materials Research for the purchase of a SQUID magnetometer at Northeastern University. He also thanks Professor Eckard Münck (Carnegie-Mellon University) for informative discussions and applied field Mössbauer spectra of complex 2. The work in the laboratory of R.D.B. is supported by the National Institutes of Health (Grant GM 48242).


Supporting Information Available. X-ray crystallographic data (in CIF format) for 1 and 2 and further information on the DFT calculations are available free of charge.


1. Andersen RA, Carmona-Guzman E, Gibson JF, Wilkinson G. Dalton Trans. 1976:2204.
2. Buttrus NH, Eaborn C, Hitchcock PB, Smth JD, Sullivan AC. Chem Commun. 1985:1380.
3. Power PP. Comments Inorg Chem. 1989;8:177.
4. Power PP. Chemtracts Inorg Chem. 1994;6:181.
5. Wolf R, Brynda M, Ni C, Long GJ, Power PP. J Am Chem Soc. 2007;129:6076. [PubMed]
6. Bartlett RA, Chen H, Power PP. Angew Chem, Int Ed. 1989;28:316.
7. Chen H, Bartlett RA, Olmstead MM, Power PP, Shoner SC. J Am Chem Soc. 1990;112:1048.
8. Nguyen T, Panda A, Olmstead MM, Richards AF, Stender M, Brynda M, Power PP. J Am Chem Soc. 2005;127:8545. [PubMed]
9. Andersen RA, Berg DJ, Fernhold L, Faegri K, Green JC, Haaland A, Leung W-P, Rypdal K. Acta Chem Scand. 1988;42a:554.
10. (a) Andersen RA, Faegri K, Green JC, Haaland A, Lappert MF, Leung WP, Rypdal K. Inorg Chem. 1988;27:1782. (b) Olmstead MM, Power PP, Shoner SC. Inorg Chem. 1991;30:2547.
11. Chen H, Bartlett RA, Dias HVR, Olmstead MM, Power PP. J Am Chem Soc. 1989;111:4338.
12. Ellison JJ, Ruhlandt-Senge K, Power PP. Angew Chem, Int Ed. 1994;33:1178.
13. Ellison JJ, Ruhlandt-Senge K, Hope H, Power PP. Inorg Chem. 1995;34:49.
14. Wehmschulte RJ, Power PP. Organometallics. 1995;14:3264.
15. Kays DL, Cowley AR. Chem Commun. 2007:1053. [PubMed]
16. Müller H, Seidel W, Görls M. Angew Chem, Int Ed. 1995;36:325.
17. Bartlett RA, Power PP. J Am Chem Soc. 1987;109:7563.
18. Au-Yueng HY, Lam CH, Lam CK, Ang WY, Lee HK. Inorg Chem. 2007;46:7695. [PubMed]
19. La Pointe AM. Inorg Chem Acta. 2003;345:359.
20. Reiff WM, La Pointe AM, Witten EH. J Am Chem Soc. 2004;126:10206. [PubMed]
21. Reiff WM, Schulz CE, Whangbo MH, Seo JI, Lee YS, Potratz GR, Spicer CW, Girolami GS. J Amer Chem Soc. 2009;131:404–405. [PubMed]
22. Pangborn AB, Giardello MA, Grubbs RH, Rosen RK, Timmers FJ. Organometallics. 1996;15:1518.
23. Tilley TD, Gavenonis J. Organometallics. 2000;21:5549.
24. Twamley B, Hwang CS, Hardman NJ, Power PP. J Organmet Chem. 2000;609:152.
25. Ruhlandt-Senge K, Ellison JJ, Wehmschulte RJ, Pauer F, Power PP. J Am Chem Soc. 1993;115:11353.
26. Ghosh AK, Bischoff A, Cappiello A. Eur J Org Chem. 2003:821.
27. Hope H. Prog Inorg Chem. 1995;41:1.
28. Blessing RH. Acta Crystallogr. 1995;51A:33.
29. Sheldrick GM. SADABS, version 2.10, Siemens Area Detector Absorption Correction. Universität Göttingen; Göttingen, Germany: 2003.
30. Sheldrick GM. SHELXTL, version 6.1. Bruker AXS, Inc; Madison, WI: 2002.
31. Frisch MJ, et al. Gaussian 03, revision A.1. Gaussian, Inc; Pittsburgh, PA: 2003.
32. Bersuker IB. Chem Rev. 2001;101:1067. [PubMed]
33. Wang SG, Schwarz WHE. J Chem Phys. 1998;109:7252.
34. Drago RS. Physical Methods for Chemists. Harcourt Brach Jovanovich; New York: 1993. p. 484.
35. Pavel EG, Kitajima N, Solomon EI. J Am Chem Soc. 1998;120:3949.
36. Andres H, Bominar EL, Smith JM, Eckert NA, Holland PL, Münck E. J Am Chem Soc. 2002;124:3012. [PubMed]
37. Figgis BN, Lewis J, Mabbs FE, Webb GA. J Chem Soc A. 1967:442.
38. Kahn O. Molecular Magnetism. VCH-Wiley; New York: 1993. pp. 31–52.
39. Greenwood NN, Gibb TC. Mössbauer Spectroscopy. Chapman and Hall; London: 1971.
40. La Macchia G, Gagliardi L, Power PP, Brynda M. J Am Chem Soc. 2008;130:5104. [PubMed]
41. Reiff, W. M. Unpublished results for Fe{C(SiMe3)3}2.
42. Dadi D, Whangbo MH. Inorg Chem. 2005;44:4407. [PubMed]
43. Surerus KK, Hendrich MP, Christie PD, Rottgardt D, Orme-Johnson WH, Münck E. J Am Chem Soc. 1992;114:8579.
44. Merrill WA, Olmstead MM, Wright RJ, Power PP. Manuscript in Preparation, Abstracts of Papers, 229th ACS National Meeting; San Diego, CA, United States. March 13–17; 2005.