Effects of ligand topology and iron coordination number on the electronic structure of Fe^III-μ-oxo-Cr^III complexes supported by tetramethylcyclam

Abstract

⁵⁷Fe Mössbauer, electron paramagnetic resonance, and ⁵⁷Fe nuclear resonance vibrational spectroscopies are applied to characterize three different Fe^III–O–Cr^III complexes derived from the tetramethylcyclam (TMC, 1,4,8,11-tetramethyl-1,4,8,11-tetraazacyclotetradecane) ligand. [(CH₃CN)(TMC)Fe^III-O_anti-Cr^III(OTf)₄(NCCH₃)] (1) features an anti configuration where the bridging O-atom is located on the opposite face of all four methyl groups of TMC, which leads to a six-coordinate iron center. Similarly, due to the presence of a pyridine appended to one of the methyl groups of TMC, [(TMC-Py)Fe^III–O_anti–Cr^III(OTf)₄(NCCH₃)] (3) also exhibits an anti configuration with a six-coordinate iron center. But for [(TMC)Fe^III–O_syn–Cr^III(OTf)₄(NCCH₃)] (2), the iron center is five-coordinate due to a syn configuration where the bridging O atom is on the same face of all four methyl groups of TMC. The detailed spectroscopic characterizations reported here reveal that all three complexes exhibit S = 1 spin ground states, which result from antiferromagnetic coupling between an S = 5/2 Fe^III center and an S = 3/2 Cr^III center. However, the detailed magnetic properties derived from Mössbauer and EPR measurements and vibrational properties of these complexes, particularly the symmetric and asymmetric Fe-O-Cr stretching modes, closely reflect the ligand topology difference (anti vs. syn) and the coordination number of the iron center. These spectroscopic properties are used to guide the density functional theory calculations to show how the ligand topology affects the electronic structures of these complexes. Thus, the current study provides a comprehensive geometric and electronic structure correlation of these unique hetero-dinuclear complexes.

1. Introduction

Enzymes containing dinuclear metallocofactors play important roles in catalyzing redox reactions or transferring electrons in nature. In those reactions, the two transition-metal ions are often structurally connected through bridging ligands (such as oxo, hydroxo, and/or carboxylic ligands) and cooperatively generate reactive intermediates. Typical enzyme examples include soluble methane monooxygenase (sMMO) with a diiron center [1–3], tyrosinase with dicopper center [4,5], flavodiiron enzyme (FDP) with diiron center [6,7], ribonucleotide reductase (RNR) with a diiron center in Class 1a RNR, a dimanganese center in Class 1b RNR and an iron-manganese hetero-dinuclear center in Class 1c RNR [3,8–10]. In addition to the iron-manganese hetero-dinuclear center found in Class 1c RNR, other types of hetero-dinuclear centers also exist in nature, such as heme-copper oxidase [5,11]. Oxygen Evolving Center (OEC) in photosystem II (PS II) is another example where a Mn–O–Ca moiety of OEC has been found to play a key role in O–O bond formation during water oxidation reaction [12–15]. These structurally and functionally unique dinuclear metallocofactors have captivated inorganic chemists and spectroscopists for a long time. In order to understand the structure-function relationships of these metallocofactors, many synthetic complexes mimicking the natural occurring analogues have been synthesized and spectroscopically characterized [3,16–19]. Among them, μ-oxo-bridged-heterodimetallic complexes have been the focus of many studies due to their unique asymmetric molecular structure and their relevance to the similar heterodimetallic centers found in heme-copper oxidase, RNR and OEC of PSII.

For a long time, different strategies have been proposed and investigated to prepare μ-oxo-bridged-heterobimetallic molecules supported by non-heme ligands. In general, the synthesis strategies can be cataloged in two classes: non-redox and redox reactions. The former approach was exemplified by Wieghardt et al., who synthesized the Fe^III-O-M^III (M = Mn, Cr) complexes via hydrolysis between starting materials: mononuclear Fe^III complexes and a M^III complex supported by 1,4,7-triazacyclononane (TACN) and 1,4,7-trimethyl-1,4,7-triazacyclononane (Me₃TACN) [20,21]. The μ-oxo ligand is derived from H₂O. As another example, the Que. group used 2,6-bis[(bis(2-pyridylmethyl) amino)methyl]-4-methylphenol (BPMP) as the ligand to synthesize an Fe^II-Mn^II complex by adding Fe and Mn salt into the solution containing BPMP ligand [22]. The Fe-O-Mn core is supported by a phenoxy bridge. The latter approach, the redox reaction, has been used to synthesize Fe^III-O-Cr^III complexes, and can be further cataloged into two sub-branches: 1) use Fe^IV-oxo as the oxidant and Cr^II as the reductant; or 2) use Cr^IV-oxo as the oxidant and Fe^II as the reductant. The first strategy has been successfully developed by the Que. group based on their extensive expertise on handling Fe^IV-oxo species supported by various non-heme ligands [23,24]; the other approach, with longer history, was developed by Oyaizu et al. and Liston et al. on di-porphyrin systems [25,26]. All of these Fe-O-Cr complexes showed attractive properties on optical and magnetic spectroscopies, but detailed discussions on the connections between structure/ligand topology and electronic properties of these complexes are rare.

Herein, we present detailed spectroscopic characterizations of three different Fe^III-oxo-Cr^III complexes (Scheme 1) that are derived from tetramethylcyclam (TMC, 1,4,8,11-tetramethyl-1,4,8,11-tetraazacyclotetradecane) ligand with orientation-selective reactions and different coordination numbers on iron [23,24]. These complexes are [(CH₃CN) (TMC)Fe^III-O_anti-Cr^III(OTf)₄(NCCH₃)] (1), [(TMC)Fe^III-O_syn-Cr^III(OTf)₄(NCCH₃)] (2), [(TMC-Py)Fe^III-O_anti-Cr^III(OTf)₄(NCCH₃)] (3). The spectroscopic analyses derived from Mössbauer, electron paramagnetic resonance, and nuclear resonance vibrational spectroscopy suggest that the spectroscopic properties and the electronic structures of these complexes can be closely correlated with the coordination numbers of the iron center (5-coordinate vs. 6-coordinate) and the topology of the supporting ligand.

Scheme 1.

Molecular structures of the three Fe^III-O-Cr^III complexes 1, 2, 3.

2. Experimental section

2.1. Materials

All reagents and solvents were purchased from commercial sources and used as received unless specified. The preparations for Cr^II(OTf)₂•2CH₃CN [23], Fe^II(TMC)(OTf)₂ [27], Fe^II(TMC-Py)(OTf)₂ [28], iodosylbenzene (C₆H₅IO) [29], and 2-(tert-butylsulfonyl) iodosylbenzene (2-^tBuSO₂-C₆H₄IO) [30,31] were carried out following the published procedures. All moisture- and oxygen-sensitive compounds were prepared using standard Schlenk-line techniques. Complexes 1 and 2 were obtained by following the previous published procedures [23,24]. Complex 3 was obtained by following a similar synthetic procedure published in [32,33] with additional details shown in Fig. S1 in the supplementary information. All the samples are ⁵⁷Fe enriched complexes (~ 90 % ⁵⁷Fe enrichment) in frozen CH₃CN solution with the Fe concentrations of 3.0 mM (complex 1), 2.8 mM (complex 2) and 3.5 mM (complex 3) for the Mössbauer and EPR samples and 4.6 mM (complex 1) and 3.5 mM (complex 2) for the NRVS samples.

2.2. Instrumentation

Elemental analyses were carried out by Atlantic Microlab (Norcross, GA, USA). UV – vis absorption spectra were recorded on an HP 8453 diode array spectrometer. Low temperature visible spectra were obtained using a cryostat from UNISOKU Scientific Instruments, Japan.

Mössbauer spectra were recorded with home-built spectrometers using Janis Research Super-Varitemp dewars. The applied magnetic fields are parallel to the γ radiation. Mössbauer spectral simulations were performed using the WMOSS software package (SEE Co., Edina, MN, USA). Isomer shifts are quoted relative to Fe metal at 298 K. Parallel (9.36 GHz) mode X-band EPR spectra were recorded on a Bruker E500 spectrometer equipped with an Oxford ESR 910 cryostat and a Bruker bimodal cavity. The modulation amplitude and frequency were 1 mT and 100 kHz for all spectra. The microwave frequency was calibrated with a frequency counter and the magnetic field was calibrated with a calibrated Hall probe. The temperature was calibrated with a carbon glass resistor. EPR signals were quantified by relative to a 1.2 mM Cu(II) EDTA standard in 10 % glycerol. The Mössbauer and EPR spectral simulations were performed by using SpinCount software. All Mössbauer and EPR figures were prepared using SpinCount software [34].

⁵⁷Fe nuclear resonance vibrational spectroscopy experiments were conducted at beamline 3-ID-B at the Advanced Photon Source (APS), Argonne National Laboratory, IL, USA. The instrument was equipped with a liquid helium closed-cycle cryostat and the sample was cooled with a copper cold finger, which was maintained at a cryogenic temperature of approximately 20 K. The temperature of the sample, as calculated from individual spectra using the ratio of anti-Stokes to Stokes intensity according to $S\left(E\right)=S(-E)\mathrm{e}\mathrm{x}\mathrm{p}.(E/kT)$, were generally in the range of 40 K to 80 K. The monochromators provide a high energy resolution of ~1 meV by employing a cryo-cooled silicon (1,1,1) double crystal with 1.1 eV bandpass, followed by two separate Si(4,0,0) and Si (10,6,4) channel-cut crystals in a symmetric geometry [35]. Single avalanche photodiode (APD) detector was used at both forward and perpendicular direction. A solid K₂MgFe(CN)₆ sample was used as energy calibration standard with a reference point of its Fe-C(ν7) vibrational mode reported at 74 meV [36,37]. The resolution function was collected by recording the delayed nuclear fluorescence at the forward direction simultaneously when the scattering photons were recorded at the perpendicular direction. The spectra were recorded between −50 meV and + 120 meV, 3 s/step for each scan with a step size of 0.25 meV. Each scan required about 50 min, and all scans were added and normalized to the intensity of the incident beam. In total, 10 scans and 16 scans have been carried out for complexes 1 and 2, respectively, to obtain enough data statistics. The program PHEONIX-3.0.3 was employed to process the data and derive the ⁵⁷Fe PVDOS [38].

2.3. Computational details

Density functional theory calculations were carried out using Gaussian 16 Rev. C01 software package [39]. Geometry optimizations were performed using broken-symmetry DFT method, in which the initial guess was set as a high-spin ferric center antiferromagnetically coupled with a high-spin chromic center, yielding a total spin state of $S=1$. All subsequent analyses were based on the optimized geometries. The Mössbauer isomer shift values for ⁵⁷Fe were computed using our self-developed calibration scheme [40] based on the B3LYP functional [41,42] and the 6–311G basis sets. For all other calculations, the B3LYP functional with Grimme’s D3 dispersion correction and a conductor-like polarizable continuum model (CPCM) for acetonitrile was employed [43]. All atoms were described using Ahlrichs’ def2TZVP basis set [44,45]. The magnetic coupling constant was calculated using Yamaguchi approach [46].

$$J=\frac{2\left(E_{FM}-E_{BS}\right)}{{⟨{\widehat{S}}^2⟩}_{FM}-{⟨{\widehat{S}}^2⟩}_{BS}}$$ (1)

Frequency analysis was followed by optimization to generate normal modes of vibration and their associated frequencies. The ⁵⁷Fe PVDOS was generated by using an in-house program that extracts the Cartesian normal mode displacements obtained by DFT frequency calculations and converts them into ⁵⁷Fe mode composition factor $e_{Fe,\alpha }^2$ based on the following equation [47]:

$$e_{\mathrm{F}\mathrm{e},\alpha }^2=\frac{m_{\mathrm{F}\mathrm{e}}{r}_{\mathrm{F}\mathrm{e},\alpha }^2}{{\sum }_{\alpha }{m}_i{r}_{i,\alpha }^2}$$ (2)

where $m_i$ and $r_{i,\alpha }^2$ are the mass of atom $i$ and its mean square motion in mode $\alpha$. The obtained ⁵⁷Fe mode composition factor was further used to reconstruct ⁵⁷Fe PVDOS spectra to compare with experimental data according to the following equation [48]:

$$D_{\mathrm{F}\mathrm{e}}\left(\bar{\nu }\right)={\sum }_{\alpha }{e}_{\mathrm{F}\mathrm{e},\alpha }^2\ell \left(\bar{\nu }-\overline{{\nu }_{\alpha }}\right)$$ (3)

where the line shape function was generated by using a $10{\mathrm{c}\mathrm{m}}^{-1}$ gaussian broadening to mimic the experimental energy resolution, as well as the line broadening due to the structural inhomogeneity in frozen solutions.

3. Results

3.1. Mössbauer and EPR analysis on complexes 1, 2 and 3

To determine the electronic structures of complexes 1, 2, and 3, we first carried out Mössbauer measurements under variable field and variable temperature conditions. The Mössbauer data were analyzed by using an S = 1 spin Hamiltonian that assumes a strong magnetic coupling between an $S=5/2$ Fe^III and an $S=3/2$ Cr^III spin center. The spin Hamiltonian is described by Eq. (4a–4c). Since the hyperfine interaction is much smaller than zero-field splitting (ZFS), second-order perturbation theory can be applied, allowing the electronic spin part to be solved prior to the nuclear spin part. Unless specified otherwise, all Mössbauer simulations were carried out with the slow relaxation limit approximation.

$$\widehat{H}=\widehat{\mathit{S}}\bullet \mathit{D}\bullet \widehat{\mathit{S}}+\beta \mathit{B}\bullet \mathit{g}\bullet \widehat{\mathit{S}}+\widehat{\mathit{S}}\bullet \mathit{A}\bullet \widehat{\mathit{I}}+\frac{eQ}{6I(2I-1)}\widehat{\mathit{I}}\bullet \mathit{V}\bullet \widehat{\mathit{I}}-g_n{\beta }_n\mathit{B}\bullet \widehat{\mathit{I}}$$ (4a)

where $\widehat{S}$: electron spin operator, with Cartesian components ${\widehat{S}}_x,{\widehat{S}}_y,{\widehat{S}}_z;\mathit{D}$: zero-field splitting (ZFS) tensor, with diagonal elements $D_{\mathrm{x}\mathrm{x}},D_{\mathrm{y}\mathrm{y}},D_{\mathrm{z}\mathrm{z}};\beta$: Bohr magneton; $\mathit{g}$: electron g-tensor; $\mathit{B}$: external magnetic field vector; $\mathit{A}$: hyperfine tensor; $\widehat{\mathit{I}}$: nuclear spin operator, with Cartesian components ${\widehat{I}}_x,{\widehat{I}}_y,{\widehat{I}}_z;e$: elementary charge; $Q$: nuclear electric quadrupole moment; $I$: nuclear spin quantum number; $\mathit{V}$: electric field gradient (EFG) tensor, with diagonal elements $V_{\mathrm{x}\mathrm{x}},V_{\mathrm{y}\mathrm{y}},V_{\mathrm{z}\mathrm{z}};g_n$: nuclear g-factor; ${\beta }_n$: nuclear magneton.

$${\widehat{H}}_e=D\left[{\widehat{S}}_z^2-\frac{1}{3}S(S+1)+\frac{E}{D}\left({\widehat{S}}_x^2-{\widehat{S}}_y^2\right)\right]+\beta \mathit{B}\bullet \mathit{g}\bullet \widehat{\mathit{S}}$$ (4b)

where $D=\frac{3}{2}{D}_{\mathrm{z}\mathrm{z}},E=\frac{1}{2}\left(D_{\mathrm{x}\mathrm{x}}-D_{\mathrm{y}\mathrm{y}}\right)$ with $0\le \frac{E}{D}\le \frac{1}{3}$.

$${\widehat{H}}_{Nuc}=<\widehat{\mathit{S}}>\mathit{A}\bullet \widehat{\mathit{I}}+\frac{eQ{V}_{zz}}{12}\left[3{\widehat{I}}_z^2-I(I+1)+\eta \left({\widehat{I}}_x^2-{\widehat{I}}_y^2\right)\right]-g_n{\beta }_n\mathit{B}\bullet \widehat{\mathit{I}}$$ (4c)

where $<\widehat{\mathit{S}}>$: expectation value of $\widehat{\mathit{S}}$ operator; $\eta =\frac{V_{\mathrm{x}\mathrm{x}}-V_{\mathrm{y}\mathrm{y}}}{V_{\mathrm{z}\mathrm{z}}}$ with $\left|V_{\mathrm{z}\mathrm{z}}\right|\ge \left|V_{\mathrm{y}\mathrm{y}}\right|\ge \left|V_{\mathrm{x}\mathrm{x}}\right|$.

The zero field Mössbauer spectra of all three compounds exhibited a quadrupole doublet with $\delta =0.53$ and $\left|\mathrm{\Delta }{\mathrm{E}}_{\mathrm{Q}}\right|=0.88\mathrm{m}\mathrm{m}/\mathrm{s}$ for 1, $\delta =0.43$ and $\left|\mathrm{\Delta }{\mathrm{E}}_{\mathrm{Q}}\right|=0.89\mathrm{m}\mathrm{m}/\mathrm{s}$ for 2, $\delta =0.51$ and $\left|\mathrm{\Delta }{\mathrm{E}}_{\mathrm{Q}}\right|=0.94\mathrm{m}\mathrm{m}/\mathrm{s}$ for 3 (Fig. 1). These parameters are typical of those from high-spin ($S=5/2$) ferric species. However, the observation of quadrupole doublets at 4.2 K indicates that these three compounds have an integer spin ground state ($S=0,1,2,\dots$). This is confirmed by variable magnetic field measurements. The spectra recorded on the sample containing compound 1 can be reasonably simulated with an $S=1$ spin Hamiltonian, representing ~80 % of the total iron in the sample (Fig. 1A). The $S=1$ spin state of compound 1 is most likely originated from an antiferromagnetic coupling between an S_Fe(III) = 5/2 Fe(III) and an S_Cr(III) = 3/2 Cr(III), thus consistent with compound 1 as a Fe^III–O–Cr^III dimer species. The axial zero-field splitting(ZFS) parameter ($D=8.3{\mathrm{c}\mathrm{m}}^{-1}$) can be well determined by the relative intensities of the spectral features originated from the $\mid -1^{\text{'}}>$ level (the features indicated by the black arrows in Fig. 1A) and the $\mid 0^{\text{'}}>$ level (the major magnetic features in the Fig. 1A). The ⁵⁷Fe hyperfine coupling tensor determined by the $S=1$ spin Hamiltonian is rather isotropic ($\mathit{A}/g_{\mathrm{n}}{\beta }_{\mathrm{n}}=(-36.5,-36.2,-36.0)\mathrm{T}$), which is consistent with the involvement of an S = 5/2 Fe(III) species in 1. By using the spin projection factor $\mathit{A}=7/4{\mathit{a}}^{\mathrm{F}\mathrm{e}(\text{III})}$ for antiferromagnetic coupled system with $S_{\mathrm{A}}=5/2$ and $S_{\mathrm{B}}=3/2$ [49], the intrinsic ⁵⁷Fe hyperfine coupling constants can be deduced to be ${\mathit{a}}^{\mathrm{F}\mathrm{e}(\text{III})}/g_{\mathrm{n}}{\beta }_{\mathrm{n}}=(-20.8,-20.6,-20.5)\mathrm{T}$. This local A value further confirms the nature of the iron center in 1 as an S = 5/2 Fe(III). The variable field measurements also revealed the sign of $\mathrm{\Delta }{\mathrm{E}}_{\mathrm{Q}}$ and the asymmetry parameter $\eta$ for 1 (see Table 1 for Mössbauer parameters determined from spectral simulations). Under the same strong coupling assumption, our spectroscopic simulations with $S=1$ spin Hamiltonian showed that complex 3 have very similar spin Hamiltonian parameters as those of complex 1 with $\mathrm{D}=8.1{\mathrm{c}\mathrm{m}}^{-1}$ and the ⁵⁷Fe hyperfine coupling constants as $\mathit{A}/g_{\mathrm{n}}{\beta }_{\mathrm{n}}=(-36.5,-36.2,-36.0)\mathrm{T}$ with the corresponding intrinsic ⁵⁷Fe hyperfine coupling constants as ${\mathit{a}}^{\mathrm{F}\mathrm{e}(\text{III})}/g_{\mathrm{n}}{\beta }_{\mathrm{n}}=(-20.8,-20.6,-20.5)\mathrm{T}$ (Fig. 1C and Table 1). These results not only confirm compound 3 as another Fe^III-O-Cr^III dimer species, and the highly similar Mössbauer parameters exhibited by compounds 1 and 3 are also consistent with the fact that the similar pseudo-C_4v 6-coordinate geometry around the Fe^III centers is found in 1 and 3.

Fig. 1.

4.2 K Mössbauer spectra measured on the samples containing compound 1 (A), 2 (B) and 3 (C) under different applied magnetic fields indicated in the figure. The applied magnetic fields are parallel to the γ radiation. Black vertical bars represent the experimental data, the red solid lines represent the spectral simulation (See Table 1 for simulation parameters). The arrows indicates the spectral features originated from the excited $|-1^{\prime }>$ level of the $S=1$ spin manifold. The grey solid lines in panel B are spectral simulations of compound 1 that represents 20 % absorption area. The grey solid line in panel B is spectral simulation of a high-spin ferrous species ($\delta =0.95\mathrm{m}\mathrm{m}/\mathrm{s},\left|\mathrm{\Delta }{\mathrm{E}}_{\mathrm{Q}}\right|=2.76\mathrm{m}\mathrm{m}/\mathrm{s}$), representing ~15 % absorption area. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Table 1.

Mössbauer and EPR simulation parameters obtained by using an S = 1 spin hamiltonian.^#

	Complex	D (cm−1)	E/D%	δ (mm/s)	ΔEQ (mm/s)	η	[Ax, Ay, Az]/gnβn (T)
Mössbauer	1	8.3	<0.05	0.53 (0.51)	−0.88 (−0.82)	0.8 (0.35)	−36.5, −36.2, −36.0
	2	11.9	<0.15	0.43 (0.45)	−0.89 (−0.83)	0.2& (0.23)	−32.4, −34.4, n.d.
	3	8.1	<0.05	0.51 (0.49)	−0.94 (−0.88)	0.7& (0.37)	−35.5, −35.5, −35.5
		D (cm−1)	E/D	σ(D)*	σ(E/D)*	LW(Gauss)	Concentration (mM)
EPR	1	7.0	0.02	2.0	0.016	9.0	3.0

^%For complexes 1 and 3, the simulation parameters listed in the table were obtained by using E/D = 0.02. For complex 3, the simulation parameters were obtained by using E/D = 0.

^#The italic values are derived from DFT calculations.

^&Euler angle in the zyz convension is used for the EFG tensor with α = 90°.

^*We applied a Gaussian distribution on these parameters to obtain a more natural simulation result. Hereby the σ indicates the standard deviation of that parameter.

In addition, there is 20 % of unidentified species in the sample containing 1. Currently, we do not have good explanation for this additional species. Its spectroscopic features cannot be satisfactorily simulated as originated from complex 2. The possibility of the diamagnetic di-ferric species was also considered, but the data do not support such an assignment.

The Mössbauer data of complex 2 are distinct from those of 1 and 3. First, the variable field measurements on the sample containing compound 2 revealed that this sample contained ~20 % of 1, which is clearly demonstrated in Fig. 1B. Apart from the difference in isomer shift (0.43 mm/s for 2 vs. 0.53/0.51 mm/s for 1/3), no excited state signals can be observed in the 4.2 K Mössbauer data of 2 under all field conditions we measured. This indicates that 2 must have a larger D value than those of 1 and 3. Our spectroscopic analysis by using an $S=1$ spin Hamiltonian confirmed this observation, estimating D ~ 12.0 cm⁻¹ for 2. In addition, the simulation also determined the ⁵⁷Fe hyperfine coupling tensor of 2 to $\mathit{A}/g_{\mathrm{n}}{\beta }_{\mathrm{n}}=(-32.4,-34.4$, n.d.) T. By assuming a strong antiferromagnetic coupling between Fe(III) and Cr(III) for 2, the deduced intrinsic ⁵⁷Fe hyperfine coupling constants to be ${\mathbf{a}}^{\mathrm{F}\mathrm{e}(\text{III})}/g_{\mathrm{n}}{\beta }_{\mathrm{n}}$ = (−18.5, −19.6, n.d.) T. For complex 2, the principal value of the hyperfine tensor along the z direction (${\mathrm{A}}_{\mathrm{z}}$) cannot be determined (n.d.) by using the spectra measured at 4.2 K. The magnetic splittings of 2 observed are proportional to the internal field, ${\mathbf{B}}_{\text{int}}=-<\mathit{S}>\cdot \mathbf{A}/g_{\mathrm{n}}{\beta }_{\mathrm{n}}$. At 4.2 K with large positive D, only the $|0\text{'}>$ spin level is thermally populated at 4.2 K, and for the $\left|0^{\text{'}}\right.⟩$ level even at a strong applied magnetic field condition, such as 7 T used in this study, therefore, A_z cannot be determined. The differences between Mössbauer parameters (e.g. isomer shift, zero-field splitting) exhibited by complex 2 and those in complexes 1 and 3 are consistent with their differences in structure. As previously demonstrated, Fe(III) is five-coordinated in 2 with μ-oxo located on the same face of all four methyl groups from TMC (syn isomer), whereas in 1 and 3, Fe(III) is six-coordinated with μ-oxo located on the opposite face of all four methyl groups (anti isomer).

To further validate our characterization of the axial zero-field splitting parameter (D) of complexes 1 and 3, we measured their Mössbauer spectra at elevated temperatures above 4.2 K (Fig. 2). As expected, the intensities of the spectral features originated from the excited states $\left(\left|-1^{\prime }\right.⟩,\left|+1^{\prime }\right.⟩\right.$ states of the S = 1 spin manifold are increased due to their increased population at elevated temperatures. The D values determined by the spectral data measured at 4.2 K can satisfactorily reproduce the relative intensities of the spectral features originated from the ground state and the two excited states recorded at 10 K and 15 K. This indicates the reliability of the D values determined by Mössbauer data collected at 4.2 K. We also noticed that the spectra recorded above 10 K exhibited partial intermediate relaxation. This is evidenced by the mismatch of the intensities between the spectral features from the $\left|+1^{\prime }\right.⟩$ state and the corresponding simulations (see Fig. 2) and the additional spectral area at the center of the spectra that is not completely reproduced by the simulations. Despite these mismatches, we can still estimate the D values of complex 1 and 3 by the population of $\left|-1^{\prime }\right.⟩$ state based on the slow relaxation approximation (see Fig. 2).

Fig. 2.

Mössbauer spectra measured on the samples containing compound 1 (A) and 3 (B) under 4 T applied magnetic field and at 10 K or 15 K as indicated in the figure. The applied magnetic fields are parallel to the γ radiation. Black vertical bars represent the experimental data, the red solid lines represent the spectral simulation (See Table 1 for simulation parameters). The arrows indicate the spectral features originated from the excited $\left|-1^{\prime }\right.⟩$ and $\left|+1^{\prime }\right.⟩$ levels of the $S=1$ spin manifold. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Finally, Mössbauer spectra of this type of $S=1$ spin systems cannot accurately determine the E/D values but can provide an upper limit. For the $\mid 0^{\prime }>$ spin sublevel, the ground spin sublevel observed here ($D>0$), E/D controls the anisotropy of the x and y direction of the spin expectation values ($<{\mathit{S}}_{\mathrm{x}}>$ and $<{\mathit{S}}_{\mathrm{y}}>$). Thus, E/D value correlates with the anisotropy of the x and y principal components of the ⁵⁷Fe A tensors ($A_{\mathrm{x}}$ and $A_{\mathrm{y}}$) in the simulation since the internal field observed by Mössbuaer spectra can be expressed as ${\mathbf{B}}_{\text{int}}=-<\mathit{S}>\cdot \mathbf{A}/g_{\mathrm{n}}{\beta }_{\mathrm{n}}$. For the $\left|+1^{\prime }\right.⟩$ and $\left|-1^{\prime }\right.⟩$ spin sublevels, the excited sublevels observed here, E/D also controls the magnetization behavior of the z direction of the spin expectation value (), thus the magnetization behavior of the Mössbauer features originated from the $\left|+1^{\prime }\right.⟩$ and $\left|-1^{\prime }\right.⟩$ spin sublevels bears information on E/D. For complexes 1 and 3, the spectral features from all three spin sublevels are observed (Figs. 2 and 3), therefore, E/D values can be better estimated to be 0 < E/D < 0.05. For complex 2, only the spectral features originated from the $\mid 0^{\prime }>$ sublevel are observed, therefore the E/D value is less defined (E/D < 0.15).

Fig. 3.

Parallel-mode EPR spectra of complexes 1 (A) at various temperature with 2 mW microwave power. Black trace: experimental data; red trace: simulation results with the parameters listed in Table 1–S1. The small resonance signal around g = 8 (pointed by black arrows in C) has been shown to be the S = 2 Cr^II starting materials (See Fig. S3 in Supporting Information). The “signal*T vs T” plot is shown in (B) The thick dot represents the actual data point and the dashed line represents the simulation curves with a D = 7 cm⁻¹. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Based on D and E/D values determined by Mössbauer spectroscopy for complex 1, 2 and 3, we anticipated that 1 and 3 could exhibit an $S=1$ integer spin EPR signal at X-band frequency (9.36 GHz). For an $S=1$ spin system, the most likely observed EPR transition at X-band frequency should be originated from the $|\pm 1^{\prime }\rangle$ levels of the $S=1$ spin manifold. However, $S=1$ X-band EPR signal is generally difficult to observe [50]. The energy gap between the $|\pm 1^{\prime }\rangle$ levels is equal to $\Delta =2D(E/D)=2E$, and the transition probability between these two levels is proportional to ${\mathrm{\Delta }}^2=4{\mathrm{E}}^2$. It means that Δ needs to be smaller than the X-band microwave energy (0.3 cm⁻¹) to observe an EPR transition at the X-band frequency for $S=1$ complexes, which is the case for complexes 1 and 3 based on Mössbauer derived D and E/D value. In the same time, E cannot be too small, otherwise the transition probability would be too small to enable an observable signal.

To test whether $S=1$ EPR signal can be observed for these complexes, we recorded EPR spectra of complexes 1 at various temperatures. The EPR spectra displayed a broad feature centered at g ~ 4, which is consistent with an $S=1$ EPR signal due to the transition between the two states $|-1^{\prime }\rangle$ and $|+1^{\prime }\rangle$ sublevels (Fig. 3). We ruled out the possibility that the observed signal was originated from molecular O₂ that may be trapped in the sample (see Fig. S2). Temperature dependent measurements of these EPR signals confirmed that the D values as 7.0 cm⁻¹ for complex 1 by simulating the signal*T vs. T plot. With determined D values, we also directly simulated the EPR signals by fixing D value while varying E/D, σ(D), σ(E/D), line width and spin concentration. The final simulation parameters are listed in Table 1. These ZFS values obtained from EPR simulations are consistent with those obtained from the simulations of Mössbauer spectra. Furthermore, the spin concentrations obtained from EPR simulations are also consistent with the concentrations of complex 1 in the EPR samples, which further solidifies that the observed S = 1 EPR signal comes from 1.

To better connect the Mössbauer results with the molecular structures of these complexes, we conducted density functional theory (DFT) calculations. For complexes 1 and 2, the initial structures were adopted from the previously published DFT calculated structures of these two complexes (obtained by the TPSS/de2-TZVP level of theory [24]). For complex 3, the initial structure was obtained by modifying the structure of complex 1. Geometry optimizations were performed by using broken-symmetry (BS) DFT with the computational details described in the experimental method section. The initial guess and the final result are consistent with a BS state that one S = 5/2 Fe^III is antiferromagnetically coupled with one S = 3/2 Cr^III, yielding a total spin of 1 for all three complexes. Our DFT optimized structures (Fig. 4) are highly similar with those of the previously published DFT results and the Fe-ligand bond metrics are also consistent with those determined by the published EXAFS data on complexes 1 and 2 (Table 2) [23,24]. For example, d[Fe-μO] is calculated to be 1.79 Å for complex 1 and 1.74 Å for complex 2. The shortening of this bond length in 2 is due to the 5-coordinate Fe center, which misses an axial ligand trans to the bridging O atom and features a syn configuration (the O atom is in the same face as the four methyl groups of the TMC ligand). In contrast, the longer Fe-μO bond length in 1 is due to the 6-coordinate Fe center, which exhibits a MeCN ligand trans to the bridging O atom and features an anti configuration (the O atom is in the opposite face as the four methyl groups of the TMC ligand). The difference in iron coordination number in 1 and 2 also leads to a different Fe–Cr distance (d[Fe—Cr]) with the calculated 3.63 Å in 1 and 3.54 Å in 2, and a different in Fe-O-Cr angle (∠[Fe-O-Cr]) with the calculated ~180 degree in 1 and ~ 170 degree in 2. As a comparison, based on EXAFS analysis, d[Fe-μO] has been determined to be 1.81 Å in 1 and 1.77 Å in 2, and d[Fe—Cr] has been determined to be 3.65 Å in 1 and 3.58 Å in 2. For complex 3, although no crystal structure or EXAFS data is available, the DFT optimized structure shows many similarities to that of complex 1 (Table 2). Particularly, d[Fe-μO], d[Fe—Cr], and ∠[Fe-O-Cr] are almost identical for complexes 1 and 3 and are distinct from those of complex 2, which suggest the topology of the TMC ligand directly governs the overall complex structures. (See Fig. 4.)

Fig. 4.

DFT optimized structures of complexes 1 (left), 2 (middle), and 3 (right). The Fe-O-Cr angle and the Fe—O bond length are shown for each structure. For clarity, all hydrogen atoms are omitted and all carbon atoms and fluorine atoms are shown in sticks.

Table 2.

Computationally and experimentally obtained structural metrics of complexes 1 and 2.

	DFT (B3LYP)#			DFT (TPSS)*		EXAFS*
	1	2	3	1	2	1	2
d[Fe-Neq]	2.24, 2.24, 2.20, 2.20	2.23, 2.24, 2.19, 2.18	2.24, 2.27, 2.18, 2.17	2.22, 2.22, 2.19, 2.19	2.21, 2.21, 2.17, 2.17	–	–
d[Fe-Neq] avg	2.22	2.21	2.22	2.21	2.19	–	2.16
d[Fe-Nax]	2.32	–	2.41	2.26	–	–	–
d[Fe-N] avg	2.24	–	2.25	2.22	–	2.17	–
d[Fe—O]	1.79	1.74	1.80	1.82	1.76	1.81	1.77
d[Fe—Cr]	3.63	3.54	3.64	3.62	3.52	3.65	3.58
∠[Fe-O-Cr]	179.4	171.0	179.4	177.1	172.6	–	–

^#This work.

^*[23,24].

Based on the DFT geometry optimized structures of these complexes, Mössbauer parameters, specifically, the isomer shift and quadrupole splittings were also calculated. The calculated values closely agree with the experimental ones (Table 1). In particular, all three complexes exhibit very similar quadrupole splittings (−0.88 mm/s for 1, −0.89 mm/s for 2, and − 0.94 mm/s for 3), which suggest that the total valence electrons on the Fe 3d-based molecular orbitals and the electron distribution on those orbitals are high similar irrespective to the iron center coordination numbers. Indeed, the Mulliken population analysis from the DFT calculations shows that the total electrons on all 5 Fe 3d-based molecular orbitals are almost the same for both complexes (5.93 for 1, 5.88 for 2, and 5.94 for 3) and the number of electrons on each orbital between these two complexes are also very similar (Table S1), which leads to highly similar calculated quadrupole splitting values (−0.82 mm/s for 1, −0.83 mm/s for 2, and −0.88 mm/s for 3), consistent with the experimental observations. Despite a similar electron distribution on the Fe 3d orbitals for three complexes, the isomer shift of complex 2 is quite different from that of complexes 1 and 3 (0.53 mm/s for 1, 0.51 mm/s for 3, and 0.43 mm/s for 2 observed experimentally, and 0.51 mm/s for 1, 0.49 mm/s for 3, and 0.45 mm/s for 2 calculated by DFT). This difference is most likely due to the shorter Fe-μO bond length observed for the 5-coordinate 2 (1.74 Å) than that for the 6-coordinate 1 and 3 (1.79 Å and 1.80 Å). The shorter bond length may lead to higher direct s electron donation to the iron center, thus driving the isomer shift to a lower value. The magnetic coupling constant (J) between the Fe and the Cr centers were also calculated, which yielded J = 230 cm⁻¹ for 1, J = 276 cm⁻¹ for 2, and J = 220 cm⁻¹ for 3 (in $H_{ex}=J{\widehat{\mathit{S}}}_{Fe}\bullet {\widehat{\mathit{S}}}_{Cr}$ notation). This strong antiferromagnetic coupling between the two metal centers is expected for the μO bridged di-metal complexes and further justifies the use of a total S = 1 spin Hamiltonian for the Mössbauer and EPR analyses.

3.2. Nuclear resonance vibrational spectroscopy (NRVS) on 1 and 2

To gain more insights of molecular geometry of these complexes, ⁵⁷Fe partial vibrational density of state (PVDOS) have been measured on complexes 1 and 2 by using ⁵⁷Fe nuclear resonance vibrational spectroscopy (NRVS) (Fig. 5).

Fig. 5.

⁵⁷Fe PVDOS of samples containing complexes 1 and 2 derived from NRVS measurements. The frequencies of major spectral features are indicated in the figure.

For complex 1, three major PVDOS features have been observed below 600 cm⁻¹, which are centered 221 cm⁻¹, 266 cm⁻¹ and 378 cm⁻¹. The peak at 221 cm⁻¹ is the strongest PVDOS feature. The second strongest feature is at observed 266 cm⁻¹, with a high-energy shoulder at ~300 cm⁻¹. In addition, a series of minor peaks have been observed at 150 cm⁻¹, 176 cm⁻¹, and 509 cm⁻¹. In the energy region above 600 cm⁻¹, a very minor PVDOS feature has been observed at 768 cm⁻¹, which could be attributed to the asymmetric Fe-O-Cr stretching vibration (v_as[Fe-O-Cr]) that has been previously assigned by resonance Raman measurements for complex 1 (770 cm⁻¹ observed by resonance Raman [23]). Complex 2 exhibits a similar set of PVDOS features, but some of the major features are located at different energies with different relative intensities. Specifically, the peak at 224 cm⁻¹ and 262 cm⁻¹ are highly similar to those at 221 cm⁻¹ and 266 cm⁻¹ observed for complex 1. In contrast, the high-energy shoulder at ~300 cm⁻¹ disappears in complex 2, instead, a peak at 278 cm⁻¹ shows up, making the peak at 262 cm⁻¹ the strongest feature for complex 2. In addition, a PVDOS feature is observed at 396 cm⁻¹, which may correspond to the 378 cm⁻¹ feature observed in complex 1. Finally, a relatively strong PVDOS feature is observed at 876 cm⁻¹, which could be attributed to the asymmetric Fe-O-Cr stretching vibration (v_as[Fe-O-Cr]) that has previously been assigned by resonance Raman measurements for complex 2 (877 cm⁻¹ observed by resonance Raman [24]). The significant blue shift of v_as[Fe-O-Cr] in going from complex 1 to complex 2 is most likely due to the shorter Fe—O bond in 2 due to the syn coordination of the TMC ligand towards the iron center, which leads to a 5-coordinate iron geometry (Table 2).

To better understand the PVDOS spectra of complexes 1 and 2, we performed frequency calculations based on the geometry optimized structures of these complexes by DFT. Our calculations can well reproduce the ⁵⁷Fe PVDOS spectra for both complexes, suggesting the optimized geometries provide an opportunity for reliable structural comparison between these two complexes (Figs. 6 and 7).

Fig. 6.

⁵⁷Fe PVDOS spectra of complex 1 obtained by NRVS measurements and by DFT calculations together with the calculated normal mode pictures of the selected ⁵⁷Fe PVDOS features. The experimental spectrum is shown in black and the DFT calculated spectrum is shown in red with the red sticks indicate the energies of different calculated normal modes with the stick height representing the intensity of the mode composition factor, e², for individual normal mode. The selected normal mode pictures are shown in the lower panel with the corresponding energy labeled. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 7.

⁵⁷Fe PVDOS spectra of complex 2 obtained by NRVS measurements and by DFT calculations together with the calculated normal mode pictures of the selected ⁵⁷Fe PVDOS features. The experimental spectrum is shown in black and the DFT calculated spectrum is shown in red with the red sticks indicate the energies of different calculated normal modes with the stick height representing the intensity of the mode composition factor, e², for individual normal mode. The selected normal mode pictures are shown in the lower panel with the corresponding energy labeled. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

To gain detailed understanding of the impact of coordination mode on the molecular vibrations, vibration assignments are performed for the major ⁵⁷Fe PVDOS features for both complexes by comparing with the normal modes of an idealized C_4v M(X)₅Y moiety (M = Fe, X = N, and Y = O) for 1 or an idealized C_4v M(X)₄Y moiety for 2. The normal modes of vibration of the M(X)₅Y and the M(X)₄Y models are shown in Fig. S3 and S4 in the supporting information. For complex 1 (Fig. 6), the experimentally observed 221 cm^{−1 57}Fe PVDOS feature is calculated at 213 cm⁻¹, which is mainly contributed from three vibrational modes calculated at 210 cm⁻¹, 213 cm⁻¹ and 215 cm⁻¹. These are N-Fe-N in-plane bending modes combined with axial N-Fe-O bending modes, which correspond to the normal mode v₉(E) of the C_4v M(X)₅Y moiety (Fig. S3). In addition, there is a calculated N-Fe-N out-of-plane bending mode located at 204 cm⁻¹, which correspond to the normal mode v₃(A₁) of the C_4v M(X)₅Y moiety (Fig. S3). The experimentally observed 266 cm⁻¹ feature is calculated at 264 cm⁻¹, which is contributed from three major vibrational modes at 264 cm⁻¹, 269 cm⁻¹, and 258 cm⁻¹. Both the calculated 264 cm⁻¹ and 258 cm⁻¹ modes can be best described as the in-plane asymmetric Fe—N stretching motion combined with the N-Fe-O bending motion, which correspond to the normal mode v₄(E) of the C_4v M(X)₅Y moiety (Fig. S3). While the calculated 269 cm⁻¹ mode is the symmetric Fe—N stretching motion combined with an Fe—O out-of-plane motion, which corresponds to the v₁₁(A₁) mode of the C_4v M (X)₅Y moiety (Fig. S3). Next the experimentally observed 378 cm⁻¹ feature is calculated at 390 cm⁻¹, which is contributed by two major vibrational modes at 386 cm⁻¹ and 394 cm⁻¹. These vibrations are mainly the Fe-O-Cr symmetric stretching modes (v_s[Fe-O-Cr]). Finally, the asymmetric Fe-O-Cr stretching mode (v_as[Fe-O-Cr]) is calculated at 794 cm⁻¹, which is slightly overestimated in energy than the observed v_as[Fe-O-Cr] at 768 cm⁻¹. The overall low intensity of this mode is because only the bridging oxo moves significantly but not the Fe and the Cr, which reduces the ⁵⁷Fe PVDOS.

For complex 2 (Fig. 7), the experimentally observed 224 cm⁻¹ feature is calculated at 212 cm⁻¹, which are contributed from four vibrational modes calculated at 210 cm⁻¹, 214 cm⁻¹, 220 cm⁻¹ and 205 cm⁻¹. The modes at 205 cm⁻¹, 210 cm⁻¹, and 214 cm⁻¹ can be best described as the in-plane N-Fe-N bending motion, which correspond to the normal mode v₄(E) of the C_4v M(X)Y moiety (Fig. S4). While the 220 cm⁻¹ mode is mainly the Fe-O-Cr bending mode. The observed 262 cm⁻¹ feature with the 278 cm⁻¹ shoulder is reproduced as a broad ⁵⁷Fe PVDOS feature with a peak at 251 cm⁻¹ and two shoulders at 267 cm⁻¹ and 286 cm⁻¹ by the DFT calculations. A series vibrational modes contribute to this broad feature with two major modes with strong intensity located at 251 cm⁻¹ and 286 cm⁻¹. These two features are the asymmetric Fe—N stretching modes that correspond to the v₉(E) mode of the C_4v M(X)₄Y moiety. In the idealized C_4v symmetry, this asymmetric Fe—N stretching mode is two-fold degenerate. But in complex 2, the molecule exhibits a lower symmetry so that this degenerate mode is split into two modes showing 35 cm⁻¹ energy difference. In addition, the experimentally observed 396 cm⁻¹ feature is calculated at 409 cm⁻¹, which is contributed by the Fe-O-Cr symmetric stretching modes (v_s[Fe-O-Cr]). Finally, the asymmetric Fe-O-Cr stretching mode (v_as[Fe-O-Cr]) is calculated at 880 cm⁻¹, which is very close to the observed v_as[Fe-O-Cr] at 876 cm⁻¹.

Overall, the major ⁵⁷Fe PVDOS features from the two complexes (the 221 cm⁻¹ and 266 cm⁻¹ features observed for complex 1 and the 224 cm⁻¹ and 262 cm⁻¹ features observed for complex 2) are originated from similar vibrational modes with the lower energy features mainly representing the N-Fe-N bending modes combined with N–Fe–O bending motions and the higher energy features mainly representing the asymmetric Fe—N stretching modes. The main ⁵⁷Fe PVDOS features that distinguish these two complexes are the symmetric and asymmetric Fe–O–Cr stretching modes, which are observed at 378 cm⁻¹ and 768 cm⁻¹ for complex 1 and at 396 cm⁻¹ and 876 cm⁻¹ for complex 2. The clear blue shifts of these two vibrational modes in going from complexes 1 and 2 are the direct consequence of iron coordination number. The 6-coordinate Fe center in complex 1, especially the axial CH₃CN ligand, weakens the Fe–μ-O bond, thus leading to a longer Fe—O distance and lower Fe–O–Cr stretching frequencies.

4. Discussion and conclusion

In this work, we studied three hetero-dinuclear Fe^III–O–Cr^III complexes that differ on the orientation of the TMC ring and coordination numbers. With detailed magnetic spectroscopic studies via Mössbauer and EPR, vibrational analysis via ⁵⁷Fe NRVS, and DFT calculations, we have found that the orientation of the TMC ring governs the properties of these complexes in multiple dimensions even though they have very similar overall structure. When the TMC methyl groups are syn to the oxo bridge in the (TMC)Fe–O–Cr species (complex 2), the iron center favors a 5-coordinate square pyramidal geometry, resulting in a shorter Fe–μ-O distance and a slightly bent Fe–O–Cr angle (171 degrees for complex 2) based on the DFT results. In contrast, when the TMC methyl groups are anti to the oxo bridge (complexes 1 and 3), the iron center prefers to be six-coordinate, with a longer Fe–μ-O distance and a close to linear Fe–O–Cr angle (179 degrees for 1 and 3). The geometry metrics of the DFT optimized structures of complexes 1 and 2 closely agree with the previously published EXAFS data. The different structural features among these complexes are reflected by all the spectroscopic properties. Specifically, the Mössbauer analysis revealed that for all three complexes, the Fe(III) (S = 5/2) is antiferromagnetically coupled with the Cr (III) (S = 3/2) to give an S = 1 spin ground state, which is expected for a μ-oxo bridged di-metal system and is further supported by DFT calculations with the calculated magnetic coupling constant J > +200 cm⁻¹ (in the $H_{ex}=J{\widehat{\mathit{S}}}_{Fe}\bullet {\widehat{\mathit{S}}}_{Cr}$ notation) for all three complexes. The Mössbauer isomer shifts for the 6-coordinate complexes 1 and 3 are higher (δ = 0.53 mm/s for 1 and δ = 0.51 mm/s for 3) than that of 5-coordinate complex 2 (δ = 0.43 mm/s for 2), which may be due to a higher direct s electron donation to the iron center because of the shorter Fe-μO distance in 2, which leads to a lower isomer shift. However, the difference of the coordination geometry around the iron center does not impact the Mössbauer quadrupole splittings ($\mathrm{\Delta }{\mathrm{E}}_{\mathrm{Q}}$ = −0.88 mm/s for 1, $\mathrm{\Delta }{\mathrm{E}}_{\mathrm{Q}}$ = −0.89 mm/s for 2, and $\mathrm{\Delta }{\mathrm{E}}_{\mathrm{Q}}$ = −0.94 mm/s for 3). These observations suggest that the electron population on the Fe 3d-based molecular orbitals is similar among these three complexes, which is supported by the population analysis from the DFT calculations. Additionally, the coordination geometry differences around the iron center in these complexes are also reflected by vibrational features revealed by ⁵⁷Fe PVDOS spectrum of complexes 1 and 2 derived from ⁵⁷Fe NRVS data. Although the strong ⁵⁷Fe PVDOS features located in the energy region of 200–300 cm⁻¹ can be assigned to similar normal modes for both complexes, the clear differences have been observed in the energy regions of 300–400 cm⁻¹ and > 600 cm⁻¹ where the symmetric Fe-O-Cr and the asymmetric Fe-O-Cr stretching modes are observed at 378 cm⁻¹ and 768 cm⁻¹ for complex 1 and at 396 cm⁻¹ and 876 cm⁻¹, respectively. The clear blue shifts of these modes for complex 2 are the direct consequence of a shorter Fe-μO bond length due to 5-cooridnation, which is calculated to be 1.75 Å. For complex 1, this bond length is calculated to be 1.79 Å.

The above-mentioned results emphasize the importance of the ligand topology in determining the coordination sphere of the iron center and the corresponding electronic and geometric structures. Beyond the structural impact, the ligand topology also affects chemical reactivity of the iron complexes. For example, Que., Swarts, and coworkers have recently demonstrated that a pair of Fe(IV)═O complexes with TMC-syn or TMC-anti ligand topology showed significant difference in oxygen atom transfer (OAT) reactivity [51]. Specifically, TMC-syn Fe(IV) = O exhibits 10²- to 10³-fold faster OAT rates towards R₂S=O substrates than observed for TMC-anti Fe(IV) = O. TMC-syn Fe(IV) = O further shows oxidation towards all olefin substrates tested, while TMC-anti Fe(IV) = O only reacts with cis-cyclooctene, the most electron-rich of the olefins studied. Based on computational studies, it is concluded that the 5-coordinate Fe center in the TMC-syn conformation provides more exposure of the Fe(IV) = O moiety towards the substrate, thus significantly reducing the reaction barrier and increasing the reaction rate.

In conclusion, three Fe^III–O–Cr^III complexes with the iron center supported by the tetramethylcyclam (TMC, 1,4,8,11-tetramethyl-1,4,8,11-tetraazacyclotetradecane) ligand are examined in detail by various spectroscopic techniques (⁵⁷Fe Mössbauer, EPR, and ⁵⁷Fe NRVS) and computational modeling. The unique feature of these heterodinuclear metal complexes is the difference in ligand topology around the iron center (TMC-syn vs. TMC-anti configuration), which leads to the difference in iron coordination number (5-coordination vs. 6-coordination), iron-ligand distance, and even Fe^III-O-Cr^III bond angle. These ligand topology induced structural changes are nicely reflected by the corresponding spectroscopic features (magnetic and vibrational properties), which can be further connected to the electronic structural changes via density functional theory calculations. The current study emphasizes on how the ligand configuration change around the metal center could lead to drastic change in metal center structure without the change of ligand composition and on how the combined spectroscopic and computational studies could derive the essential geometric and electronic structure correlations of these unique complexes. The methodology used here also could serve as a useful strategy to examine the conformation change of the iron centers in metalloenzymes during enzyme catalysis.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.jinorgbio.2025.113054.

Data availability

Data will be made available on request.