Achieving One-Electron Oxidation of a Mononuclear Nonheme Iron(V)-Imido Complex

Abstract

A mononuclear nonheme iron(V)-imido complex bearing a tetraamido macrocyclic ligand (TAML), [Fe^V(NTs)(TAML)]⁻ (1), was oxidized by one-electron oxidants, affording formation of an iron(V)-imido TAML cation radical species, [Fe^V(NTs)(TAML^+•)] (2); 2 is a diamagnetic (S = 0) complex, resulting from the antiferromagnetic coupling of the low-spin iron(V) ion (S = 1/2) with the one-electron oxidized ligand (TAML^+•). 2 is a competent oxidant in C–H bond functionalization and nitrene transfer reaction, showing that the reactivity of 2 is greater than that of 1.

High-valentiron-oxo and -imido species have been invoked as key intermediates in enzymatic and synthetic oxidative transformation reactions.^1,4 In biomimetic studies, heme and nonheme iron-oxo complexes have been synthesized, characterized, and investigated in various oxidation reactions over the past several decades.^2,3 Compared to the heme and nonheme iron-oxo species, the chemistry of iron-imido analogs is less clearly understood. Moreover, unlike the iron(IV and V)-oxo systems, most of the synthetic iron-imido complexes reported so far are limited to iron(IV)-imido complexes,⁵ and a mononuclear iron(V)-imido complex bearing a highly negatively charged TAML, [Fe^V(NTs)(TAML)]⁻ (1), was successfully synthesized by some of us very recently.⁶ Encouraged by the successful synthesis of 1, we attempted to oxidize it further to obtain a species such as [Fe^VI(NTs)-(TAML)]. Interestingly, we observed that the iron(V)-imido complex was oxidized at the TAML site, but not at the iron center. Herein, we report for the first time the synthesis, spectroscopic characterization, and reactivity studies of a novel iron(V)-imido TAML cation radical species, [Fe^V(NTs)-(TAML^+•)] (2) (see Scheme 1).

Scheme 1.

Synthesis and Reactivity of [Fe^V(NTs)(TAML^+•)]

The iron(V)-imido TAML complex, [Fe^V(NTs)(TAML)]⁻ (1), was synthesized by following reported procedures (see Experimental Section in Supporting Information (SI)).⁶ The cyclic voltammogram of 1 exhibited one reversible wave centered at 0.86 V vs SCE in CH₃CN, with the reduction and oxidation peak potentials at 0.82 and 0.90 V, respectively (SI, Figure S1). Given the electrochemical property of 1, the oxidation of 1 was performed with one-electron oxidants, such as [Fe^III(bpy)₃]³⁺ (E_ox = 1.06 V vs SCE), tris(4-bromophenyl)-ammoniumyl hexachloroantimonate, [(4-BrC₆H₄)₃N]SbCl₆ (TBPA, E_ox = 1.08 V vs SCE), and [Ru^III(bpy)₃]³⁺ (E_ox = 1.24 V vs SCE).⁷ Addition of the oxidants to the CH₃CN solution of 1 at −40 °C immediately changed the solution color from dark green to deep brown (SI, Experimental Section). The intermediate, denoted as 2, was metastable (t_1/2 ~ 4 h) at −40 °C, allowing us to characterize it using various spectroscopic techniques, such as UV–vis, cold spray time-of-fiight mass spectrometry (CSI MS), electron paramagnetic resonance (EPR), resonance Raman (rRaman), nuclear magnetic resonance (NMR), Mössbauer, and X-ray absorption spectroscopy/extended X-ray absorption fine structure (XAS/EXAFS), along with density functional theory (DFT) calculations.

The UV–vis spectrum of 2 exhibited two distinct absorption bands at 545 nm (ε = 10 000 M⁻¹ cm⁻¹) and 750 nm (ε = 4000 M⁻¹ cm⁻¹) (Figure 1a and SI, Figure S2). Titration experiments, monitored at 750 nm as a function of the amount of [Ru(bpy)₃]³⁺ added (Figure 1a, inset), indicated that the complete conversion of 1 to 2 required a stoichiometric amount of the [Ru(bpy)₃]³⁺ oxidant. While the X-band EPR spectrum of 1 exhibited an isotropic signal at g = 2.00,⁶ 2 is EPR silent (SI, Figure S3). The latter result suggests that (i) a low-spin (S = 1/2) iron(V) in 1 was oxidized to iron(VI) (S = 0) or (ii) a one-electron oxidized TAML was antiferromagnetically or ferromagnetically coupled with (S = 1/2) iron(V) (vide infra). CSI MS of 2 in positive mode revealed two prominent ion peaks at m/z of 618.1 and 659.2, with mass and isotope distribution patterns corresponding to {Na[Fe(NTs)-(TAML)]}⁺ (calculated m/z of 618.1) and {Na[Fe(NTs)-(TAML)(CH₃CN)]}⁺ (calculated m/z of 659.1), respectively (SI, Figure S4). When 2 was generated with ¹⁵N-labeled 1, which was synthesized using PhI¹⁵NTs, one-mass unit shift from m/z of 618.1 to 619.1 and from m/z of 659.2 to 660.2 was observed. The latter result demonstrates that 2 contains one NTs group (SI, Figure S4). The rRaman spectrum of 2, upon 441.6 nm-excitation in CH₃CN at −40 °C, exhibited an isotopically sensitive doublet feature centered at 796 cm⁻¹, which shifted to a new doublet centered at 770 cm⁻¹ upon ¹⁵N-labeling of 2 (2-¹⁵NTs; SI, Figure S5). The isotopic shift of 26 cm⁻¹ is in good agreement with the calculated value of 21 cm⁻¹ for a diatomic Fe–N oscillator (Hooke’s Law). It is notable that the rRaman spectrum of 1 showed a Fe–N stretching vibration at 817 cm⁻¹ with a doublet feature,⁶ which is ~20 cm⁻¹ higher than that of 2 (vide infra). ¹H NMR and 2D ¹H–¹H COSY experiments were performed in CD₃CN at −40 °C (SI, Figures S6–S8). All peaks in the ¹H NMR spectra (SI, Figure S6) were in the diamagnetic region, indicating that 2 is a diamagnetic species (S = 0).

Figure 1.

(a) UV–vis spectra of [Fe^V(NTs)(TAML)] ⁻ (1, black line) and [Fe^V(NTs)(TAML^+•)] (2, blue line); 2 was synthesized by reacting 1 (0.20 mM) with 1.0 equiv of [Ru(bpy)₃]³⁺ (0.20 mM) in CH₃CN at −40 °C. Inset shows plot of the absorbance change at 750 nm due to 2 upon addition of [Ru(bpy)₃]³⁺ to 1 (0.20 mM) in increment of 0.2 equiv. (b) Mössbauer spectra (black dotted lines) with fits (blue lines) for 2 recorded at 4.2 K and 0, 4, and 7 T.

Mössbauer spectra of 2 were collected at 4.2 K using fields of 0, 4, and 7 T applied parallel to the γ-ray (Figure 1b). In the absence of applied field, the spectrum consists of a quadrupole doublet centered at δ = −0.34 mm s⁻¹ with a quadrupole splitting of ΔE_Q = 3.62 mm s⁻¹. These parameters are very similar to those of 1 (δ = −0.40 mm s⁻¹, ΔE_Q = 3.83 mm s⁻¹),⁶ suggesting that the Fe ions in 1 and 2 have the same oxidation state. Application of a magnetic field splits the doublet, but the whole spectrum stays confined to a narrow range (−2 to +2 mm s⁻¹), indicating that 2 possesses an overall spin of S = 0 and the set of three spectra could be simulated perfectly under this assumption. Thus, the Mössbauer data suggest that the electronic structure of 2 is best described as a low-spin (S = 1/2) iron(V) center antiferromagnetically coupled to a radical spin located on the ligand.

Fe K-edge XAS data on solution samples of 1 and 2 are presented in Figure 2a. The data show very little difference in the rising-edge energy or pre-edge energy position, indicating no change in the oxidation state of the Fe centers in 1 and 2, as shown in the Mössbauer study. The Fe K-pre-edge intensity is a sensitive probe of the Fe–N bond distance. This is because the short Fe–N bond is along the molecular z-axis, which leads to a strong Fe 3d_z²–4p_z mixing, which increases with the increase in distortion along the Fe–N bond (i.e., shortening of the Fe–N bond).^8,9 The very small difference in the two pre-edge intensities indicates that the bond distances are not dramatically different in 1 and 2. To confirm this structural assessment, EXAFS data were measured on 1 and 2 and a comparison is shown in Figure 2b and SI, Figure S9. A qualitative comparison shows very similar structures for 1 and 2, with loss of long-range multiple scattering at high R (R′ ~ 3–4 Å) in 2 relative to 1. This is consistent with perturbation in TAML due to oxidation. FEFF fits for 2 indicate that the data are consistent with a first shell with 1 Fe–N at 1.66 Å and 4 Fe–N at 1.85 Å. The second shell was fit with single and multiple scattering contributions from TAML. No long distance shell (between R′ ~ 3–4 Å) was observed. For comparison, the EXAFS data for 1 were also fit using the same protocol used here and the best fit was obtained using 1 Fe–N at 1.67 Å and 4 Fe–N 1.86 Å. It is important to note that the EXAFS resolution is ±0.02 Å, indicating that the first-shell structures of 1 and 2 are very similar. Interestingly, however, a difference in intensity in the Fourier transforms at high R is observed, which is replicated over several measurements. This indicates a change in TAML due to oxidation. Multiple-scattering components from the outer edge of TAML were required to obtain a good fit to 1 (SI, Table S1), suggesting that the TAML ring in 1 is oriented in such a way as to maximize this longer-range contribution to the EXAFS. This typically occurs when the ring is more ordered or planar. Thus, the XAS/EXAFS data indicate that the oxidation of 1 occurs on TAML, not at the Fe center.

Figure 2.

(a) XAS data of 1 (black line), and 2 (blue line). Inset shows the expansion of pre-edge region. (b) Non-phase shift corrected Fourier transforms of 1 (black line) and 2 (blue line) (see SI, Figure S9).

Insight into the electronic structure of 2 was gained from DFT calculations (SI, Figure S10). Relying on the EXAFS and Mössbauer experiments, the hybrid B3LYP* with a decreased contribution of exact exchange (15%),¹⁰ compared to the well-known B3LYP functional (20%), proved to yield a satisfactory agreement with experiment (SI, Table S2). A singlet ground state was observed, with structural parameters in good agreement with EXAFS data, with a Fe-NTs distance of 1.68 Å and an averaged Fe–N (TAML) distance of 1.88 Å. The calculated Mössbauer parameters are also consistent with experiment, with ΔE_Q = 3.49 mm s⁻¹ (exp. 3.62 mm s⁻¹) and δ = −0.30 mm s⁻¹ (exp. −0.34 mm s⁻¹). The group spin densities unambiguously identify this electronic structure as being a low-spin Fe^V ion antiferromagnetically coupled to a radical on TAML (SI, Table S3). A plot of the spin densities shows that the latter radical is localized on the o-phenylenediamine part, as a result of its conjugated character (SI, Figure S11). Finally, the Kohn–Sham orbital diagram of this Fe^V(NTs) TAML cation radical species is fully consistent with the picture described above (SI, Table S4). The Fe electronic configuration is d_xy²d_xz^α1 corresponding to a Fe^V low-spin ion. The LUMO is localized on the α N (p_z) orbitals of the o-phenylenediamine part of TAML, resulting in a negative spin density on TAML (SI, Figure S12). Such a radical character on TAML site in high-valent transition metal TAML complexes was previously discussed in theoretical studies.¹¹ Notably, the Fe^VI d² configuration could be obtained as a very low-lying excited state, but the resulting structural and Mössbauer parameters are not consistent with experiments. On the basis of the spectroscopic characterization and DFT calculations, we conclude that an iron(V)-imido TAML cation radical complex, [Fe^V(NTs)(TAML^+•)] (2), was formed in one-electron oxidation of the iron(V)-imido TAML complex, [Fe^V(NTs)-(TAML)]⁻ (1).

The reactivity of 2 was investigated in C–H bond functionalization and nitrene transfer reaction and then compared with that of 1. Upon addition of xanthene to a CH₃CN solution of 2, absorption bands at 545 and 750 nm disappeared, and the decay rate increased proportionally with the increase of xanthene concentration (SI, Figure S13); a second-order rate constant (k₂) of 1.3(2) × 10 M⁻¹ s⁻¹ at 15 °C was determined. This k₂ value of 2 was slightly greater (~2.5 times) than that of 1 (e.g., 4.7 M⁻¹ s⁻¹ at 15 °C).⁶ A kinetic isotope effect (KIE) value of 7(1) was obtained in the xanthene oxidation by 2 (SI, Figure S14 and Table S5); a KIE value of 11(1) was reported in the oxidation of xanthene by 1.⁶ We also determined the k₂ values in the oxidation of other substrates, such as 9,10-dihydroanthracene (DHA), indene, and fluorene (SI, Figure S15 and Table S5), and a linear correlation between the k₂ values and the C–H bond dissociation energies (BDEs) of substrates was observed (Figure 3a). The observation of a large KIE value and a good correlation between the reaction rates and the C–H BDEs of substrates leads us to conclude that a hydrogen atom (H atom) abstraction from the C–H bonds of substrates by 2 is the rate-determining step.

Figure 3.

(a) Plot of log k₂′ against C–H BDEs of substrates in the amination reaction by 2 at 15 °C (SI, Table S5). (b) Plots of log k₂ against the E_ox values of para-X-substituted thioanisole derivatives in the sulfimidation reaction by 2 at −40 °C (SI, Table S6).

Product analysis of the xanthene and fluorene oxidation by 2 revealed the formation of xanthene-NHTs and fluorene-NHTs as major products (~90% yield in both cases). We also found the formation of [Fe^IV(TAML)] as the decay product of 2; [Fe^IV(TAML)] was synthesized independently as an authentic sample to compare the spectroscopic data of the decay product of 2 (SI, Figures S16–S19). Taken together, the amination reaction by 2 occurs via a H atom abstraction of substrates to give aminated products and [Fe^IV(TAML)] as the decay product of 2.

In the nitrene transfer reaction by 2, para-X-substituted thioanisoles (X = OMe, Me, H, Cl, and CN) were used as substrates under stopped-flow kinetic conditions at −40 °C. In the sulfimidation of thioanisole by 2, a k₂ value of 3.3(2) × 10 M⁻¹ s⁻¹ was determined at −40 °C, which was then calculated to be 4.4 × 10² M⁻¹ s⁻¹ at 15 °C (SI, Figures S20 and S21). By comparing the k₂ values of 1 and 2, we found that the reactivity of the iron(V)-imido complex was markedly enhanced (over 10⁴ folds) by the one-electron oxidation of the supporting ligand (e.g., 2.6 × 10⁻² M⁻¹ s⁻¹ for 1 versus 4.4 × 10² M⁻¹ s⁻¹ for 2).⁶ The analysis of the reaction solution revealed the formation of PhS(=NTs)CH ₃ (~95% yield) and [Fe^IV(TAML)] as products (SI, Figures S22 and S23). In addition, as often proposed in the sulfoxidation of thioanisoles by high-valent metal-oxo species,¹² a large negative slope of −9.1, which was obtained by plotting log k₂ values against one-electron oxidation potentials (E_ox) of para-X-substituted thioanisoles (Figure 3b; SI, Figure S24 and Table S6), suggests that the sulfimidation of thioanisoles by 2 proceeds via electron transfer, followed by nitrene transfer.

In order to reconcile the reactivity differences of 1 and 2 in the amination and sulfimidation reactions, we calculated their H atom affinity (HAA) for aminations and electron affinity (EA) for sulfimidation (SI, Table S7). The EAs for 1 and 2 are 103.9 and 122.8 kcal mol⁻¹, respectively, a trend which is in line with the experimental results obtained in the sulfimidation of thioanisole by 1 and 2. The calculated HAAs of 1 and 2 for the amination reactivity were found to be very close, being 81.9 and 85.2 kcal mol⁻¹, respectively, which is consistent with their similar reactivity in the amination reactions. The group spin densities of 1–H and 2–H show a similar Fe^IV-NHTs configuration (SI, Table S8), explaining the closeness of their HAAs.

In summary, an iron(V)-imido TAML cation radical complex, [Fe^V(NTs)(TAML^+•)] (2), was synthesized for the first time by oxidizing an iron(V)-imido TAML complex, [Fe^V(NTs)(TAML)]⁻ (1), with one-electron oxidants; 2 can be considered formally as an iron(VI)-imido species as Cpd I is considered formally as an iron(V)-oxo species in heme systems.³ The one-electron oxidized species, 2, showed slightly increased reactivity in amination reaction (i.e., ~2.5 times increase) but markedly enhanced reactivity in nitrene transfer reaction (i.e., ~17 000 times increase). Future studies will focus on detailed mechanisms and reactivity comparison of 1 and 2 in various oxidation reactions.