Mechanistic Investigation of Oxygen Rebound in a Mononuclear Nonheme Iron Complex

Abstract

An iron(III) methoxide complex reacts with para-substituted triarylmethyl radicals to give iron(II) and methoxyether products. Second-order rate constants for the radical derivatives were obtained. Hammett and Marcus plots suggest the radical transfer reactions proceed via a concerted process. Calculations support the concerted nature of these reactions involving a single transition state with no initial charge transfer. These findings have implications for the radical “rebound” step invoked in nonheme iron oxygenases, halogenases, and related synthetic catalysts.

The preferential formation of C−O bonds by heme and nonheme iron enzymes typically occurs through C−H bond cleavage by a high-valent iron(IV)-oxo species, followed by a radical “rebound” step in which the newly formed carbon radical (R•) and Fe(OH) intermediate combine to give R−OH and a reduced Fe product.^1–4 However, alternate outcomes are sometimes observed, including desaturation or decarboxylation (e.g., heme: Cytochrome P450 (CYP) OleT; nonheme: AsqJ, NapI, VioC, UndA).^5–12 In the nonheme iron halogenases (e.g., CytC3, WelO5, SyrB2),^13–17 the radical selectively combines with a halide ligand rather than an OH ligand, leading to halogenation. Interestingly, the nonheme iron enzyme isopenicillin N synthase (IPNS) may operate through a similar pathway, in which a C_Val radical combines selectively with a coordinated thiolate ligand, instead of a bound OH group, to give the final thiazolidine ring.¹⁸ Studies on biomimetic, high-valent metal-oxo complexes showed that different outcomes can occur from the carbon radical stage, including conventional hydroxylation as well as simple radical dissociation away from the Fe(OH) species.^4,19,20 The factors that control the rates and selectivities of these reactions remain poorly understood.

The synthesis of a heme-like Fe(OH) corrole complex provided us with a platform to examine reactions with carbon radicals (trityl radical derivatives).^21,22 These model reactions could be compared to the rebound hydroxylation process seen in CYP, involving protonated Compound II (Cpd-II, Scheme 1).²¹ The heme-based CYP undergoes reduction of a formal Fe^IV(OH)(porphyrin) to an Fe^III(porphyrin). The analogous nonheme chemistry, exemplified by the hydroxylation performed by TauD shown in Scheme 1, occurs via the lower-valent Fe^III(OH)-to-Fe^II transformation.^1,23 Recently, we showed that a nonheme Fe^III(OMe) complex reacts with trityl radical via homolytic cleavage of the Fe−OMe bond to give Ph₃COMe and Fe^II.²⁴ However, no kinetic data were obtained and the mechanism was not examined in detail. To our knowledge, experimentally determined rates of the rebound reaction in nonheme enzymes or models are not known.

Scheme 1.

Comparison of Heme and Nonheme Iron Enzymes

Herein, we show that [Fe^III(N3PyO^2Ph)(OCH₃)](ClO₄) (1) reacts with a series trityl radical derivatives, and a detailed kinetic study provides key insights regarding the mechanism of these reactions. Density functional theory (DFT) calculations help support the observed reactivity.

In our initial report, we employed triphenylmethyl radical (Ph₃C•), a stable carbon radical, for reaction with 1. In the current work, we varied the electronic character of the radical derivatives by para substitution.²⁵ A series of para-X-substituted radicals (X = OCH₃, tBu, Ph, CN) were prepared. Addition of (4-tBu-C₆H₄)₃C• (5 equiv) to 1 (Scheme 2) led to the conversion of dark purple 1 to yellow-orange within 5 min. Analysis by ¹H NMR spectroscopy of the same reaction mixture in toluene-d₈ showed formation of the methoxy group of (p-^tBu-C₆H₄)₃COCH₃. Quantitation gave a 70% yield of methoxyether (see Figure S8 in the Supporting Information). The ⁵⁷Fe-labeled 1 in THF/toluene revealed a broad doublet in the Mössbauer spectrum (δ = 0.5 mm s⁻¹, |ΔE_Q| = 1.29 mm s⁻¹), which is indicative of an Fe^III complex in an intermediate relaxation regime, as reported earlier.²⁴ This doublet disappears following the addition of (4-^tBu-C₆H₄)₃C•, giving rise to two new high-spin Fe^II subcomponents: δ = 1.16 mm s⁻¹, |ΔE_Q| = 2.68 mm s⁻¹ (80% of total fit) and δ = 1.34 mm s⁻¹, |ΔE_Q| = 3.29 mm s⁻¹ (20% of total fit) (see Figure S10 in the Supporting Information). The latter component exhibits parameters that are similar to the Fe^II species formed after reaction with the unsubstituted Ph₃C•,²⁴ although this previous reaction was operated in pure THF. The solvent mixture employed here (toluene/THF 1:2) likely leads to different solvent binding equilibria that produces two, related high-spin Fe^II products, which, in turn, gives rise to the two overlapping Mössbauer signals. There is no evidence of any Fe^III starting material remaining. When acetonitrile is added to the product mixture, the spectrum resolves to a single high-spin Fe^II component, δ = 1.16 mm s⁻¹, ΔE_Q = 2.66 mm s⁻¹, matching that for [Fe^II(N3PyO^2Ph)(CH₃CN)]⁺,²⁴ and indicating that 1 is quantitatively reduced to iron(II) via the radical reaction. The ¹H NMR and Mössbauer spectra indicate that (4-^tBu-C₆H₄)₃C• reacts with 1 in good yield via the radical reaction shown in Scheme 2.

Scheme 2.

Reaction of 1 with para-X-Substituted Triphenylmethyl Radicals

Mechanistic information was obtained from kinetic studies. Complex 1 was reacted with (p-X-C₆H₄)₃C• derivatives under pseudo-first-order conditions (excess radical) in toluene at 23 °C. The consumption of 1 could be followed by UV-vis spectroscopy, leading to decay curves that give a pseudo-first-order rate constant (k_obs) (see Figures S2–S5 in the Supporting Information). Measuring the k_obs values at different radical concentrations led to second-order rate constants (k₂) (Figure 1A) with a range of k₂ values between 2–25 M⁻¹ s⁻¹ (X = CN < Ph < tBu < OMe), which increased with the electron-donating ability of the para-substituent.

Figure 1.

(A) Plot of k_obs versus [(p-^tBu-C₆H₄)₃C•], where the slope of the best fit (red line) gives k₂ = 13.4(1) M⁻¹ s⁻¹. (B) Eyring plot of (ln(k_obs/T) vs 1/T for the reaction of 1 and (p-^tBu-C₆H₄)₃-C• from −10 °C to 25 °C. (C) Hammett plot. (D) Marcus plot.

Although there are no other analogous rate constants available for nonheme iron complexes for comparison, we can compare these rate constants to those measured previously for a heme-like corrole complex. The iron corrole Fe(OH)(ttppc) (ttppc = tris(2,4,6-triphenylphenyl corrole) undergoes similar radical reactivity with (p-X-C₆H₄)₃C• (X = −OCH₃, −^tBu, −Ph, Cl) to give (p-X-C₆H₄)₃COH and Fe^III (ttppc), with rate constants ranging from 12.6(1) to 357(4) M⁻¹ s⁻¹.²¹ The ttppc complex is sterically encumbered by large triphenylphenyl groups, but appears to react significantly faster than 1; e.g., Fe(OH)(ttppc) reacts 14 times faster than 1 with the para-OMe derivative. The physical oxidation state of the iron center in the ttppc complex is not easily assigned, because of the possible noninnocent behavior of the corrole ligand, but the overall redox level is one unit above the nonheme system (formally Fe^IV(OH)), and this difference in redox levels may, in part, help to explain the enhanced reactivity of the corrole complex. Alternatively, the difference in reactivity may be due to the steric demands of the axial OMe versus OH ligand, or may arise from an inherent difference in Fe−OMe versus Fe−OH homolytic bond strengths. Further work is needed on both heme and nonheme systems to resolve these fundamental questions.

A Hammett plot (Figure 1C) consisting of log k₂ vs 3σ ⁺, where σ ⁺ is the Hammett parameter for the para-X substituents in (p-X-C₆H₄)₃C•. The rates decrease linearly with the σ ⁺ values, indicating that 1 is behaving as an electrophile, as expected. However, the slope is small (ρ = −0.25). In fact, it is less than half the slope seen for the Hammett plot for Fe(OH)(ttppc) (ρ = −0.55) with the same trityl radical derivatives. Both ρ values suggest little charge separation in the transition state,^21,26 and 1 is even less sensitive to the electrophilicity of the radical derivatives.

The reaction for 1 with the carbon radicals could occur as a concerted process in which C−O bond formation occurs with concomitant Fe−O bond cleavage and reduction of Fe^III to Fe^II, or it could proceed following a stepwise electron-transfer/cation transfer (ET/CT) pathway shown in Scheme 3. The possibility of an ET/CT mechanism was addressed by a Marcus plot (Figure 1). The plot shows reasonable linearity, but the slope (ρ = −0.098) is quite small, showing only a weak dependence on the redox potentials of the radical substrates. In comparison, a rate-limiting, outer-sphere ET process gives typical slopes on the order of −0.5.^21,27 The same plot for the reaction of Fe(OH)(ttppc) also gave a small slope (ρ = −0.15). This analysis indicates the mechanism for 1 is best-described as a concerted process (i.e., the diagonal path in Scheme 3). The same Marcus analysis has also been applied to metal-mediated H atom transfer reactions, with small slopes indicating a concerted proton electron-transfer (CPET) process.^28,29

Scheme 3.

Concerted versus ET/CT Pathways

The pseudo-first-order rate constants (k_obs) were measured between −10 °C and 25 °C, and a plot of ln(k/T) versus 1/T (Figure 1) gave activation parameters: ΔH^‡ = 13.2(1.6) kcal mol⁻¹ and ΔS^‡ = −22.1(0.4) cal mol⁻¹ K⁻¹, and ΔG^‡ = 19.7(1.7) kcal mol⁻¹ at 298 K. These values are consistent with a bimolecular rate-determining step.

Reaction of 1 and trityl cation, Ph₃C⁺, gives very different results. Combining 1 and (Ph₃C)BF₄ (4 equiv) in THF at 23 °C led to a reaction that was over within seconds, as seen by UV-vis. The ether product (Ph₃COCH₃) was formed in 80% yield (Figure S9 in the Supporting Information), and the iron complex did not change oxidation state, as seen by UV-vis (Figure S7 in the Supporting Information) and Mössbauer spectroscopy (see Figure S10 in the Supporting Information). The nucleophilic reactivity of 1 with trityl cation is consistent with the reactivity observed for other metal-alkoxide complexes.³⁰ It appears as though heterolytic cleavage of the Fe−O bond in 1 is facile, compared to homolytic cleavage, although the origins of this difference are not well understood at this time.

DFT calculations on [Fe(OCH₃)(N3PyO^2Ph)]⁺, and the related hypothetical hydroxide complex [Fe(OH)(N3PyO^2Ph)]⁺, led to optimized geometries for the sextet spin ground states of these species (see Figure S11 in the Supporting Information). The metrical parameters for ⁶[Fe(OCH₃)(N3PyO^2Ph)]⁺ match reasonably well with the reported crystal structure, except for Fe−N1(py), which is elongated by ~0.10 Å in the DFT structure. The initial reactant complex (Re) involves the association of the Fe^III complex with a radical substrate whose unpaired spin is aligned parallel (septet) or antiparallel (quintet) to the Fe d electrons. The quintet states were uniformly lower in energy. No charge transfer in the Re was observed. Attempts to interchange molecular orbitals to generate a charge-transfer state (e.g., ⁵[Fe(OH)(N3PyO^2Ph)]⁰−(4-Cl-C₆H₄)₃C⁺) for ⁵Re_OH,Cl) always converged back to the structure with a radical on the carbon moiety. The [Fe^II(N3PyO^2Ph)⁺ + ROCAr₃)] (R = H, Me) products (Pr) were also calculated, and their relative energies showed similar exothermicities (⁵Pr(ΔE + ZPE) − (⁵Re(ΔE + ZPE)) = −7 to −8 kcal mol⁻¹) for the Fe^III(OMe) and Fe^III(OH) complexes, suggesting similar reaction pathways for these species. Transition state (TS) structures were located for the Fe^III(OMe) and Fe^III(OH) complexes and the p-Cl⁻ radical, and free energies of activation with the Wertz entropy correction give values of 23.9 and 25.0 kcal mol⁻¹, respectively. The calculated enthalpy values (see Table S9 in the Supporting Information) are relatively close to the experimental value obtained from the Eyring analysis.

Transition state (TS) structures were located for the Fe^III(OMe) and Fe^III(OH) complexes and the p-Cl⁻ radical, and they are indeed similar in both geometry and relative energies, although one of the py donors appears dissociated from the metal in both structures (Figure 2) due to steric clash, as a result of substrate approach. However, this Fe-py bond is restored during product formation, giving the expected Fe^II product. The structures shown in Figure 2 indicate relatively early transition states, consistent with facile, exothermic reactions. The DFT results show that a concerted, radical group transfer mechanism is favored for both the Fe^III(OMe) and Fe^III(OH) complexes, and the electronic changes are similar for both species. Calculations for other heme and nonheme iron-hydroxo complexes have also shown a single transition state, implicating a concerted process.^31–33

Figure 2.

Transition-state structures for [Fe^III(OCH₃)(N3PyO^2Ph)]⁺ (left) and [Fe^III(OH)(N3PyO^2Ph)]⁺ (right) with (4-Cl-C₆H₄)₃C•. The C−O and Fe−O distances (in Å), and imaginary frequencies, are listed for each structure.

The combined experimental and computational data suggest that the radical rebound step in nonheme iron enzymes or synthetic catalysts may proceed by a concerted process, with little or no charge transfer prior to C−O bond formation. The sizable reaction barrier for the rebound process for complex 1 is significantly different from the almost-barrierless rebound processes predicted for nonheme iron enzymes. However, the model complex reacts via a bimolecular process, whereas the radicals in the enzymes are trapped in the active site pocket and react directly via a first-order process. In addition, the trityl radical derivatives are significantly more stable than typical primary or secondary alkyl radicals generated in the biological systems. The conclusion that a concerted “rebound” pathway is operative for 1 suggests that radical rebound in nonheme iron enzymes also may be concerted, which could help determine product selectivity. The substrate radical must be held in a close, appropriate orientation for a concerted pathway to lead to a productive reaction. Obtaining such information from model systems should further our understanding of the selectivity of rebound processes in nonheme iron enzymes such as the α-KG halogenases, IPNS, and related systems.