Nonheme Diiron Oxygenase Mimic That Generates a Diferric–Peroxo Intermediate Capable of Catalytic Olefin Epoxidation and Alkane Hydroxylation Including Cyclohexane

Abstract

Herein are described substrate oxidations with H₂O₂ catalyzed by [Fe^II(IndH)(CH₃CN)₃](ClO₄)₂ [IndH = 1,3-bis(2′-pyridylimino)isoindoline], involving a spectroscopically characterized (μ-oxo)(μ-1,2-peroxo)diiron(III) intermediate (2) that is capable of olefin epoxidation and alkane hydroxylation including cyclohexane. Species 2 also converts ketones to lactones with a decay rate dependent on [ketone], suggesting direct nucleophilic attack of the substrate carbonyl group by the peroxo species. In contrast, peroxo decay is unaffected by the addition of olefins or alkanes, but the label from H₂ ¹⁸O is incorporated into the the epoxide and alcohol products, implicating a high-valent iron–oxo oxidant that derives from O–O bond cleavage of the peroxo intermediate. These results demonstrate an ambiphilic diferric–peroxo intermediate that mimics the range of oxidative reactivities associated with O₂-activating nonheme diiron enzymes.

Nonheme diiron enzymes are involved in many oxidative metabolic pathways.¹ This class includes those that hydroxylate strong C–H bonds like soluble methane monooxygenase (sMMO)² and deoxyhypusine hydroxylase (hDOHH), which helps to regulate cell proliferation in humans.³ These enzymes activate O₂ at diiron(II) active sites that form (μ-1,2-peroxo)diiron(III) centers called P.^4a,5 For sMMO, P converts into a diiron(IV) intermediate Q that is directly responsible for methane oxidation.^4b An analogous O–O bond cleavage step is proposed for hDOHH,^5b but the putative high-valent oxidant has not been trapped. Yet another diiron enzyme, cyanobacterial aldehyde deformylating oxygenase (cADO), converts fatty aldehydes into alkanes, a reaction initiated by nucleophilic attack of the corresponding peroxo intermediate on the aldehyde functionality of the substrate.^6,7 These examples illustrate the mechanistic diversity of the peroxodiiron(III) unit in this family of enzymes.

In recent years, synthetic peroxodiiron(III) complexes have served to model such nonheme diiron enzyme intermediates.¹ Kodera et al. have described a (μ-oxo)(μ-1,2-peroxo)diiron-(III) complex supported by a bis-TPA [TPA = tris(pyridyl-2-methyl)amine] ligand that undergoes O–O bond cleavage to oxidize the benzylic C–H bonds of toluene.⁸ More recently, Kaizer et al. have reported a peroxodiferric species supported by 2-(4-thiazolyl)benzimidazole ligands that exhibits ambiphilic reactivity in aldehyde deformylation, oxidation of the O–H bonds of phenols, and oxidative demethylation of DMA.⁹ In this study, we focus on another (μ-oxo)(μ-1,2-peroxo)diiron(III) intermediate that exhibits even greater oxidative versatility.

Previously, [Fe^II(IndH)(CH₃CN)₃](ClO₄)₂ [1; IndH = 1,3-bis(2′-pyridylimino)isoindoline] has been found to react with H₂O₂ at 25 °C in acetonitrile (MeCN) to form a transient green species 2 (Figure 1 left). This species has been identified as [Fe^III₂(μ-O)(μ-1,2-O₂)(IndH)₂(solv)₂]²⁺ and shown to oxidize thioanisoles and benzyl alcohols.^10a Remarkably, we herein demonstrate 2 to be an even more versatile and powerful oxidant that converts cyclic ketones to lactones, epoxidizes olefins, and even oxidizes cyclohexane.

Figure 1.

Left: Spectral changes upon cyclohexanone oxidation by 2 in MeCN at 5 °C with [1] = 0.5 × 10⁻³ M and [H₂O₂] = 2.0 mM. Inset: Plot of the rate constants for 2 decay versus [cyclohexanone]. Right: Oxidation of cyclohexane (0.1 M) by 1 (1 mM) and H₂O₂ (0.1 M) at 25 °C, monitoring the change in [2] (●) at 680 nm and the concentration of cyclohexanone formed (mM, ■) versus time.

In a previous publication, we reported that 2 decayed over 10 min at 5 °C with k_decay = 1.16(5) × 10⁻³ s⁻¹ (Figure S1) and with ΔH^‡ = 81(5) kJ mol⁻¹ and ΔS^‡ = −10(10) J mol⁻¹ K⁻¹ (Figure S2a), reflecting a unimolecular decay process.^10a,b In our current study, we find that the addition of cyclohexanone accelerates its decay at rates that depend linearly on [cyclohexanone] (Figure 1, left), indicating a direct reaction between 2 and cyclohexanone with k₂ = 0.4 M⁻¹ s⁻¹. Caprolactone is formed with up to 12.5 TON (relative to 1; see Table 1), demonstrating catalytic conversion of cyclohexanone to caprolactone. Under these conditions, the decay of 2 exhibits ΔH^‡ = 22(1) kJ mol⁻¹ and ΔS^‡ = −170(10) J mol⁻¹ K⁻¹ (Figure S2a), parameters similar to those for PhCHO oxidation by 2.^10b When the experiment is carried out in the presence of H₂ ¹⁸O, no ¹⁸O is incorporated into the lactone product (Figure S2b), analogous to Baeyer–Villiger oxidations.¹¹ Complex 2 thus acts as a nucleophilic oxidant (Scheme 1, top left) and represents a rare example of a diferric–peroxo complex that performs the role proposed for the corresponding intermediate in the cADO mechanism.⁷

Table 1.

Substrate Oxidation Rates at 25 °C, Products, and Turnover Numbers for 1-Catalyzed Reactions^a

substrate [DC–H in kcal mol−1]	k2 (M−1 s−1)	product [% 18O]b	TON
cyclohexanone	0.4	caprolactone	13
PhSMe	10.5	PhS(O)Me [43]b	56
cyclooctene	4.2	epoxide [44]b	36
1-octene	1.2	epoxide	20
Ph3CH [81]	0.072	Ph3COH [50]b	10
PhCH(CH3)2 [84]	0.054	cumyl alcohol	7
PhCH2CH3 [87]	0.019	PhC(O)Me	12
PhCH3 [90]	0.008	PhCHO (KIE 37; Hammett Hρ = −0.42)	5
c-C8H16 [96]	0.0062	C8H14O	7
c-C6H12 [99]	0.0045	C6H10O	3

^a0.1 mmol of substrate, 0.25 mmol of H₂O₂, 1 μmol of 1, and 10 mL of CH₃CN.

^b% ¹⁸O label incorporated into oxidation products. Reaction conditions: 4 mM 1, 4 mM H₂O₂, 250 mM substrate, 25 °C.

Scheme 1.

Proposed Oxidation Mechanism via 2

More significant than acting as a nucleophilic oxidant, 2 also carries out catalytic electrophilic hydrocarbon oxidation (Table 1). Unlike its reaction with cyclohexanone, 2 decays at a rate of ~0.02 s⁻¹ at 25 °C, which is independent of the nature of the hydrocarbon and its concentration (Figure S3). These results show that 2 itself cannot be the actual oxidant but must evolve into a more powerful species to generate the oxidized products (Figure 1, right). Interestingly, the second-order rate constants associated with hydrocarbon oxidation show a linear dependence on [substrate] (Figure 2, left) and span over a 1000-fold range, with O-atom transfer to a C═C bond being much faster than C–H bond cleavage (Table 1). These observations indicate that the rate-determining step for these catalytic reactions must involve substrate oxidation. Kinetic measurements of toluene oxidation (see the Table S2 data with the gray background and Figure S4) show a first-order (not half-order) dependence on [2], excluding the possibility that 2 dissociates into two mononuclear Fe^IV═O units.

Figure 2.

Left: Plot of the product formation rates versus [substrate] in reactions of cyclooctene (red ●), 1-octene (red ◆), cumene (black ■), and cyclohexane (black ▲) with 0.05 mM 1 and 250 mM H₂O₂ in CH₃CN at 25 °C. Right: Linear correlation between log k_ox per substrate H atom and the corresponding C–H BDEs.

The oxidative reactivity of 2 resembles that of nonheme Fe^IV═O complexes.¹² As shown in Figure 2 (right), there is a linear correlation between the logarithms of the product formation rate constants normalized per equivalent target H atom for substrates that undergo C–H bond attack versus C–H bond dissociation enthalpies (BDEs), showing that C–H bond cleavage is an important component of the product formation step. In addition, a nonclassical H/D kinetic isotope effect (KIE) of 37 for toluene is observed (Figure S4), concurring with Kodera’s observations for a related diiron catalyst supported by a dinucleating ligand.^8b Such large nonclassical values are associated with nonheme iron(IV) oxo complexes¹² and are much larger than the classical values for catalytic alkane hydroxylations by mononuclear Fe(L)/H₂O₂ systems involving Fe^V═O oxidants.¹³

Further insight into the nature of the oxidant has been obtained from H₂ ¹⁸O-labeling experiments, which show significant label incorporation into the oxidized products. As much as 40–50% ¹⁸O is incorporated into the PhS(O)Me, cyclooctene oxide, and Ph₃COH products, which represent three different types of reactions (Table 1). Because labeled oxygen from H₂ ¹⁸O cannot exchange into a peroxide moiety without prior O–O bond cleavage, the results show that the actual oxidant in these electrophilic reactions must involve a species formed subsequent to O–O bond cleavage of 2 and capable of undergoing label exchange with H₂ ¹⁸O. When examined as a function of the H₂ ¹⁸O concentration, the % ¹⁸O incorporated into cyclooctene oxide increases linearly with [H₂ ¹⁸O] but plateaus above 0.6 M H₂ ¹⁸O (Figure 3, left). This saturation behavior implicates a reversible H₂O binding step prior to O-atom transfer to the substrate. Indeed, the high degree of ¹⁸O-label incorporation observed indicates reaction conditions that approach complete equilibration within the putative Fe^IV(O)(OH₂) unit. Moreover, ¹⁸O-label incorporation into cyclooctene oxide is found to be independent of the cyclooctene concentration (10–300 equiv; Figure S6), suggesting that ¹⁸O-label exchange is much more facile than substrate oxidation.

Figure 3.

Left: Plot of % ¹⁸O incorporation into product versus [H₂ ¹⁸O] in cyclooctene oxidation (300 equiv) with 1 and H₂O₂ (5 equiv) in CH₃CN at 25 °C. Right: % ¹⁸O labeling of the epoxide product of cyclooctene (200 equiv) by 1 (4 mM), H₂O₂ (2 equiv), and H₂ ¹⁸O (250 equiv), with added PyO (black ●), 4-MeO-PyO (red ■), or 4-NO₂-PyO (blue ◆).

The observed saturation behavior for ¹⁸O incorporation in cyclooctene oxidation (Figure 3, left) resembles that reported for Fe(TPA)-catalyzed hydrocarbon oxidations with H₂O₂ as the oxidant.¹⁴ For the latter, the percentage of incorporation of ¹⁸O from H₂ ¹⁸O into cyclohexanol and cyclooctane-1,2-diol products increased linearly with [H₂ ¹⁸O] and maximized above 0.3 M with added H₂O. These results were rationalized by H₂ ¹⁸O exchange into the site cis to the hydroperoxo unit on the [(TPA)Fe^III(OOH)(solv)]²⁺ intermediate. Subsequent O–O bond heterolysis formed a putative cis-Fe^V(O)(¹⁸OH) oxidant, which, in turn, underwent oxo–hydroxo tautomerization to introduce ¹⁸O to the high-valent Fe═O unit, thereby accounting for the observed partial ¹⁸O labeling of the products. An analogous mechanism can be formulated for 2, which has an available site on each Fe center for label exchange with H₂ ¹⁸O to allow ¹⁸O incorporation into the putative high-valent O═Fe^IV–O–Fe^IV═O oxidant 3 (Scheme 1). Oxidant 3 is analogous to the oxidant that Kodera postulated for his diferric–peroxo intermediate, which is supported by an octadentate 6,6′-(ethylene-bridged)-bis(TPA) ligand,⁸ but the Kodera oxidant does not have solvent-exchangeable sites that allow exchange with H₂ ¹⁸O to afford the labeled products.

This proposed mechanism is further supported by the drop in % ¹⁸O incorporation found for the cyclooctene oxide product with increasing equivalents of pyridine-N-oxide (PyO) added into the reaction mixture, reflecting the competition between PyO and H₂O for the labile sixth site on each Fe (Figure 3, right). Furthermore, introducing 1 equiv of the more basic 4-MeO-PyO instead of PyO results in a lower % ¹⁸O incorporation, while adding 1 equiv of the less basic 4-NO₂-PyO has very little effect. These results show that PyO competes with H₂ ¹⁸O for binding to 2 in Scheme 1. As a control, PyO added to the reaction mixture in place of H₂O₂ produces no epoxide, showing that PyO does not act as an oxidant in this reaction.

Given that 2 can carry out both nucleophilic and electrophilic transformations, we have also conducted competitive oxidations between cyclohexanone and 1-octene or toluene. Consistent with our mechanistic hypothesis, electrophilic oxidation products decrease in yield with higher [cyclohexanone], raising the yield of caprolactone (Figure 4). These results show that nucleophilic oxidation of cyclohexanone competes with the O–O bond cleavage step required to generate the electrophilic oxidant responsible for 1-octene or toluene oxidation.

Figure 4.

Yields of 1-octene oxide (red ■, left) and PhCHO (red ■, right) from the oxidations of 1-octene (left) and toluene (right) by 1/H₂O₂ versus added cyclohexanone. Caprolactone, the cyclohexanone oxidation product, is represented by blue ● in both panels. Reaction conditions: 0.1 mM 1, 0.25 M H₂O₂, 0.1 M substrate in CH₃CN, 25 °C.

Among the many diferric–peroxo complexes characterized thus far as models for peroxo intermediates in nonheme diiron enzymes,¹ 2 stands out as the only synthetic diferric–peroxo species reported to date found to oxidize cyclohexane and the only one that carries out both nucleophilic and electrophilic oxidations under catalytic conditions. Specifically, 2 catalytically converts cyclohexanone to caprolactone in competition with the epoxidation of C═C bonds and the hydroxylation of aliphatic C–H bonds as strong as those in cyclohexane (Table 1). In fact, caprolactone formation is favored over epoxidation and hydroxylation (Figure 4). Whereas the rate of 2 decay is dependent on the cyclohexanone concentration, it is unaffected by the addition of either olefins or alkanes. These results show that 2 reacts directly with cyclohexanone to initiate its lactonization but must evolve into a more powerful oxidant that oxidizes the latter substrates. In support, ¹⁸O from H₂ ¹⁸O is incorporated into the epoxide and alcohol products (Table 1), indicating prior cleavage of the O–O bond of 2 to form the high-valent iron–oxo species that oxidizes the hydrocarbons. Diferric–peroxo intermediate 2 is thus quite a versatile reagent, with its diverse reactivity supporting the notion that nonheme diiron enzymes share a common diferric–peroxo intermediate that can be tuned to perform the electrophilic functions of hydroxylases like sMMO and the nucleophilic functions of deformylating enzymes like cADO.¹