Dioxygen-Derived Nonheme Mononuclear Fe^III(OH) Complex and Its Reactivity with Carbon Radicals

Abstract

A new tetradentate, monoanionic, mixed N/O donor ligand (BNPA^Ph2O⁻) with second coordination sphere H-bonding groups has been synthesized for stabilization of terminal Fe^III(OH)(X) complexes. The complex [Fe^II(BNPA^Ph2O)(OTf)] (1) reacts with O₂ to give a mononuclear terminal Fe^III(OH) complex, [Fe^III(OH)(BNPA^Ph2O)(OTf)] (2), both of which were characterized by X-ray diffraction, electrospray ionization mass spectrometry, UV–vis, ¹H and ¹⁹F nuclear magnetic resonance, ⁵⁷Fe Mössbauer, and electron paramagnetic resonance spectroscopies. Treatment of 2 with carbon radicals (Ar₃C·) gives Ar₃COH and the Fe^II complex 1, in direct analogy with the elusive radical “rebound” process proposed for nonheme iron enzymes.

Dioxygen activation at nonheme iron centers is of fundamental importance for a range of enzymatic and synthetic catalysts.^1–3 An archetypal example is seen in the alpha-ketoglutarate (α-KG) dependent nonheme iron enzyme taurine dioxygenase (TauD). The mechanism of TauD involves the binding of O₂ to Fe^II, followed by electron transfer from the Fe^II/α-KG unit to cleave the O–O bond and give a ferryl (Fe^IV(O)) intermediate. The Fe^IV(O) species then cleaves the substrate C–H bond to give an Fe^III(OH) complex and substrate radical, which recombines via an “oxygen rebound” mechanism to give the final hydroxylated product and the reduced Fe^II active site (Scheme 1).^4–8 A related class of nonheme Fe enzymes are the α-KG-dependent halogenases, which exhibit an interesting mechanistic divergence at the ferric hydroxide stage, in which a coordinated halide, or other pseudohalide X (e.g., X = Cl⁻, Br⁻, N₃⁻, NO₂⁻),^9–13 can preferentially recombine with the substrate radical in lieu of the hydroxide group. Recent efforts to model the latter processes have been described.^14–18 The factors that may bias an Fe^III(OH)(X) center toward rebound of X⁻ versus OH⁻ are still not well understood. A similar conundrum arises in the proposed mechanism for isopenicillin N synthase (IPNS), in which a nascent carbon radical recombines with a ligated sulfur center instead of the OH⁻ ligand.^19–21

Scheme 1.

Selective Rebound Step in Different Nonheme Iron Enzymes

Our research group has made significant efforts toward constructing inorganic analogs of sulfur-ligated thiol dioxygenases,^22–26 a subclass of nonheme Fe dioxygenases. Part of these efforts has involved discerning the general structural requirements for synthesizing mononuclear Fe^II complexes that facilitate O₂ activation. In the course of this work, we have also become interested in examining the factors that contribute to the control of sulfur versus OH rebound in IPNS, as well as more generally, the factors that influence Fe^III(OH) reactivity in hydroxylases and halogenases. We recently described the observation of C–O bond formation with a porphyrinoid, formally Fe^IV(OH) complex,²⁷ and a nonheme Fe^III(OMe) complex,²⁸ in reactions with carbon radicals. However, reactivity between an isolated nonheme Fe^III(OH) complex and a carbon radical, in direct analogy with the enzymatic processes described in Scheme 1, has not been achieved in any synthetic system to date. Such studies are hampered by the inherent difficulty of synthesizing well-defined, mononuclear Fe^III(OH) complexes, which arises in part from their tendency to convert to O(H)-bridged, diferric species.^29,30 Only a relatively small number of mononuclear, nonheme, terminal Fe^III(OH) complexes have been structurally characterized.^31–39

Herein we describe a new tetradentate ligand that was designed to fulfill three criteria: (1) stabilization of an Fe^III(OH) unit, (2) O₂ activation at an Fe^II center, and (3) inclusion of an open site cis to the OH group. This ligand led to the isolation of an Fe^III(OH) complex that was obtained via activation of O₂. We can react this complex with exogenous carbon radicals to give the hydroxylated product and the reduced ferrous form, providing a direct analog of the radical “rebound” step in nonheme iron oxygenases.

The new tetradentate ligand BNPA^Ph2OH, and Fe^II complex Fe^II(BNPA^Ph2O)(OTf) (1), were prepared as shown in Scheme 2. The neopentylamine substituents were included as potential stabilizing H-bond donors,⁴⁰ a strategy that has been used by others to stabilize metal/oxygen species.^{32,33,35,41–44} In addition, an anionic alkoxide donor was installed to facilitate O₂ activation. Metalation with Fe^II followed by crystallization gave yellow-green crystals of 1 suitable for X-ray diffraction (XRD). The structure of 1 (Figure 1) reveals a 5-coordinate (τ = 0.67)⁴⁵ mononuclear iron(II) complex, with a triflate ligand occupying the fifth site. The neopentyl amine groups are oriented toward the triflate, forming two hydrogen bonds with N1(H)–O2 = 3.026(2) Å and N5(H)–O2 3.116(2) Å.

Scheme 2.

Synthesis of BNPA^Ph2OH and Complex 1

Figure 1.

Displacement ellipsoid plots (50% probability level) for 1 and 2 at 110(2) K. Hydrogen atoms (except for N–H and O–H) have been omitted for clarity.

Complex 1 is pale yellow in CH₃CN, with UV–vis maxima at 323 nm (ε = 9550 M⁻¹ cm⁻¹) and 420 nm (ε = 990 M⁻¹ cm⁻¹). The ¹H nuclear magnetic resonance (NMR) spectrum of 1 in CD₃CN reveals paramagnetic peaks from +100 to −20 ppm (Figure 2), indicative of a high-spin (hs) Fe^II complex. The ¹⁹F NMR spectrum in CD₃CN shows a sharp peak at −78 ppm, whereas a broad peak at −74 ppm is observed in tetrahydrofuran (THF)-d₈ (Figure S25), indicative of free and bound triflate ligand, respectively. These data suggest the CD₃CN displaces the OTf⁻ ligand and the H-bonded site is relatively labile, which would allow for O₂ binding. Mössbauer spectroscopy on solid 1 at 80 K shows a sharp quadrupole doublet with δ = 1.03 and |ΔE_Q| = 2.42 mm s⁻¹, also typical of a hs-Fe^II center (Figure S15).

Figure 2.

¹H NMR spectra (CD₃CN) of 1 (top), 2 (middle), and reaction of 2 with excess (Ph₃C)₂ (bottom) after removal of THF and dissolution in CD₃CN.

Complex 1 was reacted with excess, dry O₂ gas in CH₃CN at 23 °C, causing an immediate color change from pale yellow to dark orange, with new UV–vis features appearing at 315 nm (ε = 9000 M⁻¹ cm⁻¹), 365 nm (ε = 2400 M⁻¹ cm⁻¹), and 440 nm (ε = 950 M⁻¹ cm⁻¹). The new dark orange species is air-stable, and the CH₃CN can be removed in vacuo to give an orange solid, which was dissolved in toluene/THF and layered with pentane to afford orange crystals in 24 h that were suitable for XRD.

Analysis by XRD revealed a six-coordinate Fe^III(OH) complex, Fe^III(BNPA^Ph2O)(OH)(OTf) (2) (Figure 1). A terminal hydroxide ligand occupies the hydrogen bonding site, and the axial triflate ligand of 1 has shifted to an equatorial position cis to the OH⁻ ligand in 2. The hydroxide ligand forms two intramolecular hydrogen bonds with the neopentyl amine substituents, with N1(H)–O2 = 2.831(3) and N5(H)–O2 = 2.857(2) Å. There is a third H-bonding interaction between the OH⁻ and the OTf⁻ ligands, with O2(H)–O5 = 2.832(1) Å. The Fe–O(H) bond length is 1.880(1)Å, in the range of other H-bonded and non-H-bonded Fe^III(OH) complexes.^31–35 However, a proper comparison with a non-H-bonded analog of 2 is currently not available. The Fe–N_py and Fe–N_amine distances are shorter by 0.1–0.2 Å when compared to the Fe^II analog 1, and the Fe–O1 distance is also contracted by 0.05 Å.

Complex 2 shows paramagnetic peaks in the ¹H NMR spectrum from 150 to 1 ppm (Figure 2). The ¹⁹F NMR spectrum for 2 in CD₃CN shows a sharp peak at −74.6 ppm, which corresponds to free OTf⁻, and indicates the OTf⁻ ligand is likely displaced by solvent. In contrast, the same spectrum in the more weakly coordinating THF-d₈ shows no peaks from +200 to −200 ppm, consistent with OTf⁻ remaining coordinated (Figure S26). The EPR spectrum (20 K) reveals signals at g = 9.01, 4.23, which is consistent with an S = 5/2 ferric ion (Figure S14). Analysis of a crystalline sample of 2 by Mössbauer spectroscopy at 80 K reveals a broad quadrupole doublet which can be fit with δ = 0.47 mm s⁻¹ and |ΔE_Q| = 0.97 mm s⁻¹ (Figure S16). The broadened spectrum comes from 2 likely being in an intermediate spin relaxation regime.^46,47 The Mössbauer spectrum of 2 at 5 K shows a 6 line hyperfine splitting pattern which is expected for a high-spin Fe^III species (Figure S17).

Density functional theory (DFT) was employed to obtain optimized structures for 1 and 2 at the BP86/6–311G*/6–31G* (for C and H atoms) level of theory. These optimized structures matched well with the structures from XRD for 1 and 2, and were then used for the calculation of Mössbauer spectroscopic parameters. The calculations gave δ = 1.05 and |ΔE_Q| = 2.61 mm s⁻¹ for 1, and δ = 0.47 and |ΔE_Q| = 1.42 mm s⁻¹ for 2. These parameters are in reasonable agreement with experiment for the solid-state samples.

Production of 2 by addition of excess O₂ in acetonitrile followed by electrospray ionization mass spectrometry (ESI-MS) reveals the major peak at m/z = 636.85 (positive ion mode), corresponding to [2–OTf]⁺. In comparison, the use of isotopically labeled ¹⁸O₂ leads to the major peak shifted by +2 units to m/z = 638.85. Addition of excess H₂¹⁸O to 1 in CH₃CN, followed by exposure to excess, natural abundance O₂, does not lead to any shift in the major peak at m/z 636.85. These ¹⁸O-labeling experiments show that the hydroxide ligand must originate from dioxygen.

A suggested mechanism is given in Scheme S1. Initial binding of O₂ gives an Fe^III(superoxo) species that then leads to a peroxo-bridged dimer, followed by homolytic O–O cleavage to give an Fe^IV(O) species. The latter species then abstracts hydrogen from an exogenous source (e.g., solvent) to give 2. Precedent for this mechanism is described for both heme and nonheme Fe and Mn complexes.^33,48–50 Carrying out the reaction of 1 with O₂ in the presence of excess 9,10-dihydroanthracene, a potential H atom donor, leads to production of anthracene in ~90% yield (Scheme 3), consistent with H atom abstraction by the putative ferryl intermediate. Attempts to observe intermediates in the reaction between 1 and O₂ were carried out at lower temperature. However, at temperatures as low as −90 °C, only the formation of 2 was observed as seen at 23 °C, and at lower temperatures (e.g., −105 °C) no reaction occurs.

Scheme 3.

Activation of Dioxygen by 1

The similarity of complex 2 to the enzymatic species shown in Scheme 1 motivated us to examine its reactivity with carbon-based radicals. Addition of Ph₃C· to 2 in THF at 23 °C causes a color change from dark orange to yellow over 4 h. The ¹H NMR spectrum shows complete loss of 2, and the appearance of peaks for 1 (Figure 2). A new peak at δ 5.50 ppm was also formed, arising from the alcohol (Ph₃COH) in 84% (±3) yield (Scheme 4).

Scheme 4.

Reaction between 2 and Triarylmethyl Radicals

Changing to the more electron-rich (p-OMe-C₆H₄)3C· (1 equiv) causes faster conversion (~45 min) of 2 to 1 as seen by ¹H NMR spectroscopy. Quantification of (p-OMe-C₆H₄)₃COH gives a yield of 87% (±2) (Scheme 4), consistent with 1:1 stoichiometry between 2 and the radical.

The formation of the products (p-X-C6H₄)₃COH (X = H, OMe) was also confirmed by ESI-MS analysis (Figures S22 and 23). Incorporation of ¹⁸OH into 2 by exchange with H₂¹⁸O was quantitative, and reaction of (p-X-C₆H₄)₃C· (X = H, OMe) with ¹⁸O-labeled 2, led to a two unit shift in m/z by ESI-MS analysis (Figures S28 and 29, ¹⁸O incorporation, 99%)). These results indicate that there is direct transfer of the OH ligand to the carbon radicals.

The reaction with the p-OMe derivative was monitored by Mössbauer spectroscopy (Figure 3). The spectrum for ⁵⁷Fe-labeled complex 1 in 2-MeTHF(80 K) gives δ = 1.14 mm s⁻¹ and |ΔE_Q| = 2.24 mm s⁻¹ (Figure 3a). The spectrum for complex 2 (Figure 3b) is broad, with δ = 0.47 mm s⁻¹ and |ΔE_Q| = 1.24 mm s⁻¹ (see Figure S18 for fitting details). Reaction between 2 and (p-OMe-C₆H₄)₃C· gives a quadrupole doublet matching 1 (δ = 1.14 mm s⁻¹; |ΔE_Q| = 2.23 mm s⁻¹) (Figure 3c), which accounts for 90% of the total Fe.

Figure 3.

Zero field ⁵⁷Fe Mössbauer spectra (80 K) of (a) 1, (b) 2 in 2-MeTHF, and (c) reaction of 2 with (p-OMe-C₆H₄)₃C· (at 23 °C) in 2-MeTHF. Black circles = experimental data; red line = best fit.

Electrochemical analysis of 2 shows a reversible wave in the cyclic voltammogram with E_1/2 = −0.69 V (ΔE_pp = 170 mV) vs Fc⁺/Fc in CH₃CN. This reduction potential is in line with other Fe^III(OH) complexes (Table S8). The redox potentials for (p-X-C₆H₄)₃C⁺/(p-X-C₆H₄)₃C· are −0.11 V (X = H) and −0.58 V (X = OMe), in the same solvent,⁵¹ indicating that outer-sphere electron-transfer (ET) from (p-X-C₆H₄)₃C· to 2 should be endergonic by 13.4 kcal mol⁻¹ for X = H, and 2.5 kcal mol⁻¹ for X = OMe. Thus, ET to 2 is thermodynamically unfavorable for both carbon radicals, suggesting that a concerted process may be operative (Scheme S2).

In conclusion, the incorporation of an anionic donor combined with H-bond groups in the second coordination sphere of a new tetradentate ligand provided access to an Fe^II complex that activates O₂ and yields a stable, terminal Fe^III(OH) complex. Reaction of the latter complex with trityl radical derivatives led to the first direct observation⁵² of a model “rebound” reaction to give the hydroxylated carbon product and the one-electron-reduced Fe^II complex. It may be somewhat surprising that smooth transfer of OH^• from Fe to the radical occurs, because of the significant steric encumbrance and tight H-bond network in 2. Determining what inherent factors (e.g., H-bonding, sterics, redox potential, donor atoms) influence O₂ activation and Fe(OH)/radical reactivity will be the focus of future studies.

In addition, the new BNPA^Ph2O⁻ ligand leaves a potential labile equatorial site on the metal to build Fe(OH)(X) complexes. Efforts to replace the equatorial OTf⁻ in 2 with other anionic donors are underway.