Proton-Coupled Electron Transfer Reactivity Controls Iron Versus Sulfur Oxidation in Nonheme Iron Thiolate Complexes

Abstract

Reaction of the 5-coordinate Fe^II(N₄S) complexes, [Fe^II(iPr₃TACN)(abt^x)](OTf) (abt = aminobenzenethiolate, X = H, CF₃) with a one-electron oxidant and an appropriate base leads to net H-atom loss, generating new Fe^III(iminobenzenethiolate) complexes that were characterized by single-crystal X-ray diffraction (XRD), as well as UV-vis, EPR, and Mössbauer spectroscopies. The spectroscopic data indicate that the iminobenzenethiolate complexes have S = 3/2 ground states. In the absence of a base, oxidation of the Fe^II(abt) complexes leads to disulfide formation instead of oxidation at the metal center. Bracketing studies with separated proton-coupled electron-transfer (PCET) reagents show that the Fe^II(aminobenzenethiolate) and Fe^III(iminobenzenethiolate) forms are readily interconvertible by H⁺/e⁻ transfer, and provide a measure of the bond dissociation free energy (BDFE) for the coordinated N–H bond between 64 – 69 kcal mol⁻¹. This work shows that coordination to the iron center causes a dramatic weakening of the N–H bond, and that Fe- versus S- oxidation in a nonheme iron complex can be controlled by the protonation state of an ancillary amino donor.

SYNOPSIS TOC

One-electron oxidation of the thiolate-ligated [Fe^II(iPr₃TACN)(abt^x)]⁺ (X = H, CF₃) complexes in the presence of an appropriate base results in the formation of [Fe^III(iPr₃TACN)(ibt^x)]⁺ via PCET. In contrast, oxidation in the absence of base leads to disulfide, indicating that the amino ligands offer a protective effect for the thiolate donors during redox activity. The N–H bonds of the amino ligands are dramatically weakened by ~30 kcal mol⁻¹ via coordination to the redox-active nonheme iron(II) centers.

Graphical Abstract

Introduction

Nonheme iron oxygenases and oxidases are a major class of metalloenzymes that utilize non-porphyrinic Fe centers to selectively carry out a diverse array of difficult oxidative reactions, including hydroxylation, halogenation, C–C bond cleavage, cyclization, S-oxygenation, and C–S coupling.^1–3 A subset of these enzymes are those with sulfur-ligation, which includes the thiol dioxygenases (e.g., CDO),^4–6 isopenicillin N synthase (IPNS),^7–8 the sulfoxide synthases (e.g., EgtB, OvoA),^9–10 and the persulfide dioxygenases (e.g., ETHE1).^11–12 These enzymes are proposed to bind O₂ at a mononuclear iron center cis to a coordinated sulfur donor. However, the steps following O₂ binding diverge for the different enzymes, leading to either C–S or O–S bond formation, or both.³ The proposed mechanisms of O₂ activation for these enzymes are closely related, and subtle factors likely dictate the operative reaction pathways.^13–16

In comparison, synthetic, sulfur-ligated nonheme iron complexes react with O₂ or its reduced surrogates (e.g., H₂O₂, O₂^•−) to give sulfur mono-,¹⁷ di-,^18–23 or tri- oxygenation,^24–25 disulfide formation,^{18, 22} C–S bond formation,²⁶ or metal center oxidation.^27–29 These reactions are influenced by many factors, such as ligand architecture, redox potentials, metal coordination number and geometry, and reaction conditions (e.g., solvent, temperature). Determining the underlying principles that govern the outcome of these oxidation reactions should help improve our understanding of both nonheme iron enzymes and synthetic iron oxidation catalysts.

Recent work in our laboratory has focused on the construction of Fe(thiolate) complexes bearing derivatives of the neutral, tridentate ligand triazacyclononane (R₃TACN; R= Me, iPr) and a bidentate, thiolate co-ligand to model key structural and functional properties of nonheme iron enzymes with sulfur ligation.^{18, 30–31} Our ability to synthetically tune both the R₃TACN and bidentate donor ligand provides a means to control the steric profile and electron-donating characteristics of the coordination environment about the metal center.

We previously reported a series of Fe^II(N₄S(thiolate)) complexes with Me₃TACN or iPr₃TACN and 2-aminobenzenethiolate (abt^x; X = H, CF₃) ligands, as models of the thiol dioxygenases.¹⁸ We demonstrated that the Me₃TACN complexes, Fe^II(Me₃TACN)(abt)(OTf) (1) and Fe^II(Me₃TACN)(abt^CF3)(OTf) (2), react with dioxygen in methanol at low temperature (−95 °C) to generate S-oxygenated, sulfinate products, analogous to native thiol dioxygenase reactivity (Figure 1). However, the analogous iPr₃TACN complexes [Fe^II(iPr₃TACN)(abt^x)]⁺ (X = H, CF₃) rapidly decompose in MeOH under anaerobic conditions, preventing a comparative study of their O₂ reactivity with that observed for the Me₃TACN derivatives. In contrast, the iPr-substituted Fe^II complexes are stable in aprotic solvents such as acetonitrile. Here we report the reactivity of the iPr-substituted complexes, [Fe^II(iPr₃TACN)(abt)](OTf) (3) and [Fe^II(iPr₃TACN)(abt^CF3)](OTf) (4) with O₂ in CH₅CN. Instead of S-oxygenation, we show that 3 and 4 react with O₂ in the presence of a base via proton-coupled electron transfer (PCET), with the Fe-bound aniline N–H bond cleaving to yield an Fe^III-iminobenzenethiolate species. Similar reactivity is observed with different combinations of bases and outer-sphere oxidants. However, in the absence of a base, oxidation of 3 or 4 generates disulfide, indicating that the aniline protonation state determines whether Fe or S is the site of oxidation.

Figure 1.

Reaction of Fe(Me₃TACN)(abt^x)(OTf) complexes with O₂.

Results and Discussion

Addition of excess dry O₂ gas to colorless solutions of 3 or 4 in CH₃CN results in a gradual color change to red-purple. Analysis of the reaction mixture of 3 and excess O₂ in CD₃CN by ¹H NMR spectroscopy reveals a mixture of unreacted 3, the disulfide compound, 2,2′-di-sulfanediyldianiline, and an unknown paramagnetic species, that likely corresponds to the purple color. Given the persistence of the purple species over 1 h at 23 °C, we anticipated that this product could be an S-oxygenate as opposed to an Fe/O₂ adduct (e.g. Fe^III(O₂^•−)), which are usually only observable at lower temperatures. We attempted to generate this species in high yield by reaction of 3 with the O-atom donor isopropyl 2-iodylbenzoate (IBX-ester). Addition of 0.5 equiv of IBX-ester (2 O atoms per IBX-ester) to 3 in CH₃CN at −40 °C led to immediate formation of a red-purple species. The UV-vis spectrum of the reaction mixture showed bands at 370, 545, and 940 nm. Layering of the reaction mixture with Et₂O and allowing to stand at −23 °C led to maroon needles after several days. X-ray structure determination gave the molecular structure shown in Figure 2. The structure reveals that the complex is a 5-coordinate iron species with formula [Fe(iPr₃TACN)(ibt)](OTf) (ibt = iminobenzenethiolate) (5), with a deprotonated amine group and an intact sulfur donor. The triflate counterion is hydrogen bonded to the ibt NH group, similar to the H-bonding interaction between OTf and the neutral aniline donor in the starting material 3.¹⁸ The overall charge of the complex implies that this product is formally a ferric iminobenzenethiolate (ibt) complex, although ibt-type ligands are known to be non-innocent and other electronic descriptions such as an iron(II) (iminothiobenzosemiquinone radical) are possible (vide infra).

Figure 2.

(a) Displacement ellipsoid plots (50% probability level) of the cations for 5 and 6 at 110(2) K and 293(2) K, respectively. H atoms (except for the H atom attached to N4) were removed for clarity. (b) UV-vis and EPR spectra for 5 – 6. UV-vis spectra were collected in CH₃CN at 23 °C. Insets: EPR spectra (20 K, black line) of 5 – 6 dissolved in butyronitrile. Simulations shown as red lines. Conditions: microwave freq = 9.41 GHz; microwave power = 2.0 mW (for 5) and 0.20 mW (for 6); modulation amp. = 10 G; receiver gain = 5 × 10³.

The conversion of 3 to 5 occurs via the loss of one proton and one electron, suggesting that 3 reacts with IBX-ester via a proton-coupled electron transfer (PCET). This result was not anticipated because iodylarenes are generally known as O-atom transfer agents.³² However, iodylarenes have also been employed as single electron-transfer (SET) agents, oxidizing certain organic substrates by 1H⁺/1e− processes.^{33 1}H NMR spectroscopy indicated that the unknown species generated from the reaction of 3 with O₂ gave identical spectral signatures to crystalline 5 and likely also formed via PCET (eq 1–2). We hypothesize that a reduced product of O₂ (e.g., superoxide) may serve as a base in these reactions and facilitate the PCET chemistry.

$$\mathbf{3}+{\text{O}}_2\to \mathbf{5}+\text{RSSR}+{\left[{\text{Fe}}^{\text{II}}\left({\text{iPr}}_3\text{TACN}\right)\right]}^{2+}$$ (1)

$$3+0.5\text{IBX-ester}\to 5$$ (2)

We thus hypothesized that 5 could be formed quantitatively by a combination of O₂, or a suitable one electron oxidant, with an appropriate base. Addition of stoichiometric or excess (1 – 40 equiv) triethylamine (pK_a = 18.8 in CH₃CN)³⁴ to 3 results in no reaction, as determined by ¹H NMR spectroscopy (Figure S6). However, bubbling of excess O₂ through a solution of 3 and Et₃N in CH₃CN at −35 °C results in the rapid formation of 5 as determined by UV-vis spectroscopy (eq 3, Figure S8).

$$\mathbf{3}+{\text{O}}_2+{\text{Et}}_3\text{N}\to \mathbf{5}$$ (3)

Running the same reaction with 1 equiv of the outer-sphere, one-electron oxidant ferrocenium triflate ([Fc]OTf) and Et₃N also leads to the formation of 5 (eq 4, Figure S11). In contrast, addition of [Fc]OTf (1 equiv) at −35 °C in the absence of base leads to significant release of disulfide and Fe^II(iPr₃TACN)(OTf)₂,³⁵ as determined by NMR and Mössbauer spectroscopies (eq 5, Figures S13–S15).

$$3+{\text{Fc}}^{+}+{\text{Et}}_3\text{N}\to \mathbf{5}$$ (4)

$$3+{\text{Fc}}^{+}\to \text{RSSR}+{\left[{\text{Fe}}^{\text{II}}\left({\text{iPr}}_3\text{TACN}\right)\right]}^{2+}$$ (5)

This observation suggests that oxidation of the metal in the absence of base to generate the Fe^III(abt) complex is unfavorable, and instead results in one-electron sulfur oxidation. The requirement for both base and oxidant to form 5 is consistent with a PCET reaction.

Complex 5 can also be produced by addition of the H-atom abstractor 2,4,6-tri-tert-butylphenoxyl radical (ttbp•) to a solution of 3 in CH₃CN at −35 °C (Figure S16). Immediate formation of purple 5 is observed. Layering of the reaction mixture with Et₂O and allowing to stand at −35 °C produced crystalline 5. This method was used to produce 5 for subsequent spectroscopic analyses.

Analogous reactivity was seen for the CF₃-substituted 4. Addition of Et₃N (1 equiv) to a solution of 4 in CH₃CN results in formation of a light yellow solution, likely due to deprotonation of the aniline donor. However, lutidine (pK_a = 14.13 in CH₃CN),³⁴, a weaker base than Et₃N, does not react with 4. Addition of either excess O₂ or 1 equiv of [Fc]OTf to a solution of 4 and lutidine in CH₃CN produces a red-purple species (λ_max = 390, 530, and 870 nm). These UV-vis features are qualitatively similar to those of 5, and are assigned to the analogous product, [Fe(iPr₃TACN)(ibt^CF3)](OTf) (6). In the spectra for both 5 and 6, the long-wavelength band near 900 nm may be assigned to a sulfur-to-iron(III) charge-transfer band.³⁶ However, a similar band in Ni(abt^x) complexes has been associated with thiyl radical character on the abt^x ligand.³⁷ Complex 6 is also generated via H-atom abstraction from 4 mediated by the ttbp•. Layering with Et₂O at −35 °C afforded maroon needles of 6 suitable for X-ray structure determination. These results, together with those for the para-H derivative, are summarized in Scheme 1.

Scheme 1.

The molecular structures for 5 and 6 are shown together for comparison in Figure 2. Similar to 5, complex 6 is five-coordinate, with molecular formula [Fe(iPr₃TACN)(ibt^CF3)](OTf). Both complexes 5 and 6 exhibit geometries that are intermediate between square pyramidal (sp) and trigonal bipyramidal (tbp) (τ₅ = 0.45 for both 5 and 6; where τ₅ = 0.0 for sp and τ₅ = 1.0 for tbp).³⁸ In both cases, the geometry is closer to square pyramidal compared to the parent Fe^II complexes 3 and 4, which exhibit τ₅ = 0.62 for 3 and τ₅ = 0.63 and 4.¹⁸

Ligand non-innocence in abt complexes is well-established,^39–46 and complexes 5 – 6 could lie anywhere between the two extremes of a ferrous(iminothiobenzosemiquinone radical) (Fe^II(itbs•)) and a ferric(iminobenzenethiolate) (Fe^III(ibt)) formulation (Scheme 2). We note that the covalent nature of Fe^III–thiolate bonds makes it difficult to distinguish between an Fe^III(thiolate) and a coupled Fe^II(thiyl) species, but it is possible to distinguish between a closed shell aryl ring and the imino-semiquinone form via bond distance comparisons.^39–46

Scheme 2.

The metal–ligand bonds in 5 – 6 are contracted compared to the parent ferrous complexes, which is expected for a metal-based oxidation. Complexes 5 and 6 have Fe–S bond lengths of 2.2257(6) Å and 2.2358(9) Å, respectively, which are 0.13 – 0.14 Å shorter than the Fe–S bonds in 3 – 4. These Fe–S bond lengths are similar to those for other reported Fe^III–SR complexes.^39,47–49 A t-butyl-substituted Fe^III₂(abt)₂(ibt)₂ dimeric complex displayed Fe^III–S distances of ~2.21 Å for both the Fe^III–S_abt and Fe^III–S_ibt bonds, which are similar to the Fe–S bonds in 5 – 6.³⁹ In contrast, a related ferrous–iminothiobenzosemiquinone radical complex, Fe^II(abt)₂(itbs•) has a much longer Fe–S_itbs• distance of 2.351 (1) Å.³⁹ The average Fe–N_TACN bond lengths are 2.13 Å and 2.12 Å for 5 – 6, respectively, which is about 0.1 Å shorter than those in the ferrous precursors. There is also significant contraction of the Fe–N_abt bonds, with d(Fe–N_abt) = 2.2555(13) and 2.2656(19) for 3 and 4, respectively, compared to d(Fe–N_ibt) = 1.853(2) and 1.842(3) for 5 and 6, respectively.

The aryl C–C bond lengths in abt complexes have been used to distinguish between the two descriptions in Scheme 2, with alternating long/short bonds indicative of an itbs^x• structure, and more uniform C-C distances pointing to an ibt^x complex.^{39,41–42,40} The aryl C–C bond distances in both 5 and 6 are relatively uniform, supporting an Fe^III(ibt^x) structure. The average C-aryl C-C bond lengths of 1.3953(8) Å for 5 and 1.390(11) Å for 6 matches with those reported for the parent Fe^II complexes (Table S4). Taken together, the structural parameters indicate a closed-shell ligand for 5 – 6.

X-band electron paramagnetic resonance spectroscopy (EPR) corroborated the iron(III) assignments. The EPR spectrum of crystalline 5 in butyronitrile at 20 K exhibits features that were simulated with g_eff = 5.14, 2.71, 1.81, consistent with an Fe^III (S = 3/2) center (Figure 2). Complex 6 displays a similar EPR spectrum with g_eff= 5.20, 2.63, 1.78. Inspection of a rhombogram indicates that each complex has an E/D ~ 0.22 and is consistent with an S = 3/2 species.

Samples of ⁵⁷Fe-enriched 5 and 6 were prepared in situ by H-atom abstraction with ttbp• from 3 or 4 enriched in ⁵⁷Fe (95.5% isotopic purity).¹⁸ The Mössbauer spectra for 5 and 6 are shown in Figure 3, together with the spectra for the starting complexes 3 and 4.¹⁸ These species exhibit nearly identical Mössbauer parameters (5: δ = 0.36 mm s⁻¹, |ΔΕ_Q| = 3.00 mm s⁻¹; 6: δ = 0.34 mm s⁻¹, |ΔE_q| = 3.06 mm s⁻¹). The low isomer shift values for 5 – 6 are consistent with their formulations as Fe^III(ibt) complexes. In contrast, Fe^II(itbs•) complexes have been shown to exhibit much higher isomer shifts, typical of ferrous complexes.⁴⁶ The Mössbauer and EPR spectra further support the description of 5 – 6 as ferric (S = 3/2) centers.

Figure 3.

Zero-field ⁵⁷Fe Mossbauer spectra of (a) 3(⁵⁷Fe), (b) 4(⁵⁷Fe), (c) 5(⁵⁷Fe), and (d) 6(⁵⁷Fe) in CH₃CN. Data shown as black points. Overall fits are shown in red. Conditions and fitting parameters are described in the Supporting Information.

Further insight into the electronic structure of 5–6 was provided by density functional theory (DFT) studies. Geometry optimization of each complex was performed at the B3LYP/6–311G* level of theory and gave similar results to other functional/basis set combinations. The calculated geometries gave bond metrics that were a close match with the experimental XRD data. As seen for the XRD data, the aryl C–C bond lengths are all nearly identical, with an average bond length of 1.401 Å for 5 and 1.399 Å for 6. The average C–C bond lengths are nearly identical to those for the respective Fe^II complexes (Table S4). Analysis of the Löwdin spin density revealed spin density of +2.45 and + 2.5 on Fe for 5 and 6, respectively, and little spin density on the ligand aryl ring or sulfur donors. The spin density analysis is thus consistent with the assignment of 5 – 6 as Fe^III(ibt) complexes. The Mössbauer parameters for the geometry optimized structures of 3 – 6 were calculated using a method previously used for complexes 1 – 4.¹⁸ The calculated parameters for 3 – 4 (3: δ = 0.90 mm s⁻¹, |ΔE_Q| = 2.34 mm s⁻¹; 4: δ = 0.92 mm s⁻¹, |ΔE_Q| = 2.28 mm s⁻¹) are nearly identical with those previously calculated for geometries optimized via a different functional (BP86), and also match the experimental data.¹⁸ The calculated parameters for 5 (δ = 0.32 mm s⁻¹, |ΔE_Q| = 2.33 mm s⁻¹) and 6 (δ = 0.30 mm s⁻¹, |ΔE_Q| = 2.42 mm s⁻¹) indicate that the calculated isomer shift values are in good agreement with the experimental data. The calculated quadrupole splitting values for 5 – 6 are lower than expected. However, it has been previously noted that errors in calculated ΔE_Q are common for complexes with experimental ΔE_Q of larger magnitude.⁵⁰ The relative values for δ (5 > 6) and |ΔE_Q| (6 > 5) are reproduced by the computed Mössbauer parameters.

Further insight into the base-dependent oxidation of 3 – 4 was provided by DFT. The hypothetical one-electron oxidized complexes [Fe^III(iPr₃TACN)(abf^x)]²⁺, in which the aniline donors remain protonated, were examined computationally in the S = 3/2 and S = 5/2 spin states. Löwdin spin population analysis for each complex revealed a significant amount of spin density on the sulfur donor, contrasting that seen for the monodeprotonated ferric complexes 5 – 6, which show essentially no spin on the sulfur atoms (Table S6). These observations suggest that upon one-electron oxidation of 3 – 4 in the absence of a base, thiyl character is imparted to the sulfur, making it more susceptible to disulfide formation. These calculations are consistent with the observed disulfide products formed upon oxidation of 3 – 4 in the absence of base.

Proton-Coupled Electron Transfer.

The thermodynamics of PCET reactions can be broken up into individual proton transfer (pK_a) and electron transfer (E°) steps using Hess’s law, as seen in the thermodynamic “square-scheme” shown in Scheme 3.⁵¹ This thermodynamic analysis has been used previously⁵¹ to rationalize patterns of reactivity and determine the relative importance of redox potentials and pK_a values in predicting PCET reactions. Efforts were made to determine the thermodynamics of each of the separated steps in the square schemes for 3/4 and 5/6.

Scheme 3.

The step involving deprotonation of 3 was monitored by ¹H NMR spectroscopy. Organic bases of varying strength (TBD, DBU, TMG, PhTMG; conjugate acid pK_a values = 20.8 – 26.0 in CH₃CN)^{52, 34} were added to 3 in CD₃CN. For each of these reactions, addition of 1 equiv of base leads to complete loss of the peaks associated with 3, and formation of a new pattern of paramagnetic peaks. However, this pattern was different for each of the bases studied (Figures S17–S20). Despite this discrepancy, addition of a slight excess of a weak acid ([Et₃NH]BF₄) regenerates 3 regardless of which base is used, suggesting that the reaction is fully reversible. The ambiguity in the identity of the product(s) may arise from differences in H-bonding with the abt amine donor or coordination of the base. These observations rule out using these bases for quantitative titrations to determine the pK_a of 3 and also prevent us from establishing an upper limit for the pK_a. However, weaker bases (1 – 20 equiv), including pyridine, lutidine, 2,4,6-trimethylpyridine, DMAP, Et₃N, or quinuclidine (pK_a values: 12.5 – 19.51)^{34, 53} do not show any reaction with 3 as seen by ¹H NMR spectroscopy, implying a lower limit for pK_a (3) > 19.5.

Reaction of 4 with a series of organic bases leads to similar results. Complex 4 exhibits reactivity with bases of pK_a = 17.95 – 26.03, but as seen for 3, the NMR spectra show different final patterns depending upon the identity of the base. For each reaction, addition of lutidinium triflate ([LutH]OTf) causes reformation of 4, consistent with a reversible acid-base reaction. In contrast, no reaction is observed for bases with pK_a < 14.98, implying a lower limit of pK_a (4) > 15.0. In comparison with 3, these results are consistent with the electron-withdrawing nature of the CF₃ group in 4.

The Fe^III(ibt)/Fe^II(ibt) redox potentials of 5 – 6 were determined by cyclic voltammetry in CH₃CN (Figure 4). The cyclic voltammogram of 5 exhibits a quasi-reversible redox event at E_1/2 = −0.84 V vs Fc⁺/Fc and a second irreversible event with E_p,a = +0.24 V. Upon scanning the full potential range shown in Figure 4, the E_1/2 = −0.84 V redox event becomes irreversible following the first scan. The event at −0.84 V is assigned to the Fe^III(ibt)/Fe^II(ibt) redox couple. The second event at E_p,a = +0.24 V likely corresponds to a ligand oxidation that leads to decomposition. Cyclic voltammetry of 6 reveals two redox events at E_1/2 = −0.70 V and E_1/2 = +0.33 V vs Fc⁺/Fc that are qualitatively similar to those of 5 but are shifted to higher potentials and, unlike 5, are both reversible. The anodically shifted redox events for 6 compared with those of 5 are expected for the electron-withdrawing nature of the −CF₃ substituent in 6. The redox event at E_1/2 = −0.70 V is assigned to the Fe^III(ibt)/Fe^II(ibt) couple, and the higher potential couple at E_1/2 = +0.33 V is assigned to a ligand-based redox process.

Figure 4.

Cyclic voltammograms of 5 (5.1 mM, top) and 6 (3.7 mM, bottom) in CH₃CN at 23 °C, with [ⁿBu₄N]OTf (0.1 M) as the supporting electrolyte. Working electrode: platinum; reference electrode: Ag wire; counter electrode: Pt wire. Scan rate: 200 mV s⁻¹.

The thermodynamic quantities of E° and pK_a in Scheme 3 can be related to the bond dissociation free energy for the N–H bond cleaved in 3 or 4 via equation 6 derived from the work of Bordwell⁵⁴ and frequently used for transition metal-mediated PCET reactions.^51,55–59 Inserting the lower limit pK_a values for 3 – 4 and the measured redox potentials for 5 – 6 into eq 6 (where C_G = 54.9 kcal mol⁻¹ and represents the free energy of formation of the hydrogen atom in CH₃CN)⁵¹ gives a lower limit of BDFE_N–H = 62 kcal mol⁻¹ for 3 and 59 kcal mol⁻¹ for 4.

$$\text{BDFE}(\text{N}-\text{H})=1.37\text{p}{K}_{\text{a}}+23.0E\circ +{\text{C}}_{\text{G}}$$ (6)

Further assessment of BDFEs for the N–H bonds in 3 – 4 was performed using bracketing experiments with H⁺/e⁻ donors (Figure 5). It has been shown that PCET reactions can be performed using separated acid/reductant combinations, which exhibit an “effective” BDFE based on their individual pK_a values and redox potentials.^{51, 60-61}

Figure 5.

Reactions with separated PCET reagents.

The acid/reductant combination of lutidinium tetrafluoroborate ([LutH]BF₄) (pK_a = 14.13 in CH₃CN)³⁴ and decamethylferrocene (Fc*) (E_1/2 = −0.51 V in CH₃CN)⁶² has a BDFE_eff = 62.5 kcal mol⁻¹. Addition of Fc* as the e⁻ donor and [LutH]BF₄ as the H⁺ donor to 5 in CH₃CN leads to an immediate color change from purple to pale green. The disappearance of the bands associated with 5 in the UV-vis spectrum are accompanied by formation of new features near 780 nm indicative of [Fc*]⁺. ¹H NMR spectroscopy confirmed that 5 was quantitatively converted to 3 together with [Fc*]⁺ and Lut (Figure S33). The reverse reaction between [Fc*]⁺/Lut and 3 does not occur. These data indicate that 3 has an N-H(BDFE) > 62.5 kcal mol⁻¹. Similar results were obtained using the H⁺/e⁻ combination of 2,4,6-trimethylpyridinium triflate ([^Me3PyH]OTf) (pK_a = 14.98)³⁴ and Fc* (BDFE_eff = 63.6 kcal mol⁻¹).

In contrast, addition of 4-dimethylaminopyridinium tetrafluoroborate ([DMAPH]BF₄) (pK_a = 17.95 in CH₃CN),³⁴ and decamethylferrocene (Fc*) (E_1/2 = −0.51 V in CH₃CN) (BDFE_eff = 67.7 kcal mol⁻¹) to 5 in CH₃CN results in only partial decay of the bands associated with 5 by UV-vis spectroscopy. Monitoring the reaction by ¹H NMR reveals both 3 and 5 in the reaction mixture. The reverse reaction, in which DMAP and [Fc*]OTf are added to a solution of 3, also results in only partial formation of 5. Taken together, these results suggest that 5 and DMAPH⁺/[Fc*] are in equilibrium, consistent with BDFE(N–H) for 3 near 68 kcal mol⁻¹. Reaction of 5 with the acid/reductant pair [Et₃NH]BF₄/Fc* (BDFE_eff = 68.9 kcal mol⁻¹) results in only slight decay of the peaks associated with 5, suggesting a BDFE for 3 that is less than 69 kcal mol⁻¹.

The analogous experiments performed with 6 showed identical results, with 6 reacting with [^Me3PyH]⁺/Fc* to produce 4 and with [DMAPH]⁺/Fc* partially forming 4 (Figures S42–S51). Overall these studies suggest that both 3 and 4 have highly activated N–H bonds between 64 – 69 kcal mol⁻¹.

Support for the weak BDFE values for 3 – 4 was provided by DFT calculations performed with B3LYP/6–311G* as described above. Calculation of the N–H bond strengths of 3 and 4 were derived from the geometry optimization and frequency calculations on 3 – 6 in the gas phase using methods that have been reported previously.^{30, 63} The BDFE(N–H) for 3 and 4 were calculated to be 68.3 kcal mol⁻¹ and 72.1 kcal mol⁻¹, respectively. Similar bond strengths were obtained using the B3LYP/def2-tzvp basis set/functional combination. The BDFE value for 4 is shghtly higher than the experimental data, possibly due to the lack of solvation. The calculated BDE values, obtained without considering entropy contributions, are 65.7 and 69.5 kcal mol⁻¹, for 3 and 4, respectively. Overall, these data support the observation of relatively weak N–H bonds.

Conclusions

Oxidation of the aminothiolate-ligated nonheme iron(II) complexes 3 and 4 can follow one of two pathways. In the presence of an appropriate base, a PCET reaction occurs that involves metal-centered oxidation to give the iron(III) iminothiolate-ligated complexes 5 and 6. This PCET process, which should involve net loss of an H atom, was confirmed by reaction of 3 – 4 with the H-atom abstractor 2,4,6-tri-tert-butylphenoxyl radical, which produces 5 – 6. In contrast, the one-electron oxidation of 3 – 4 in the absence of a base leads to disulfide formation, consistent with purely ligand-centered oxidation. The new complexes 5 and 6 were characterized by UV-vis, EPR, and Mössbauer spectroscopies, and their structures were determined by single crystal XRD. The data show that 5 and 6 are Fe^III complexes with S = 3/2 spin ground states, in which the aromatic ring of the ibt ligand is closed shell and does not exhibit non-innocent behavior. However, it should be noted that metal-thiolate bonds can be highly covalent, and we cannot rule out some contribution of Fe^II(RS^•) (thiyl radical) character in these complexes, similar to what we have seen for a Co^III(Me₂SiS₂²⁻)/Co^II(Me₂SiS₂^•−) complex.³⁰

Electrochemical and pK_a measurements were carried out to analyze the thermodynamics of these PCET reactions. These experiments provided a lower limit for the N-H BDFE of the metal-bound amino donors in 3 and 4 of 62 and 59 kcal mol⁻¹, respectively. Additional experiments involving separated PCET reagents gave a bracketed range of 64 – 69 kcal mol⁻¹ for the N-H BDFEs of 3 – 4, consistent with the lower limits set by the electrochemical/pK_a measurements. Although the N–H BDFE values for the abt^x ligands are not known, the N–H BDFE for aniline is 94 kcal mol⁻¹ in CH₃CN.⁵¹ The ortho thiolato group may contribute in part to the lower N–H BDFE for 3 – 4 compared to aniline, but this effect is expected to be small,^{51, 64} suggesting that N-H bonds in 3 and 4 are dramatically weakened by ~30 kcal mol⁻¹ from coordination to the iron center. This work adds to the growing body of data that shows that metal ion coordination can significantly weaken a range of different types of N–H bonds.^58,65–68 For example, the Ni complexes [SNS]Ni(PR₃) (R = Cy, Ph), in which [SNS] is an amino-dithiolate-based ligand similar to abt, are susceptible to ligand-based H-atom abstraction involving a coordinated amine group. The BDFEs for the N–H bonds in the [SNS] complexes were measured to be 63.9 and 62.4 kcal mol⁻¹ for [SNS]Ni(PCy₃) and [SNS]Ni(PPh₃), respectively.⁵⁸ Similarly, an Os^III(NH₂Ph) complex was measured to have a BDE = 66 kcal mol⁻¹. ⁶⁸

There is a large effect seen for the −CF₃ substituent on the redox properties of 4, in which the E_1/2(Fe^III/Fe^II) value is 140 mV more positive than that seen for unsubstituted 3. However, this large shift in redox potential has little effect on the PCET reactivity and N–H BDFE for 4. The effect on the E_1/2 value is apparently offset by the change in pK_a for the abt^CF3 N–H bond, leading to the bracketed pK_a range for 4 being several units lower than that for 3. These combined effects lead to similar N–H bond strengths for the two complexes (see Figure S52), and show that further tuning of the N-H bonds through ligand substitution maybe difficult to accomplish.

DFT calculations on the hypothetical one-electron oxidized, amino-thiolate forms of 3 and 4 reveal significant spin density on the sulfur donor, consistent with disulfide formation being the preferred one-electron oxidation pathway. However, the experimental results presented here show that the protonation state of an ancillary amine ligand can exert a protective effect over the sulfur donor, providing a proton-coupled oxidation pathway in which the thiolate ligand remains intact and the locus of oxidation shifts from the sulfur to the nonheme iron center. We hypothesize that redox-active, sulfur-ligated nonheme iron sites in biology could take advantage of similar effects, in which other ligands in the iron coordination sphere help direct Fe versus S oxidation.

Experimental Section

General Considerations.

All syntheses and manipulations were conducted in an N₂-filled drybox (Vacuum Atmospheres, O₂ < 0.2 ppm, H₂O < 0.5 ppm) or using standard Schlenk techniques under an atmosphere of Ar unless otherwise noted. Complexes 3 – 4 were synthesized as previously reported.¹⁸ iPr₃TACN was synthesized according to a reported procedure.⁶⁹ 2-aminothiophenol was purchased from Alfa Aesar, degassed by three freeze-pump-thaw cycles, and stored over 3 Å molecular sieves. Fe(OTf)₂•2MeCN⁷⁰ and ⁵⁷Fe(OTf)₂•2MeCN⁷¹ were prepared according to a literature procedures. ⁵⁷Fe metal (95.93%) was purchased from Cambridge Isotope Laboratories. Ferrocene and its derivatives were purified by sublimation prior to use. PhTMG was prepared according to a literature procedure.⁷² All organic bases were degassed with multiple freeze pump thaw cycles and stored over 3 Å molecular sieves prior to use. Organic acids were prepared according to a literature procedure.⁷³ Ferrocenium salts were prepared based on a modified version of a reported procedure and recrystallized by vapor diffusion of pentane into solutions of the salts in DCM prior to use.⁷⁴ 2,4,6-tri-tert-butylphenol was crystallized from hot ethanol and dried over P₂O₅ prior to use and 2,4,6-tri-tert-butylphenoxyl radical was prepared using a modified version of a reported procedure and verified for purity by EPR and UV-vis spectroscopies.⁷⁵ All other reagents were purchased from commercial vendors and used without further purification. Acetonitrile, acetonitrile-d₃, and hexamethyldisiloxane were distilled from CaH₂. Tetrahydrofuran and tetrahydrofuran-d₈ was dried over Na/benzophenone and subsequently distilled. Diethyl ether was obtained from a PureSolv solvent purification system (SPS). All solvents were degassed by a minimum of three freeze-pump-thaw cycles and stored over freshly activated 3 Å molecular sieves in the drybox following distillation.

Instrumentation.

The ¹H and ¹⁹F NMR spectra were measured on a Bruker 300 MHz or a Bruker 400 MHz spectrometer. Solution magnetic susceptibilities were determined by Evans method using hexamethyldisiloxane as an internal standard.^70–77 Chemical shifts were referenced to reported solvent resonances.⁷⁸ UV-vis experiments were carried out on Agilent 5453 diode-array spectrophotometer using a 1 cm cuvette or a Cary bio-50 or Cary 60 UV-vis spectrophotometer equipped with a Unisoku USP-203A cryostat using a 1 cm modified Schlenk cuvette. Midwest Microlab (Indianapolis, IN) conducted elemental analyses on samples prepared and shipped in ampules sealed under vacuum. EPR measurements were performed on a Bruker X-band EPR in 5 mm quartz EPR tubes (Wilmad). EPR Spectral simulations were performed using EasySpin.⁷⁹ Mössbauer spectra were recorded on a spectrometer from SEE Co. (Edina, MN) operating in the constant acceleration mode in a transmission geometry. The sample was kept in an SVT - 400 cryostat from Janis (Wilmington, MA), using liquid He as a cryogen for 5 K data collection and liquid N₂ as a cryogen for 80 K measurements. Isomer shifts were determined relative to the centroid of the spectrum of a metallic foil of α-Fe collected at room temperature. Data analysis was performed using version F of the program WMOSS (www.wmoss.org) and quadrupole doublets were fit to Lorentzian line-shapes.⁸⁰ Cyclic voltammetry experiments were performed in a N₂ atmosphere glovebox using a Princeton Applied Research Versastat II potentiostat and a three-electrode setup (1 mm platinum working electrode, Pt wire counter electrode, and Ag wire pseudoreference electrode) with electrodes purchased from BASi, Inc., and/or CH Instruments, Inc.

DFT Computational Studies.

All geometry optimizations and Mossbauer calculations were performed in the ORCA-4.0.1.2 or ORCA-4.1.2 program package.⁸¹ Initial geometries were obtained from X-ray crystallographic models. Optimized geometries were calculated using the BP86 functional^82–83 or with B3LYP⁸⁴ functional, in combination with the D3 dispersion correction,^85–86 which gave satisfactory results in reproducing the experimentally derived bond metrics. The 6–311g* basis set^87–89 was used for all Fe, N, O, F, and S atoms and the 6–31g* basis set^90–91 was used for all C and H atoms. To reduce computational costs, the Resolution of Identity (RIJCOSX) approximations⁹² in tandem with the SARC/J auxiliary basis set were employed.⁹³ Due to SCF convergence difficulties in some cases, damping parameters were altered using the Slowconv function in ORCA. Frequency calculations at the same level of theory confirmed that all optimizations had converged to true minima on the potential energy surface (i.e., no imaginary frequencies). The optimized structures were used for Mössbauer calculations. Mössbauer parameters were computed using the B3LYP⁸⁴ functional and a combination of CP(PPP)⁹⁴ for Fe and def2-TZVP^95–96 for all other atoms. The angular integration grid was set to Grid4 (NoFinalGrid), with increased radial accuracy for the Fe atom (IntAcc 7). To simulate solid state effects, a continuum solvation model was included (COSMO) with methanol designated as solvent, which has been shown to lead to accurate predictions of Mössbauer parameters.⁹⁷ The isomer shift was obtained from the electron density at the Fe nucleus, using a linear fit function δ_calc = α(ρ(0) − C) + β. For the methodology described here, α = −0.44024 mm s⁻¹ a.u.³, β = 2.1042 mm s⁻¹, and C = 11813 a.u.³, which was based on a previously reported calibration curve.¹⁸ The calibrated quadrupole splitting was obtained from a linear fit function: |ΔE_Q|_calibrated = η(|ΔE_Q|_calc) – B₀ with η = 0.84003 B₀ = −0.0019275 mm s⁻¹, as previously described.¹⁸ Bond dissociation free energies were calculated at 1 atm and 25 °C using the geometry optimization and frequency calculations performed with B3LYP. Corrections for vibrational, zero-point energy, and contributions from translational, rotational, and vibrational modes to the energy and entropy of the H-atom transfer were accounted for. The electronic energy of H^• used in the calculation of the BDFE values is 313.1 kcal mol⁻¹.⁶³

[Fe^III(iPr₃TACN)(ibt)](OTf) (5).

A solution of 3 (34 mg; 0.058 mmol) in CH₃CN (~1 mL) was cooled to −35 °C. In a separate vial, a solution of ttbp• (21 mg; 0.081 mmol) was dissolved in Et₂O (~1 mL) and diluted with CH₃CN (~1 mL) then cooled to −35 °C. The chilled solution of ttbp• was added dropwise to the solution of 3 resulting in an immediate color change from colorless to purple. The solution was allowed to react at −35 °C for 2 h with occasional mixing. Cold Et₂O was layered on the solution, which was then allowed to sit at −35 °C for 2 days resulting in the formation of purple needles that were washed with cold Et₂O (18.1 mg, 54% yield). UV-vis (CH₃CN, 23 °C): λ_max = 370 nm (ε = 3200 M⁻¹ cm⁻¹), 545 nm (ε = 2300 M⁻¹ cm⁻¹), 940 nm (ε = 700 M⁻¹ cm⁻¹). ¹H NMR: (CD₃CN, 400 MHz): δ 25.73, 11.55, 5.44 ppm. ¹⁹F NMR: (CD₃CN, 300 MHz): δ −79.47 ppm. Evans method (CD₃CN, 400 MHz): μ_eff =4.11 μ_Β. Anal. Calcd for C₂₂H₃₈N₄O₃S₂F₃Fe: C, 45.28; H, 6.56; N, 9.60. Found: C, 45.36; H, 6.71; N, 9.44. Mössbauer (CH₃CN, 80 K): δ = 0.36 mm s⁻¹, |ΔE_Q| = 3.00 mm s⁻¹.

[Fe^III(iPr₃TACN)(ibt^CF3)](OTf) (6).

A solution of 4 (48 mg; 0.073 mmol) in CH₃CN (~1 mL) was cooled to −35 °C. In a separate vial, a solution of ttbp• (27 mg; 0.10 mmol) was dissolved in Et₂O (~1 mL) and diluted with CH₃CN (~1 mL) then cooled to −35 °C. The chilled solution of ttbp• was added dropwise to the solution of 4 resulting in an immediate color change from colorless to purple. The solution was allowed to react at −35 °C for 2 h with occasional mixing. Cold Et₂O was layered on the solution, which was then allowed to sit at −35 °C for 2 days, resulting in the formation of purple needles (41 mg, 87% yield). UV-vis (CH₃CN, 23 °C): λ_max = 390 nm (ε = 3500 M⁻¹ cm⁻¹), 530 nm (ε = 2500 M⁻¹ cm⁻¹), 870 nm (ε = 800 M⁻¹ cm⁻¹). ¹H NMR: (CD₃CN, 400 MHz): δ 16.99, 12.07 ppm. ¹⁹F NMR: (CD₃CN, 300 MHz): δ −55.38, −78.70 ppm. Evans method (CD₃CN, 400 MHz): μ_eff =4.20 μ_Β. Anal. Calcd for C₂₃H₃₇N₄O₃S₂F₆Fe: C, 42.40; H, 5.72; N, 8.60. Found: C, 41.79; H, 5.89; N, 8.79. Mössbauer (CH₃CN, 80 K): δ = 0.34 mm s⁻¹, |ΔΕ_Q| = 3.06 mm s⁻¹.

Preparation of 5(⁵⁷Fe) – 6(⁵⁷Fe) for Mössbauer spectroscopy.

A solution of ⁵⁷Fe-enriched 3(⁵⁷Fe) (2.1 g, 0.0031 mmol) or 4(⁵⁷Fe) (2.1 mg, 0.0032 mmol) in CH₃CN (300 μL) was cooled to −35 °C. To the solution was added 3 equiv ttbp• in chilled CH₃CN (100 μL) that was solubilized with several drops of Et₂O, resulting in a color change from colorless to purple. The resulting solution was allowed to react at −35 °C with occasional manual mixing and transferred to a Mössbauer cup and frozen. The sample was stored under liquid nitrogen until it was loaded into the Mössbauer spectrometer.