Factors Affecting Hydrogen Atom Transfer Reactivity of Metal–Oxo Porphyrinoid Complexes

CONSPECTUS:

There has been considerable interest in hydrogen atom transfer (HAT) reactions mediated by metal/oxygen species because of their central role in metalloenzyme function as well as synthetic catalysts. This Account focuses on our progress in synthesizing high-valent metal–oxo and metal–hydroxo porphyrinoid complexes and determining their reactivities in a range of HAT processes. For these studies we have utilized corrolazine and corrole ligands, which are a ring-contracted subclass of porphyr inoid compounds designed to stabilize high-valent metal complexes. The high-valent manganese complex Mn^V(O)-(TBP₈Cz) (TBP₈Cz = octakis(4-tert-butylphenyl)corrolazine) provided an early example of a well-characterized low-potential oxidant that exhibits a redox potential indicating it to be a relatively weak oxidant yet can still be effective at abstracting H atoms from 20 certain C−H/O−H bonds. Approximating the thermodynamics of the HAT reactivity of the Mn^V(O) complex and related species with the help of a square scheme approach in which HAT can be formally separated into proton (pK_a) and electron transfers (E°) indicates that affinity for the proton (i.e., the basicity) is a key factor in promoting HAT. Anionic axial ligands have a profound influence on the HAT reactivity of Mn^V(O)(TBP₈Cz), supporting the conclusion that basicity is a critical parameter in determining the reactivity. The influence of Lewis acids on Mn^V(O)(TBP₈Cz) was examined, and it was shown that both the electronic structure and reactivity toward HAT were significantly altered. High-valent Cr(O), Re(O), and Fe(O) corrolazines were prepared, and a range of HAT reactions were studied with these complexes. The chromium and manganese complexes form a rare pair of structurally characterized Cr^V(O) and Mn^V(O) species in identical ligand environments, allowing for a direct comparison of their HAT reactivities. Although Cr^V(O) was the better oxidant as measured by redox potentials, Mn^V(O) was significantly more reactive in HAT oxidations, pointing again to basicity as a key determinant of HAT reactivity. The iron complex, Fe^IV(O)(TBP₈Cz^+•), is an analogue of the heme enzyme Compound I intermediate and was found to be mildly reactive toward H atom abstraction from C–H bonds. In contrast, Re^V(O)(TBP₈Cz) was inert toward HAT, although one-electron oxidation to Re^V(O)(TBP₈Cz^+•) led to some interesting reactivity mediated by the π-radical-cation ligand alone. Other ligand modifications, including peripheral substitution as well as novel alkylation of the meso position on the Cz core, were examined for their influence on HAT. A highly sterically encumbered corrole, tris(2,4,6-triphenylphenyl)corrole (ttppc), was employed for the isolation and structural characterization of the first Mn^IV(OH) complex in a porphyrinoid environment, Mn^IV(OH)(ttppc). This complex was highly reactive in HAT with O–H substrates and was found to be much more reactive than its higher-oxidation-state counterpart Mn^V(O)(ttppc), providing important mechanistic insights. These studies provided fundamental knowledge on the relationship between structure and function in high-valent M(O) and M(OH) models of heme enzyme reactivity.

Graphical Abstract

1. INTRODUCTION

Hydrogen atom transfer (HAT) is a fundamental chemical reaction involving the transfer of a combination of a proton and an electron from one group to another. This reaction is a key step in a variety of chemical and biological processes. These processes typically involve oxidation of substrate via a transition metal complex that has an oxidizing metal center to accept the electron and a basic ligand to accept the proton. One example, in particular, is the oxidation of organic substrates by high-valent metal–oxo species. In the heme monooxygenase cytochrome P450 (CYP), hydroxylation of a C–H bond involves an initial HAT to an iron(IV)–oxo π- radical cation, also known as Compound I (Cpd-I) followed by a faster radical recombination step (Scheme 1).^1,2 However, alternative pathways have been observed for certain CYPs where, instead of radical recombination, the initial HAT step is followed by a second HAT to an Fe^IV(OH) intermediate, protonated Compound-II (Cpd-II), resulting in desaturation of the substrate instead of hydroxylation (Scheme 1).³ An analogous but reverse HAT reaction may be critical to water oxidation at the Mn-containing oxygen-evolving cluster (OEC) in Photosystem II (Scheme 1).^4,5

Scheme 1.

Hydrogen Atom Transfers in CYP and the OEC^a

^aReproduced from ref 33. Copyright 2018 American Chemical Society.

Our group has been interested in elucidating the fundamental structural and electronic factors that control HAT reactions involving high-valent metal–oxo complexes in biologically relevant ligand environments. A basic scheme showing HAT pathways is presented in Scheme 2.

Scheme 2.

HAT in Metal–Oxo Systems

We developed corrolazine (Cz) ligands, which are ring-contracted forms of tetraazaporphyrins (porphyrazines), to stabilize high-valent metal–oxo species.⁶ Corrolazines and the analogous ring-contracted corroles contain the typical tetrapyrrolic macrocyclic chelate of a porphyrinoid molecule with a trianionic charge upon full deprotonation of the internal pyrrole N–H groups. The corrolazine core has proven to be particularly stable toward oxidation, allowing us to isolate and fully characterize a range of high-valent metal–oxo complexes. This Account focuses on the HAT reactivity of a range of metal–oxo and metal–hydroxo species prepared by our laboratory with corrolazine and corrole ligands.

2. A Mn^V(O) COMPLEX THAT ABSTRACTS HYDROGEN ATOMS

Early work was conducted in our laboratory with an octa-tert butylphenyl-substituted corrolazine (TBP₈Cz = octakis(4-tert butylphenyl)corrolazinato³⁻), which was synthesized via ring contraction of a porphyrazine precursor (Scheme 3).⁷ This ligand led to the synthesis and isolation of the high-valent manganese–oxo complex Mn^V(O)(TBP₈Cz), which was stable enough for isolation.^8,9 However, the stability of this complex varies in solution and is highly dependent on the identity of the solvent; it was not until more than 12 years later that we discovered the appropriate solvent system (toluene/acetonitrile) for growing single crystals of Mn^V(O)(TBP₈Cz) for structural characterization by X-ray diffraction (XRD) (Figure 1).¹⁰ With the synthesis of Mn^V(O)(TBP₈Cz) in hand, our lab was the first to be able to examine directly the reactivity of a Mn^V(O) porphyrinoid complex in HAT reactions.¹¹ Reaction of the Mn^V(O) complex with a series of para-X-substituted phenols (4-X-2,6-tBu₂C₆H₂OH, X = C(CH₃)₃, H, Me, OMe, CN) resulted in the formation of a Mn^III complex and the corresponding phenoxyl radical (Figure 1), indicating that Mn^V(O)(TBP₈Cz) was competent to extract hydrogen atoms from phenol O–H bonds. However, the immediate product from HAT should be a Mn^IV(OH) complex, but no evidence for this species could be obtained; instead, only the two- electron-reduced product Mn^III(TBP₈Cz) was observed.

Scheme 3.

Corrolazine Synthesis

Figure 1.

Reactivity of Mn^V(O)(TBP₈Cz) with phenol derivatives, showing the postulated Mn^IV(OH) intermediate.

There are two possible mechanisms that could explain the absence of the Mn^IV(OH) complex, which are shown in Scheme 4. One possibility is that the Mn^IV(OH) complex is more reactive toward the phenol substrates than the starting Mn^V(O) complex, making the second HAT step fast in comparison with the initial slow HAT step (mechanism A). A second possibility is that the Mn^IV(OH) complex is unstable and rapidly disproportionates into Mn^III and Mn^V(O) species (Scheme 4), with the Mn^V(O) complex being the sole oxidant in the system (mechanism B).

Scheme 4.

Possible Mechanisms of H-Atom Abstraction by a Mn^V(O) Complex^a

^aReproduced from ref 33. Copyright 2018 American Chemical Society.

The rate of conversion of Mn^V(O)(TBP₈Cz) to Mn^III(TBP₈Cz) was monitored by UV–vis spectroscopy under pseudo-first-order conditions with excess phenol substrate. Isosbestic conversion of the Mn^V(O) complex to the Mn^III complex was seen for all of the substituted phenol derivatives, leading to second-order rate constants ranging from 6.0 × 10⁻³ to 2.9 M⁻¹ s⁻¹. The modification of the para substituents in the phenol derivatives led to a linear correlation of rate constant versus the Hammett σ parameter with ρ = −1.26. A plot of the ArO–H bond dissociation energy, BDE(ArO–H), versus the logarithm of the second-order rate constant (k″) also showed a linear relationship with a slope of −0.39, with the relative rate decreasing as the O–H bond strength increases. Deuteration of the OH group in 2,4,6-tri-tert-butylphenol (TTBP) led to a kinetic isotope effect (KIE) of k_H/k_D = 5.9. All of these data provided compelling evidence for a concerted HAT process involving HAT from the phenol derivative to Mn^V(O)(TBP₈Cz). We hypothesized that the Mn^V(O) complex would also be able to oxidize C–H bond substrates with BDE(C–H) values similar to the BDE(O–H) values of the phenol derivatives. Both cyclohexadiene (CHD) and dihydroanthracene (DHA), which have relatively weak C–H bonds (BDFE_CHD(C–H) = 67.8 kcal/mol (gas) and BDFE_DHA(C–H) = 76.0 kcal/mol (DMSO))¹² reacted with the Mn^V(O) complex to give the Mn^III product, although at significantly lower rates (k″ ~ 10⁻⁵ M⁻¹ s⁻¹). Much lower reaction rates for C–H versus O–H bonds with metal complexes in HAT is expected.¹³

Although we were unable to isolate or even spectroscopically observe the Mn^IV intermediate during HAT reactions with Mn^V(O)(TBP₈Cz), we were able to identify a nicely reversible wave in the cyclic voltammogram that could be assigned to one-electron reduction of Mn^V(O)(TBP₈Cz).⁹ The electro-chemistry showed that the Mn^V(O) complex exhibited a low reduction potential (E_½(Mn^V/Mn^IV) = 0.05 V vs SCE), which led to an intriguing hypothesis regarding HAT reactivity. Perhaps this “low-potential” oxidant was effective at abstracting H atoms from O–H/C–H bonds because the basicity of the incipient [Mn^IV(O)]⁻ species was high enough to provide a sufficient overall thermodynamic driving force for HAT.

At about this time, similar observations were emerging regarding the reactivity of the high-valent iron–oxo intermediates in heme enzymes such as CYP and copropor phyrinogen oxidase (CPO). An important early contribution came from Green, Dawson, and Gray,¹⁴ who provided evidence for the idea that Cpd-II, the Fe^IV(O)(porph) species in heme enzymes, was far more basic than previously thought when the axial ligand was a cysteinate residue. This basicity led to protonation of Cpd-II to give Fe^IV(OH)(porph) under physiological conditions. Importantly, the high basicity also suggested that the redox potential for Cpd-I (Fe^IV(O)(π-cation-radical porph)) could remain relatively low and still allow for effective H-atom abstraction from inert C–H substrates because the basicity of Cpd-II would compensate and lead to a favorable driving force for HAT. The idea of a “low-potential” high-valent metal–oxo oxidant in CYP could help to explain a conundrum in the field: how could deleterious oxidation of the surrounding protein matrix by Cpd-I be avoided? We realized that we could examine these ideas in more detail by taking advantage of the Cz ligand and the access to well-characterized high-valent metal–oxo porphyrinoid species such as the Mn^V(O) complex. With such complexes in hand, we could determine the relative electronic and structural influences that are important for controlling the outcomes of HAT reactions.

3. BASICITY AS A FACTOR

Following the literature on HAT reactions, especially from Mayer,^12,15 led us to consider breaking down the HAT step into the elementary proton transfer (PT) and electron transfer (ET) steps, as shown in Scheme 5. This “square scheme” approach was shown to be especially useful in transition-metal- mediated oxidations.

Scheme 5.

Square Scheme for HAT to Mn^V(O)(TBP₈Cz)

The driving force (ΔG°) for the HAT reaction (diagonal) can be equated to the sum of the ΔG° values for the individual ET (horizontal) and PT (vertical) steps, as shown in Scheme 5. This analysis leads to eq 1, which was initially provided by Bordwell et al.¹⁶ for estimating homolytic bond dissociation free energies (BDFEs) in solution from equilibrium acidities and electrochemical data:

$$\text{BDFE}(\text{in kcal}/\text{mol})=1.37{\text{pK}}_{\text{a}}+23.06E^{°}+{\text{C}}_G$$ (1)

The strength of the O–H bond formed in the Mⁿ⁻¹(OH)species can be related to the physical parameters of redox potential (E°) and basicity (pK_a) of the relevant metal–oxo species in the scheme. The larger the bond strength (we will use BDFE in place of BDE because the Bordwell equation strictly refers to free energies; see Warren et al.¹²) of the O–H bond formed, the greater the driving force for HAT. In addition, the square scheme shows explicitly that if E° for the M^V(O)/M^IV(O) reduction is relatively low, the driving force for HAT can be maintained by a relatively high pK_a, i.e., a high basicity for M^IV(O). The reactivity of Mn^V(O)(TBP₈Cz) with the phenol and C–H derivatives suggested that HAT was thermodynamically favored for these substrates. With the assumption that the reaction is thermodynamically favored, the BDFE of the putative Mn^IV(O–H) bond should be greater than that of the X–H bond being cleaved in the substrate, i.e., ~80 kcal/mol. This estimate of the BDFE for the [Mn^IV(OH)]⁻ species combined with E_½(Mn^V(O)/[Mn^IV(O)]⁻) = 0.02 V vs SCE in PhCN (−0.38 V vs Fc⁺/Fc)⁹ and use of the Bordwell equation with a C_G value of 54.9 (in CH₃CN)¹² leads to a predicted pK_a of ≥24.7 for [Mn^IV(OH)]⁻. This estimate is valid provided that the assumption that the C_G value in CH₃CN is a reasonable approximation of the C_G value for PhCN. This value can be compared to that for the basic Cpd-II in CYP, which has pK_a = 12 for the ferryl heme.¹⁷ We reasoned that the comparison of the pK_a value of [Mn^IV(OH)]⁻ estimated in organic solvent with that of the pK_a value for Cpd-II in aqueous solution was reasonable because the Cpd-II active site is buried in a highly hydrophobic pocket, and therefore, the local environment is much closer to an organic solvent than it is to an aqueous medium. Basicity appeared to be a key factor in promoting HAT for Mn^V(O)(TBP₈Cz), and further studies supported this conclusion (vide infra).

The electron transfer properties of Mn^V(O)(TBP₈Cz) were next examined and provided further insights into overall oxidative reactivity, including HAT, and the related process of hydride transfer.¹⁸ Reaction of the Mn^V(O) complex with decamethylferrocene (Fc*) led to the two-electron-reduced Mn^III product (Figure 2a). Spectral titrations confirmed the 2:1 reductant:Mn complex stoichiometry (Figure 2b). As in the HAT chemistry, a Mn^IV intermediate was not observed, and it was postulated that rapid protonation of the putative basic Mn^IV(O)⁻ species occurred from exogenous water, giving a transient Mn^IV(OH) which was then rapidly reduced by a second equivalent of reductant. The data were consistent with the mechanism in Scheme 6, which shows a proton-coupled electron transfer (PCET) mechanism for the reduction of Mn^V(O) by ferrocene derivatives.

Figure 2.

(a) UV–vis changes for the reaction between Fc* and Mn^V(O)(TBP₈Cz) in PhCN. (b) Spectral titration showing a 2:1 stoichiometry of electron transfer.

Scheme 6.

Mechanism of Electron Transfer from [Fe(C₅H₄Me)₂] to Mn^V(O)(TBP₈Cz)

The reactions of Mn^V(O)(TBP₈Cz) with NADH analogues as potential hydride donors were examined. Addition of the acridine derivatives shown in Scheme 7 to the Mn^V(O) complex caused the rapid formation of [Mn^III(OH)-(TBP₈Cz)]⁻ and acridinium (AcrH⁺) products. Kinetic analyses, including a KIE of $k_{{\text{AcrH}}_2}/k_{{\text{AcrD}}_2}=6.9$ (Figure 3), indicated that the rate-determining step involves C–H bond cleavage. A significant decrease in rate was also seen for the ethyl-substituted acridine (Figure 3), which would be expected if an initial HAT step in Scheme 7 were the rate-determining step, as opposed to hydride transfer.

Scheme 7.

Mechanism of Hydride Transfer from NADH Analogues (AcrHR) to Mn^V(O)(TBP₈Cz)

Figure 3.

Plots of the observed pseudo-first-order rate constants (k_obs) for the reaction of NADH analogues with Mn^V(O)(TBP₈Cz).

4. AXIAL LIGAND EFFECT

The heme enzymes CPO and CYP are unique in that they contain rare thiolate ligation at the axial position, in contrast to the more common His ligation in other enzymes (e.g., peroxidases, heme dioxygenases). Studies of CPO and CYP suggested that the anionic axial ligand is linked to the increase in basicity exhibited by Cpd-II, which in turn facilitates HAT (vide supra). The addition of thiols to Mn^V(O)(TBP₈Cz) as potential axial donors was problematic because of the direct oxidation of sulfur donors to disulfide. However, addition of more inert anionic axial ligands, such as F⁻ and CN⁻, led to the in situ formation of the six-coordinate Mn^V(O)(F⁻)(TBP₈Cz) and Mn^V(O)(CN⁻)(TBP₈Cz) complexes.¹⁹ Spectroscopic data indicated weak binding of the anionic donors to the position trans to the terminal oxo group.

The reactivity of the six-coordinate species was dramatically different from that of the starting five-coordinate complex. The fluoride-ligated complex reacted with the C–H substrate DHA with a second-order rate constant that was ~2100-fold greater than that measured for the five-coordinate complex. A large KIE of 10.5 for DHA versus [D₄]DHA was also obtained, supporting a rate-determining HAT step. An even more dramatic rate enhancement was observed when the anionic donor was changed from F⁻ to the more electron-donating CN^–, with a rate enhancement of ~16000-fold. Similar trends in reaction rate were seen with other C–H substrates, including CHD, indene, xanthene, and triphenylmethane, as well as for the HAT reactivity of non-heme iron(IV)–oxo complexes.²⁰

Density functional theory calculations were performed on the reactions of Mn^V(O)(TBP₈Cz) and the six-coordinate F⁻- and CN⁻-ligated derivatives (1, 1-F, and 1-CN in Figure 4) with DHA. The potential energy profiles for all three complexes show a similar reaction mechanism (Figure 4) that proceeds via two HAT steps, with the first step being rate- determining. The energy profiles show a dramatic lowering of the initial, rate-determining HAT barrier upon coordination of F⁻ or CN⁻. Thus, the calculations are in satisfying agreement with the trend in reactivity that was observed experimentally. Additional calculations suggested that the reactivity could be linked to increased basicity of the oxo group. It is interesting to note that the addition of the axial ligand has a greater impact on the increase in basicity of the metal–oxo group, compared with the lowering of the redox potential, which would have lowered the driving force for HAT and attenuated the reactivity. It should also be noted that no “multistate reactivity” was implicated for the Mn^V(O) corrolazines. Recent CASSCF calculations confirmed the singlet ground states for the five-coordinate and axially ligated Mn^V(O) complexes.²¹

Figure 4.

(left) Calculated potential energy profiles for reactions of singlet-state [Mn^V(O)(TBP₈Cz)(X)] (X = none (1), F (1-F), CN (1-CN)) with DHA; TS = transition state; Int = intermediate; Prod = product). (right) Transition state structures for the initial HAT step, with selected bond distances in angstroms, bond angles in degrees, and imaginary frequencies in wavenumbers. (bottom) Mechanism of HAT with DHA as the substrate. Adapted with permission from ref 19. Copyright 2010 John Wiley & Sons.

5. ELECTRONIC STRUCTURE

With the elucidation of the X-ray structure of the oxygen- evolving complex of Photosystem II showing a Lewis acidic Ca²⁺ ion closely associated with the catalytic center in a Mn₄CaO₅ cluster, there has been increased interest in understanding the potential influence of redox-inactive metal ions on high-valent Mn(O) species.^22–26

We found that the Lewis acid Zn²⁺ stabilizes a valence tautomer of Mn^V(O)(TBP₈Cz), formulated as Mn^IV(O–Zn)(TBP₈Cz^+•) (Scheme 8).²⁷ The paramagnetic ¹H NMR spectrum, Evans’ method measurement (μ_eff = 4.11μ_B), and EPR silence were consistent with an integer-spin ground state for the Zn²⁺ adduct. Subsequent studies found that the valence tautomer was stabilized by the nonmetallic Lewis acid B(C₆F₅)₃²³ as well as the Bronsted acids²⁴ trifluoroacetic acid (TFA) and HBAr^F (Scheme 8). The Zn²⁺ and B(C₆F₅)₃ adducts were characterized by X-ray absorption spectroscopy, revealing elongated Mn–O bonds consistent with binding of the Lewis acid to the oxo ligand. The latter species abstracted H atoms from the O–H bonds in 2,4-di-tert-butylphenol (DTBP) and 2,4,6-tri-tert-butylphenol (TTBP), acting as one-electron oxidants to give Mn^IV complexes, in contrast with the Mn^V(O) tautomer, which acts as a two-electron oxidant (Scheme 9). An increased rate of HAT reactivity up to ~100- fold was seen for the Mn^IV(O–LA)(TBP₈Cz^+•) complexes with the O–H donors (Table 1).

Scheme 8.

Reversible Binding of Lewis/Bronsted Acids Stabilizing Mn^IV(O–LA)(TBP₈Cz^+•)^a

^aReproduced from ref 24. Copyright 2016 American Chemical Society.

Scheme 9.

HAT Reactivities of Mn^V(O)(TBP₈Cz), Mn^V(O)(tpfc), and Lewis Acid Adducts

Table 1.

Rate Constants for Different Lewis Acids Ranked by Acceptor Number

	k2 (M−1 s−1)
Lewis acid (A.N.)a	DTBP	xanthene
MnV(O)(TBP8Cz)
none	2.9 ± 0.1	(1.8 ± 0.2) × 10−3
TFA (40)	–	(9 ± 1) × 10−4
Zn2+ (68)	17 ± 1	(8.1 ± 0.4) × 10−3
HBArF (102)	–	(1.9 ± 0.2) × 10−2
B(C6F5)3 (82)	107 ± 8	(5.5 ± 0.3) × 10−2
MnV(O)(tpfc)
none	45 ± 0.3	–
TFA	32.6 ± 0.5	–
Zn2+	943 ± 26	–
Sc3+	27.3 ± 0.7	–
Yb3+	42 ± 2	–
Ca2+	1386 ± 5	–
B(C6F5)3	271 ± 6	–

^aA.N. = acceptor number, which is proportional to the Lewis acid strength (see ref 21).

In 2015, a study by Abu-Omar on Mn(O) in a corrole ligand revealed different results. Mn^V(O)(tpfc) (tpfc = tris- (pentafluorophenyl)corrole) forms the valence tautomer Mn^IV(O)(tpfc^+•) upon addition of TFA. Reaction of Mn^IV(O)-(tpfc^+•) with 2,4-DTBP gives the Mn^III complex as the product (Scheme 9), a two-electron reduction process. In addition, slower HAT reactivity for Mn^IV(O)(tpfc^+•) compared with the Mn^V(O) tautomer was seen.²⁵ A follow-up study by our lab showed that Mn^IV(O–LA)(TBP₈Cz^+•) (LA = Zn²⁺, B(C₆F₅)₃, TFA, HBAr^F) acted as a one-electron oxidant with the C–H donors xanthene, DHA, and CHD.²⁴ The stoichiometry suggests that the Mn^IV product following HAT is more stable in the Cz than in the corrole. As can be seen in Table 1, Zn²⁺, B(C₆F₅)₃, and HBAr^F adducts lead to rate enhancements of HAT in comparison to the naked Mn^V(O), while a slightly lower k₂ was observed for reactivity of the TFA adduct, similar to what was found for Mn^IV(O)(tpfc^+•) with TFA. These results suggest that TFA is too weak an acid under the reaction conditions to increase the rate of HAT.

Abu-Omar also explored HAT reactivity with O–H substrates of Mn^V(O)(tpfc) and Zn²⁺, Ca²⁺, Sc³⁺, Yb³⁺, and B(C₆F₅)₃. The Ca²⁺ adduct, which was slowest to form, was found to be most reactive toward HAT, while Sc³⁺ and Yb³⁺, similar to TFA, showed fast kinetics of formation but slower HAT reactivity. It is interesting to note that the Ca²⁺ adduct exhibited superior activity given that calcium is involved in the Mn-containing OEC of Photosystem II. However, the underlying factors (e.g., steric or electronic properties) that influenced the HAT rate, and in particular what makes Ca²⁺ adduct the most reactive, could not be determined.²⁶

6. METAL SUBSTITUTION

6.1. Chromium versus Manganese

The isolation and structural characterization of Cr^V(O)-(TBP₈Cz) (Figure 5a), which is isomorphous with Mn^V(O)-(TBP₈Cz), provided a unique opportunity to compare their HAT reactivities.¹⁰

Figure 5.

(a) Crystal structure of Cr^V(O)(TBP₈Cz). (b) Cyclic voltammogram of Cr^V(O)(TBP₈Cz) in CH₂Cl₂.

While Mn^V(O) was reactive with a range of HAT substrates, the Cr^V(O) species was reactive only with TEMPOH (Table 2 and Figure 6). An E_½ of −0.43 V vs Fc⁺/Fc is seen for the Cr^V/Cr^IV redox potential (Figure 5b), which is ~100 mV more positive than that for the Mn^V/Mn^IV couple. Since the reactivity of the Cr^V(O) complex (Table 2) implies a weaker Cr^IV(O–H) bond compared with Mn^IV(O–H), eq 1 suggests that the reduced [Cr^IV(O)⁻] is 8 orders of magnitude less basic than [Mn^IV(O)⁻]. The dramatic difference in reactivity therefore can be assigned to the different basicities of the metal–oxo units.

Table 2.

Reactivities of Cr^V(O) and Mn^V(O) toward HAT Substrates

substrate	BDFE (kcal/mol)	MnV(O)	CrV(O)
HMBa	83	no	no
2,4,6-TTBP	80	yes	no
DHA	77	yes	no
xanthene	73	yes	no
TEMPOH	67	yes	yes

^aHMB = hexamethylbenzene.

Figure 6.

(a) Time-resolved UV–vis spectral changes for the reaction of Cr^V(O)(TBP₈Cz) with excess TEMPOH. (b) Plot of the absorbance at 653 nm vs time for the reaction in (a). Inset: dependence of the first-order rate constant on [TEMPOH] and the best-fit line.

6.2. Iron

The iron complex Fe^III(TBP₈Cz) functioned as a catalyst for oxidation of alkenes in the presence of pentafluoroiodosylbenzene (C₆F₅IO, PFIB),²⁸ and spectroscopic evidence suggested the formation of Fe^IV(O)(TBP₈Cz^+•), a Cpd-I analogue, during catalysis.²⁹ Quantitative oxidation of Fe^III(TBP₈Cz) to Fe^IV(O)(TBP Cz^+•) was initially accomplished by addition of PFIB at −78 °C (Scheme 10). The use of the oxidant m - chloroperoxybenzoic acid (mCPBA) in place of PFIB led to more efficient formation of the Cpd-I analogue.

Scheme 10.

Generation of Fe^IV(O)(TBP₈Cz^+•) Using Different Oxidants

Examination of the C–H activation capacity of Fe^IV(O)-(TBP₈Cz^+•) was carried out with a series of substrates with a range of BDE(C–H) values. A plot of the second-order rate constants versus the BDE(C–H) values for these substrates revealed a slope of −0.42 (Figure 7a). This plot was similar to what was seen for ([Fe^IV(O)(p-X-PyO)(tmp^+•)]⁺)³⁰ (tmp = meso-tetramesitylporphyrin) and Mn^V(O)(TBP₈Cz). A KIE of 5.7 was also found for xanthene (Figure 7b). These results support a HAT mechanism for the rate-determining step of Fe^IV(O)(TBP₈Cz^+•) in reaction with C–H substrates.

Figure 7.

(a) Plot of log k versus BDE for the reaction of Fe^IV(O)(TBP₈Cz^+•) with C–H substrates. (b) Second-order plots for xanthene (green circles) and 9,10-d₂-xanthene (blue diamonds).

The reactivity of [Fe^IV(O)(p-X-PyO)(tmp^+•)]⁺ with xanthene is approximately 50–225 times higher than that of Fe^IV(O)(TBP₈Cz^+•). This difference in reactivity was initially thought to be attributed to the extra negative charge on the corrolazine ligand, which should reduce the electrophilicity of Fe^IV(O)(TBP₈Cz^+•) in comparison with the porphyrin system. However, we have shown that addition of negatively charged axial ligands to the neutral Mn^V(O)(TBP₈Cz) greatly increases the reactivity toward C–H substrates, which suggests that a difference in ligand charge cannot by itself explain the reactivity trends observed.

6.3. Rhenium

A rhenium(V)–oxo corrolazine was the first metallocorrolazine containing a third-row metal ion.³¹ While its isoelectronic analogue Mn^V(O)(TBP₈Cz) showed both oxygen atom transfer and HAT reactivity, Re^V(O)(TBP₈Cz) was found to be unreactive in both processes. However, the Re^V(O) complex showed a reversible wave in cyclic voltammetry at +1.04 V vs SCE, indicative of Cz ring oxidation.¹⁰ Oxidizing the Re^V(O) complex with [(4-BrC₆H₄)₃N^+•][SbCl₆⁻] (E_red = +1.16 V vs SCE) gave the monocationic complex Re^V(O)-(TBP₈Cz^+•), and reaction of this species with DHA resulted in quantitative conversion to the [Re^V(O)(Cz(H))]⁺ complex, in which a meso-N atom was the presumed site of protonation. Anthracene was also formed in good yield (90%). However, further experiments revealed that instead of HAT to the Re^V(O)(TBP₈Cz^+•) complex, the mechanism involved an electron back-transfer between Re^V(O)(TBP₈Cz^+•) and (4-BrC₆H₄)₃N, and it was the amminium radical cation that oxidized DHA directly. To avoid electron back-transfer between the reduced oxidant and the oxidized rhenium complex, the more powerful oxidant Ce^IV(NH₄)₂(NO₃)₆ (E_red = +1.33 V vs SCE) was used to generate Re^V(O)-(TBP₈Cz^+•). In this case, direct HAT between the Re complex and DHA was seen (Scheme 11).

Scheme 11.

HAT Reactivity of Re^V(O)(TBP₈Cz^+•) with DHA

Thus, the inertness of the Re–O group allowed us to demonstrate that a π-radical-cation porphyrinoid ligand isolated from its metal–oxo core was capable of HAT reactivity.

7. LIGAND MODIFICATION

The influence of the Cz ligand on HAT was examined by peripheral modification of the aryl substituents. The p-tBu groups on the eight phenyl rings were substituted with p-OMe groups. This new ligand was used to generate Mn^V(O)-(MeOP₈Cz).⁷ The redox potential of the Mn^V/Mn^IV couple (−0.57 V vs Fc⁺/Fc) was shifted in the negative direction by 40 mV, and that of the Cz ring oxidation (+0.45 V vs Fc⁺/Fc) was shifted by 60 mV in the same direction. The reactivity of Mn^V(O)(MeOP₈Cz) was examined with C–H substrates, and rate constants ranging from 6.8(5) × 10⁻⁵ to 1.70(2) × 10⁻¹ M⁻¹ s⁻¹ were obtained, similar to what was seen for the p-tBu derivative. While there was little change in the reaction rates for HAT reactivity with the substrates xanthene and acridine (k_MeO/k_TBP ratios of 1.3 and 1.0, respectively), a 2-fold increase in rate was observed for the 1,4-CHD substrate. The lower redox potential of the p-OMe derivative might be expected to lead to a slightly weaker driving force for HAT. However, since the HAT rates for the p-OMe derivative were seen to be slightly greater than those for the p-tBu derivative, it is reasonable to conclude that the basicity of the unseen [Mn^IV(O)]⁻ species is enhanced and compensates for the lower redox potential to increase the driving force for HAT.

Another modification of the Cz ligand that led to changes in reactivity is alkylation of the meso-N position. We devised a synthetic strategy for the selective alkylation of a single meso-N atom of Re^V(O)(TBP₈Cz), leading to the cationic complex [Re^V(O)(N-MeTBP₈Cz)]⁺.³² This complex was capable of abstracting hydrogen atoms from weak O–H and N–H bonds, leading to a rare, air-stable 19-π-electron porphyrinoid complex, [Re^V(O)(N-MeTBP₈Cz)]^•, in which the alkylated ligand was reduced by one electron. Kinetic analysis with TEMPOH as the substrate gave k₂ = 0.76(2) M⁻¹ s⁻¹ and a KIE of 1.4. On the basis of the relatively small KIE, it was suggested that a PCET process occurs in which a weakly basic meso-N atom of the Cz ring is the initial proton acceptor but is then deprotonated by OTf⁻. Scheme 12 shows a summary of the influence of various ligand modifications on the reactivity of Re^V(O)(TBP₈Cz).

Scheme 12.

Influence of Ligand Modification on the HAT Reactivity of Re^V(O)(TBP₈Cz)

8. MISSING HAT INTERMEDIATE

We were able to synthesize an Fe(OH) complex with a formal +4 oxidation state by employing a large, sterically encumbered corrole ligand, tris(2,4,6-triphenylphenyl)corrole (ttppc).³³ We hypothesized that the ttppc ligand might allow us entry into the analogous yet elusive Mn^IV(OH) species, which is the missing intermediate in many HAT reactions with the Mn^V(O) starting complex.³³ The desired Mn^IV(OH)(ttpcc) complex was successfully synthesized and characterized by XRD (Figure 8). We also prepared and structurally characterized the related Mn^V(O)(ttppc) and Mn^III(H₂O)(ttppc) complexes.³⁴ Thus, the series of well-characterized Mn complexes shown in Figure 8 provided an opportunity to study the nature of the interconversion of these species through HAT.

Figure 8.

(top) HAT steps relating Mn(ttppc) complexes and (bottom) their crystal structures.

Reaction of Mn^V(O)(ttppc) with excess 2,4-DTBP showed complete isosbestic conversion to the Mn^III–aquo complex in minutes, with k_H = 17.4(1) M⁻¹ s⁻¹ and a KIE of 1.05 (Figure 9a–c). Product analysis showed the formation of the bis(phenol) dimer (91% by GC-FID), supporting HAT with a Mn^V(O):DTBP stoichiometry of 1:2. This stoichiometry is the same as that seen for Mn^V(O)(TBP₈Cz) and Mn^V (O)-(tpfc) (vide supra). The expected product after initial HAT is the Mn^IV(OH) species, but this species was not observed. Either of the two possible mechanisms in Scheme 4 may apply. However, with the isolation of the Mn^IV(OH)(ttppc) complex, we were able to directly examine this mechanistic question for the first time.

Figure 9.

(a) Reaction of Mn^V(O)(ttppc) and 2,4-DTBP. (b) Time-resolved UV–vis spectral changes for the reaction between Mn^V(O)(ttppc) and 2,4-DTBP. Inset: change in absorbance vs time for the growth of Mn^III(ttppc) (660 nm) (green circles). (c) Plots of k_obs versus [2,4-DTBP-OH] (black circles) and [2,4-DTBP-OD] (red squares). (d) Reaction of Mn^IV(OH)(ttppc) and 2,4-DTBP. (e) Time-resolved UV–vis spectral changes for the reaction between Mn^IV(OH)(ttppc) and 2,4-DTBP. Inset: change in absorbance vs time for the growth of Mn^III(ttppc) (660 nm) (green circles). (f) Plots of k_obs versus [2,4-DTBP-OH] (black circles) and [2,4-DTBP-OD] (red squares). Reproduced from ref 33. Copyright 2018 American Chemical Society.

Addition of excess of 2,4-DTBP to Mn^IV(OH)(ttppc) resulted in rapid conversion to Mn^III(H₂O)(ttppc) with k₂ =2.73(12) × 10⁴ M⁻ s⁻¹ (Figure 9d–f). This study clearly showed that the Mn^IV(OH) complex reacts much more rapidly with 2,4-DTBP than the Mn^V(O) complex does. The >1500-fold rate enhancement is consistent with mechanism A in Scheme 4. No evidence of disproportionation of Mn^IV(OH)-(ttppc) in solution over many hours was observed, which further rules out mechanism B. This settles the debate between the two suggested mechanisms in favor of mechanism A. Further kinetic studies were performed on a series of para- substituted 2,6-di-tert-butylphenol substrates (4-X-2,6-DTBP, X = OMe, Me, tBu, H). A Hammett plot for these substrates gave ρ⁺ = −1.4(2), and a Marcus analysis of the rate constants gave a weak slope of −0.12(1) (Figure 10). This latter slope is much smaller than expected for a rate-limiting ET reaction (−0.5) and is comparable to the slope observed for the reaction of cumylperoxyl radical and phenols (−0.05), which is a clear example of a concerted HAT.³⁵ Taken together, the data show that the Mn^IV(OH) complex abstracts hydrogen atoms fashion. in a concerted

Figure 10.

(a) Hammett and (b) Marcus plots for the reaction of Mn^IV(OH)(ttppc) and 4-X-2,6-DTBP (X = OMe, Me, tBu, H). Reproduced from ref 33. Copyright 2018 American Chemical Society.

SUMMARY

High-valent metal–oxo species of corrolazine and corrole ligands were synthesized and employed to examine factors that can influence HAT reactivity. The ability of Mn^V(O)(TBP₈Cz) to abstract H atoms from various substrates initially showed that a low-potential oxidant can still function as an effective H- atom abstractor and supported the idea that metal–oxo basicity is an important part of the driving force for HAT. These ideas were strengthened by the findings on the effects of axial ligands on Mn^V(O)(TBP₈Cz) as well as the effects of peripheral substitution (e.g., p-OMe vs p-tBu) and metal ion identity (Mn^V(O) vs Cr^V(O)). Lewis acids were found to have a significant influence on HAT reactivity as well. Although the Re^V(O)(TBP₈Cz) analogue was found to be unreactive toward HAT, the inert Re–O group allowed for the opportunity to study the isolated HAT reactivity of the corrolazine ligand itself. The isolation and structural characterization of a series of Mn(corrole) complexes with different oxidation and proto nation states allowed for the study of the interconversion of these species through HAT and settled a debate regarding the mechanism of HAT for previously reported Mn–oxo porphyrinoid systems. This work showed that the lower-valent Mn^IV(OH) complex is a much more reactive H-atom abstractor than the higher-valent Mn^V(O) analogue. The results in this Account show that an interplay of a variety of factors (e.g., metal identity, ligand modifications, electronic structure) can greatly influence the H-atom reactivity of metal–oxo/hydroxo species. Analysis of these factors for the different porphyrinoid complexes presented here has provided knowledge about the thermodynamic forces that drive HAT as well as mechanistic insights into HAT processes relevant to enzyme systems.

Biographies

Jireh Joy D. Sacramento received her B.S. degree in Biochemistry from the University of the Philippines Manila in 2010. She is currently a Ph.D. candidate at the Johns Hopkins University working with Professor David P. Goldberg.

David P. Goldberg received his B.A. degree from Williams College in 1989 and a Ph.D. degree from M.I.T in 1995. After completing a postdoctoral fellowship at Northwestern University, he moved to Johns Hopkins University in 1998, where he is currently a Professor of Chemistry.