Concerted Proton-Electron Transfer in the Oxidation of Hydrogen-Bonded Phenols

Abstract

Three phenols with pendant, hydrogen-bonded bases (HOAr-B) have been oxidized in MeCN with various one-electron oxidants. The bases are a primary amine (–CPh₂NH₂), an imidazole, and a pyridine. The product of chemical and quasi-reversible electrochemical oxidations in each case is the phenoxyl radical in which the phenolic proton has transferred to the base, ^•OAr-BH⁺, a proton-coupled electron transfer (PCET) process. The redox potentials for these oxidations are lower than other phenols, predominately from the driving force for proton movement. One-electron oxidation of the phenols occurs by a concerted proton-electron transfer (CPET) mechanism, based on thermochemical arguments, isotope effects, and ΔΔG^‡/ΔΔG°. The data rule out stepwise paths involving initial electron transfer to form the phenol radical cations [^•+HOAr-B] or initial proton transfer to give the zwitterions [⁻OAr-BH⁺]. The rate constant for heterogeneous electron transfer from HOAr-NH₂ to a platinum electrode has been derived from electrochemical measurements. For oxidations of HOAr-NH₂, the dependence of the solution rate constants on driving force, on temperature, and on the nature of the oxidant, and the correspondence between the homogeneous and heterogeneous rate constants, are all consistent with the application of adiabatic Marcus Theory. The CPET reorganization energies, λ = 23 – 56 kcal mol⁻¹, are large in comparison with those for electron transfer reactions of aromatic compounds. The reactions are not highly nonadiabatic, based on minimum values of H_rp derived from the temperature dependence of the rate constants. These are among the first detailed analyses of CPET reactions where the proton and electron move to different sites.

Proton-coupled electron transfer (PCET) is of much current interest as it is important in a variety of chemical and biological processes.^1,2 Such reactions can occur by concerted or stepwise mechanisms. The stepwise possibilities include initial transfer of the proton followed by electron transfer (PT-ET), sometimes termed proton-gated electron transfer,³ and ET followed by PT (ET-PT). Reactions in which the proton and electron transfers occur in one single kinetic step have recently been termed concerted proton-electron transfer (CPET).^4,5 CPET encompasses a range of processes that involve the transfer of an electron and a proton, including hydrogen atom transfer (HAT),⁶ and non-HAT processes where the e⁻ and H⁺ are separated in the reactants, products, and/or at the transition structure.^7–8,⁹,¹⁰,¹¹ While HAT reactions continue to be the subject of extensive study in organic radical chemistry, the second class of CPET has received less attention. This report describes studies of a set of reactions of the latter class: oxidations of intramolecularly hydrogen bonded phenols (Scheme 1). Removal of an electron from these compounds results in transfer of the phenolic proton to the base. These reactions involve movement of both e⁻ and H⁺ but cannot be described as HAT.

Scheme 1.

PCET oxidations of phenols to phenoxyl radicals are of particular importance in biological systems because of the widespread involvement of tyrosyl radicals in enzymatic processes.¹² They have been implicated as intermediates in class I ribonucleotide reductases,¹³ photosystem II,¹⁴ prostaglandin H synthases 1 and 2,¹⁵ cytochrome c oxidase,¹⁶ galactose oxidase,¹⁷ amine oxidases¹⁸ and other systems.¹² In many cases, the phenoxyl radical is generated from the phenol by outer-sphere electron transfer, with release of the proton to a nearby residue (histidine, arginine, lysine, etc.) or to a hydrogen bonded network.¹² An interesting example is the oxidation of tyrosine 160 of the D₂ subunit (Y_Z) in Photosystem II by long-range electron transfer to the light-induced chlorophyll radical cation P₆₈₀^+.19 The phenolic proton of Y_Z is likely transferred to a hydrogen-bonded histidine (His₁₉₀ of subunit D₁). This tyrosyl radical then is involved in the oxidation of the manganese cluster and eventually the conversion of water to O₂.

The HOAr-B systems examined here were designed to model such phenol oxidations with concomitant proton transfer. Related model studies include oxidation of tyrosine by a pendant photogenerated [Ru(bpy)₃]³⁺ or a photoexcited Re^I center^8,9 and electron transfer from phenol-pyridine adducts to photoexcited C₆₀.⁷ These previous studies have all involved intermolecular proton transfer (PT), in some cases to bulk solution, while the HOAr-B compounds reported here have an intramolecular PT in aprotic media. The use of aprotic media and a strong initial hydrogen bond provides the advantage of being able to keep track of the proton but may limit the generality of the conclusions. More studies are required to model biological and chemical systems with weaker hydrogen bonding interactions, and systems in which the formation of charged intermediates is more facile (perhaps with a higher local effective dielectric constant). Our studies and the model systems mentioned above all conclude that concerted proton-electron transfer is the dominant pathway under most conditions, but Hammarström and co-workers have shown that a proton-first mechanism takes over at high pH where deprotonation of tyrosine is energetically accessible.⁸ Similarly, elegant work by Okamura and others has indicated stepwise mechanisms for quinone reduction in photosystem I.²⁰

The motif of a tyrosine hydrogen-bonded to a base may be viewed as a biological redox cofactor. A variety of other electron transfer cofactors, such as iron-sulfur clusters, hemes, and quinones, have been studied and understood based on the Marcus-Hush Theory of electron transfer.²¹ We have previously shown that rate constants for hydrogen atom transfer reactions are in many cases well predicted by the Marcus cross relation.²² This report shows that Marcus Theory can also be applied to non-HAT CPET reactions, and it describes the characteristics of the HOAr-B compounds as electron transfer reagents, highlighting the influence of the PT on the thermodynamics and kinetics of electron transfer. The results are also discussed in light of the more recent and more sophisticated theoretical models of CPET.²³ A preliminary report has described the oxidation of one of the phenols, HOAr-NH₂.²⁴

Results

1. Syntheses and Characterization of Compounds

The phenol-amine HOAr-NH₂ was synthesized as outlined in Scheme 2, following literature precedents.²⁵ The tertiary -CPh₂NH₂ and ^tBu substituents in the 2, 4, and 6 positions confer stability on the derived phenoxyl radical; 2,4,6-^tBu₃C₆H₂O^•, for instance, is stable in solution.²⁶ Recently 2,4-di-tert-butyl-6-(N-methyl-2-pyrrolidyl)phenol was reported to give a persistent oxidized form, decaying over 30 min after bulk electrolysis.²⁷ Related compounds with a –CH₂– spacer between the amine and the phenol are readily available via the Mannich reaction (phenol + formaldehyde + amine),²⁸ but such compounds are susceptible to radical attack at the benzylic hydrogens (and at other C–H bonds α to the amine).²⁹ The Mannich procedure cannot be used to make tertiary substituents because of the decreased reactivity of the ketone-derived iminium cation.²⁸ HOAr-NH₂ was therefore synthesized by addition of benzophenone to the lithiated phenol, leading to the gem-diphenyl substituents.^25a Subsequent trityl chemistry leads to products.^25b,c The corresponding chemistry with gem-methyl groups is diverted by elimination from the HOArCMe₂OH intermediate under the mild acidic conditions.

Scheme 2.

Synthesis of HOAr-NH₂.

The related phenol with a 4,5-bis(4-anisyl)-2-imidazolyl substituent, HOAr-im, has been reported by Benisvy³⁰ and the pyridyl compound HOAr-py has been prepared by Fujita.³¹ In each case, the authors explored the compounds’ properties as ligands to metals. Both compounds fluoresce under ultraviolet light due to excited-state intramolecular proton transfer (ESIPT).³²

The X-ray crystal structures of HOAr-NH₂, HOAr-py, and HOAr-im (Figure 1) all show molecules with intramolecular hydrogen bonds from the phenol to the nitrogen base. There is some twisting between the phenol and pyridyl or imidazolyl rings, with inter-ring torsion angles of 22.6° for HOAr-im and 11.9–15.4° for the three crystallographically independent molecules of HOAr-py. In the two independent molecules of HOAr-NH₂, the NCCC torsion angles are 33.4 and 42.0°. Similar structures have been observed for related molecules.³³ The O•••N distances across the hydrogen bond vary from between 2.550(3) and 2.646(2) Å (Table 1), which are in the shorter portion of the known range for OH••N hydrogen bonds.^33,34 Crystal packing forces appear to play a significant role in these distances, as the two independent molecules of HOAr-NH₂ in the unit cell have O•••N distances that differ by 0.063 (4) Å; for the three molecules of HOAr-py, the O •••N distances vary by 0.012(3) Å. The imidazole derivative crystallizes with a molecule of methanol that is hydrogen-bonded to the imidazole hydrogen.

Figure 1.

ORTEP drawings of (a) HOAr-NH₂, (b) HOAr-im, and (c) HOAr-py

Table 1.

Structural, spectroscopic, and electrochemical data for phenol-bases.

phenol (HOAr-B)	dO•••N (Å)	δO–H (ppm)a	E1/2 [V] (δEp [mV])b
HOAr-NH2	2.550(2), 2.613(3)c	12.32	0.37 (143)d
HOAr-im	2.646(2)	13.42	0.42 (105)e
HOAr-py	2.561(3), 2.567(3), 2.573(3)f	14.83	0.58 (100)d

^a1H NMR data in CD₃CN.

^bE vs. Cp₂Fe^+/0.

^cTwo independent molecules in the unit cell.

^dScan rate = 200 mV s⁻¹.

^eScan rate = 100 mV s⁻¹.

^fThree independent molecules in the unit cell.

NMR spectra of the phenols in dry CD₃CN all show sharp downfield resonances for the phenolic proton, e.g., 12.32 ppm for HOAr-NH₂, typical of intramolecularly hydrogen-bonded phenols.^35a The chemical shifts for HOAr-py (14.83 ppm) and HOAr-im (13.42 ppm) are farther downfield as has been previously observed for related compounds³⁶ that have ‘resonance-assisted hydrogen bonds’ due to the conjugation between the phenol and the basic site.^34c,37 The two p-anisyl groups in HOAr-im are inequivalent, indicating that intermolecular proton transfer between imidazole nitrogen atoms is slow on the NMR timescale, presumably due in part to the strong OH••N hydrogen bond.

The UV-vis spectrum of HOAr-NH₂ contains absorptions typical of aromatic compounds^35b at 207 nm (40,000) and 287 nm (3600) (Figure 2a; the ε value is stated parenthetically after each λ_max in M⁻¹ cm⁻¹) The deprotonated phenol (⁻OAr-NH₂) is generated in MeCN by addition of excess di(tetra-n-butylammonium) succinate.³⁸⁻OAr-NH₂ has additional absorptions at 259 (6900) and 327 nm (4700) (Figure 2b), low energy absorptions that are typical of phenoxides.³⁵ A UV-vis spectrum of a saturated (16.0 mM) MeCN solution of HOAr-NH₂ in a 10.00 cm quartz cell shows no absorption maximum in the phenoxide region (inset of Figure 2a). These optical spectra provide an estimate of the equilibrium constant K_PT2 for formation of the zwitterion ⁻OAr-NH₃⁺ (eq 1). Mannich bases with strongly acidic phenols can exist in this tautomeric form, and the optical spectrum of the phenoxide (e.g., ⁻OAr-NH₂) has often been taken as a model for the low-energy part of the spectrum of the zwitterion.³⁹ With this assumption, the lack of an absorption maximum at 327 nm (ε_HOAr-NH2(327) = 1.1 M⁻¹ cm⁻¹) implies that essentially no zwitterion is present in MeCN solution, that K_PT2 < 10⁻⁴. Similarly, the UV-vis spectrum of HOAr-py shows no peak above 385 nm that would be characteristic of the proton-transferred structure.⁴⁰

Figure 2.

UV-vis spectra of (a) phenol HOAr-NH₂ and (b) phenoxide ⁻OAr-NH₂ in MeCN. The inset of spectrum (a) is the spectrum of a saturated solution of HOAr-NH₂ in a 10 cm pathlength cell.

(1)

2. Cyclic Voltammetry and Chemical Oxidations

Oxidations of HOAr-NH₂, HOAr-im and HOAr-py with near stoichiometric amounts of [N(p-C₆H₄Br)₃]^•+ yield the corresponding phenoxyl radical (Scheme 1 above), by both UV-vis and ¹H NMR spectroscopies (at ~10 μM and ~1 mM concentrations, respectively). The reactions are marked by rapid decrease of the intense absorption of the blue aminium ion at 699 nm (40,000 M⁻¹ cm⁻¹). These reactions, and most of the solution measurements in this report, were done in MeCN.

Oxidation of HOAr-im with [N(p-C₆H₄Br)₃]^•+ yields a blue solution of ^•OAr-imH⁺ with an absorption at 695 nm (8300) which decays to ~33% intensity over 1.5 h. Phenoxyl radicals typically have absorptions between 420 and 720 nm, with higher intensity for the more conjugated radicals [λ_max, nm (ε, M⁻¹ cm⁻¹)]: 2,4,6-tri-t-butylphenoxyl radical (^tBu₃ArO^•), 630 (400), benzene;⁴¹ 2,6-^tBu₂-4-Ph-C₆H₂OH, 488 (2780); 2,6-^tBu₂-4-(Me₂NC₆H₄)C₆H₂OH, 650 (6000).²⁶ Treating ^•OAr-imH⁺ in MeCN with triethylamine or excess pyridine (pK_a = 18, 12, respectively³⁸) produces a purple solution with λ_max = 544 nm (approx. 6100) due to the deprotonated phenoxyl radical ^•OAr-im (Figure 3, eq 2). This species likely still has an intramolecular hydrogen bond from the imidazole hydrogen to the oxyl radical. ^•OAr-im was prepared independently by heterogeneous PbO₂ oxidation of HOAr-im in MeCN or DMSO.⁴² This isolated ^•OAr-im had an absorption at 544 nm and contained some HOAr-im by ¹H NMR. Addition of 1 equivalent of triflic acid to solutions of ^•OAr-im formed the 695 nm absorption characteristic of ^•OAr-imH⁺ (eq 2).

Figure 3.

Visible spectra of (a) ^•OAr-imH⁺ and (b) ^•OAr-im in MeCN.

(2)

Oxidation of HOAr-py by [N(p-C₆H₄Br)₃]^•+ gives a yellow solution with λ_max = 481 nm (1600) which fades with t_1/2 ≈ 8 h. Reactions of HOAr-NH₂ with [N(p-C₆H₄Br)₃]^•+ show no absorptions above 400 nm at 100 μM, indicating a colorless radical product. A complex EPR spectrum was recorded for one of the oxidation mixtures of HOAr-NH₂ in CH₂Cl₂ (see Supporting Information of ref 24). ¹H NMR monitoring of reactions of HOAr-NH₂ with substoichiometric amounts of [N(p-C₆H₄Br)₃]^•+ in MeCN showed reduced signals for HOAr-NH₂ and the appearance of N(p-C₆H₄Br)₃. With excess [N(p-C₆H₄Br)₃]^•+, the amine is not observed because there is rapid exchange between NAr₃ and [NAr₃]^•+ by electron transfer.⁴³

The cyclic voltammograms of the three phenols in 0.1 M ⁿBu₄NPF₆/MeCN (Table 1, EFigures 4a and S20 in Supporting Information) are quasi-reversible, with almost equal anodic and cathodic currents but with peak separations (Δ_p) larger than the theoretical 59 mV. The rate constant for heterogeneous electron transfer (k_el) for HOAr-NH₂ has been determined by analysis of the CV data at different scan rates ν (Figure 4a).⁴⁴ k_el is related to ΔE_p and ν by eqs 3 and 4:

$$k_{el}^0=\psi {\left(\frac{\pi D_{\text{O}}F\nu }{RT}\right)}^{1/2}{\left(\frac{D_{\text{R}}}{D_{\text{O}}}\right)}^{\alpha /2}$$ (3)

$$ln\psi =3.69-1.16ln(\mathrm{\Delta }{E}_{\text{p}}-59)$$ (4)

where D_O and D_R = diffusion constant (cm² s⁻¹) for the oxidized and reduced forms of the analyte, α = the transfer coefficient (taken to be 0.5⁴⁵), and R and T have their standard meanings. A, D_O, and D_R were determined using chronoamperometry (see Supporting Information). k_el was found to be (3 ± 1) × 10⁻³ cm s⁻¹ from the slope of a plot of ψ vs. ν^−1/2. (Figure 4b). To support our measurements, these parameters were used to simulate the CVs using DigiSim⁴⁶ with good results (inset, Figure 4a). For comparison, Evans, Savéant, and co-workers have recently reported a much slower heterogeneous rate constant of (9 ± 5) × 10⁻⁷ cm s⁻¹ for CPET reduction of a water-superoxide complex, which exhibits a much more distorted cyclic voltammogram.⁴

Figure 4.

(a) Cyclic voltammograms of HOAr-NH₂ with ν = 20, 50, 100, and 200 mV s⁻¹. Inset: an overlay of the simulated (blue dots) and experimental CV’s at 100 mV s⁻¹. (b) Plot of ψ vs. ν^−1/2 for HOAr-NH₂ (see eqs 3 and 4 in text).

Similar quasi-reversible voltammograms have been reported for other phenols with intra-or intermolecularly hydrogen-bonded amine or pyridine bases.^7,27,30,47 In contrast, electrochemical oxidations of phenols without an attached base are irreversible in dried aprotic solvents, occurring via an EC mechanism. The chemical step (“C”) is typically proton transfer into the bulk solution which is effectively irreversible in aprotic media.⁴⁸ Thus oxidation of 2,4,6-^tBu₃ArOH is irreversible in dry MeCN even though the radical 2,4,6-^tBu₃ArO^• is stable.^48a Matsumura et al. have shown²⁷ that moving the attached base from the ortho to the para position changes the oxidation from quasi-reversible to irreversible. Oxidation of the methyl ether MeOAr-NH₂ is irreversible (E_p,a = 1.2 V, all potentials in this report are vs. Cp₂Fe^0/+ in MeCN), probably because without the stabilizing proton transfer the high energy anisyl or aminium radical cation decays rapidly. The related phenol-alcohol HOAr-OH [(2-CPh₂OH)(4,6-^tBu₂)C₆H₂OH] also shows irreversible electrochemistry (E_p,a = 1.1 V), possibly because proton transfer to the weakly basic primary alcohol is not favored and the proton is lost to the bulk solution. CV of the phenoxide ⁻OAr-NH₂, as the ⁿBu₄N⁺ salt, shows a reversible oxidation wave centered at −0.57 V, essentially equal to the E_1/2 for 2,4,6-tri-tert-butylphenoxide, −0.572 V.^48a,53

The average of the anodic and cathodic peaks for HOAr-NH₂, HOAr-im and HOAr-py are taken as the E_1/2 for the coupled proton-electron transfer (CPET), the potential for transfer of both an electron to the electrode and the phenolic proton to the amine (Scheme 1). This is the interpretation of most of the previous electrochemical studies of phenol/base systems.^7,27,30 One recent paper has interpreted the large ΔE_p for oxidation of a phenol-amine as indicating a stepwise EC (ET-PT) mechanism^47a but this is, in our view, inappropriate.^47b The assignment of E_1/2 as the energetics of CPET is supported by the thermodynamic discussion below and by the following equilibration experiment.

Oxidation of HOAr-NH₂ by [N(tol)₃]^•+ yields an equilibrium mixture with the phenoxyl radical and the tri-p-tolylamine, with equilibrium constant K₅ (eq 5). Addition of N(tol)₃ to the

$$\mathbf{HOAr}-{\mathbf{NH}}_{\mathbf{2}}+{[\text{N}{(\text{tol})}_3]}^{•+}\stackrel{K_5=2.0\pm 0.5}{\rightleftharpoons }{}^{•}\mathbf{OAr}-\mathbf{N}{{\mathbf{H}}_{\mathbf{3}}}^{\mathbf{+}}+\text{N}{(\text{tol})}_3$$ (5)

reaction mixture causes an increase in the optical absorbance due to [N(tol)₃]^•+, yielding K₅ = 2.4. Alternatively, addition of aliquots of triflic acid to solutions containing large excesses of N(tol)₃ and HOAr-NH₂ versus [N(tol)₃]^•+ quantitatively protonates HOAr-NH₂ and therefore shifts the equilibrium toward [N(tol)₃]^•+. This experiment afforded K₅ = 1.5, and also established that a second proton transfer equilibrium, eq 6, is not significant (K₆ ≪ 1).⁴⁹ It should be noted that

$${}^{•}\mathbf{OAr}-\mathbf{N}{{\mathbf{H}}_{\mathbf{3}}}^{\mathbf{+}}+\mathbf{HOAr}-\mathbf{N}{\mathbf{H}}_{\mathbf{2}}\rightleftharpoons {}^{•}\mathbf{OAr}-\mathbf{N}{\mathbf{H}}_{\mathbf{2}}+\mathbf{HOAr}-\mathbf{N}{{\mathbf{H}}_{\mathbf{3}}}^{\mathbf{+}}$$ (6)

these equilibrium experiments are only possible because of the stability of the phenoxyl radical on the chemical timescale. Together, the equilibria establish an overall equilibrium constant K₅ = 2.0 + 0.5, which implies a difference in redox potential between E(HOAr-NH₂^+/0) and E([N(tol)₃]^•+/0) of 18 ± 8 mV. This is in excellent agreement with the 20 ± 30 mV difference in the electrochemical E_1/2 values: 0.36 ± 0.02 V for HOAr-NH₂^+/0 and 0.38 ± 0.02 V for [N(tol)₃]^•+/0. The agreement validates the assignment of the phenol E_1/2 values as E° (CPET).

The oxidants used in this study include variously substituted triarylaminium ions [N(p-C₆H₄X)₃]^•+, iron(III) tris-polypyridyl complexes [Fe(N–N)₃]³⁺ (N–N = 2,2′-bipyridine or 1,10-phenanthroline derivative), and the 10-methylphenothiazinium ion [MPT]^•+. All displayed reversible cyclic voltammograms (Table 2). The [Fe(N–N)₃]^3+/2+ potentials in MeCN vary substantially with ionic strength due to differences in ion pairing between the Fe^II and Fe^III forms.⁵⁰ For [Fe(5,5′-Me₂bpy)₃]^3+/2+, the potential changes by −40 ± 4 mV/log(i). Kinetic studies using [Fe(R₂bpy)₃]³⁺ and [Fe(Me_Xphen)₃]³⁺ were done at 0.1 M ionic strength to match the electrochemical conditions. For the singly-charged [N(tol)₃]^•+, the change in potential with ionic strength was found to be minimal [3 ± 1 mV/log(i)].

Table 2.

Potentials of oxidants (in MeCN, V vs. Cp₂Fe^+/0)^a

Oxidant	E1/2	Oxidant	E1/2
[N(p-C6H4Br)3]•+	0.67	[Fe(bpy)3]3+	0.70
[N(p-C6H4OMe)(p-C6H4Br)2]•+	0.48	[Fe(5,5′-Me2bpy)3]3+	0.58
[N(tol)3]•+b	0.38	[Fe(4,7-Me2phen)3]3+	0.53
[N(p-C6H4OMe)2(p-C6H4Br)]•+	0.32	[Fe(3,4,7,8-Me4phen)3]3+	0.46
[N(p-C6H4OMe)3] •+	0.16	[MPT]•+c	0.32

^aSee Experimental for conditions.

^bTri-p-tolylaminium.

^c10-Methylphenothiazinium.

3. Kinetics

The rates of oxidation of the phenols have been monitored by stopped-flow kinetics, following the disappearance of the oxidant in reactions of [NAr₃]^•+or the appearance of [Fe(N–N)₃]²⁺ in reactions with iron oxidants. Reactions of the iron complexes were performed in MeCN containing 0.1 M ⁿBu₄NPF₆ to match the electrochemical conditions (see above). When possible, reactions were performed with a large excess of phenol relative to oxidant (>5 equiv).

The time sequences of optical spectra were globally analyzed to derive rate constants using SPECFIT^™ software⁵¹ (or, in one instance,⁴⁹ Microsoft Excel^®) (Table 3). For thermodynamically favorable reactions (K_eq » 1) run under pseudo-first order conditions, the second-order rate constant was taken as the slope of a plot of k_obs vs. [HOAr-B]. Particularly fast reactions were analyzed with second-order kinetics, and reactions with K_eq ≲ 1 were analyzed as opposing second-order reactions. In each case, the rate constant was derived from approximately 25 kinetic runs, at five different concentrations. The temperature dependence of the rate constants was measured over 30–47 K ranges (Figure 5), yielding the Eyring parameters⁵² in Table 4 below. Variations in driving force over the appropriate temperature ranges for the reactions of HOAr-NH₂ + [N(tol)₃]^•+ and HOAr-py + [Fe(5,5′-Me₂bpy)₃]³⁺ were evaluated by cyclic voltammetry of the individual reagents. The difference between the half-reaction potentials measured at 2 °C and 46 °C (for HOAr-py and [Fe(5,5′-Me₂bpy)₃]³⁺) or 49 °C (for HOAr-NH₂ and [N(tol)₃]^•+) were found to be within the propagated experimental error (±30 mV).

Table 3.

Rate constants for phenol oxidations (295 ± 2 K, MeCN).

Phenol	oxidanta	k (M−1 s−1)	Erxnb(V)
HOAr-NH2	[Fe(bpy)3]3+	(4 ± 1) ×106	0.34
	[N(p-C6H4Br)3]•+	(4 ± 2) × 107	0.31
	[Fe(5,5′-Me2bpy)3]3+	(1.5 ± 0.2) ×l05	0.22
	[Fe(4,7-Me2phen)3]3+	(3.8 ± 0.4) × 105	0.16
	[N(p-C6H4OMe)(p-C6H4Br)2]•r	(8 ± 1) × 105	0.12
	[Fe(3,4,7,8-Me4phen)3]3+	(3.0 ± 0.3) × 104	0.09
	[N(tol)3]•+	(1.1 ± 0.2) ×105	0.02
	[N(p-C6H4OMe)3]•+	(1.1±0.1) ×103	−0.20
	[N(p-C6H4OMe)2(p-C6H4Br)]•+	(2.7 ± 0.3) × 104	−0.04
	[MPT]•+	(3.2 ± 0.3) × 104	−0.04
DOAr-ND2c	[Fe(5,5′-Me2bpy)3]3+	(5.8 ± 0.6) × 104	0.22
		kH/kD = 2.6 ± 0.4d
	[N(tol)3]•+	(4.3 ± 0.4) × 104	0.02
		kH/kD = 2.5 ± 0.3d
	[N(p-C6H4OMe)3]•+	(6.9 ± 0.7) × 102	−0.20
		kH/kD = 1.6±0.2d
HOAr-im	[N(p-C6H4OMe)3]•+	(1.1 ± 0.1) × 104	−0.26
HOAr-py	[Fe(bpy)3]3+	(5.2 ± 0.8) × 106	0.12
	[Fe(5,5′-Me2bpy)3]3+	(5.8 ± 0.9) × 105	0.00
	[Fe(4,7-Me2phen)3]3+	(1.9 ± 0.4) × 106	−0.05
	[Fe(3,4,7,8-Me4phen)3]3+	(3.3 ± 0.6) × 105	−0.12
DOAr-py c	[Fe(bpy)3]3+	(1.5±0.2) ×l06	0.12
		kH/kD = 2.8 ± 0.6
	[Fe(5,5′-Me2bpy)]3]3+	(2.3 ± 0.4) × 105	0.00
		kH/kD = 2.5 ± 0.6

^aReactions with [Fe(N–N)₃]³⁺ were performed in 0.1 M Bu₄NPF₆/MeCN.

^bE_rxn = E_1/2(oxidant) − E_1/2(phenol).

^cReactions with deuterated substrates were performed in 0.5–1% v/v CH₃OD in MeCN. Rate constants are corrected for residual proton content using k_expt = k_D(1 − f_H) + f_HK_H where f_H is the fraction protonated.

^dk_HOAr-NH2/k_DOAr-ND2.

Figure 5.

Temperature dependence of the rate constants for HOAr-py + [Fe(5,5′-Me₂bpy)₃]³⁺ (▴), HOAr-NH₂ + [N(tol)₃]^•+ (●), and HOAr-im + [N(C₆H₄OMe)₃]^•+ (▪). The curve fits are to the nonadiabatic form of the Marcus equation (eq 15; see Discussion).

Table 4.

Activation parameters, adiabatic reorganization energies, and apparent non-adiabatic reorganization energies and H_rp values for phenol-base oxidations.^a

Reaction	λ12(E)b	λ12(T)b	λ11c	[λ12(nonad)]d	[Hrp]d	ΔH‡e	ΔS‡e
HOAr-NH2 + [N(Ar)3]•+f	34 ± 1	33 ± 1	53 ± 3	[29.6 ± 1.6]	[10 ± 4]	6.3 ± 0.4	−14.1 ± 1.3
HOAr-NH2 + [Fe(N-N)3]3+g	38 ± 2	h	52 ± 4	h	h	h	h
HOAr-NH2 + [MPT]•+	35 ± 1i	h	58 ± 3i	h	h	h	h
HOAr-py + [Fe(R2bpy)3]3+gj	27 ± 1	27 ± 1	30 ± 3	[23.2 ± 1.6]	[6 ± 2]	4.7 ± 0.4	−16.0 ± 1.2
HOAr-py + [Fe(Mexphen)3]3+g	22 ± 1	h	23 ± 3	h	h	h	h
HOAr-im + [N(anisyl)3]•+k	25 ± 2i	25 ± 2	36 ± 4i	[17 ± 3]	[4 ± 3]	7.0 ± 0.7	−17 ± 3

^aΔH^‡ and λ, in kcal mol⁻¹, ΔS^‡ in cal K⁻¹ mol⁻¹ (e.u.), and H_rp in cm⁻¹.

^bλ₁₂(E) and λ₁₂(T) are the adiabatic reorganization energies calculated from the dependence of k on either E° or T using eq 12 (Figures 5,6).

^cλ₁₁ is the adiabatic reorganization energy for HOAr-B/^•OAr-HB⁺ self exchange from eq 13 [using the average of λ₁₂(E) and λ₁₂(T)].

^dλ₁₂(nonadiabatic) and H_rp from eq 15 which may not be appropriate; see text.

^eDetermined by fitting k_obs vs. T data (Figure 5) to the Eyring equation.⁵²

^fTemperature dependent results for HOAr-NH₂ + [N(tol)₃]^•+, 280 ≤ T ≤ 327 K.

^gCorrected for work terms following references 64,71.

^hNot determined.

ⁱFrom a single phenol/oxidant pair.

^jTemperature dependent results for HOAr-py + [Fe(Me₂bpy)₃]³⁺, 279 ≤ T ≤ 318 K.

^kanisyl = -C₆H₄OMe; 279 ≤ T ≤ 309 K.

To determine the kinetic isotope effects, MeCN solutions of HOAr-NH₂ and HOAr-py were prepared with 0.5–1% v/v CH₃OD and the kinetics performed otherwise as above. The large molar excess of CH₃OD provided high isotopic enrichment at the exchangeable OH and NH₂ positions (the rate constants were corrected for the residual proton content in the CH₃OD). Control experiments showed that addition of 1% v/v protio-methanol (CH₃OH) does not affect the rate constant for these reactions. k_HOAr-NH2/k_DOAr-ND2 ranges from 1.6 ± 0.2 to 2.6 ± 0.4 and k_HOAr-py/k_DOAr-py = 2.5 ± 0.6 or 2.8 ± 0.6 depending on the oxidant (Table 3).

Discussion

Phenols with an intramolecular hydrogen bond react with one-electron oxidants to generate phenoxyl radicals in which the proton has transferred. The phenols and phenoxyl radicals have been characterized by spectroscopy, cyclic voltammetry, and chemical reactivity. The kinetics of oxidation have been examined for a number of phenol-oxidant pairs, and in one case electrochemically. We first discuss the thermochemistry of outer-sphere oxidation of these phenols, then the mechanistic data that implicate a concerted proton-electron transfer (CPET) pathway for the reactions. Finally, analysis of the CPET rate constants indicates that the classical Marcus theory is an excellent starting point to understand these processes.

1. Phenol Potentials

The potentials for HOAr-B (0.36–0.58 V vs. Cp₂Fe^+/0 in MeCN, Table 1) are significantly less than reported values for one-electron oxidation of phenols without a pendant base. 2,4,6-Tri-t-butylphenol, for example, has E_p,a[^tBu₃ArOH^•+/0] = 1.09 V.⁵³ Such large shifts have been suggested to be due to hydrogen-bonding effects but the analysis below shows that the shifts must be due to proton transfer.⁴⁷

Consider the oxidation of a phenol hydrogen-bonded to a base B by electron transfer without proton transfer. The effect of the hydrogen bonding is illustrated by the thermochemical cycle in Scheme 3a, which compares the potentials for the H-bonded and non-H-bonded forms (E_ArOH–B^+/0,E_ArOH^+/0). The difference between these two potentials is equal to the difference in the strengths of the hydrogen bonds in the reduced and oxidized forms (ΔG_HB/red − ΔG_HB/ox) (eq 7). Note that the absolute hydrogen bond strengths are not important, only the change upon oxidation. Since most hydrogen bonds are 3–8 kcal mol⁻¹,⁵⁴ shifts of more than ca. 5 kcal mol⁻¹ (0.2 V) would be quite unusual. A very recent experimental study in general supports these thermochemical arguments.⁵⁵

Scheme 3.

Thermochemical cycle indicating the effect on hydrogen bonding on redox potentials.

The effect of the hydrogen bonding on a CPET redox potential is similar. Progressing around the thermochemical cycle in Scheme 3b, (1) the hydrogen bond in HOAr-B is broken (−ΔG_HB/red); (2) the not-hydrogen-bonded phenol is oxidized (E_ArOH^+/0); (3) the proton is transferred (−RTΔpK_a); and (4) the hydrogen bond of ^•OAr-BH⁺ is formed (ΔG_HB/ox). The sum of these four steps is equal to the overall potential (eq 8). For the sterically crowded phenols discussed here, E_ArOH^+/0, is probably well approximated by the ^tBu₃ArOH^•+/0 potential.⁵⁶ With this approximation, the difference between the potential for ^•OAr-HB⁺/HOAr-B and that for ^tBu₃ArOH^•+/0 is the energetics of the proton transfer step (−RTΔpK_a) plus the difference in hydrogen bond strengths. As in the pure electron transfer case of Scheme 3a, it is the change in H-bond strengths rather than their absolute value that is important.

The change in hydrogen bond strength and the attendant shift of the redox potential is likely to be quite small. As noted above, the hydrogen bonded phenoxide “OAr-NH₂ has the same potential as the non-H-bonded 2,4,6-⁽Bu₃ArO”. The potentials for ^tBu₃ArOH (+1.09 V), the anisole MeOAr-NH₂ (~1.2 V), and the hydroxy-phenol HOAr-OH (~ 1.1 V) are quite similar despite what are likely very different H-bonds (OH••NCMe, O••HN, and OH••OH). Hammarström and co-workers attribute 0.10 V (2 kcal mol⁻¹) to the change in hydrogen bonding for the tyrosine-histidine pair in their model system.⁵⁷ Phenoxyl radicals are known to make strong hydrogen bonds in some systems,⁵⁸ so the H-bond in ^•OAr-BH⁺ could be stronger than that in HOAr-B, but this effect is usually small. The H-bond strengthening upon oxidation of catechols to oxyl radicals has been variously estimated as ~4 kcal mol⁻¹ to < 1 kcal mol^−1.58 The strengthening in catechols and 1,8-napthalene diols is particularly large because oxidation yields hydrogen bonds in which PT is degenerate; one report describes a ~7 kcal mol⁻¹ (0.3 eV) strengthening for 1,8-naphthalene diols.^58d For the case of HOAr-NH₂, the hydrogen bond could be stronger in the neutral phenol because of the much larger pK_a mismatch between donor and acceptor in the radial cation.⁵⁹ This would shift the potential in the opposite direction.

In sum, the difference of 0.5 – 0.7 V in redox potentials for HOAr-B vs. ^tBu₃ArOH^•+/0 are too large to be due to changes in hydrogen bond strength. These differences are primarily due to the proton transfer from the phenol radical cation to the base, step 3 in Scheme 3b. In MeCN, 2,4,6-tri-t-butylphenol radical cation has a pK_a of ca. 0⁶⁰ and protonated benzylamine has a pK_a of 17,³⁸ yielding a ΔpK_a of 17. This provides a crude prediction of a shift of 1 V (ΔE = 0.059 V × ΔpK_a), somewhat larger than the observed 0.73 V difference between E(^•ArO-NH₃⁺/HOAr-NH₂) vs. E(^tBu₃ArOH^•+/0).

2. Mechanistic Analysis

There are three reasonable mechanisms for the one-electron oxidation of the hydrogen-bonded phenols (Scheme 4). Rate-limiting outer-sphere electron transfer could yield the phenolic radical cation (^•+HOAr-B), which would be followed by fast proton transfer to give the product (ET1-PT1). Radical cations of simple phenols are well established transients, particularly in photochemical processes.⁶¹ Alternatively, pre-equilibrium proton transfer to yield the zwitterion (−OAr-BH⁺), could be followed by electron transfer (PT2-ET2). Zwitterions such as ⁻OAr-BH⁺ are well known in phenol-base chemistry, particularly when the phenolic portion is highly acidic as in p-nitrophenols.^39,40,62 Rate-limiting proton transfer is ruled out because different oxidants react at different rates and because PT between electronegative elements in general occurs at very fast rates.⁶³ The defining characteristic of the stepwise mechanisms is the formation of an intermediate with a finite lifetime. The third mechanism is the concerted transfer of both particles, CPET, defined by the absence of an intermediate along the reaction coordinate. Concerted implies that both particles move in a single kinetic step but does not imply synchronous movement of the proton and electron.

Scheme 4.

Three possible mechanisms for oxidation of phenol-base compounds.

There are three experimental markers indicating the oxidation mechanism as CPET. First, isotope effects on the oxidation of DOAr-ND₂ and DOAr-py (1.6–2.8 depending on oxidant and phenol) can only be explained through CPET. In rate-limiting electron transfer (ET1), no bond is made or broken in the ET step and, like other electron transfers,^64a,65 there would be only a small secondary isotope effect. The proton-first pathway PT2-ET2 would have an equilibrium PT isotope effect, which would also be small.

Second, the intermediates in the stepwise mechanisms appear to be too high in energy to be involved in the reactions. In the ET1-PT1 mechanism for HOAr-NH₂ + [N(tol)₃]^•+, the ET1 step (eq 9) is estimated to have E_ET1 = −0.71 V (ΔG_ET1 = 16 kcal mol⁻¹, K_ET1 = 10⁻¹²). This estimate

$$\mathbf{HOAr}-\mathbf{N}{\mathbf{H}}_{\mathbf{2}}+\text{N}{{\text{tol}}_3}^{•+}\stackrel{k_{\text{ET}1}}{\to }{}^{•+}\mathbf{HOAr}-\mathbf{N}{\mathbf{H}}_{\mathbf{2}}+\text{N}{\text{tol}}_3{E}_{\text{ET}1}=-0.71\text{ V}$$ (9)

uses E(^•+HOAr-NH₂/HOAr-NH₂) = 1.09 V, taken to be the same as E(^tBu₃PhOH^+/0)⁵³ and E(HOAr-OH^+/0) and 0.11 V below E(MeOAr-NH₂^+/0) (as noted above, hydrogen bonding effects are likely to be small). The estimated value of ΔG_ET1 is significantly higher than the observed Eyring barrier for this reaction, ΔG^‡ = 11 kcal mol⁻¹ (from k = 1.1 × 10⁵ M⁻¹ s⁻¹ and the Eyring equation with κ = 1 ;⁵² with smaller prefactors or κ < 1 the discrepancy would be larger^66,67). For ^•+HOAr-NH₂ to be a viable intermediate (eq 9), our estimate of the potential would have to be in error by more than 0.2 V. (This would also predict a dependence on driving force different than what is observed, as described below). This analysis can equivalently be framed in terms of rate constants instead of barriers. The values K_ET1 = k_ET1/k_–ET1 = 10⁻¹² (see above) and k_ET1 = k_obs = 10⁵ M⁻¹ s⁻¹ would imply an impossible k_–ET1 = 10¹⁷ M^−l s⁻¹, much faster than the diffusion limit in MeCN.⁶⁸ Similar arguments hold for the HOAr-py and HOAr-im systems.

A more complete analysis of the ET1-PT1 pathway includes precursor and successor complexes, as illustrated for the HOAr-NH₂ + [N(tol)₃]^•+ reaction in Scheme 5. The formation of the precursor and successor complexes are assumed to have equilibrium constants K_precursor = K_successor = 1 M⁻¹ as is commonly done⁶⁷ (no evidence for a precursor or successor complex is evident in optical spectra of reaction mixtures). The ET1 step generates the successor complex in which the proton has not transferred, [^•+HOAr-NH₂|Ntol₃]. In the scenario most favorable to the ET1/PT1 pathway, this intermediate partitions equally between back electron transfer to the precursor complex (k_-ET1) and forward proton transfer (k_PT1) to [^•OAr-NH₃⁺|Ntol₃], with both occurring at the fastest possible rate, ~10¹³ s⁻¹. With these assumptions, k_ET1 would have to be 2 × 10⁵ M⁻¹ s⁻¹ based on the experimentally observed k_obs = 1 × 10⁵ M⁻¹ s⁻¹ and K₅ = 2 (eq 5 above). This requires K_ET1 for ET within the precursor complex (k_ET1/k_-ET1) to be 2 × 10⁻⁸, more than 10⁴ larger than the K_ET1 based on the estimated redox potentials (see above). And even in this best-case scenario, [^•+HOAr-NH₂|Ntol₃] is barely an intermediate since the 10¹³ s⁻¹ rate constants imply a half-life of only 35 fs, roughly one vibrational period for a 1000 cm^−l mode. The reaction with [N(p-C₆H₄OCH₃)₃]^•+ provides even tighter constraints because ET1 becomes an additional 0.22 V uphill (k_ET1 is less favorable by 5 × 10⁻³) but k_obs only changes by 10⁻². The constraints on K_ET1 are also more stringent if the back electron or proton transfers have any barrier and are slower than the maximal 10¹³ s⁻¹.

Scheme 5.

Putative kinetic scheme for ET-PT involving precursor and successor complexes, assuming K_precursor = k_successor= 1 M⁻¹ and k_-ET1= k_PTl = 10¹³ s⁻¹ as the best-case scenario.

It should be added that most recent computational and experimental reports conclude that similar intermolecularly hydrogen-bonded [PhOH|base]^•+ species are not minima in the gas phase (proton transfer from O to the base proceeds without barrier).⁶⁹ If this is also the case for the solution ArOH–B^•+ species discussed here, they cannot by definition be intermediates.

The PT2-ET2 pathway is also very unlikely, based on thermochemical arguments similar to those above. The rate constant for PT2-ET2 is the product of the equilibrium constant for initial proton transfer (K_PT2, eq 1 above) and the ET rate constant from the zwitterion (k_ET2, eq 10). K_PT2 might be expected to be ~10⁻⁹ based on the difference in pK_a of amines (~18) and phenols (~27)

$$\mathbf{HOAr}-\mathbf{N}{\mathbf{H}}_{\mathbf{2}}\stackrel{K_{\text{PT}2}}{\rightleftharpoons }{}^{-}\mathbf{OAr}-\mathbf{N}{{\mathbf{H}}_{\mathbf{3}}}^{\mathbf{+}}\stackrel{K_{\text{ET}2}}{\to }{}^{•}\mathbf{OAr}-\mathbf{N}{{\mathbf{H}}_{\mathbf{3}}}^{\mathbf{+}}$$ (10)

in MeCN.³⁸ Since this ignores potential electrostatic interactions in the zwitterion, we instead use the more conservative experimental value, K_PT2 < 10⁻⁴ derived from UV-vis spectra (see above). Then the k_obs > 10⁷ M⁻¹ s⁻¹ for HOAr-NH₂ + [N(p-C₆H₆Br)₃]^•+ implies k_ET2 > 10¹¹, faster than the diffusion limit in MeCN.⁶⁸ An analogous argument can be made with barrier heights and with precursor and successor complexes.

The third argument for the CPET mechanism is the dependence of barrier on driving force. For HOAr-NH₂ + [NAr₃]^•+, a plot of ΔΔAG^‡ vs. ΔΔG° has a slope 0.53 [ΔΔG° = nF(E₁ − E_n) and ΔΔG^‡ = RTln(k₁/k_n)]. Following the discussion above, the stepwise path with rate-limiting ET1 would require ΔG^‡_ET1 ≈ ΔG°_ET1 and therefore that ΔΔG^‡/ ΔΔG° ≅1 (ΔG° ≈ λ, in the Marcus picture; see below).⁶⁷ The initial PT2 mechanism requires that ΔG^‡_ET2 ≈ 0 so ΔΔG^‡/ ΔΔG° ≅ 0 (−ΔG° ≈ λ). As discussed in the next section, ΔΔG^‡/ΔΔG° = 0.53 is close to the value of ½ predicted by Marcus theory for the concerted process in this |AG°_CPET| ≪ 2λ, situation. The dependence of barrier on driving force has previously been used by Okamura et al. to discuss stepwise vs. concerted PCET pathways.²⁰

In sum, the isotope effects, the thermochemistry, and the dependence of the rate constants on driving force are consistent only with a concerted mechanism. These conclusions are consistent with the findings of Linschitz, Hammarstrom, and Nocera who have all found CPET mechanisms for their systems, which include both aqueous and non-aqueous media.^7–89

3. Analysis using Marcus Theory

From one perspective, the phenol-bases HOAr-B are simply outer-sphere electron transfer reagents, so Marcus Theory may be appropriate to analyze these reactions. However, the inner-sphere reorganizations for HOAr-B/^•OAr-BH⁺ are unusual because they involve not only small shifts in equilibrium bond distances, as in the standard Marcus picture, but also movement of a proton across an OH••N hydrogen bond. The proton can be thought of as transferring ~0.7 Å between two minima on an adiabatic potential energy surface.⁷⁰ This would not seem to fit easily into the standard Marcus model where a single parabolic surface, defined by the reorganization energy λ, describes all of the solvent and inner-sphere reorganizations. Current, more refined theoretical formulations of CPET treat the proton transfer explicitly but are more complicated and require more parameters than are readily determined experimentally.²³ So despite its simplifications, the adiabatic Marcus equation is still the logical starting point because it predicts barriers and rate constants using only the two parameters λ and ΔG° (eqs 11,12).^67,71

$$\mathrm{\Delta }{G}^{‡}=\frac{\lambda }{4}{\left(1+\frac{\mathrm{\Delta }G°}{\lambda }\right)}^2$$ (11)

$$k=[{10}^{11}{\text{M}}^{-1}{\text{s}}^{-1}]e^{-\frac{\lambda }{4R\text{T}}{\left(1+\frac{-F{E}_{\text{rxn}}}{\lambda }\right)}^2}$$ (12)

$${\lambda }_{12}=\frac{1}{2}({\lambda }_{11}+{\lambda }_{22})$$ (13)

The kinetic and thermochemical data for HOAr-B + oxidant reactions (Table 3) can be fit by eq 12, as shown in Figure 6.⁷² The data sets are also well fit by straight lines; as noted above the slope ΔΔG^‡/ΔΔG° for the HOAr-NH₂ + [NAr₃]^•+ reactions is 0.53. The limited range of experimentally accessible driving forces (0.51 V) does not provide a test of the predicted parabolic dependence of log(k) on E_CPET.

Figure 6.

log(k) vs. E_rxn for oxidations (a) of HOAr-NH₂ by [NAr₃]^{• +} ( ) and (b) of HOAr-py by [Fe(Me_xphen)₃]³⁺ ( ) and [Fe(R₂bpy)₃]³⁺ ( ) and HOAr-NH₂ by [Fe(Me_xphen)₃]³⁺ ( ) and [Fe(R₂bpy)₃]³⁺ ( ). The curves are fits to eq 12.

The reorganization energies for the oxidations of HOAr-NH₂ derived from eq 12 are 34 ± 1 kcal mol⁻¹ (1.5 eV) for the reactions with [NAr₃]^•+, 38 ± 2 kcal mol⁻¹ (1.9 eV) for the reactions with [Fe(N-N)₃]³⁺ (N-N = R₂bpy, Me_xphen), and 35 ± 1 kcal mol⁻¹ (1.5 eV) from the single rate constant oxidation by 10-methylphenothiazinium (MPT^•+) (Table 4). Following the additivity postulate (eq 13), each of these cross reaction values (λ₁₂) is the average of the λ’s for the individual self-exchange reactions A⁺ + A → A + A⁺. The λ₁₁ for N(tol)₃/[N(tol)₃]^•+ self exchange in MeCN is 12 kcal mol⁻¹ (0.5 eV),⁴³ which is taken as characteristic of the series of NAr₃^•+ oxidants used here. λ₁₁ for MPT^•+/0 is similar (9 kcal mol⁻¹, 0.4 eV);⁷³ that for [Fe(bpy)₃]^3+/2+ in MeCN is twice as large (24 kcal mol⁻¹ or 1.0 eV).⁷⁴ Using eq 13, these values yield reorganization energies for HOAr-NH₂/^•OAr-NH₃⁺ self exchange of 53 ± 3, 53 ± 4, and 58 ± 3 kcal mol⁻¹ (2.4 ± 0.2 eV). These are the same within experimental error, which is an indication that it is appropriate to use the adiabatic Marcus equation to analyze these reactions. The single rate constant for HOAr-im + [N(p-C₆H₄OMe)₃]^•+ gives λ₁₁ = 36 ± 3 kcal mol⁻¹ (1.4 eV) for CPET self-exchange.

The rate constants for oxidation of HOAr-py fall on different lines for oxidants [Fe(R₂bpy)₃]³⁺ (R = H, Me) vs. [Fe(Me_xphen)₃]³⁺ (Me_x = 4,7-Me₂, 3,4,7,8-Me₄). The λ₁₂ values are 27 ± 1 kcal mol⁻¹ (R₂bpy oxidants) and 22 ± 1 kcal mol⁻¹ (Me_xphen oxidants). The distinction is surprising because of the similarity of these oxidants. The self-exchange rate constants for these species are similar, with the phenanthroline derivatives reacting ca. 3 times faster, although this comparison is complicated by scatter amongst the different derivatives, counter-ion and ionic strength effects, etc.^50c Taking λ₁₁ ≅ 24 and 21 kcal mol⁻¹ for [Fe(R₂bpy)₃]^3+/2+ and [Fe(Me_xphen)₃]^3+/2+, respectively, yields apparent λ₁₁ (HOAr-py) values of 30 ± 3 and 23 ± 3 kcal mol⁻¹ for the different oxidants. These values are different, as indicated by the distinct lines in Figure 6. This discrepancy suggests a deviation from the adiabatic Marcus treatment, perhaps due to non-adiabaticity or to ion pairing issues, as will be probed in future work.⁷⁵ The oxidations of HOAr-NH₂ do not show such a distinction between reactions with [Fe(R₂bpy)₃]³⁺ vs. [Fe(Me_xphen)³⁺, although there is some scatter in the rate constants (red curve of Figure 6b). Fitting these rate constants separately yields cross reaction λ₁₂ and HOAr-NH₂ self-exchange λ₁₁ values that agree within error: [Fe(R₂bpy)₃]³⁺: λ₁₂ = 39 ± 2, λ₁₁ = 53 ± 4; [Fe(Me_xphen)₃]³⁺: λ₁₂ = 37 ± 2, λ₁₁ = 53 ± 4 kcal mol⁻¹.

Reorganization energies can also be derived from the temperature dependence of the rate constants using equation 12.⁷⁶ These are given as λ₁₂(T) in Table 4, to distinguish them from the reorganization energies derived from the dependence on E_CPET, λ₁₂(E). The λ₁₂(E) and λ₁₂(T) values are the same within experimental error for the three cases where comparisons are made. This agreement, between two different kinds of analysis and involving mostly independent data sets, supports the use of the Marcus equation for these CPET reactions. Another correct prediction of the Marcus treatment is that ΔΔG^‡/ΔΔG° = 0.5 + ΔG°/2λ. Using the values in Tables 3 and 4 for the five HOAr-NH₂ + NAr₃^•+ reactions, 0.5 + λG°/2λ, ranges from 0.39 to 0.57 with an average value of 0.49. This is in good agreement with the experimental linear fit ΔΔG^‡/ΔΔG° = 0.53.

The electrochemical rate constant k_e1 provides an additional test of the applicability of the adiabatic Marcus treatment. While rigorous comparison of heterogeneous and homogenous electron transfer kinetics is complex, there is often a good correspondence between k_e1 and the homogenous self-exchange rate constant k₁₁ via eq 14.^44a,77 Equation 14 follows from the assumption that a given reagent has similar intrinsic barriers for homogeneous and heterogeneous electron transfer. The rate constants are divided by the different collision frequencies and the self-exchange k₁₁ appears as a square root because it involves two molecules and therefore two intrinsic barriers. K₁₁ for HOAr-NH₂ has not been directly determined, but is

$$\sqrt{\frac{k_{11}}{{10}^{11}{\text{M}}^{-1}{\text{s}}^{-1}}}=\frac{k_{\text{el}}}{{10}^4{\text{cm}}^{-1}{\text{s}}^{-1}}$$ (14)

calculated to be 8 M⁻¹ s⁻¹ using the adiabatic Marcus equation (eq 12) with λ₁₁ = 55 kcal mol⁻¹ (the average of the three experimentally derived values in Table 4). Then (k₁₁/10¹¹)^1/2 = 9 × 10⁻⁶, a factor of 20 larger than (k_e1/10⁴) = 3 × 10⁻⁷. This is good agreement given the approximate nature of eq 14 and that the k_e1 of 3 × 10⁻³ cm s⁻¹ lies on the cusp of the conditions where eq 14 holds - according to Swaddle, only for k_e1 ≤ 10⁻² cm s^−1.77b

The results reported here are among the first confirmations that the adiabatic Marcus equation is applicable to this class of CPET reactions, in which the proton and electron are clearly separated in the reactants or products. We and others have used Marcus Theory for CPET reactions, assuming its applicability.^8,24 The tests described here are: the equivalence of the intrinsic barriers derived from the dependence on driving force, the dependence on temperature, and from different reagents, and the agreement between electrochemical and solution rate constants. It should be noted that these tests are not especially stringent, and that there is the possibility of a deviation in the difference λ’s derived for HOAr-py with the different iron oxidants, which will be explored in more detail in future work.⁷⁵ Hammarstrom and co-workers have previously shown that similar λ’s are derived from driving force and temperature dependent measurements.⁸

The 56 kcal mol⁻¹ (2.4 eV) reorganization energy for HOAr-NH/^•OAr-NH₃⁺ self-exchange in MeCN is a quite large value.^64b,67d HOAr-NH₂ is fairly close in size to N(tol)₃ and has the same charge, yet the phenol-amine CPET λ₁₁ is 4.7 times that of the triarylamine: 56 vs. 12 kcal mol⁻¹. The 12 kcal mol⁻¹ value for N(tol)₃/[N(tol)₃]^•+ is typical of λ’s for outer-sphere electron transfer by aromatic organic molecules, usually ≤ 20 kcal mol^−1.64b HOAr-NH₂ has a much higher intrinsic barrier because ET is coupled to transfer of the proton. Hammarstrom and co-workers have reached the same conclusion in their studies, that CPET oxidations of tethered phenol and indole groups in water have much higher intrinsic barriers than the pure electron transfers from the same reagents.⁸ The similarities of our conclusions are striking in light of the differences in our systems, Hammarstrom measuring rate constants in water for intramolecular ET coupled to proton transfer to the bulk aqueous solution. The λ₁₁’s for HOAr-im and HOAr-py are smaller than that of HOAr-NH₂ but still larger than those for aromatic organic molecules. The significant differences in intrinsic barriers for the amino vs. the pyridyl and imidazolyl derivatives will be discussed in a future report.⁷⁵

The large intrinsic barriers to CPET - indicating that it is inherently difficult - would suggest that this concerted reaction would be disfavored relative to the stepwise ET-PT mechanism. If the driving forces for the competing rate limiting steps were identical, this would indeed be the case. However, in this system E°(CPET) is substantially more favorable than E°(initial ET), which leads to the lower barrier for the CPET (Figure 7). In other words, initial pure ET or PT is disfavored because of the high energy of the intermediate that would be formed. A similar argument has been advanced by Hammarstrom et al. for aqueous CPET reactions.⁸

Figure 7.

Marcus potential energy surfaces for (a) CPET, with a favorable ΔG°_CPET but a large λ_CPET; vs. (b) initial ET (to be followed by PT), with an unfavorable ΔG°_ET but a smaller λ_ET.

Extrapolation of these conclusions to a specific biological system requires caution because typically the driving forces for CPET, pure ET and pure PT are not known. The local dielectric constant and nearby protein residues can substantially affect these values (and the intrinsic barriers). When initial ET or PT are energetically competitive with CPET, as found for instance for quinone reductions in PS I, stepwise pathways are favored.²⁰ However, the concerted mechanism likely occurs in many situations, when pure ET and pure PT are high in energy.

4. Adiabatic vs. Nonadiabatic Electron Transfer

The discussion above has utilized the adiabatic Marcus equation, but many electron transfer reactions are nonadiabatic. Current theoretical descriptions of CPET use a nonadiabatic formalism.²³ In a nonadiabatic reaction, there is a low probability of crossing from the reactant to product diabatic surfaces when the system reaches the transition structure (transmission coefficient κ « 1). In adiabatic reactions, the system is well described by a ground state potential energy surface with κ ~ 1. The matrix element H_rp is a measure of this coupling and appears in the pre-exponential of the nonadiabatic Marcus equation (eq 15). Values of H_rp less than ~200 cm⁻¹ (~k_BT) normally indicate a nonadiabatic reaction.⁷⁸

$$k=K_{\text{P}}\times \frac{4{\pi }^2{H}_{\text{rp}}^2}{h\sqrt{4\pi [\lambda (\text{nonad})]k_{\text{B}}\text{T}}}{e}^{-\left[\frac{{([\lambda (\text{nonad})]+\mathrm{\Delta }G°)}^2}{\text{4}[\lambda (\text{nonad})]k_{\text{B}}\text{T}}\right]}$$ (15)

H_rp is best determined through measurements in the region where −ΔG° ≅ λ, but such measurements are not possible with this system. An alternative though problematic approach fits k as a function of T to eq 15 to obtain H_rp and the nonadiabatic reorganization energy λ₁₂(nonad).⁷² This analysis requires the assumption that the equilibrium constants for forming the precursor complexes K_P are 1 at all temperatures, in the absence of electrostatic work (note ΔG° has been found to be roughly constant with temperature). The apparent values for H_rp, 10 ± 4,6 ± 2, and 4 ± 3 cm⁻¹ and λ₁₂(nonad) are given in Table 4. These H_rp values would normally indicate a nonadiabatic reaction. However, K_P is likely to be smaller than 1 (two standard estimating approaches give K_P ≅ 0.86^64e and 0.02⁷⁹) and is likely to have a temperature dependence, becoming smaller at higher temperatures due to an unfavorable entropy. Including either K_P < 1 or such a temperature dependence would increase the value of H_rp. Thus the apparent H_rp values calculated from eq 15 with K_P = 1 are lower limits.⁸⁰ For instance, if the entropy of forming the precursor complex ΔS°_P were −10 cal K⁻¹ mol⁻¹ and K_P = 0.02 at 298 K, the derived H_rp would be 130 cm^−1.81

In sum, the CPET reactions described here appear to be at most mildly nonadiabatic. The slowness of the electron transfer reactions of HOAr-NH₂ are not due to substantial nonadiabatic character but rather to large reorganization energies. For example, HOAr-NH₂ + [N(tol)₃]^•+ is ~10⁵ slower than [N(tol)₃]^•+/0 self-exchange, both of which have ΔG° ≅ 0. This is because λ₁₂ for HOAr-NH₂ + [N(tol)₃]^•+ is substantially larger than the λ₁₁ of 12 kcal mol⁻¹ for N(tol)₃^•+/0 self exchange,⁴³ whether one uses the adiabatic λ₁₂ = 34 kcal mol⁻¹ (from eq 12), or the nonadiabatic λ₁₂[nonad] = 30 kcal mol⁻¹ (from eq 15 with K_P = 1). HOAr-py and HOAr-im have intrinsic barriers that are also large but are smaller than that for HOAr-NH₂. The origin of these barriers and the differences among these structurally similar phenol-bases will be discussed in a future publication.⁷⁵

Conclusions

One-electron oxidation of phenols hydrogen-bonded to a pendant base, HOAr-B, yield radical cations in which the phenolic proton has transferred to the base, ^•OAr-BH⁺. Three cases are reported here, with amino, pyridyl, and imidazolyl bases. These systems serve as models for hydrogen-bonded tyrosine residues in proteins, and more generally as an archetype for a class of coupled proton-electron transfer (CPET) reactions where the electron and proton travel to different sites. The redox potentials of these phenols are lower than those of simple phenols reflecting the favorable transfer of the proton to the hydrogen-bonded base.

Reactions of HOAr-B with [NAr₃]^•+ or [Fe(N-N)₃]³⁺ oxidants in MeCN follow simple bimolecular kinetics. The mechanism of oxidation involves concerted transfer of the proton and electron (CPET). Three arguments rule out the alternative step wise mechanisms of initial proton and subsequent electron transfer, or initial electron and subsequent proton transfer. First, the primary kinetic isotope effects (1.6 – 2.8) are inconsistent with the stepwise pathways. Second, the rates of oxidation are too fast to involve the high energy intermediates of the stepwise pathways (the observed barriers are lower than the estimated free energies of [⁻OAr-BH⁺] and [^•+HOAr-B]). Third, the dependence of the rate on driving force for reaction of HOAr-NH₂ + [NAr₃]^•+, ΔΔG^‡/ΔΔG°= 0.53, is consistent only with the |ΔG°| « 2λ, situation found for CPET. Based on this work and related model systems,^7–9 CPET is likely a common (albeit underappreciated) mechanism for phenol oxidations. It is favored when the phenol is hydrogen-bonded to a base and when the intermediates in the stepwise paths are high in energy. These conditions probably occur often in biological systems, although the local protein environment can have a substantial influence on the relevant energetics.

The CPET reactions are in general well described by the adiabatic Marcus equation (eq 12). Fitting the variation in k with E_CPET for a series of oxidants yields self-exchange reorganization energies λ₁₁ = 56 ± 3,27 ± 4, and 36 ± 3 kcal mol⁻¹ for HOAr-NH/^•OAr-NH₃⁺, HOAr-py/^•OAr-pyH⁺, and HOAr-im/^•OAr-imH⁺. For HOAr-NH₂, the same λ₁₁ is found for three different oxidants, as required by the Marcus treatment. For each of the phenols, the temperature dependence of the rate constants gives the same λ₁₁ as found from k vs. E_CPET. The ΔΔG^‡_CPET/ΔΔG°_CPET = 0.53 for HOAr-NH₂ + [NAr₃]^•+ reactions is very close to the value of ½ predicted by the Marcus equation for this |ΔG°_CPET| « 2λ case. The electrochemical rate constant for HOAr-NH₂ correlates well with the calculated HOAr-NH₂ self-exchange rate constant. These results support the use of simple Marcus Theory for such CPET systems, although these are not particularly stringent tests. A deviation from the adiabatic Marcus equation may have been observed in the different λ₁₁ values obtained for oxidations of HOAr-py with [Fe(R₂bpy)₃]³⁺ (27 ± 3) vs. [Fe(Me_xphen)₃]³⁺ (23 ± 3 kcal mol⁻¹).

The CPET rate constants are slower than pure ET reactions of comparable organic reagents, especially for HOAr-NH₂. This is a result of the large intrinsic barriers for CPET. A small part of this rate difference could be that the CPET reactions are more nonadiabatic but the data do not support highly nonadiabatic CPET. The 27 – 56 kcal mol⁻¹ adiabatic reorganization energies for these reactions are significantly larger than typical λ’s for ET reactions of aromatic organic compounds. For instance, [N(tol)₃]^•+/0 self exchange has λ₁₁ = 12 kcal mol^−1.43 The pyridyl and imidazolyl compounds have significantly lower intrinsic barriers than the amino derivative. Future work will probe the origins of these barriers and the differences among the different phenols.⁷⁵

Experimental

General

Unless otherwise noted, reagents were purchased from the Aldrich, solvents from Fischer, and deuterated solvents from Cambridge Isotope. MeCN was used as obtained from Burdick and Jackson (low-water brand) and stored in an argon-pressurized stainless steel drum plumbed directly into a glove box. ⁿBu₄NPF₆ was recrystallized three times from EtOH and dried in vacuo for two days at 110° C prior to use. ¹H NMR and ¹³C NMR spectra were recorded on Bruker AF300, AV300, AV301, DRX499 or AV500 spectrometers at ambient temperatures; chemical shifts are reported relative to TMS in ppm by referencing to the residual solvent signals. The UV-vis spectra were obtained on a Hewlett Packard 8453 diode array spectrophotometer and are reported as λ_max in nm (ε, M⁻¹ cm⁻¹), except for the long pathlength spectrum that was obtained on a CARY-500 instrument. The EPR spectrum was recorded on a Bruker EPX CW-EPR spectrometer operating at X-band frequency at room temperature.

The synthesis of HOAr-NH₂, equilibration experiments, crystallographic data and the EPR spectrum of ^•OAr-NH₃⁺ are given in the Supporting Information of this paper and of reference ²⁴. Preparation of MeOAr-NH₂ followed the procedure in Scheme 2 starting from the methyl-bromoaryl ether C₆H₂(OMe)(2-Br)(4,6-^tBu₂) (see Supporting Information). HOAr-im was prepared as described by Benisvy³⁰ by the condensation reaction: aldehyde + ammonium acetate + 4,4′-dimethoxybenzil. The preparation of HOAr-py used the Ni(dppe)₂ coupling of the phenol-derived Grignard and 2-bromopyridine described by Fujita³¹ except with a BBr₃ deprotection of the methyl ether.⁸²

Electrochemistry

Cyclic voltammograms were taken on an E2 Epsilon electrochemical analyzer (Bioanalytical Systems) at ca. 5 mM substrate in anaerobic 0.1 M ⁿBu₄NPF₆ acetonitrile solution, unless otherwise specified. The electrodes were as follows: working, platinum disc (unless noted otherwise); auxiliary, platinum wire; and reference, Ag/AgNO₃ (0.01 M) in electrolyte solution. All potentials are reported vs. a Cp₂Fe^+/0 internal standard. Errors are estimated to be ±0.02V. Representative CVs are included as Figure S17 of the Supporting Information.

For cyclic voltammetry at elevated and depressed temperatures a single solution was used for each analyte. Typically the electrochemical cell was prepared and degassed and CVs were collected yielding E_½ vs Ag/AgNO₃. The entire cell was then placed in a warm water bath and allowed to come to thermal equilibrium before CVs were collected. This process was repeated using an ice bath. The cell was allowed to return to ambient temperature and a final series of CVs were collected, in all cases, the E_½ was found to be within 5 mV of the initial room temperature measurements. Lastly, ferrocene was added as an internal standard and another CV was obtained. The potential of ferrocene vs Ag/AgNO₃ was found to vary by less than 5 mV over the temperatures studied so room temperature Cp₂Fe^+/0 was used as a reference for all of the CVs. A glassy carbon working electrode (φ = 3 mm) was used for HOAr-NH₂ and HOAr-py in these experiments. The driving force for the HOAr-NH₂ + [N(tol)₃]^+• and HOAr-py + [Fe(Me₂-bpy)₃]³⁺ reactions were taken as the difference of E_1/2(oxidant) – E_1/2(HOAr-B), see Table S21 in Supporting Information.

In the determination of the heterogeneous k_e1 for HOAr-NH₂, the platinum disk working electrode (φ = 1.6 mm) was polished before each scan with commercial alumina solution and rinsed with water, dilute HNO₃ (aq), ethanol, and acetonitrile before use. The uncompensated resistance R_u of the electrochemical cell was measured at each experiment, and was found to be on the order of 500 Ω which is expected to have a negligible effect on the measured potential E (< 5 mV). Simulated CVs were produced with DigiSim^® version 3.03⁴⁶ using the experimentally measured values for E₀,E_int,E_rev,E_end, ν, electrode area (planar), D_O, D_R, and k_e1. The mechanism model used was: B + e⁻ → A. The parameter a was taken to be 0.5.⁴⁵

Kinetics

Kinetics experiments were performed on an OLIS RSM-1000 stopped-flow in anaerobic MeCN. The data were analyzed with SpecFit™ global analysis software.⁵¹ Kinetics were fit to pseudo-first order, second order, or opposing second order kinetics as appropriate. To determine the isotope effects, solutions were prepared with a large excess of benchtop CH₃OD (1% v/v for HOAr-NH₂ or 0.5% v/v for HOAr-py). Control experiments showed that aerobic addition of an equivalent amount of CH₃OH to MeCN solutions did not affect the rate. The isotope effects were corrected for the OH content in the CH₃OD (determined via ¹H NMR), 7% for the experiments with DOAr-ND₂ and 4% for those with DOAr-py. Rate constants and data analyses are in the Supporting Information of this paper and of reference ²⁴.

X-ray crystallographic data and experimental descriptions are in the Supporting Information.

Supplementary Material

SI1

Supporting Information Available: Syntheses and cyclic voltammograms, and kinetic, chronoamperometry, and X-ray crystallographic data. This material is available free of charge at http://pubs.acs.org.