Intramolecular Hydrogen Bonding Enhances Stability and Reactivity of Mononuclear Cupric Superoxide Complexes

Abstract

[(L)Cu^II(O₂^•−)]⁺ (i.e., cupric-superoxo) complexes, as the first and/or key reactive intermediates in (bio)chemical Cu-oxidative processes, including in the monooxygenases PHM and DβM, have been systematically stabilized by intramolecular hydrogen bonding within a TMPA ligand-based framework. Also, gradual strengthening of ligand-derived H-bonding dramatically enhances the [(L)Cu^II(O₂^•−)]⁺ reactivity toward hydrogen-atom abstraction (HAA) of phenolic O−H bonds. Spectroscopic properties of the superoxo complexes and their azido analogues, [(L)Cu^II(N₃⁻)]⁺, also systematically change as a function of ligand H-bonding capability.

The activation of molecular oxygen with copper is of considerable interest in nature, as well as in oxidative synthetic procedures and in addressing energy concerns (e.g., fuel-cell O₂ reduction catalysts).¹ Reduction of O₂ with Cu^I leads to complexes with singly or doubly reduced dioxygen (superoxo or (hydro)peroxo, respectively); further important downstream intermediates (viz. Cu^II−O^•) form following reductive O−O cleavage.² Various Cu^I−O₂ derived species are suggested to initiate substrate attack leading to O atom incorporation i.e., net hydroxylation in copper monooxygenases such as peptidylglycine-α-hydroxylating monooxygenase (PHM) and dopamine-βmonooxygenase (DβM).^3a-e For these cases, along with the dehydrogenation of RCH₂OH in galactose oxidase (GO), copper amine oxidase iminosemiquinone H atom abstraction (HAA), and GO cofactor biogenesis (HAA from cysteine S− H),^3f a mononuclear cupric superoxide (Cu^II−O₂^•−) intermediate is implicated in effecting key H atom abstraction (HAA) chemistry.^3b,c,4

Thus, there have been considerable efforts in the design of appropriate synthetic ligands to probe the factors affecting the formation, stabilization, and reactivity of these metastable [LCu^II(O₂^•−)]⁺ species.^2b,c,5 For example, increasing electron donation to Cu,⁵ providing steric interactions,⁶ controlling coordination geometry,⁷ and employing cryogenic conditions and suitable solvents^5–8 have each been shown to stabilize the superoxide species and prevent its subsequent reaction with a second equiv of the Cu^I precursor to yield thermodynamically favored binuclear peroxo-bridged dicopper(II)complexes(Eqs1 and 2).⁹

$${\text{LCu}}^{\text{I}}+{\text{O}}_2\underset{k_{-1}}{\overset{k_1}{\rightleftharpoons }}{\text{LCu}}^{\mathrm{II}}-{\text{O}}_2{}^{•-}{K}_1$$ (1)

$${\text{LCu}}^{\mathrm{II}}{-\text{O}}_2{}^{•-}+{\text{LCu}}^{\mathrm{I}}\underset{k_{-2}}{\overset{k_2}{\rightleftharpoons }}{\text{LCu}}^{\mathrm{II}}-{\text{O}}_2{}^{2-}-{\text{Cu}}^{\text{II}}\text{L}{K}_2$$ (2)

H-bonding in the secondary coordination sphere is known to stabilize binuclear [{(L)Cu^II}₂(O₂²⁻)]²⁺ and mononuclear [(L)Cu^II(OOH)]⁺ complexes with N₃O and N₄ ligand architecture.¹¹ Considering extensive recent efforts in studying H-bonding influences on model complexes¹² and in biological systems (for example, this plays a crucial role in stabilizing and activating O₂-intermediates,¹³ effecting nitrite reduction,¹⁴ and facilitating neural NOS activity¹⁵), the importance of H-bonding on [(L)Cu^II(O₂^•−)]⁺ chemistries still remains a fundamental and critical area for study.

In this report, we demonstrate that the imposition of progressively enhancing H-bonding moieties in a series of TMPA-based L−Cu complexes {Figure 1; pK_as of related relevant organic groups: pentafluorobenzylamine (~7) < benzylamine (9.3),¹⁶ and note that pivalamide and other amides are more acidic than 2° amines} (i) inhibits the formation of corresponding binuclear trans-μ−1,2-peroxo [{(L)Cu^II}₂(O₂²⁻)]²⁺ complex, in favor of monocopper [(L)-Cu^II(O₂^•−)]⁺ species, while (ii) also significantly enhances the HAA reactivity of the superoxo species toward phenolic substrates.

Figure 1.

[(L)Cu^II(O₂^•−)]⁺ complexes without (A, L = TMPA; tris(2pyridyl-methyl)amine) and with internal H-bonding substituents (B−D, L = BA, F₅BA, and MPPA), and corresponding resonance Raman spectra (λ_ex = 413.1 nm) of complexes A−D in frozen 2methyltetrahydrofuran (MeTHF) solutions prepared with ¹⁶O₂ (blue) and ¹⁸O₂ (red). Labels indicate the center of the Fermi doublets. See Supporting Information for tabulation of ν(Cu−O) values.¹⁰

When a 0.4 mM solution of [(TMPA)Cu^I(MeCN)]B(C₆F₅)₄ in MeTHF was subjected to dioxygen bubbling at −135 °C and allowed to equilibrate for 30 min, the resulting mixture consisted oftheknown⁹ end-on superoxide(i.e.,Cu^II−O−O^•−)complex A {[(TMPA)Cu^II(O₂^•−)]⁺} (63% based on Cu (see Section 4.3(a) in the SI); λ_max = 423 nm, ε = 5600 M⁻¹cm⁻¹; 752 nm)¹⁰ and the known trans-peroxide dicopper(II) complex [{(TMPA)Cu^II}₂(O₂²⁻)]²⁺ (37%; λ_max = 525, 610 nm) (Figure 2).¹⁰ With its ligand derived weak H-bonding (vide infra),¹⁷ O₂-bubbling of [(BA)Cu^I)]⁺ significantly increased relative amounts of B {[(BA)Cu^II(O₂^•−)]⁺} (83% (see Section 4.3(b) in the SI); λ_max = 418 nm, ε = 4100 M⁻¹ cm⁻¹; 750 nm), compared to [{(BA)Cu^II}₂(O₂²⁻)]²⁺ (17%; λ_max = 500, 605 nm).¹⁰ However, dramatic stabilization is observed with utilization of F₅BA, a new ligand capable of stronger H-bonding than in BA, cupricsuperoxide [(F₅BA)Cu^II(O₂^•−)]⁺ (414 nm, ε = 4000 M⁻¹ cm⁻¹; 748 nm), is generated exclusively.¹⁰ Employing MPPA,^11c,18 possessing the strongest H-bonding moiety, complex D {[(MPPA)Cu^II(O₂^•−)]⁺} is formed. These superoxo species, C and D, are very stable in solution (−135 °C); they are fully formed, and there appears to be no hint of any peroxide analog present.^10,19

Figure 2.

UV−vis spectra of oxygenated mixtures of ~0.4 mM of [(L)Cu^I]B(C₆F₅)₄ (L = TMPA, BA, F₅BA, and MPPA) at −135 °C in MeTHF after 30 min. H-bonding stabilizes superoxide species and prevents trans-peroxide formation (see Figures S1−S4).¹⁰

This systematic H-bonding enhancement of [(L)-Cu^II(O₂^•−)]⁺ stabilization is further supported by corresponding UV−vis data obtained at [(L)Cu^I]⁺ initial concentrations of 0.1 mM (see Figures S5−S8).¹⁰ Analogous to synthetic and protein derived oxy-heme adducts (formally Fe^III−O₂^•− species), the primary effect of the H-bonding (e.g., for the distal His residue in hemo- or myoglobins) is to dramatically decrease the rate of O₂ dissociation, hence lowering k₋₁ thereby increasing K₁ (Eq 1). With stronger H-bonding, [(L)Cu^II(O₂^•−)]⁺ species becomes systematically more stable across B−D. Without the H-bond in [(TMPA)Cu^II(O₂^•−)]⁺(A), k−₁ is significant, more Cu(I) becomes available, and [(L)Cu^II(O₂^•−)]⁺ readily reacts to yield the thermodynamic [{(L)Cu^II}₂(O₂²⁻)]²⁺ product (Eq 2).

Resonance Raman (rR) spectroscopy was used to characterize the effect of H-bonding in the series of superoxide complexes A−D in frozen MeTHF solution. Laser excitation (413.1 nm) of each superoxide complex (~1 mM) revealed two sets of ^16/18O isotope-sensitive enhanced modes at around 460 and 1120 cm⁻¹ (Figure 1), which are characteristic of ν(Cu−O) and ν(O−O), respectively (the latter appears as a Fermi doublet (Figure 1), consistent with previous observations²¹), of end-on cupric superoxides (see Section 5 in the SI for details).^{5a,c,6a,7,18,21b,c}

Interestingly, a systematic trend is observed for the A−D complexes where the ν(O−O)values¹⁸ increasedwithenhancing H-bonding ability of the systematically stronger N−H dipoles present across B−D. Analogous to previously reported cases,^18,21c we presume the observed incremental trend in ν(O−O) across B−D is caused by enhanced structural vibrational coupling and alignment of increased O−O dipole opposite to the systematically increasing N−H dipole. [(TMPA)Cu^II(O₂^•−)]⁺ (A), without any H-bonding moiety, exhibited the lowest (weakest) O−O stretch as expected.

In parallel, the doubly occupied superoxide π_σ* (LMCT donor MO) undergoes more electrostatic stabilization compared to the Cu^II-d_z² (LMCT acceptor MO) with increasing H-bonding across the A−D [(L)Cu^II(O₂^•−)]⁺ series, which increases the energy of the LMCT (π_σ* → Cu^II-d_z²) transition;¹⁰ hence, a systematic blue shift is observed in the higher energy LMCT absorption, ranging from 423 to 410 nm across A−D (Figure 2 and Table 1).

Table 1.

Summary of Spectroscopic Features (UV-vis, IR, and rRaman) for [(L)Cu^II(O₂^•−)]+ and [(L)Cu^II(N₃⁻)]⁺ Complexes

	TMPA	BA	F5BA	MPPA
rR O−O stretch {Δ18O2} (cm−1)	1119 {−61}	1123 {−64}	1126 {−64}	1130 {−63}
λmax CuII−O2•− (nm)	423	418	414	410
λmax CuII−N3− (nm)	413	402	399	395
IR N≡N stretch (cm−1)	2046	2049	2052	2063

To obtain better insight into the H-bonding in these [(L)Cu^II(O₂^•−)]⁺ complexes, we examined the structures of the azido analogs, [(BA)Cu^II(N₃⁻)]ClO₄, [(F₅BA)Cu^II(N₃⁻)]ClO₄, and [F₅BA)Cu^II(OClO₃)(acetone)]ClO₄ (for syntheses, see the SI).¹⁰ Azide ion bound to Cu^II complexes have been previously used to model both electronic and connectivity structural aspects for coordination of a peroxide dianion.²² With respect to a cupric-superoxide complex, a Cu^II−N₃ moiety would be expected to have a very similar Cu^II−N−N−N binding mode (e.g., bond distances and the <Cu^II−N−N angle). The crystal structure of [(BA)Cu^II(N₃⁻)]⁺ (Figure 3) experimentally determined via single-crystal X-ray crystallography revealed a weak but clearly observable intramolecular H-bond found between the benzyl amino N−H hydrogen atom and the azido N atom (2.17(2) Å) proximal to the Cu(II) ion. The N_amino−H…N_azido distance is 2.965(2) Å with <N_amino−H···N_azido = 163(2)°. Similar H-bonding is observed in the F₅BA analog mentioned above.¹⁰ This interaction, also observed in other Cu^II(X) (X = N₃⁻, ⁻OOH, or peroxo-Cu) instances with TMPA based ligands possessing internal H-bonding groups,^11,23 supports the notion that the H-bonding in complexes B−D occurs to the proximal O atom, as depicted in Figure 1.

Figure 3.

Displacement ellipsoid plot (50% probability level) of the cationic [(BA)CuII(N3−)]+ at 110(2) K.Alist of relevant bond distances and angles is also provided. The H-bonding between the benzylamineand the proximal azide nitrogen is highlighted

To further support our suppositions concerning the Hbonding in these [(L)Cu^II(O₂^•−)]⁺ complexes, we examined the antisymmetric N₃⁻ IR stretch (see Section 7 in the SI)¹⁰ for thefull series of complexes [(L)Cu^II(N₃⁻)]⁺. The characteristic ν_N−N values (Table 1) increased progressively with H-bonding strength across A−D, presumably due to an increase in the triplebond character of the azido ligand, resulting from removal of electron density from the copper-bound N_azido atom, mirroring the increase in rRaman O−O stretches in A−D. Additionally, the azido → Cu^II LMCT band shifts progressively to higher energy with increased ligand H-bonding (Table 1, also see Section 7 in the SI),¹⁰ consistent with the increasing trend in O₂^•−→ Cu^II LMCT energy across A−D.

Elucidation of the oxidative capability and scope of reactivity for [(L)Cu^II(O₂^•−)]⁺ complexes is a major objective. We expected that systematic increased H-bonding to the O₂^•− should stabilize its π* orbital and also increase its electrophilicity. Indeed, we have found a striking difference in the reactivity toward phenols (electrophilic HAA) with varying O−H bond strengths, across the A−D series.

Reactions with 2,6-di-tert-butyl-4-methoxyphenol (DTBP) withaweak O−Hbond(BDE= 82.0kcal/molinDMSO²⁴) were first investigated. [(TMPA)Cu^II(O₂^•−)]⁺ (A) reacts with a second-order rate constant of 3.11 × 10⁻¹ M⁻¹ s⁻¹ (−135 °C, MeTHF; Figure 4) as determined by monitoring absorbance decrease at 752 nm due to A. [(TMPA)Cu^IIOOH]⁺ (~80% yield, see Figure S42) and the corresponding phenoxyl radical were observed, demonstrating a net H atom transfer from phenol to superoxo.^5b,10 Importantly the corresponding peroxodicopper(II) complex [{(TMPA)Cu^II}₂(O₂²⁻)]²⁺ is observed to be completely unreactive toward phenols (see Figure S27).¹⁰

Figure 4.

(a) UV−vis monitoring of the reaction of A and 2,6-di-tertbutyl-4-methoxyphenol; (b) contrasting rates of oxidation by A−D (k₂ values are given in the text); (c) UV−vis spectra of reaction of A with 4methoxyphenol, showing no reaction; (d) reaction of D with 4methoxyphenol and (inset) kinetic trace of the decay of 741 nm band of D. See Section 6 in the SI for details.

The cupric superoxo complexes possessing intramolecular Hbonds (B−D) react more efficiently with DTBP under the same conditions (Figure 4b), with k₂ = 5.73 × 10⁻¹ M⁻¹ s⁻¹ (B), 11.47 × 10⁻¹ M⁻¹ s⁻¹ (C), and 9.88 × 10⁻¹ M⁻¹ s⁻¹ (D). In each case, the phenoxy radical was directly observed by both EPR (g ≈ 2.0) and UV−vis (407 nm) spectroscopies (see Section 9.2 in the SI).¹⁰ Formation of the hydroperoxo product (86% yield) by C via net HAA from DTBP is also verified by spectrophotometric quantitative analysis of the amount of H₂O₂ released upon acidification; see Section 9.3 in the SI. Although the spectroscopic data imply that D is the most activated in the series, it is less reactive toward DTBP than C. We presume that this apparent anomaly is due to the larger steric interaction of the tert-butyl group in D, inhibiting bulky substrate approach and decreasing the advantage D should have, with its strongest Hbonding pivalamido group. Notably, C could also exert π−π stacking interactions with DTBP, facilitating a much favorablesubstrate approach compared to D, hence resulting in its slightly enhanced reaction rate.

Further, we tested our [(L)Cu^II(O₂^•−)]⁺ complexes A−D toward 4-methoxyphenol (MP), a substrate with a stronger O− H bond (BDE = 87.6 kcal/mol in DMSO²⁴). [(TMPA)-Cu^II(O₂^•−)]⁺ was completely unreactive toward 250 equiv of MP at −135 °C (Figure 4c). [(BA)Cu^II(O₂^•−)]⁺ (B) exhibited exceedingly slow reactivity (completion took >6 h), and [(F₅BA)Cu^II(O₂^•−)]⁺ (C) reacted considerably faster.¹⁰ However, the kinetic behavior was complicated in these cases, and the final copper complex product could not be identified, similar to the observation by Tolman and co-workers²⁵ in MP HAA by a Cu^III-hydroxide complex. The MP reaction with [(MPPA)Cu^II(O₂^•−)]⁺ (D) was much cleaner and followed clear secondorder decay (k₂ = 2.33 × 10⁻² M⁻¹ s⁻¹). The corresponding hydroperoxo-Cu(II) product was also observed in UV−vis spectroscopy (Figure 4d). This is the first example of a much stronger phenol being oxidized by a mononuclear cupric superoxide compared to the previously reported cases.

In conclusion, we report for the first time how H-bonding strength in the ligand secondary coordination sphere can be tuned to systematically stabilize cupric superoxide ([LCu^II(O₂^•−)]⁺) species, control the superoxide−peroxide equilibrium, and concurrently activate the superoxide toward enhanced exogenous O−H substrate oxidation. This was achieved by systematically increasing the electrophilicity of [LCu^II(O₂^•−)]⁺ via implementation of differing H-bonding moieties. The structures of the [(L)Cu^II(N₃⁻)]⁺ analogs suggest that H-bonding occurs to the proximal O-atom in ([LCu^II(O₂^•−)]⁺) complexes. Further use of H-bonding or other ligand modifications in order to generate cupric−superoxide complexes with greater oxidative capabilities is being explored.