Structural Characterization of the [CuOR]²⁺ Core

Abstract

Formal Cu(III) complexes bearing an oxygen-based auxiliary ligand ([CuOR]²⁺, R = H or CH₂CF₃) were stabilized by modulating the donor character of supporting ligand L^Y (L^Y = 4-Y, N,N′-bis(2,6-diisopropylphenyl)-2,6-pyridinedicarboxamide, Y = H or OMe) and/or the basicity of the auxiliary ligand, enabling the first characterization of these typically highly reactive cores by NMR spectroscopy and X-ray crystallography. Enhanced lifetimes in solution and slowed rates of PCET with a phenol substrate were observed. NMR spectra corroborate the S = 0 ground states of the complexes, and X-ray structures reveal shortened Cu–ligand bond distances that match well with theory.

Understanding the molecular structures, spectroscopic properties, and reactivity of copper–oxygen complexes^1–5 is important for gaining insight into the mechanisms by which copper enzymes and other catalysts function.^6,7 Among the various complexes studied, those comprising the [CuOH]²⁺ core supported by dicarboxamide ligands (Figure 1)^8–14 are notably reactive, attacking C–H and O–H bonds via proton-coupled electron transfer (PCET) processes and undergoing electron transfer at high rates (cf. the rate constant for the reaction of L^HCuOH with 1,2-dihydroanthracene 50 M⁻¹ s⁻¹ at −25 °C;⁹ electron-transfer self-exchange rate constant ~10⁴ M⁻¹ s⁻¹ at −88 °C¹³). The formulations of the complexes are supported by UV–vis spectroscopy (diagnostic absorptions with ligand-to-metal charge-transfer (LMCT) character), EPR silence, resonance Raman spectroscopy (ν_Cu–O ≈ 630 cm⁻¹),¹³ EXAFS (avg Cu–N,O ≈ 0.1 Å shorter than for the [Cu^IIOH]⁺ precursor),⁸ and theory.^8,15

Figure 1.

Complexes studied previously and the compounds that comprise the focus of this work. R′ = Me or aryl groups.

Perturbations of the [CuOH]²⁺ core have been effected by the installation of remote substituents on the flanking aryl rings (L^XCuOH, Figure 1).^11,12 These perturbations are reflected by shifts in the LMCT energies (lower), redox potentials (higher), basicities (lower), and PCET reaction rates (faster) that may be rationalized by electron withdrawal by the X groups. Changing the pyridyl group to a piperidine (L^pipCuOH) has the opposite effects, attributed to greater electron donation by the amine donor. In no case has a complex with the [CuOH]²⁺ core been structurally defined by X-ray crystallography, in large part due to its high reactivity and poor thermal stability (t_1/2 ≈ minutes at −80 °C in THF). These characteristics and the fact that decomposition yields paramagnetic Cu(II) species have also inhibited efforts to obtain NMR spectra that would be useful in confirming the proposed S = 0 ground state. Significantly slower reactions and greater stability were observed for the LCuZ derivatives (Z = halide^8,16 or carboxylate¹⁷) featuring a less-basic X-type reactive moiety, of which the halide complexes were characterized recently by X-ray diffraction and NMR spectroscopy.¹⁶ Inspired by that success, we hypothesized that more complete characterization of the [CuOH]²⁺ core might be attained if its stability could be enhanced. Toward this end, we targeted two modifications: the placement of an electron-donating para-methoxy pyridyl group (Y = OMe; L^OMe) and the use of a less-basic alkoxide moiety (CF₃CH₂O⁻). On the basis of previous work,^9,11,12 we hypothesized that these changes would enhance the solution stability, thereby facilitating handling and characterization. We now report the confirmation of these hypotheses through the synthesis of the targeted derivatives and their characterization by X-ray crystallography and NMR spectroscopy, which with comparison to results from theory provides new insights into the molecular and electronic structures of complexes with [CuOH]²⁺ and [CuOCH₂CF₃]²⁺ cores.

The reaction of proligand H₂L^OMe with Cu(OTf)₂ in the presence of NaOMe and CH₃CN led to the isolation of key precursor L^OMeCu(CH₃CN) (77%), which upon treatment with NBu₄OH led to [NBu₄][L^OMeCuOH] (76%). [NBu₄]-[L^YCuOCH₂CF₃] (Y = H or OMe) complexes were obtained from the respective hydroxide precursors via protonolysis with HOCH₂CF₃. The complexes L^OMeCu(CH₃CN), [NBu₄]-[L^OMeCuOH], and [NBu₄][L^YCuOCH₂CF₃] (Y = H or OMe) were characterized by UV–vis and X-band EPR spectroscopy, CHN analysis, and X-ray crystallography (SI). The complexes feature a slightly distorted square-planar geometry (τ₄ = 0.15–0.22)¹⁸ with Cu–N,O bond distances typical for such Cu(II) species (Figures S43–S45; Table 1) and characteristic S = 1/2 rhombic signals with Cu and N hyperfine patterns in EPR spectra (Figures S7–S9).

Table 1.

Bond Distances (Å) Obtained by X-ray Crystallography for [CuOR]²⁺ (Cu^III) and [CuOR]⁺ (Cu^II) Species and by DFT Geometry Optimization for [CuOR]²⁺ (Theory)^a

bond	species	LOMe, OH	LH, OCH2CF3	LOMe, OCH2CF3
Cu-N2	CuIII	1.841(3)	1.864(13)	1.856(3)
	theory	1.845	1.875	1.866
	CuII	1.927(2)	1.925(4)	1.927(1)
	CuIII−CuII	−0.086	−0.061	−0.071
Cu-N1,3	CuIII	1.900(3)	1.950(12)	1.948(3)
	theory	1.916	1.960	1.960
	CuII	2.027(2)	2.058(4)	2.020(1)
	CuIII−CuII	−0.127	−0.108	−0.072
Cu-O1	CuIII	1.799(3)	1.812(12)	1.849(5)
	theory	1.783	1.818	1.817
	CuII	1.867(2)	1.857(4)	1.838(2)
	CuIII−CuII	−0.068	−0.045	0.011
mean	CuIII−CuII	−0.102	−0.081	−0.051

^aAveraged values are presented for carboxamide Cu–N₁ and Cu–N₃ bonds. N₂ is pyridyl donor. Estimated standard deviations are in parentheses.

The expected electronic perturbations of the L^OMe and −OCH₂CF₃ moieties are reflected in cyclic voltammograms (THF, 0.2 M Bu₄NPF₆), which contain quasi-reversible waves associated with [CuOR]²⁺/[CuOR]⁺ couples (Table 2). Good reversibility of the waves for the −OCH₂CF₃ complexes is apparent at scan rates as slow as 10 mV/s (Figure S20), consistent with low reactivity for the oxidized species (vide infra). A more modest enhancement of reversibility is seen for [NBu₄][L^OMe CuOH] relative to the analog supported by L^H (Figure S18).⁵ The replacement of −OH with −OCH₂CF₃ shifts the E_1/2 by +0.191 V on average, whereas the alteration of L^Y had a lesser effect, with an average difference of only −0.049 V induced by the introduction of the p-OMe substituent.

Table 2.

Reduction Potentials for the [CuOR]^2+/+ Couple and Rate Constants for Reactions with ^ttbPhOH

compound	E1/2 (mV)a	k2 (M−1 s−1)b
LHCuOH	−167c	15 900
LOMeCuOH	−202	11 800
LCuOCH2CF3	+37	3.0
LOMeCuOCH2CF3	−25	1.5

^aConditions: THF, 0.2 M Bu₄NPF₆, values vs Fc^+/0.

^bConditions: −25 °C in DFB using either 1 equiv (L^YCuOH) or 50 equiv (L^YCuOCH₂CF₃).

^cThis value is ~100 mV lower than reported previously under the same conditions,¹¹ which we ascribe to the previous use of external Fc referencing rather than the internal referencing used herein. (See SI IV for details.)

The chemical oxidation of [NBu₄][L^OMe CuOH] or [NBu₄]-[L^YCuOCH₂CF₃] (Y = H or OMe) was performed by adding 1 equiv of FcBAr^F₄ or AcFcBArF₄, respectively, in THF (−80 °C) or in 1,2-difluorobenzene (DFB, −25 °C). Immediate color changes (deep violet or blue) and intense electronic absorption features (Figure 2) diagnostic of the formation of [CuOH]²⁺ and [CuOCH₂CF₃]²⁺ cores were observed, and reversible one-electron processes were confirmed through titrations and the sequential cyclic additions of oxidant and decamethylferrocene (Figures S22–S27).¹³ TDDFT UV–vis transitions exhibit λ_max values in agreement with experiment that are predominantly HOMO to LUMO (Figure 2) and shift from 546.1 and 537.1 nm in L^HCuOH and L^OMeCuOH to 578. 7 and 573. 6 nm in L^HCuOCH₂CF₃ and L^OMeCuOCH₂CF₃, respectively, as a result of the stabilization of the LUMO (Table S4, Figures S50–S58). Also in agreement with experiment (and precedent^17a), additional transitions with partial HOMO-to-LUMO character contribute at longer wavelengths to the spectra for the −OCH₂CF₃ complexes. Resonance Raman spectra of frozen solutions (λ_ex = 561 nm) of all L^YCuOR complexes contain a signal at ~635 cm⁻¹ that we assign as ν_Cu‑OR by analogy to data acquired for L^HCuOH and theory (Figures S45–S49 and S59, Table S15).¹³

Figure 2.

(Left) UV–visible absorption spectra of the indicated compounds in DFB at −25 °C. (Right) Orbitals for L^OMeCuOCH₂CF₃ plotted with an isovalue of 0.04 au from the B98 functional.

To evaluate how ligand variation influences stability and reactivity, we compared the reactions of L^YCuOH and L^YCuOCH₂CF₃ (Y = H or OMe) with 2,4,6-tri-tert-butylphenol (^ttbPhOH) to yield the stable phenoxyl radical.^11,19,20 Second-order rate constants were measured using either 1 equiv (L^YCuOH) or 50 equiv (L^YCuOCH₂CF₃) of ^ttbPhOH at −25 °C in DFB (Table 2). The data show significantly higher reactivity for the hydroxide complexes (≳5000-fold), a difference that may be attributed to the higher basicity of the −OH vs −OCH₂CF₃ moieties and/or steric inhibition for the latter. Modest decreases in the rate constants for the cases where Y = OMe at parity for −OR may be rationalized by stabilization of the [CuOR]²⁺ core by the electron-donating methoxide substituent. Monitoring the room-temperature decays of the four [CuOR]²⁺ species in the absence of substrate in THF and DFB revealed complicated kinetic traces, but trends in the overall lifetimes paralleled the ^ttbPhOH reactivity trends (SI; cf. t_1/2 ≈ 4 h vs <1 h for L^OMeCuOCH₂CF₃ vs LCuOH in DFB).

The enhanced stability of the new complexes led us to attempt characterization by X-ray crystallography. We discovered that the complex prepared by the treatment of [NBu₄][L^OMeCuOH] with FcBAr^F₄ in DFB could be isolated as suitable deep-purple crystals via the layered diffusion of pentane at −30 °C. Similar attempts with [NBu₄][L^YCuOCH₂CF₃] (Y = H or OMe) failed to give suitable crystals, likely in part due to the formation of highly intractable viscous residues containing [NBu₄][BAr^F₄]. To circumvent this issue, we employed reactants that would yield insoluble inorganic salts as byproducts. Thus, we reacted LCu-(CH₃CN)⁴ and L^OMeCu(CH₃CN) with NaOCH₂CF₃ and then oxidized the resulting crude materials with AcFcSbF₆ in CH₂Cl₂ or DFB. After the removal of a light-colored precipitate (presumably NaSbF₆), suitable crystals of the oxidized products were obtained at −30 °C.

Representations of the X-ray structures of L^OMeCuOH and L^OMeCuOCH₂CF₃ (Figure 3) as well as L^HCuOCH₂CF₃ (Figure S42) show similar square-planar geometries compared to their [CuOR]⁺ progenitors (τ₄ = 0.11–0.15), but they are neutral species as expected for one-electron oxidation products. Comparison of metal–ligand bond distances between the oxidized and reduced forms (Table 1) indicates in all but one case shortening upon oxidation, by as much as 0.127 Å. The average Cu–N/O bond contraction in L^OMeCuOH, 0.102 Å, is in excellent agreement with previously reported EXAFS analyses of L^HCuOH (0.1 Å).⁸ The trifluoroethoxides show somewhat less contraction and in L^OMeCuOCH₂CF₃ the Cu–O bond even lengthens slightly, by 0.011 Å, but disorder in the trifluoroethoxide ligand imparts an inherent inaccuracy to the O atom’s position. Gas-phase geometry optimizations (SI) for the S = 0 ground states of the oxidized species are in excellent agreement with the experimentally determined values (theory in Table 1). Overall, the bond length differences between the precursor and oxidation products are consistent with the loss of an electron from orbitals spanning the Cu center and/or its immediate environment, as reported for less-reactive complexes LCuZ (Z = F, Cl, Br).¹⁶ Furthermore, the lack of significant structural changes associated with this redox event agrees with the low reorganization energy of 0.95 eV previously measured for the [L^HCuOH]⁻/L^HCuOH couple.¹⁴

Figure 3.

Representations of the X-ray structures of (a) L^OMeCuOH and (b) L^OMeCuOCH₂CF₃, showing all nonhydrogen atoms as 50% thermal ellipsoids.

While theoretical calculations support a closed-shell S = 0 ground state for the [CuOH]²⁺ core (Table S5),⁸ the only experimental corroboration has come from a dearth of signal in the X-band EPR spectrum, an observation consistent with either the S = 0 or S = 1 ground state. Acquiring NMR spectra, which would distinguish the two spin states, presents challenges due to the formation of paramagnetic Cu(II) decay species. We were nonetheless able to observe sharp peaks in the diamagnetic region of ¹H NMR spectra for both L^YCuOH species in 1,2-dichlorobenzene-d₄ at −15 °C (Figure S28–S29), although broadening due to decomposition was evident for Y = H. The ¹H NMR spectrum of the more robust complex L^HCuOCH₂CF₃ in THF-d₈ at −80 °C (Figure S30) displayed negligible broadening, but some resonances were obscured by solvent/byproduct signals. Since L^OMeCuOCH₂CF₃ could be isolated in neat form as a crystalline solid, NMR spectra of isolated material were collected (CD₂Cl₂, −15 °C), and all expected ¹H NMR resonances and J couplings (Figure 4) as well as ¹³C{¹H} NMR peaks (Figure S32) were observed. The sharpness of the observed ¹H and ¹³C{¹H} NMR features in the diamagnetic chemical shift region confirms an S = 0 ground state, in agreement with predictions.⁸

Figure 4.

¹H NMR spectrum of L^OMeCuOCH₂CF₃ (CD₂Cl₂, −15 °C).

In conclusion, we prepared and characterized a new set of complexes with [CuOR]^+/2+ cores using a modified supporting ligand and/or core (R = CH₂CF₃). Both changes attenuate the PCET reactivity of the oxidized state, the former by lowering its oxidizing potential and the latter by lowering the basicity of the proton-accepting site. While each modification has the opposite effect on the opposite property (e.g., the less-basic proton acceptor also leads to a more oxidizing species), the dominant impact is on electronics for the supporting ligand and basicity for the core, in line with previously observed reactivity trends and demonstrating how PCET reactivity can be tuned. These stabilization effects were sufficient to permit, for the first time, the successful characterization of complexes with [CuOR]²⁺ cores by X-ray crystallography and NMR spectroscopy. These data are consistent with several predictions made about L^HCuOH, in particular, the conservation of geometry with minimal reorganization upon oxidation,¹⁴ the EXAFS-derived contraction of Cu–L bonds by ~0.1 Å on oxidation,⁸ the calculated geometries, and the diamagnetic S = 0 ground state.⁸ To the best of our knowledge, the complexes with R = CH₂CF₃ are the first alkoxo analogs bearing the formal Cu(III) oxidation state, and with the discovery that they share many similarities with their hydroxide counterparts, including PCET reactivity with phenols, the potential for steric and electronic tuning that is unavailable with hydroxide makes this class of compounds promising for further research into the bond-activation properties of high-valent copper species.