Photoinitiated Oxidative Addition of CF₃I to Gold(I) and Facile Aryl-CF₃ Reductive Elimination

Abstract

Herein we report the mechanism of oxidative addition of CF₃I to Au(I), and remarkably fast C_aryl–CF₃ bond reductive elimination from Au(III) cations. CF₃I undergoes a fast, formal oxidative addition to R₃PAuR′ (R = Cy, R′ = 3,5-F₂-C₆H₄, 4-F-C₆H₄, C₆H₅, 4-Me-C₆H₄, 4-MeO-C₆H₄, Me; R = Ph, R′ = 4-F-C₆H₄, 4-Me-C₆H₄). When R′ = aryl, complexes of the type R₃PAu(aryl)(CF₃)I can be isolated and characterized. Mechanistic studies suggest that near-ultraviolet light (λ_max = 313 nm) photoinitiates a radical chain reaction by exciting CF₃I. Complexes supported by PPh₃ undergo reversible phosphine dissociation at 110 °C to generate a three-coordinate intermediate that undergoes slow reductive elimination. These processes are quantitative and heavily favor C_aryl–I reductive elimination over C_aryl–CF₃ reductive elimination. Silver-mediated halide abstraction from all complexes of the type R₃PAu(aryl)(CF₃)I results in quantitative formation of Ar–CF₃ in less than 1 min at temperatures as low as −10 °C.

Introduction

Reports of organogold complexes undergoing redox processes are typically limited to slow oxidative additions and reductive eliminations.^1,2 However, organogold complexes are not necessarily unreactive; we recently showed that diaryl Au(III) complexes undergo remarkably fast aryl–aryl reductive elimination at temperatures as low as −50 °C.³ These recent findings from our group, as well those established by Vicente,⁴ Hashmi,⁵ and Lloyd-Jones,⁶ suggest that the barrier for challenging reductive eliminations might be substantially diminished at Au(III). C_aryl–CF₃ bond reductive elimination is typically a slow process requiring elevated temperatures and long reaction times, due to ground state stabilization afforded by exceptionally strong bonding between transition metals and CF₃ ligands.⁷ For instance, (dppbz)Pd(2-Me-C₆H₄)(CF₃) (dppbz = 1,2-bis(diphenylphosphino)benzene) is stable at 130 °C for 3 days,⁸ while (dppp)Pd(Ph)(CF₃) (dppp = 1,3-diphenylphosphinopropane) and (dppe)Pd(Ph)(CF₃) (dppe = 1,2-diphenylphosphinoethane) yield only 10% PhCF₃ after 3 days at 145 °C.⁹ Reductive eliminations at temperatures between 50 and 80 °C can be achieved at Pd(II) by employing bulky ligands, such as Xantphos¹⁰ and Brettphos.¹¹ Notably, while aryl-CF₃ reductive eliminations from Pd(IV) often require similarly high temperatures,^12a Sanford has shown that they can occur at temperatures as low as 23 °C over 1 h.^12b Despite advances in catalytic trifluoromethylation, C_aryl–CF₃ reductive elimination still remains a challenging step. Given the importance of trifluoromethylated arenes in pharmaceuticals and agrochemicals,¹³ we were prompted to investigate potentially low-barrier C_aryl–CF₃ bond reductive elimination at Au(III).

To access complexes of the type R₃PAu(aryl)(CF₃)I, we were drawn to Puddephatt’s report of the oxidative addition of CF₃I to Me₃PAuMe to afford cis/trans mixtures of Me₃PAuMe₂(CF₃) and Me₃PAuI.¹⁴ In one case, Me₃PAu(Me)(CF₃)I was obtained exclusively, but its preparation could not be reproduced by the authors. Because reaction times varied from 5 min to 1 day, and rates dramatically slowed in the presence of galvinoxyl, the authors concluded that a free-radical chain mechanism was operative, with ^•CF₃ as the propagating species.

Results and Discussion

Prior investigations by our group revealed that oxidation of Ph₃PAu(4-F-C₆H₄) rapidly generates 4,4′-difluorobiphenyl through a mechanism involving aryl group transfer.³ However, the use of the bulkier PCy₃ prevents transfer of the arene ligand, instead resulting in clean, rapid oxidation of Cy₃PAu(4-F-C₆H₄) (1a) to the isolable Au(III) complex cis-(Cy₃P)Au(4-F-C₆H₄)Cl₂ (2) (eq 1).¹⁵

1

Therefore, we began our investigations of Au(I) oxidation by CF₃I using 1a, with the fluorinated arene ligand also providing a convenient ¹⁹F NMR handle. Treatment of 1a in CD₂Cl₂ with CF₃I (25 equiv) afforded the product of formal CF₃I oxidative addition 3a in 1 h in good yield (eq 2 and Table 1). Both the CF₃ and PCy₃ ligands (doublet at δ = −24.5 and quartet at δ = 25.6 in the ¹⁹F and ³¹P NMR spectra, respectively) provide diagnostic NMR signals (Table 2). The substantial coupling (³J_P–F = 63 Hz) between fluorine and phosphorus are characteristic of a trans relationship between the CF₃ and phosphine ligands.¹⁴ X-ray analysis of crystals of 3a confirmed this stereochemical relationship around the square planar Au(III) (Figure 1A); other than the homoleptic anion [Au(CF₃)₄]⁻,¹⁶ complex 3a contains a rare example of a crystallographically characterized Au(III)–CF₃ bond. Complex 3a is not only stable to air and water but can be purified by column chromatography as well.

2

Table 1. Photoinitiated Oxidative Addition of CF₃I to Electronically Diverse Au(I) Aryl Complexes 1a–1f, 11a, and 11b.

Au(I) reactant	PR3	Ar	product	yield (%)
1a	PCy3	4-F-C6H4	3a	64
1b	PCy3	3,5-F2-C6H3	3b	44
1c	PCy3	C6H5	3c	59
1d	PCy3	4-Me-C6H4	3d	38
1e	PCy3	4-MeO-C6H4	3e	44
1f	PCy3	2-Me-C6H4	−	NR
11a	PPh3	4-F-C6H4	12a	71
11b	PPh3	4-Me-C6H4	12b	63

Table 2. ³¹P{¹H} and ¹⁹F NMR Data for Complexes 3a–3e, 12a, and 12b.

complex	PR3	Ar	δ 31P{1H} (ppm)	δ 19F (ppm)
3a	PCy3	4-F-C6H4	25.6 (q, 3JP–F = 63 Hz)	–24.5 (d, 3JP–F = 63 Hz)
3b	PCy3	3,5-F2-C6H3	26.1 (q, 3JP–F = 63 Hz)	–22.0 (d, 3JP–F = 64 Hz)
3c	PCy3	C6H5	25.5 (q, 3JP–F = 62 Hz)	–22.7 (d, 3JP–F = 62 Hz)
3d	PCy3	4-Me-C6H4	25.5 (q, 3JP–F = 62 Hz)	–23.6 (d, 3JP–F = 62 Hz)
3e	PCy3	4-MeO-C6H4	23.3 (q, 3JP–F = 63 Hz)	–20.6 (d, 3JP–F = 64 Hz)
12a	PPh3	4-F-C6H4	20.0 (q, 3JP–F = 68 Hz)	–21.0 (d, 3JP–F = 68 Hz)
12b	PPh3	4-Me-C6H4	20.4 (q, 3JP–F = 67 Hz)	–21.3 (d, 3JP–F = 67 Hz)

Figure 1.

(A–F) Thermal ellipsoid representations of 3a–3d, 12a, and 12b at the 50% probability level. Hydrogens have been omitted for clarity. Atoms are color-coded: gray (carbon), yellow (fluorine), gold (gold), purple (iodine), orange (phosphorus). See Supporting Information (SI) for bond lengths and angles.

Mechanism of Oxidative Addition of CF₃I

The reaction of 1a and CF₃I represents a rare oxidation of Au(I) to Au(III) that directly installs potentially reactive Au(III)–carbon bonds.¹ During our attempts to monitor the oxidative addition by ¹⁹F NMR, we found that no reaction occurred when the reaction mixture was placed inside the dark NMR spectrometer. However, when the reaction mixture was exposed to ambient fluorescent light for 5 min, the formation of 3a was detected (∼20%). Given the reliance of numerous methods on CF₃I as a trifluoromethyl source,¹⁷ we investigated its photochemical reactivity. Actinometry experiments were carried out to determine the overall quantum yield, using the Norrish II fragmentation of valerophenone as a standard.¹⁸ The oxidative addition of CF₃I to 1a was complete after 20 s of irradiation by a Hg vapor lamp (2 mM aq. K₂CrO₄ optical filter; transmittance λ_max = 313 nm), while the fragmentation of valerophenone (Φ = 1) took place over 24 h under identical conditions. This rate difference, in addition to the ability of ambient light to bring the reaction to full conversion over variable reaction times (between 15 min and 1 h), supports a radical chain reaction as the mechanism of Au(I) oxidation by CF₃I.

The reaction of excess CF₃I and 1a is also fast in THF, but the conversion is never greater than 65% (52% yield of 3a), even when irradiated by a Hg vapor lamp for 1 h (vida infra). Notably, an excess of fluoroform (HCF₃) is generated in THF, regardless of the light source (only DCF₃ is formed when THF-d₈ is used). GC-MS analysis of reaction mixtures reveals several products of THF oxidation, likely formed by H^• abstraction by ^•CF₃.

Several control experiments, using HCF₃ production relative to a standard as a probe to detect ^•CF₃ generation, support the involvement of Au(I) during the initiation of the chain reaction. The UV absorption of CF₃I is centered at 270 nm but tails beyond 350 nm.¹⁹ When irradiated at 313 nm, CF₃I undergoes fast, reversible C–I bond homolysis. However, in the absence of 1a, only negligible amounts of HCF₃ are observed when THF solutions of CF₃I are irradiated for 30 min, indicating that carbon/iodine radical recombination is substantially faster than H^• abstraction from THF. Similarly insignificant quantities of HCF₃ are observed when 20 equiv (relative to CF₃I) of the H^• donors 1,4-cyclohexadiene, 9,10-dihydroanthracene, or triphenylmethane are added (Figure 2A). Additionally, Cy₃PAu(2-(CH₂CH=CH₂)C₆H₄) (4), containing a pendent olefin to either capture a putative Au(II) intermediate and/or ^•CF₃, is fully consumed upon irradiation in the presence of excess CF₃I (Figure 2B). This oxidation affords multiple Au(III) products of indiscriminate ^•CF₃ addition to the terminal olefin and gold atom (and HCF₃ when THF is used as solvent) (see SI). Because 2-allylbromobenzene (5) does not react with CF₃I when irradiated under similar conditions (no HCF₃ is observed after 5 min, and less than 2% after 30 min), we conclude that the Au(I) aryl complex is necessary for chain initiation. These results are also consistent with an initiation mechanism involving [CF₃I]^•–, which generates iodide and ^•CF₃ following C–I bond homolysis.

Figure 2.

Control experiments to assess involvement of Au(I) in the initiation of the radical chain mechanism. (A) Irradiation of CF₃I solutions containing H^• donors to detect CF₃H in the absence of gold. (B) Radical trapping using an olefin with and without a pendant gold center.

We envisioned two possible initiation mechanisms for generating ^•CF₃ as a propagating species from [CF₃I]^•– in a chain reaction: (1) initial photoexcitation of 6 followed by electron transfer to CF₃I, or (2) initial photoexcitation of CF₃I followed by electron transfer from 6 (Scheme 1).

Scheme 1. Possible Initiation Mechanisms Involving Photoexcitation of Either Au(I) Complex 6 (Mechanism 1) or CF₃I (Mechanism 2).

Au(I) aryl complexes are well-known chromophores, and their photophysical properties have been investigated previously.²⁰ While 1a absorbs weakly above 310 nm (the cutoff for many laboratory fluorescent lamps¹⁹), excitation at 320 nm (ε = 37 M^–1 cm^–1) results in a weak, broad luminescence from 340 to 460 nm, classified as fluorescence based on the lifetime of excited species 1a* (<10 ns, quantum yield of fluorescence = 0.03).²¹ Despite the short lifetime of 1a*, CF₃I effectively quenches its fluorescence (Stern–Volmer quenching constant K_SV = 30 M^–1, Figure 3). Although this energy transfer could conceivably generate ^•CF₃ and initiate a chain reaction (mechanism 1, Scheme 1), when CF₃I is removed from fluorimetry samples under vacuum, fluorescence is restored to the same intensity prior to introduction of the gas, indicating that consumption of Au(I) has not occurred.

Figure 3.

Stern–Volmer plots of fluorescence quenching of 1a by different concentrations of CF₃I (blue boxes) and Au(III) complex 3a (blue triangles) in CH₂Cl₂. Concentrations of Au(III) are in mol/L and CF₃I concentrations are in mmol/L.

Surprisingly, fluorescence quenching by the Au(III) complex 3a is more than 2 orders of magnitude more effective (K_SV = 4270 M^–1) than quenching by CF₃I (Figure 3). If propagating species terminate frequently, some critical concentration of Au(III) product exists that may impede productive energy transfer from an excited species, halting reinitiation of the chain reaction.

In light of Puddephatt’s report, Au(I) alkyl complexes, such as Me₃PAuMe, clearly react with CF₃I.¹⁴ However, there is no mention of the dependence of light on this process, although if the reaction is photoinitiated, mechanism 1 would seem especially unlikely given the absence of a chromophoric aryl ligand in Puddephatt’s examples. To test this hypothesis, we irradiated Cy₃PAuMe (9) in the presence of CF₃I (Scheme 2). While 9 does not absorb above 300 nm (see SI), the reaction is quantitative in CD₂Cl₂ when irradiated with ambient light, and does not proceed in the dark. The oxidized product is unobservable, eliminating CH₃I to generate Cy₃PAuCF₃ at room temperature.^22,23 In THF, the reaction generates excess HCF in THF₃, presumably also from solvent H^• abstraction by ^•CF₃.

Scheme 2. Photochemical Oxidative Addition of CF₃I to 9 in CD₂Cl₂ and Spontaneous Reductive Elimination of CH₃I.

If initiation mechanism 2 is operative, then ^•CF₃ could be generated by irradiating CF₃I solutions containing electron donors other than Au(I), such as phosphines (Scheme 3).²⁴ Indeed, irradiation of PMe₃ or PCy₃ in the presence of CF₃I results in formation of [Me₃P-CF₃]I (10a, ²J_P–F = 63 Hz) or [Cy₃P-CF₃]I (10b,²J_P–F = 42 Hz);²⁵ neither reaction proceeds in the dark. Consistent with quenching of [CF₃I]* by Au(III), the oxidation of PCy₃ in THF stalls at roughly 45% conversion (by ³¹P NMR) in the presence of 25 mol % Au(III) complex 3a.

Scheme 3. Photochemical Oxidation of Trialkylphosphines by CF₃I.

PPh₃ does not react with CF₃I (eq 3), presumably due to its lower oxidation potential relative to PMe₃ and PCy₃. When PCy₃ and PPh₃ are irradiated together with CF₃I, only PCy₃ is consumed, suggesting that PPh₃ neither initiates the chain nor reacts with ^•CF₃ during propagation. Contrary to our initial hypothesis that bulky phosphine ligands prevent aryl group transfer upon Au(I) oxidation, we found that Ph₃PAu(4-F-C₆H₄) (11a) undergoes quantitative photoinitiated reaction with CF₃I in CD₂Cl₂ to generate 12a (eq 4). Since PPh₃ is unreactive toward CF₃I, oxidation of 11a cannot be initiated by small amounts of dissociated PPh₃ (we cannot disprove the analogous mechanism for PCy₃-supported complex 1a.) Complex 12a was characterized by X-ray crystallography and shown to be isostructural to 3a (Figure 1B).

3

On the basis of these results, we propose that while photoexcited [CF₃I]* undergoes rapid C–I bond homolysis and recombination, it also oxidizes Au(I) aryl and alkyl complexes by accepting electrons into a low-lying SOMO to generate radical anion [CF₃I]^•– (mechanism 2, Scheme 1). Homolysis of the C–I bond of [CF₃I]^•– generates iodide and ^•CF₃, which oxidizes (R₃P)AuR′ (6) to Au(II) intermediate 7. Iodine atom abstraction of CF₃I by 7 affords Au(III) complex 8 and regenerates ^•CF₃. In THF, oxidation of 6 by ^•CF₃ is competitive with solvent H^• abstraction to make HCF₃ and terminate the radical chain. At sufficiently high concentrations, the Au(III) product (8) quenches [CF₃I]* before it can reinitiate the radical chain reaction.

Promisingly, the photoinitiated oxidative addition of CF₃I is general for electronically diverse complexes of the type Cy₃PAu(aryl) (Tables 1 and 2). The resulting Au(III) products (see Figure 1 for their crystallographic analyses) can be purified by chromatography on silica. Complex 1b (aryl = 3,5-F₂-C₆H₃), which is more electron-deficient than 1a (aryl = 4-F-C₆H₄), reacts smoothly with CF₃I to afford 3b. While complexes with more electron-rich ligands such as 1c (aryl = C₆H₅) and 1d (aryl = 4-Me-C₆H₄) also react with CF₃I to afford 3c and 3d, respectively, the most electron-rich complex 1e (aryl = 4-MeO-C₆H₄) decomposes to Au nano particles and several CF₃-containing Au(III) complexes in solution and solid state (no products of C_aryl–I or C_aryl–CF₃ reductive elimination can be detected). Au(III) product 3e is detectable, however, and its decomposition can be slowed substantially by addition of MeCN upon concentration of the reaction, allowing its solution-state characterization. The mechanism of decomposition has not yet been identified, although we speculate that the electron-rich arene may encourage PCy₃ dissociation at room temperature and subsequent aryl group transfer.

The complex 1f (aryl = 2-Me-C₆H₄) does not react with CF₃I at all, suggesting that CF₃I oxidative addition is sensitive to the sterics of the aryl ligand and that relaxation of [CF₃I]* is faster than oxidation of the metal center to initiate the radical chain. Unsurprisingly, no HCF₃ is observed when 1f is irradiated in THF for 20 min.

Reductive Elimination from Au(III) Complexes

We next probed C_aryl–CF₃ reductive eliminations from Au(III). To our surprise, 12a undergoes quantitative C_aryl–I reductive elimination in toluene-d₈ at 110 °C to afford 4-fluoroiodobenzene and Ph₃PAuCF₃ over 20 min (Scheme 4).²² No 4-fluoro(trifluoromethyl)benzene is observed by ¹⁹F NMR or GC. This process is highly sensitive to free phosphine, stalling completely in the presence of PPh₃ (0.1 or 1.0 equiv) at 110 °C for 12 h. Treatment of 12a with PPh₃-d₁₅ at room temperature results in immediate formation of 12a-d₁₅, presumably via an associative process.^2a−2d

Scheme 4. Behavior of Au(III) Complex 12a and 12b in the Presence of Free PPh₃ and at Elevated Temperatures.

More electron-rich aryl ligands, such as 4-methylphenyl (12b), do not significantly affect the relative rates of C_aryl–I and C_aryl–CF₃ reductive elimination (Scheme 4). At 110 °C, complex 12b undergoes mostly C_aryl–I reductive elimination within 10 min to afford 4-methyliodobenzene.²⁶ Both C_aryl–I and C_aryl–CF₃ reductive eliminations are also completely inhibited in the presence of PPh₃ (0.1 or 1.0 equiv), while PPh₃-d₁₅ reacts immediately at room temperature to afford 12b-d₁₅, also via associative ligand exchange. These observations are consistent with a mechanism involving highly reversible PPh₃ dissociation from 12a and 12b, followed by slow C_aryl–I reductive elimination from 13a or 13b, respectively.

Clearly, the behaviors of 12a and 12b are similar to Au(III)alkyl complexes studied by Kochi, which not only reductively eliminate C_alkyl–C_alkyl bonds between 70 and 100 °C via a dissociative mechanism but also undergo associative ligand exchange at ambient temperature with excess phosphine.^2a−2d Unsurprisingly, analogous PCy₃-stabilized complexes 3a and 3d are stable at 110 °C for at least 12 h, presumably due to the greater σ-donating ability of PCy₃ relative to PPh₃. Phosphine exchange with excess P(n-Bu)₃, PBn₃, or PCy₃ does not occur even at these temperatures, precluding not only the lower-barrier associative exchange mechanism observed with the PPh₃-supported systems (attributed to the larger cone angle of PCy₃ relative to PPh₃), but also PCy₃ dissociation to form a three-coordinate complex.

Because C_aryl–I reductive elimination is significantly faster than C_aryl–CF₃ reductive elimination, a cycle for gold-catalyzed trifluoromethylation must necessarily involve iodide abstraction from the Au(III) product of CF₃I oxidative addition. Despite the apparent kinetic stabilities of the Au(III) complexes 3a–3e, 12a, and 12b, they all undergo quantitative C_aryl–CF₃ reductive elimination in less than 1 min upon treatment with AgSbF₆at room temperature.

To consider the effects of the phosphine ligand on the silver-mediated C_aryl–CF₃ reductive elimination of Au(III), we used variable-temperature NMR to follow the reductive elimination from 3a and 12a in the presence of AgSbF₆. PCy₃-substituted complex 3a undergoes very fast (quantitative conversion in less than 1 min) C_aryl–CF₃ reductive elimination at −10 °C, while the analogous PPh₃-stabilized 12a reacts similarly fast at room temperature (eq 5). At lower temperatures, several bridging species (most likely dimers) are observed by ¹⁹F NMR upon halide abstraction in both cases. If C_aryl–CF₃ bond reductive elimination can only occur from a monomeric three-coordinate intermediate, then 12a might be expected to undergo slower reductive elimination due to slower dimer dissociation and/or a dimer–monomer equilibrium that more favors the dimer, based on the smaller cone angle and weaker σ-donation of PPh₃ relative to PCy₃.

5

Conclusion

These results reported herein support the oxidative addition of CF₃I to Au(I) via a photoinitiated chain reaction. The reactions are fast at room temperature for both Au(I) aryl and alkyl complexes. Aryl-CF₃ reductive elimination is typically a high-barrier process but occurs in seconds at room temperature from a Au(III) cation. The Au(I)aryl species may be regenerated via one of the numerous transmetalation strategies available involving carbon nucleophiles.²⁷ For instance, excess (4-F-C₆H₄)SnMe₃ (10 equiv) undergoes fast, quantitative transmetalation with [Cy₃PAu]SbF₆ at room temperature to afford 1a, thereby closing a hypothetical catalytic cycle based on the three elementary steps shown in Scheme 5. Silver-free halide abstraction from Au(III) complexes could conceivably enable a practical and mild cycle for gold-catalyzed trifluoromethylation of aryl nucleophiles, although deleterious reactions between starting material and metalloradical intermediates and ^•CF₃ must be mitigated, as well as competitive aryl–aryl homocoupling.

Scheme 5. Oxidation of 1a, Aryl-CF₃ Reductive Elimination, and Regeneration of 1a Supports the Feasibility of a Mild, Catalytic Trifluoromethylation.

While we initially set out to probe C_aryl–CF₃ reductive elimination at Au(III), we also explored the oxidative addition of CF₃I to Au(I), a process with potential implications beyond gold chemistry. The possibility of photoinitiated oxidation of transition metals or main group elements by CF₃I should not be discounted in methods employing this reagent as a trifluoromethyl source, particularly since ambient fluorescent laboratory lighting is sufficient to initiate a chain in the presence of a suitable reductant. The results presented also suggest that substrate photoexcitation may provide a low-barrier avenue to kinetically challenging oxidative additions by Au(I), providing access to potentially reactive Au(III) complexes.²⁸

Supporting Information Available

Experimental details, characterization data, and crystallographic information (cif). This material is available free of charge via the Internet at http://pubs.acs.org.

The authors declare no competing financial interest.

Funding Statement

National Institutes of Health, United States