Reductive C–C Coupling from Molecular Au(I) Hydrocarbyl Complexes: A Mechanistic Study

Organometallic gold complexes are used in a range of catalytic reactions, and they often serve as catalyst precursors that mediate C–C bond formation. In this study, we investigate C–C coupling to form ethane from various phosphine-ligated gem-digold(I) methyl complexes including [Au2(μ-CH3)(PMe2Ar′)2][NTf2], [Au2(μ-CH3)(XPhos)2][NTf2], and [Au2(μ-CH3)(tBuXPhos)2][NTf2] {Ar′ = C6H3-2,6-(C6H3-2,6-Me)2, C6H3-2,6-(C6H2-2,4,6-Me)2, C6H3-2,6-(C6H3-2,6-iPr)2, or C6H3-2,6-(C6H2-2,4,6-iPr)2; XPhos = 2-dicyclohexylphosphino-2′,4′,6′-triisopropylbiphenyl; tBuXPhos = 2-di-tert-butylphosphino-2′,4′,6′-triisopropylbiphenyl; NTf2 = bis(trifluoromethyl sulfonylimide)}. The gem-digold methyl complexes are synthesized through reaction between Au(CH3)L and Au(L)(NTf2) {L = phosphines listed above}. For [Au2(μ-CH3)(XPhos)2][NTf2] and [Au2(μ-CH3)(tBuXPhos)2][NTf2], solid-state X-ray structures have been elucidated. The rate of ethane formation from [Au2(μ-CH3)(PMe2Ar′)2][NTf2] increases as the steric bulk of the phosphine substituent Ar′ decreases. Monitoring the rate of ethane elimination reactions by multinuclear NMR spectroscopy provides evidence for a second-order dependence on the gem-digold methyl complexes. Using experimental and computational evidence, it is proposed that the mechanism of C–C coupling likely involves (1) cleavage of [Au2(μ-CH3)(PMe2Ar′)2][NTf2] to form Au(PR2Ar′)(NTf2) and Au(CH3)(PMe2Ar′), (2) phosphine migration from a second equivalent of [Au2(μ-CH3)(PMe2Ar′)2][NTf2] aided by binding of the Lewis acidic [Au(PMe2Ar′)]+, formed in step 1, to produce [Au2(CH3)(PMe2Ar′)][NTf2] and [Au2(PMe2Ar′)]+, and (3) recombination of [Au2(CH3)(PMe2Ar′)][NTf2] and Au(CH3)(PMe2Ar′) to eliminate ethane.

Proposed mechanisms for Au-mediated C−C bond formation include reductive elimination from Au III intermediates (Scheme 1). 28 For example, reductive elimination from (R) 2 Au(X)(L) (L = phosphine; R = Me, Et, or n Pr; X = anionic ligand such as Cl, OTf, NO 3 , O 2 CCF 3 , or another alkyl ligand) was investigated by Kochi and co-workers, and reductive eliminations from [(Me) 2 Au(L) 2 ] + complexes have been reported. 18,29,30 The proposed mechanism involves initial phosphine dissociation followed by C−C reductive elimination from the three-coordinate R 2 Au III X intermediate (Scheme 1a). When R = Me, isotopic labeling studies with (Me) 2 AuX(L) and (CD 3 ) 2 Au(X)(L) (L = phosphine) indicate kinetically competitive intermolecular transfer of Me between two Au centers, but these alkyl transfers appear to occur only in nonpolar solvents. 23 Further, the putative binuclear Au intermediates responsible for alkyl transfer were not directly implicated in the C−C coupling reactions. From the starting complexes (Me) 2 Au(X)(L) (L = phosphine), it was proposed that larger phosphines facilitate ethane reductive elimination. 30 Alternatively, Kochi has proposed that ethane formation could result from digold alkyl intermediates, but to our knowledge, such reactions were not directly observed. 31,32 Other examples of ethane formation through bimolecular reductive elimination from M−CH 3 species include Ni II , 33 Cu I , 34 and Ru II . 35 The formation of C−C bonds from (NHC)Au I −R (NHC = N-heterocyclic carbene; R = Ph, Me, or p-tolyl) occurs upon addition of electrophiles (R′X), such as PhI, MeI, and MeOTf, to form R−R′ as well as homocoupled products R−R and R′− R′ (Scheme 1b). 36−38 The proposed mechanism involves formal trans oxidative addition of the electrophile (R′X) to form an (NHC)Au III (R′)(X)(R) intermediate, followed by competitive (a) C−C reductive elimination to form R′−R and (b) intermolecular transfer of RX from a Au(III) intermediate to (NHC)Au I −R to yield (NHC)Au III (R) 2 (X) followed by C−C reductive elimination to give the homocoupled product R−R. Related to these processes, the addition of F + -donors to Au(I) hydrocarbyl compounds also promotes C−C coupling reactions. 39−42 Mixed-valent gold hydrocarbyl complexes have also been proposed as intermediates responsible for the CC bond formation. 17 For example, Toste and co-workers reported a fast biaryl C−C bond reductive elimination from a mixed-valent bimetallic Au I /Au III complex [ClAu]PNP[AuCl(C 6 H 4 -4-F) 2 ] (PNP = Ph 2 PN(CH 3 )PPh 2 ) (Scheme 1c). 17 In this study, the Au(I) complex [Au(C 6 H 4 -4-F)]PNP[Au(C 6 H 4 -4-F)] is oxidized with PhICl 2 to generate a symmetric bimetallic Au(II) species, [ClAu(C 6 H 4 -4-F)]PNP[Au(C 6 H 4 -4-F)Cl]. The latter isomerizes to a mixed-valent Au I /Au III complex, [ClAu]PNP-[AuCl(C 6 H 4 -4-F) 2 ], which undergoes reductive elimination to form a biaryl product. Similarly, O'Hair and co-workers reported a concerted redox couple mechanism from a reaction between allylic halides (CH 2 = CHCH 2 X, X = Cl, Br, and I) and a gem-digold(I) compound, [(dppm) 2 Au 2 Ph] + (dppm = bis(diphenylphosphino)methane, (Ph 2 P) 2 CH 2 ). 43 2 ). 46 The thermal stabilities of phosphine-ligated gem-digold hydrocarbyl complexes have been reported to depend on the steric properties of the ancillary ligands. 46  Herein, we explore the formation of ethane from one of the simplest possible gold-based systems, namely, Au(CH 3 )-(PPh 3 ). To enable reliable mechanistic investigations, we extended our preliminary observations on triphenylphosphineligated systems to bulkier terphenyl and biaryl phosphines that provide kinetic stabilization of key digold intermediates. In particular, we have focused on C−C coupling reactions from gem-digold methyl complexes with a general formula [Au 2 (μ-CH 3 )(PMe 2 Ar′) 2 ][NTf 2 ] and [Au 2 (μ-CH 3 )(XPhos) 2 ][NTf 2 ] ( Figure 1). We studied the impact of the phosphine ligand on the stability of digold complexes, especially the influence on ethane elimination.
Terminal ethyl and phenyl complexes Au(C 2 H 5 )-(PMe 2 Ar Xyl2 ) (2a) and Au(C 6 H 5 )(PMe 2 Ar Xyl2 ) (3a) were synthesized with the aim of exploring the possibility of C−C bond heterocoupling with different hydrocarbyl substituents bound to gold (see below). These compounds were prepared by a similar procedure to their methyl analogues and characterized by spectroscopic techniques and single-crystal X-ray diffraction (Figures 3 and 4). The σ Au−C bond distances are 2.079(8) Å (2a) and 2.087(7) (3a) Å, which are similar to neutral Au−CH 3 bond distances discussed above. Complex 2a cocrystallizes in a 1:1 ratio with a molecule of Au(C 2 H 5 )(PMe 2 Ar Mes2 ) (2b, see Figure S1), whose geometric parameters are comparable to those of 2a. This is due to the fact that the crystals were obtained from a phosphine exchange experiment between 2a and free PMe 2 Ar Mes2 that was conducted as part of our mechanistic investigations (see below, Figure S1). The bond distance between C1 and C2 in the Au-ethyl fragment of 2a is 1.411(15) Å, lying between the carbon−carbon lengths of ethylene (1.34 Å) and ethane (1.54 Å). The electrophilic nature of gold may enhance the C−C bond strength and thus shorten bond length compared to a typical C−C single bond. The structure of complex 3a is similar to those of compounds 1 and 2a and does not require further discussion.
Ethane Elimination from Au(CH 3 )(PPh 3 ). For the sake of simplicity and considering the widespread utilization of PPh 3based gold complexes, we commenced our studies by exploring ethane elimination from Au(CH 3 )(PPh 3 ). This compound is stable at moderate temperatures as heating at 40°C caused no apparent alteration when monitoring by 1 H and 31 P{ 1 H} NMR spectroscopy, and no ethane formation was detected. However, in the presence of 1 equiv of Au(PPh 3 )(NTf 2 ), Au(CH 3 )-    Figure S8). 46 However, this compound is only detectable at temperatures below −40°C, and it rapidly evolves to the final products above this temperature.
T h o u g h t h e t r a n s i e n t n a t u r e o f [Au 2 (μ-CH 3 )(PPh 3 ) 2 ][NTf 2 ] prevented us from exploring its role in further detail, our initial kinetic investigations using 1 equiv of the related Au(PPh 3 )(NO 3 ) revealed a second-order dependence on neutral Au(CH 3 )(PPh 3 ) for ethane elimination ( Figure S6). Nonetheless, we could carry out these studies with related methyl complexes based on biaryl and terphenyl phosphines, as discussed in the following sections. Since we observed the formation of gold nanoparticles during ethane elimination, we decided to probe for a possible catalytic role of Au nanoparticles in the C−C coupling reaction, particularly considering their catalytic role in related processes. 53−55 However, using independently prepared gold nanoparticles (i.e., Au/TiO 2 and Au/Fe 2 O 3 ) as catalysts did not promote methyl C−C coupling at a comparable rate (t 1/2 ≈ 1 day at 25°C ). Thus, it seems unlikely that Au nanoparticles play a catalytic role in the formation of ethane. In addition, we tested for the possibility of a radical-mediated pathway. To do so, we combined equimolar amounts of Au(CH 3 )(PPh 3 ) and [Au-(PPh 3 )][NTf 2 ] in the presence of excess toluene (10 equiv) as a radical probe. Under these conditions, the formation of CH 3 • radicals should be quenched by toluene by means of hydrogen atom abstraction from the benzylic position. 56 This process would have released methane, which was not observed during our experiments, thus favoring the likelihood of a nonradical route.
Synthesis of Cationic gem-Digold Methyl Complexes. To probe if gem-digold methyl complexes are relevant intermediates during C−C coupling reactions, bulky terphenyl and Buchwald phosphines were used. Some of us have recently demonstrated that gem-digold methyl species are kinetically stabilized by large phosphine substituents, 46 which should facilitate kinetic investigations. Indeed, using the aforementioned bulky phosphines enabled the isolation and characterization of various uncommon gem-digold methyl complexes of type [Au 2 (μ-CH 3 )(PMe 2 Ar′) 2 ][NTf 2 ] (4a−4d). These were synthesized in high yields by mixing a 1:1 molar ratio of a Au(I) methyl complex Au(CH 3 )(PMe 2 Ar′) and the corresponding Au(I) bis(trifluoromethyl sulfonyl)imide (Scheme 4). Alternatively, the addition of 0.5 equiv of [Ph 3 C][B-(C 6 F 5 ) 4 ] to neutral Au(I) methyl complexes Au(CH 3 )-(PR 2 Ar′) leads to the same gem-digold species in comparable yields.
Overall, these observations indicate that kinetic analysis by 1 H and 31 P{ 1 H} NMR spectroscopy monitoring is facilitated by larger phosphine ligands compared to PPh 3 . For instance, heating complex 4e in dichloroethane at 90°C enabled us to monitor by NMR spectroscopy its evolution to [Au(XPhos) 2 ]-[NTf 2 ] (5e) with concomitant release of ethane and formation of Au(0) ( Figure S5). The thermolysis of 4e follows a secondorder dependence on the digold complex with k obs = 5.2(1) × 10 −4 M −1 s −1 at 90°C (Table 2), as previously observed for the PPh 3 -based system. In the case of the more hindered compound 4f, this reaction does not take place at 100°C, and intractable digold decomposition occurs at temperatures above 100°C where the formation of methane, instead of ethane, was observed ( Figure S14). This finding indicates that C−C coupling is likely not viable in the most sterically constrained digold system studied herein. This seems to be consistent with a second-order dependence on digold complex concentration during ethane formation, which might imply the need for more than two gold nuclei in close proximity along the reaction coordinate (see below for additional discussion). Similar to complex 4e, ethane elimination from terphenyl-ligated gemdigold methyl complexes follows a second order dependence on 4a−4d (Figure 7a; see the Supporting Information). Kinetic studies provide rates for ethane elimination from the more sterically hindered 4c and 4d of k obs = 4.8(3) × 10 −3 and 2.0(1) × 10 −2 M −1 s −1 at 50°C, respectively. In contrast, the rates of ethane elimination from 4a and 4b had to be analyzed at lower temperatures (0°C), resulting in rates of k obs = 9.8(3) × 10 −2 and 4.9(1) × 10 −1 M −1 s −1 at 0°C, respectively. The corresponding half-life (t 1/2 ) values associated with these kinetic parameters at the working temperatures are approximately 260 (4a, 0°C), 800 (4b, 0°C), 5600 (4c, 50°C), and 13 000 (4d, 50°C) s. Table 2 collects the corresponding activation barriers for C− C coupling from the methyl-bridged complexes 4a−4e, which  To complete these studies, we monitored the evolution of ethane from the gem-digold complex 4a in the temperature interval from −20 to 10°C. An Eyring analysis provided activation parameters of ΔH ⧧ = 20.5 ± 1.3 kcal/mol and ΔS ⧧ = 11.9 ± 4.8 e.u. (Figure 7b), which correspond to ΔG 298 ⧧ = 16.9 ± 2.7 kcal/mol. To obtain a deeper insight into the nature of the Au species involved in C−C coupling processes, we first considered whether dissociation of complexes 4 into their monometallic components, 59 namely, neutral methyl compounds 1 and triflimide species of type Au(PR 2 Ar′)(NTf 2 ), might be relevant. To check the viability of such equilibria, we explored exchange processes of the methyl bridge in compound 4a. In a first experiment, we examined the exchange between 1a and 4a at variable temperatures. For experimental convenience, we accessed an equimolar mixture of both species by adding 0.33 equiv of [Ph 3 C][B(C 6 F 5 ) 4 ] to 1a at −40°C. Under these conditions, one-third of the neutral methyl compound is  transformed by methyl abstraction into a cationic gold species that is immediately trapped by unreacted 1a to provide gemdigold 4a. Variable temperature 1 H and 31 P{ 1 H} NMR spectroscopy analysis revealed dynamic behavior in solution ( Figure S10), which we attribute to the exchange equilibrium depicted in Scheme 6a. It was possible to identify 4a by a 31 P{ 1 H} NMR resonance at 0.1 ppm recorded at −85°C, whereas a broad signal at 21.1 ppm was assigned to 1a. These signals coalesce at approximately −40°C, while the major component when reaching 25°C is the homoleptic bisphosphine compound 5a that accompanies ethane formation. We further investigated this dynamic behavior by DFT methods (see the Supporting Information for details). Calculations indicate that dissociation of the dinuclear species [Au 2 (μ-CH 3 )(PMe 2 Ar Xyl2 ) 2 ][NTf 2 ] (4a) into the corresponding fragments, Au(CH 3 )(PMe 2 Ar Xyl2 ) (1a) and Au-(PMe 2 Ar Xyl2 )(NTf 2 ), is only slightly endergonic (ΔG = +0.5 kcal/mol), in agreement with our experimental results. The kinetic profile of ethane evolution in these equimolar mixtures is identical, within the experimental error, to that of pure 4a.
This suggests that, even if carbon−carbon coupling takes place from a trimetallic species involving the participation of compounds 1, the required dissociation of gem-digold methyl compounds 4 into compounds 1 and [Au(PR 2 Ar′)] + is not likely kinetically relevant.
Substituting methyl compound 1a by its related ethyl (2a) and phenyl (3a) derivatives showed the formation of crosscoupling products (Scheme 6b,c). In the case of 2a, the formation of propane and butane was apparent by 1 H NMR spectroscopy, while in the reaction between 4a and 3a the formation of ethane, biphenyl, and toluene was detected in comparable amounts. GC-MS analysis of solution and gas headspace provided further evidence for cross coupling, since variable amounts of ethane, propane, and butane were measured from the reaction between 2a and 4a ( Figure  S21). In both cases, the main homogeneous gold-containing species when reaching room temperature is 5a.
To gather more information on the exchange between bridging and terminal hydrocarbyl substituents present in gemdigold and neutral compounds, we examined the reaction depicted in Scheme 6b at variable temperatures ( Figure 8). A solid mixture of 2a and 4a in equimolar amounts was dissolved in CD 2 Cl 2 at −40°C to allow the exchange to take place and then cooled to −85°C. At the latter temperature, the exchange process is halted, and a variety of gold-containing products are identified by 31 P{ 1 H} NMR. These include the neutral hydrocarbyl compounds 1a and 2a and their corresponding gem-digold species 4a and [Au 2 (μ-CH 2 CH 3 )(PMe 2 Ar Xyl2 ) 2 ] (6a), whose broad resonances were recorded at 21.2, 21.9, 0.1, and 1.4 ppm, respectively. Also, sharp signals due to Au(PMe 2 Ar Xyl2 )(NTf 2 ) and [Au(PMe 2 Ar Xyl2 ) 2 ][NTf 2 ] (5a) were identified at −4.2 and 10.8 ppm, respectively. The latter likely results from local solution warm-up during sample handling. Increasing the temperature to −40°C results in coalescence of all prior resonances except for that of 5a, which is clearly not involved in the exchange process. Further raising the temperature to 25°C leads to full consumption of gold precursors and quantitative formation of bisphosphine compound 5a along with the appearance of solid Au(0). Similarly, rapid exchange between 4a and 3a is evinced by immediate conversion of an equimolar mixture of those compounds into 1a and [Au 2 (μ-C 6 H 5 )(PMe 2 Ar Xyl2 ) 2 ] (7a) ( Figure S11).
Having in mind that the above dynamic behavior reveals the presence of compounds 1, 4, and Au(PR 2 Ar′)(NTf 2 ) in solution, and also considering the fact that ethane evolution follows a second-order dependence on bridging methyl complexes 4, we considered three possible routes (Scheme 7). In the first, reductive coupling from two neutral gold methyl compounds of type 1 may take place, similar to prior work by Kochi and co-workers (Scheme 7a). 31,32 However, it is important to highlight two distinctive features of our studies that contrast with those prior reports. First, reductive coupling from Au(CH 3 )(PPh 3 ) only occurred at high temperatures (∼100°C), while C−C bond formation from bridging digold complexes 4 is more facile. In fact, the C−C coupling reaction readily proceeds at temperatures as low as −60°C in the case of the PPh 3 -based system (Figures S8 and S9). Second, whereas a first-order dependence on gold was demonstrated for reductive coupling from Au(CH 3 )(PPh 3 ), 31,32 with phosphine dissociation toward "AuMe" as the rate-determining step, we have determined a second-order dependence on digold compounds 4 during ethane evolution. These observations suggest different operating mechanisms in the two cases, a notion that is further supported by DFT methods based on the PMe 2 Ar Xyl2 system. In agreement with Kochi's findings, the computed reaction free energy for phosphine dissociation at 1a is +32.1 kcal/mol, much higher than the experimentally determined activation free energy for the overall process (ΔG 298 ⧧ = 16.9 ± 2.7 kcal/mol, see above). Phosphine dissociation from 4a to yield [Au 2 (μ-CH 3 )-   Figure S20 (including a molecule of CH 2 Cl 2 in the calculation to compensate the unsaturation of the metal center results in higher reaction free energies: +33.3 and +42.6 kcal/mol for phosphine dissociation from 1a and 4a, respectively)]. As anticipated, these data confirm a dissimilar C−C coupling mechanism for compounds 4 compared to that exhibited by monometallic gold-alkyl species. An alternative route consists of two gem-digold methyl fragments 4 approaching to facilitate the C−C coupling event (Scheme 7b). However, a third pathway to consider, in view of the Coulombic repulsion derived from the approach of two cationic species in route b, is the reaction between 4 and its corresponding neutral methyl species 1 formed by dissociation of a second molecule of 4 into their monometallic fragments (Scheme 7c). First, we explored computationally the direct coupling of methyl groups between two molecules of 4a (route a) as well as between 4a and 1a (route b) by relaxed potential energy scans. These studies indicate that those pathways are unfeasible, both in the singlet and triplet state. We also evaluated the possibility of accessing the hypothetical Au(III) species [Au(CH 3 ) 2 (PMe 2 Ar Xyl2 )][NTf 2 ] from the above routes, since reductive coupling of ethane with such a complex should be accessible. 18,29,30 In fact, we found a feasible barrier (+16.1 kcal/mol) for ethane formation from the hypothetical Au(III) complex [Au(CH 3 ) 2 (PMe 2 Ar Xyl2 )][NTf 2 ] ( Figure  S15). However, [Au(CH 3 ) 2 (PMe 2 Ar Xyl2 )][NTf 2 ] would be formed alongside the digold(0) species [Au 2 (PMe 2 Ar Xyl2 ) 2 ], with these species being 42.1 kcal/mol higher in energy than its precursors, rendering this pathway inaccessible under the reaction conditions ( Figure S15). Similarly, CH 3 + transfer 36 from 4a to 1a presents a computed transition state at +47.0 kcal/mol (TS1 in Figure S16). In addition, we examined reductive coupling from the hypothetical trinuclear species derived from the above CH 3 + transfer; a transition state at +33.6 kcal/mol was found (TS2 in Figure S16), further suggesting that this pathway is unaffordable. Finally, we considered the possible involvement of gold carbene (AuCH 2 ) and hydride (AuH) species, 60 potentially formed by hydride abstraction from Au-methyl complexes. However, the free energy cost to access these high-energy intermediates is calculated to be at least +36.2 kcal/mol ( Figures S17 and S18), incompatible with the determined activation parameters. To further rule out this mechanistic route, we carried out an additional experiment with isotopically labeled [Au(CD 3 )(PMe 2 Ar Xyl2 )] (1a-d 3 ; see the Supporting Information for details). Treating an equimolar mixture of 1a and 1a-d 3 with 2 equiv of Au(PMe 2 Ar Xyl2 )(NTf 2 ) yielded an approximate statistic mixture of CH 3 CH 3 , CH 3 CD 3 , and CD 3 CD 3 ( Figure S12), without further observable H/D scrambling, thus excluding the involvement of gold methylidene species.
Having ruled out the most direct mechanisms involving 1a and 4a, we decided to interrogate the participation of compounds Au(PR 2 Ar′)(NTf 2 ), especially in consideration of the experimental results indicating that such complexes are accessible under the reaction conditions (see above). These compounds serve as a source of electrophilic [Au(PR 2 Ar′)] + fragments 61,62 and, as such, might facilitate or drive phosphine dissociation from other Au complexes. Potential phosphine dissociation is implicated from straightforward ligand exchange experiments (see Figure S13), and since it was proposed as the rate-limiting step in Kochi's earlier system, 31,32 it is conceivable that it could also play a role for C−C coupling from compounds 4. To examine this, we monitored ethane evolution from 4a in the presence of 3 equiv of Au-(PMe 2 Ar Xyl2 )(NTf 2 ), though this excess of gold-triflimide did not have notable effects on the rate of ethane formation. This was, however, not surprising in line with our computational results, where the larger barrier originates after binding of [Au(PMe 2 Ar Xyl2 )] + to 4a. Nonetheless, even if Au-(PMe 2 Ar Xyl2 )(NTf 2 ) is required to facilitate phosphine dissociation, its presence may also affect the observed rate of ethane evolution in an opposite manner by reducing the concentration of Au(CH 3 )(PMe 2 Ar Xyl2 ) (1a) in solution, the latter species also being required for C−C coupling. This is because [Au 2 (μ-CH 3 )(PMe 2 Ar Xyl2 ) 2 ][NTf 2 ] (4a) is in dynamic equilibrium in solution with 1a and Au(PMe 2 Ar Xyl2 )-(NTf 2 ), as discussed above. To circumvent the influence of added Au(PMe 2 Ar Xyl2 )(NTf 2 ) on that equilibrium, we investigated the effect of adding 5 equiv of BPh 3 as an alternative and less disruptive Lewis acid that could facilitate phosphine dissociation. While ethane evolution proceeded at a rate (t 1/2 = 340 s) comparable to that of pure 4a (t 1/2 = 260 s), we did observe a distinctive change in the kinetic profile. More precisely, this experiment revealed a first-order kinetic dependence on 4a ( Figure S7), in contrast to the secondorder profile observed when the consumption of the latter was monitored in pure form.
Next, we directed our efforts to examining, by computational means, the role of Au(PMe 2 Ar Xyl2 )(NTf 2 ) on the pathways and energetics for the formation of ethane (Figure 9). Since we attribute a Lewis acidic role to this fragment, as supported by our experiments with BPh 3 , we first studied BH 3 as a simplified Lewis acid. Thus, we examined the reaction between BH 3 and complex Au(CH 3 )(PMe 2 Ar Xyl2 ) (1a). The formation of a Au− BH 3 adduct is slightly exergonic (ΔG = −0.9 kcal/mol), from which the transition state for the formation of a P−B bond (TS3) lies at +16.2 kcal/mol above the independently computed 1a and BH 3 , giving the product at −7.4 kcal/mol ( Figure S19). Encouraged by this result, we studied the analogous process with cation [Au(PMe 2 Ar Xyl2 )] + instead of BH 3 as the Lewis acid. 63 A transition state for that process (TS4) was found at +29.3 kcal/mol, leading to the formation of a species of formula [(PMe 2 Ar Xyl2 ) 2 AuAu(CH 3 )] + , A in Figure 9, that lies at +18.5 kcal/mol and represents a form of masked "AuMe" stabilized by aurophilic and metal−arene interactions with the [Au(PMe 2 Ar Xyl2 ) 2 ] + fragment. Nonetheless, the large barrier renders this process inaccessible from 4a, in agreement with the experimentally determined second-order dependence on its concentration.
To account for the second-order dependence on 4a, we considered its initial dissociation into 1a and Au(PMe 2 Ar Xyl2 )-(NTf 2 ), the latter providing 1 equiv of cation [Au-(PMe 2 Ar Xyl2 )] + amenable to bind a second molecule of 4a. The resulting trigonal dicationic adduct [Au 3 (μ-CH 3 )-(PMe 2 Ar Xyl2 ) 3 ] 2+ (B) plus 1a are only 1.2 kcal/mol above two molecules of 4a (Figure 9). From trimetallic adduct B, the transition state for the formal transfer of a phosphine ligand between gold atoms was found at +21.3 kcal/mol (TS5), close enough to the experimentally determined value for the overall process of ethane evolution. This transition state gives trinuclear species C at +10.7 kcal/mol, from which dissociation of 5a is thermodynamically accessible (ΔG = +2.5 kcal/mol). This would render the bimetallic intermediate [Au 2 (CH 3 )-(PMe 2 Ar Xyl2 )] + ( Figure S20), which is reminiscent of the proposed highly reactive "AuMe" fragment proposed by Kochi. 31,32 From such a reactive fragment, masked as [Au 2 (CH 3 )(PMe 2 Ar Xyl2 )] + , it is expected that the approach of 1a would result in ethane elimination and formation of colloidal gold, not necessarily in that order.
Our combined experimental/computational approach led us to propose the mechanistic picture for C−C coupling at gemdigold compounds 4 depicted in Scheme 8. Compounds 4 readily dissociate in solution to form 1 and Au(PR 2 Ar′)-(NTf 2 ), the latter functioning as a Lewis acid to favor phosphine migration from a second molecule of 4 by forming a trimetallic intermediate of type B. 64 Following the release of diphosphine compounds 5, the resulting masked "AuMe" fragment reacts with 1a to liberate ethane with concomitant formation of elemental Au, eventually leading to the formation of Au nanoparticles. In this picture, phosphine migration from 4a constitutes the rate-limiting step of the overall process, in analogy to the previously proposed mechanism for reductive coupling from Au(CH 3 )(PPh 3 ). 31,32 In contrast, the remarkable acceleration observed for C−C coupling in compounds 4 compared to 1 seems to be the result of stabilization of key intermediates by the presence of aurophilic interactions combined with the Lewis acidic character of [Au(PR 2 Ar′)] + that enables phosphine migration, thus representing an example of rate acceleration by polymetallic entities compared to monometallic counterparts. 65−67 ■ CONCLUSIONS Au-mediated C−C coupling processes have rapidly emerged as versatile and powerful strategies for organic synthesis. Despite numerous reports on the synthetic applicability of gold catalysts, mechanistic understanding has evolved at a slower pace. Previous studies have placed the Au(I)/Au(III) redox couple at the heart of all these transformations, while mechanistic investigations on C−C coupling processes without the apparent advent of Au(III) species is lacking. Herein, we have demonstrated that gem-digold methyl complexes [Au 2 (μ-CH 3 )(PR 2 Ar′) 2 ][NTf 2 ] (4) promote the homocoupling of the bridging methyl fragments to produce ethane at a remarkably higher rate than from its parent neutral species Au(CH 3 )-(PR 2 Ar′) (1). We have also demonstrated that this approach permits the heterocoupling of the bridging methyl group with ethyl and phenyl fragments. The stability of compounds 4 toward reductive homocoupling is highly dependent on the steric bulk of the phosphine ligand. Whereas the system based on PPh 3 readily liberates ethane at −40°C, those bearing terphenyl phosphines (PMe 2 Ar′) exhibit considerably enhanced stability, which is further increased by the use of the more hindered XPhos and t BuXPhos, the latter being unable to mediate C−C coupling even at 90°C. Our kinetic studies revealed second-order dependence on gem-digold methyl complexes 4 during ethane evolution, whereas a distinctive change toward a first-order dependence on the latter was ascertained in the presence of excess BPh 3 as an external Lewis acid. On the basis of our experimental studies combined with DFT computational methods we have proposed a mechanism that involves rapid dissociation of a molecule of [Au 2 (μ-CH 3 )(PMe 2 Ar′) 2 ][NTf 2 ] (4) toward Au(PMe 2 Ar′)(NTf 2 ) and Au(CH 3 )(PMe 2 Ar′) (1). While Au(PMe 2 Ar′)(NTf 2 ) mediates phosphine migration from a second molecule of 4 via a trimetallic intermediate, compound 1 is proposed to react with the resulting highly reactive and masked "AuMe" fragment to effect the C−C coupling event, most likely by a multinuclear gold species. These studies highlight the relevance of multimetallic mechanisms in mediating uncommon transformations, herein also boosting the rate at which the C−C coupling transformation occurs.

■ EXPERIMENTAL SECTION
General Methods. Unless otherwise noted, all reactions and manipulations were performed under a nitrogen atmosphere in a glovebox or using standard Schlenk techniques with dried and degassed solvents. All solvents were purified via a solvent purification system or by common distillation techniques: Dichloromethane (CH 2 Cl 2 ) was distilled under nitrogen over CaH 2 . Toluene (C 7 H 8 ), benzene (C 6 H 6 ), n-hexane (C 6 H 14 ), and n-pentane (C 5 H 12 ) were distilled under nitrogen over sodium. Tetrahydrofuran (THF) and diethyl ether were distilled under nitrogen over sodium/benzophenone. Benzene (C 6 D 6 ) was dried over sodium, while CDCl 3 and CD 2 Cl 2 were dried over molecular sieves (4 Å) and distilled under nitrogen. Compounds PMe 2 Ar′, 68 AuCl(THT) (THT = tetrahydrothiophene), 69 46 (4c) were prepared according to previously reported procedures. Compounds 1e and 1f were prepared according to the general method described below in yields of around 75%, exhibiting identical spectroscopic data to those previously reported. Au(CH 3 )(XPhos) 78 and Au(CH 3 )( t BuXPhos) 80 were prepared by an alternative method of the published procedures and fully characterized. Methyl(triphenylphosphine)gold(I), chloro-(dimethylsulfide)gold(I), silver bis(trifluoromethanesulfonyl)imide acetonitrile adduct, chlorotriphenylphosphinegold(I), 2-dicyclohexylphosphino-2′,4′,6′-triisopropylbiphenyl (XPhos), and 2-di-tert-butylphosphino-2′,4′,6′-triisopropylbiphenyl ( t BuXPhos) were purchased from STREM Chemicals and were used as received. Other chemicals were purchased from Sigma-Aldrich and used as received. All new compounds have been characterized by 1 H NMR spectroscopy, 31 P NMR spectroscopy, 13 C NMR spectroscopy, and elemental analysis (see Figure 10). Solution NMR spectra were recorded on Varian Inova 600 or 500 MHz or on Bruker AMX-300, DRX-400, DRX-500, and Avance III 800 MHz spectrometers. Spectra were referenced to external SiMe 4 or using the residual proton solvent peaks as internal standards ( 1 H NMR experiments), or the characteristic resonances of the solvent nuclei ( 13 C NMR experiments), while 31 P was referenced to H 3 PO 4 . Spectral assignments were made by routine one-and two-dimensional NMR experiments where appropriate. For elemental analyses, the LECO TruSpec CHN elementary analyzer and PerkinElmer 2400 Series II analyzer were utilized. GC analysis was performed using a Shimadzu GCMSQP2010-Plus instrument equipped with a PoraBOND-Q capillary column (25 m, 0.25 mm i.d., 3.0 μm film thickness, Agilent Technologies). Helium carrier gas was supplied at a head pressure of 10 psi to provide an initial flow rate of 1.4 mL/min. A 1 mL injection with a split ratio of 1:10 was employed. GC temperature was initially held at 40°C for 1 min and gradually increased to 120°C at 5°C/ min. Full-scan mass spectra were collected from 5 to 70 m/z at a data acquisition rate of 3.5 spectra/s. The MS transfer line was held at 250°C , and the ion source temperature was 200°C. Samples analyzed by transmission electron microscopy (TEM) were prepared by dispersing the powders in cyclohexane or hexanes (99.5%, anhydrous, Sigma-Aldrich) and sonicating for 1 min before mounting on Cu-supported holey carbon grids. The Au samples were imaged using an FEI Titan 80-300 operating at 300 kV. The structures of compounds 1a, 1e, 1f, 2a, 3a, 4e, 4f, and Au( t BuPhos)(NTf 2 ) have been authenticated by Xray diffraction studies and their corresponding CIF files deposited in the Cambridge Crystallographic Data Centre with nos. 2024182− 2024189.
General Synthesis of Compounds 1. A suspension of the corresponding gold chloride precursor AuCl(PR 2 Ar′) (0.20 mmol) in toluene (10 mL) was cooled to −78°C, and a solution of MeMgX (X = Cl or Br; 2.5 equiv) in toluene was added dropwise. The mixture was allowed to warm up slowly for 16 h. The volatiles were removed in vacuo, and the residue was extracted with benzene for 1a−1d or pentane for 1e−1f. Evaporation of the organic solvent led to compounds 1a−1f as white powders in around 60−80% yields. Isotopologue 1a-d 3 was synthesized by the same procedure using freshly prepared CD 3 MgI. 81  General Synthesis of Compounds 4. A solid mixture of the corresponding methyl gold precursor 1a−1f (0.0175 mmol) with 1 equiv of its parent compound [Au(PR 2 Ar′)][NTf 2 ](0.0175 mmol) was dissolved in CD 2 Cl 2 (0.6 mL) under nitrogen at −50°C to rapidly yield the desired methyl-bridged complex 4a−4f in a quantitative NMR spectroscopic yield. Characterization of the less stable compounds 4a and 4b was carried out by multinuclear NMR spectroscopy at low temperature without further purification. Compounds 4c−4f were obtained as colorless microcrystalline substances by precipitation with pentane at −20°C (4c, 4d) or 25°C (4e, 4f) in around 90% yields. Alternatively 4a−4f can be prepared in comparable by treating compounds 1a−1f with 1/2 equiv of [Ph 3 C][B(C 6 F 5 ) 4 ] in dichloromethane by an otherwise identical procedure. Spectroscopic and analytical data for selected compounds (others can be found in the SI). Compound 4a. 1  General Procedure to Measure Kinetic Constants. Kinetic studies were carried using an identical procedure to that described for the general synthesis of compounds 4, in J-Young NMR tubes under nitrogen atmosphere, and monitoring the disappearance of the in situ formed gem-digold methyl compounds 4 by 1 H and 31 P{ 1 H} NMR. Each kinetic experiment was run in triplicates, and average data are given.
Synthesis and characterization of new compounds, X-ray diffraction data, information on kinetic studies, variable temperature analysis and exchange experiments, DFT calculations, and MS and NMR spectra (PDF) XYZ coordinates (XYZ)