Correction: Electrostatic control of regioselectivity via ion pairing in a Au(i)-catalyzed rearrangement

Correction for ‘Electrostatic control of regioselectivity via ion pairing in a Au(i)-catalyzed rearrangement’ by Vivian M. Lau et al., Chem. Sci., 2014, 5, 4975–4979.


Introduction
Ion pairing has been widely exploited to control the rate and selectivity of reactions that involve charged species. 1- 5 In the strategies that have been developed to date, counterions have been used to promote phase transfer, create new steric and chemical environments, bind weakly to reactive centers, participate directly in chemical reactions, or a combination of the above. Because of its proximity to a reactive species in an ion pair, a counterion could in principle affect the selectivity of a reaction through electrostatic interactions that differentiate competing transition states. Although the major electrostatic interaction is the charge-charge attraction that holds an ion pair together, transition states that have significantly different charge distributions could be (de)stabilized to different extents by the local electric field generated by a counterion. [6][7][8][9] Here we show that ion pairing changes the regioselectivity of a Au(I)catalyzed aryl alkynyl sulfoxide rearrangement by favoring the product resulting from a more polar transition state through electrostatic interactions.

Results and discussion
Previous studies have shown that Au(I) complexes catalyze a rearrangement of aryl alkynyl sulfoxides to dihydrobenzothiepinones, a transformation that replaces an aryl C-H bond with a C-C bond. [10][11][12] To study regioselectivity for this reaction, we prepared 3-Cl aryl alkynyl sulfoxide 1a, for which C-H functionalization can occur at either the 2-or 6position. We first assessed the effects of ligand structure on selectivity using a series of common phosphine (R 3 P) and N-heterocyclic carbene (NHC) ligands (Table 1). 1a was reacted with 2 mol% R 3 PAuCl or NHCAuCl precatalyst and 2 mol% NaBAr F 4 (Ar F = 3,5-(CF 3 ) 2 C 6 H 3 ) in CH 2 Cl 2 at room temperature for 17 h. NaBAr F 4 serves as a Clabstractor to generate an active cationic R 3 PAu(I) or NHCAu(I) catalyst. These conditions gave clean conversion to the two expected regioisomeric products 2a and 3a in moderate to good yields and recovery of the starting material. For NHCAu(I) catalysts, increasing the steric demand of the NHC (IMes < IPr < SIPr) 13 increased the 3a:2a ratio from 0.7:1.0 to 1.9:1.0, favoring functionalization at the more accessible C-H bond (Table 1, entries 1-3). For the R 3 PAu(I) catalysts with R 3 P = Ph 3 P, (otol) 3 P, SPhos, or XPhos, the selectivity was completely insensitive to ligand structure. The ratio of 3a:2a was 0.6:1.0 with all of these catalysts despite their substantially different steric and electronic properties (Table 1, entries 4-7). 13 Overall, the ligand screen highlights the difficulty of controlling regioselectivity for this aryl C-H functionalization by changing the ligand structure.
To test whether ion pairing could affect selectivity, we performed the reaction with different counterions in solvents that span a range of dielectric constants (ε). Nitrile complexes [NHCAu(NCR)]X and [R 3 PAu(NCR)]X (X = anion) were used as pre-catalysts in these experiments to obviate Clabstraction. Spontaneous dissociation of NCR generates the catalytically active cationic Au(I) complex. The reactions were performed with 2.5 mM 1a and 2 mol% catalyst loading at room temperature. The reactions were stopped after 4 h to determine the product ratio using 1 H NMR. Unoptimized yields varied from 22% to 88% depending on the ligand (Table S2). Recovered substrate 1a accounted for essentially all of the remaining material.
The dielectric constant of the solvent affected the regioselectivity obtained with each of the Au(I)-catalysts in a counterion-dependent manner. The results for NHCAu(I) complexes are shown in Figure 1a. With [IPrAu(NCPh)]BAr F 4 , the solvent had little effect on selectivity. The 3a:2a ratio ranged from 0.8:1.0 to 1.2:1.0 across seven solvents with ε ranging from 2.4 (toluene) to 20.7 (acetone). With [IPrAu(NCMe)]SbF 6 , however, a significant solvent dependence was observed. For solvents with ε ≥ 8.9 (CH 2 Cl 2 , (CH 2 Cl) 2 , acetone), the 3a:2a ratio was similar to the ratio with the BAr F 4 complex. For solvents with ε ≤ 6.0, the ratio increased monotonically as ε decreased, reaching 2.7:1.0 in toluene.
The same trend was observed with [IPrAu(NCMe)]BF 4 , but the increase in the ratio for ε ≤ 6.0 was attenuated. Thus, the counterion determined the dependence of the selectivity on ε, with the magnitude given by the order Magnitude of the dielectric response is dependent on the ligand and identity of counterion X -.
was decreased below 8 ( Figure 1b). The changes were significantly larger with SbF 6 than with PF 6 or BF 4 and they varied with the phosphine structure in the order XPhos ≈ (otol) 3 P > Ph 3 P (Table S2). With XPhos and (o-tol) 3 P ligands and SbF 6 counterion, the 3a:2a ratio increased by a factor of 4.3-5.0 in going from ε = 8.9 (CH 2 Cl 2 ) to ε = 2.4 (toluene). Only the solvent's ε impacted selectivity and not its molecular properties (dipole moment, coordinating ability, etc.) The product ratios obtained in solvent mixtures matched the expected value for their calculated ε.
The dependence of the product ratio on ε and the choice of counterion suggests that the selectivity-determining step proceeds from an ion-paired intermediate in low-ε solvents. Previous diffusion NMR studies of many organometallic complexes have shown that ion pairing is strongly favored in CDCl 3 and less polar solvents. 14,15 Weaker, but still substantial, ion pairing has been observed for cationic R 3 PAu(I) and NHCAu(I) complexes paired with BF 4 in CD 2 Cl 2 . 16 yielded an r H value for the cation that increased from 6.6 Å in CD 2 Cl 2 to 7.2 Å in CDCl 3 , and an r H for the anion that increased from 6.6 Å to 7.3 Å (  14,15,17,18 indicates that the equilibria strongly favor the ion paired forms for the NHCAu(I) and R 3 PAu(I) complexes in CDCl 3 and all less polar solvents. The monotonic increase of the 3a:2a ratio as ε is decreased below 8 for all complexes therefore is not likely the result of a significant increase in the extent of ion pairing but instead reflects an increased strength of the effect of the counterion on the product-determining transition states (see below).
To gain further insight into the origin of the ion pairing effect, we explored its dependence on the aryl substituent. Additional 3-substituted aryl alkynyl sulfoxides 1b -1f were reacted with [(o-tol) 3 PAu(NCMe)]SbF 6 to yield regioisomeric products 2b -2f and 3b -3f. Figure 2 shows the product ratios, 3x:2x, in CH 2 Cl 2 , CHCl 3 , and toluene. The magnitude of the change in product ratio from high ε (CH 2 Cl 2 ) to low ε (toluene) exhibited a strong dependence on the substituent. Mesubstituted 1b showed essentially no response to the change in ε. For the rest of the substrates, the 3x:2x ratio increased from high e to low e, with the magnitude of the change given by the order OMe<Br<F<Cl<CF 3 (Table 3). For CF 3 -substituted 1f, the ratio increased by a factor of 6.3. In all of these cases, a smaller increase was obtained in CHCl 3 compared to toluene, consistent with the trend evident in Figure 1.
The absence of a correlation between the size of the substituent and the magnitude of the selectivity change from high-ε to low-ε solvent indicates that ion pairing does not principally affect selectivity via steric interaction. The results in Table 2. Solvent dependence of the diffusion coefficient D (10 -10 m 2 /s) and hydrodynamic radius r H (Å) of representative Au(I) complexes.  To probe electrostatic differences between the reactions with different substrates, DFT calculations were performed for putative product-determining transition states for each substrate. Recent experimental and computational studies provide strong evidence that C-C bond formation occurs through an irreversible [3,3]-sigmatropic rearrangement of a cationic vinyl Au intermediate I (Table 3). 12,20,21 We calculated the [3,3] transition state leading to each regioisomer for the substrates in Figure 2 with a model Me 3 P-Au(I) catalyst. We then calculated a dipole moment (ρ) for each transition state using the center of nuclear charge as the origin. Figure 3 shows the charge density maps for the regioisomeric transition states for substrates 1b and 1f. The difference in the magnitude of the dipole moments (∆ρ) indicates the extent to which the transition states differ in their charge distributions. The calculations revealed a strong correlation between ∆ρ and the magnitude of the change in the product ratio upon switching from CH 2 Cl 2 to toluene: ∆ρ ≈ 0 with CH 3 -substituted sulfoxide and increased in the order OMe < F < Cl < Br < CF 3 ( Table 3).
Ion pairing favored the isomer (3a, 3c-3f) that is formed from the more polar product-determining transition state. This result indicates that the paired anion electrostatically stabilizes the transition state leading to the major product to a greater extent than it stabilizes the competing transition state. The strength of the electrostatic interactions that energetically differentiate the two transition states depends on the ε of the medium surrounding the ion pair, which explains why selectivity continues to rise as ε is decreased below 5 even though it is unlikely that the extent of ion pairing changes appreciably in this regime. The ion pairing effect is in contrast to the absence of a response to solvent polarity when ε ≥ 8. Increasing ε does not significantly favor the pathway proceeding through the more polar transition state whereas ion pairing in a low-ε medium does.
In addition to the substrate, the strength of the ion pairing effect depends on the structure of both the counterion and the ligand. This dependence most likely reflects changes to the placement(s) of the counterion in the ion pair. Maximum electrostatic differentiation of transition states requires placing the counterion as close as possible to the complex and in a position where it can afford the greatest stabilization to the more polar transition state. No effect is seen when pairing with BAr F 4 because its large radius places negative charge too far away. The difference between SbF 6 and PF 6 or BF 4 suggests that SbF 6 is better positioned in the ion pair. The relatively small ion pairing effect for Br-substituted 1d given the large ∆ρ for this substrate may also reflect poor counterion placement. Additional nuclear Overhauser effect NMR studies 17,22,23 and molecular dynamics simulations will be necessary to shed light on these important structural details. Tuning the ligand and counterion structure to adjust ion placement may substantially increase the selectivity afforded by this approach.

Conclusions
In summary, we have demonstrated that ion pairing can control selectivity by preferentially stabilizing more polar  transition states. This strategy may be applicable to diverse synthetic challenges because many reactions involve competing pathways with significantly different charge distributions.