Gold(I) and Palladium(II) Complexes of 1,3,4-Trisubstituted 1,2,3-Triazol-5-ylidene “Click” Carbenes: Systematic Study of the Electronic and Steric Influence on Catalytic Activity

The synthesis of a small family of six electronically and sterically modified 1,3,4-trisubstituted 1,2,3-triazol-5-ylidene gold(I) chloride complexes is described. Additionally, the corresponding trans-[PdBr2(iPr2-bimy)(1,3,4-trisubstituted 1,2,3-triazol-5-ylidene)] complexes are also generated and used to examine the donor strength of the 1,3,4-trisubstituted 1,2,3-triazol-5-ylidene ligands. All compounds have been characterized by 1H and 13C NMR and IR spectroscopy, high-resolution electrospray mass spectrometry (HR-ESI-MS), and elemental analysis. The molecular structures of four of the gold(I) and four of the palladium(II) complexes were determined using X-ray crystallography. Finally, it is demonstrated that these 1,2,3-triazol-5-ylidene gold(I) chloride complexes (Au(trz)Cl) are able to catalyze the cycloisomerization of 1,6-enynes, in high yield and regioselectivity, as well as the intermolecular direct etherification of allylic alcohols. Exploiting the Au(trz)Cl precatalysts allowed the etherification of allylic alcohols to be carried out under milder conditions, with better yield and regioselectivity than selected commercially available gold(I) catalysts.

Recently, homogeneous gold catalysis has become an extremely popular area of research because the soft, carbophilic Lewis acidic nature of Au(I) ions enables the mild activation of unsaturated C−C bonds. 15 While much of the early work exploited phosphine-containing gold(I) complexes, the now ubiquitous N-heterocyclic carbene ligands (NHCs, A−G; Figure 1) have become increasingly popular for the generation of these types of catalysts. 16 Au(I)−NHC complexes often display enhanced stability and catalytic activity in comparison to the phosphine analogues due to the greater σ-donor strength of the carbene ligands. Because of our interest in "click" coordination chemistry, 17 we recently reported the synthesis of the Au(trz)Cl complex 7a (Scheme 1) and showed that it was catalytically active. 18 Herein we build on this initial result and exploit the modularity of the CuAAC "click" reaction to generate a small family of sterically and electronically tuned Au(trz)Cl catalysts. Additionally, an analogous series of Pd(II) bis-carbene complexes were synthesized to enable the variation of the 1,3,4-trisubstituted 1,2,3-triazol-5-ylidene's σ-donor strength to be directly probed. Finally, the effect of this systematic modification of the Au(trz)Cl complexes is explored in two different gold(I)-catalyzed reactions: (1) the cyclo- Scheme 1 a a Legend: (i) (a) 6a−f, Ag 2 O, Me 4 NCl, CH 2 Cl 2 /CH 3 CN (1/1), room temperature, 6−14 h, (b) Au(SMe 2 )Cl, room temperature, 2 h; (ii) (a) [PdBr 2 (iPr 2 -bimy)] 2 , nBu 4 NBr, CHCl 3 , reflux, 3 h, (b) 6a−f, Ag 2 O, CH 2 Cl 2 , room temperature, 18 h. characterized by elemental analysis, HR-ESI-MS, IR, and 1 H and 13 C NMR spectroscopy. The elemental analyses of the complexes were consistent with the proposed formulations, and this was further supported by HR-ESI-MS. The gold complexes 7a−f display major signals due to [Au(trz)Cl + Na] + , [Au(trz) − Cl − Au(trz)] + and [Au(trz) 2 ] + ions, while the palladium complexes show major peaks consistent with [PdBr(iPr 2bimy)(trz)] + , [PdCl(iPr 2 -bimy)(trz)] + , and [PdBr 2 (iPr 2 -bimy)-(trz) + Na] + (Supporting Information). The 1 H NMR spectra of 7a−f and 8a−f were also consistent with complex formation. The 1 H NMR spectra of the triazolium salts 6a−f contain a C trz −H proton signal between 7.5 and 9.0 ppm which is absent in the spectra of the metal complexes (7a−f and 8a−f), indicative of deprotonation and carbene formation. 7, 18,21 Additionally, the signals due to the N-methyl protons of the triazole units experience an upfield shift on complex formation. In the complexes with phenyl substituents the o-phenyl proton signals undergo a downfield shift, due to the proximity of the deshielding metal centers. The 13 C NMR spectra of the complexes display 1,3,4-trisubstituted 1,2,3-triazol-5-ylidene carbon signals at ∼160 ppm, consistent with previous studies. 7 ,18,21 The molecular structures of four of the gold complexes (7b− d,f; Figure 2, Table 1, and the Supporting Information) were determined using X-ray crystallography. 22 The crystals were grown via vapor diffusion of either diethyl ether or petroleum ether into a dichloromethane solution of one of the complexes. The Au(I) ions of the complexes are bound to a 1,3,4trisubstituted 1,2,3-triazol-5-ylidene and a chloride ligand in the expected linear fashion (C−Au−Cl bond angles range from 176.3 to 178.8°). The Au−C trz (1.972−2.001 Å) and  bond lengths are similar to those observed in other gold(I) triazolylidene complexes. 7a,e,18,22,23 The extended solid-state structure of complex 7b contains dimers (Supporting Information) that are held together by weak hydrogen-bonding interactions between the methoxy oxygen atom and the aryl proton situated ortho to the methoxy group of the adjacent molecule (C−H···O = 2.690(3) Å, C···O = 3.592(6) Å). Additionally, π−π interactions are observed between the triazole ring and the six-membered methoxyphenyl ring (centroid···centroid = 3.748 Å).
Surprisingly, no aurophilic interactions are observed in any of the structures (7a 18 or 7b−d,f); the shortest gold···gold distance (7d, Au−Au = 3.612 Å) 26 is greater than the sum of the van der Waals radii of two Au(I) centers (3.60 Å).
Ligand Donor Properties. The mild CuAAC "click" methodology used to generate the 1,3,4-trisubstituted 1,2,3triazol-5-ylidene ligands potentially provides a facile way to tune both steric and electronic properties of the resulting carbene complexes. The M−C trz bond lengths (Tables 1 and 2) were examined to see if there is a correlation between the electronic nature of the trz ligand and the metal−carbene bond length in the solid-state upon side-arm (wingtip) substitution of the compounds 7a−f and 8a−f. The Au−C trz bond lengths of the gold(I) complexes 7b (Aryl-OMe) < 7a (Aryl-H) < 7c (Aryl-NO 2 ) follow the expected trend with the more electron rich methoxy-substituted complex 7b displaying a shorter Au− C trz bond than the parent complex 7a. Similarly, the electronpoor nitro-substituted complex 7c has a longer Au−C trz bond than the parent complex 7a. However, the observed differences are of a similar magnitude to the experimental uncertainty ( Table 1). The correlation breaks down with complexes 7d,f, where the observed Au−C trz bond lengths are longer (7d) and shorter (7f) than would be predicted on the basis of inductive arguments. Furthermore, the Pd−C trz bond lengths for complexes 8a−d are all essentially identical within the experimental uncertainty (Table 2). Therefore, there is no obvious correlation between the observed M−C trz bond lengths in the solid-state and the electronic nature of the trz ligand. However, it is noted that the M−C distance can be affected by other parameters such as crystal-packing effects.
As the solid-state data provided no useful information on the donor strengths of the various trz ligands, the 13 C NMR spectra of palladium complexes 8a−f were used to provide insight into   (2) 177 the ligand's donor properties. Huynh and co-workers previously showed that these benzimidazol-2-ylidene−dibromopalladium-(II) complexes can be used to probe the σ-donor strength of the ligands trans to the benzimidazol-2-ylidene. 21,28 They have found that there is a direct relationship between σ-donor strength of the trans ligand and the chemical shift of the benzimidazole carbene carbon in the 13 C NMR spectra of the dibromopalladium(II) complexes. 21,28 Additionally, this system has previously been used to show that the mesoionic trz ligands are stronger donors than imidazol-2-ylidenes. 21 The 13 C NMR spectra of palladium complexes 8a−f were obtained in CDCl 3 solution at 298 K. Consistent with what Huynh and co-workers previously reported, 21 the benzimidazol-2-ylidene reporter peaks were observed at approximately 180 ppm ( Figure 4).
The parent palladium(II) complex 8a displays the peak for the benzimidazolylidene carbon at 180.26 ppm. Consistent with expectations, the benzimidazolylidene carbon signal of the more electron-rich methoxy-substituted complex 8b has shifted downfield (δ 180.44 ppm), relative to the ligand in 8a, suggesting that the 4-MeOC 6 H 5 -trz ligand is more electron donating that the parent Ph-trz ligand. Similarly, the reporter carbon of the nitro-substituted complex 8c is observed upfield (δ 178.95 ppm) relative to 8a, indicating that the presence of the electron-withdrawing functionality reduces the trz ligand's donor properties, consistent with expectation. Replacing the benzyl substituent of the parent with a phenyl ring generating the diphenyl-substituted complex 8e also leads to a reduction of the trz ligand σ-donor strength (δ 179.87 ppm), as is expected upon the removal of the electron-donating methylene linker.
The observed positioning of the benzimidazolylidene reporter carbon signals in the dibenzyl-substituted complex 8d (δ 179.70 ppm) and the dimesityl-substituted complex 8f (δ 178.99 ppm) was unexpected. The data suggest that these ligands are weaker σ donors than would be expected on the basis of electronic arguments. Changing the phenyl substituent of the parent complex (8a) to a benzyl in 8d would be expected to lead to an increase in the electron-donating properties of the trz-d ligand due to the presence of a second inductively donating methylene group. Likewise, the presence of the three methyl groups on the mesityl substituents of complex 8f should make this trz-f ligand more electron-donating than the structurally similar diphenyl-trz-e. The observed 13 C shifts of 8a,e suggest that the trz ligands in complexes 8d,f are weaker σ donors than trz-a and trz-e. As the substituents on the trz ligand of 8d,f are larger (bulkier) than those on the other examples, it is postulated that steric effects lead to this observed weakening of the σ-donor properties. 1 H NMR spectroscopic and X-ray crystallographic data provide some support for this theory. The 1 H NMR spectra of all palladium complexes show characteristic septet signals representing the two tertiary proton signals of the benzimidazolylidene isopropyl groups. In most cases we see no separation of these signals, indicative of freely rotating ligands in solution. The exception is 8d, which displays two distinct signals for the isopropyl groups indicative of hindered rotation about the Pd−trz bond, presumably due to steric factors. In addition to this, the solid-state structures of 8a−c have a coplanar arrangement of the heterocyclic ligands, whereas in 8d the aforementioned ligands twist out of this plane (C2−C1− C21−N4 = 40.0(2)°). Although the solid-state structure of 8f was not obtained, molecular models (Supporting Information) show the presence of steric clashes that could weaken interaction of trz-f with the Pd(II) ion. While not completely as expected, these results indicate that electronic alteration of the side-arm substituents (wingtip groups) does affect the donor properties of the trz ligands and suggests that CuAAC "click" chemistry could be exploited to modulate these properties in a facile fashion.
Catalysis with Gold(I) 1,3,4-Trisubstituted 1,2,3-Triazol-5-ylidene Complexes. With the family of new Au(trz)Cl complexes 7a−f in hand, we were keen to investigate their application in catalysis. In particular, we wished to observe what effect, if any, changing the substituents on the triazolylidene ligand would have on catalysis. Thus, the enyne 9 was subjected to skeletal rearrangement 30 catalyzed by various Au(trz)Cl precatalystsa typical test reaction for catalytic activity of new gold complexes 31 (Table 3). Our initial few results were rather disappointing, as they showed poor selectivities, poor yields, and (within error) fairly similar results (entries 1−4). However, suspecting that 11 may form from 10 over time, the reactions were repeated with a much shorter reaction time (1 min vs 15 min before), and to our delight, the selectivities and yields improved significantly (entries 5−10). Electronic tuning seems to do little to the catalytic activity: the parent Bn,Ph-substituted Au(trz)Cl 7a reacts with almost the same excellent selectivity and yields (entry 5) as the electronrich (7b, entry 6) and electron-poor (7c, entry 7) versions. Next, the effect of sterics around the trz was probed. Changing from the parent Bn,Ph-substituted trz 7a to the more flexible dibenzyl-substituted 7d (entry 8) causes a drop in selectivity (13:1 vs >20:1) but not yield (93% vs 92%). Having diphenyl substitution (7e, entry 9) retains the excellent >20:1 selectivity but causes a drop in yield (72%). Finally, the more hindered dimesityl-substituted 7f provides the best result in this series, with an excellent 98% yield and >20:1 selectivity of 10:11. Therefore, it seems that for the skeletal rearrangement 9 → 10, steric tuning on the trz ligand has more influence than electronic tuning. Increased steric protection around the Au center provided by the Mes substituents in 7f appears to be beneficial for the performance of the catalyst in this test reaction. As a control, the reaction in entry 5 was also repeated with the AgCl filtered out of the mixture of Au(trz)Cl 7a and AgSbF 6 , prior to introduction of 9 in order to ensure that the silver is not playing a crucial role in the reaction. 32 The reaction behaves in exactly the same manner regardless of the presence or absence of AgCl in the reaction, confirming that silver is not playing a significant role in this reaction. A mercury drop test 33 was also carried out on the reaction shown in entry 10, resulting in full conversion, suggesting that the catalytic activity is not due to the formation of heterogeneous nanoparticles. Finally, reducing the catalyst loading to 2 mol % still produces an excellent 93% yield within 1 min (entry 11) and shows that it compares favorably with results from commonly used gold catalysts (entries 12 and 13).
Next, we were keen to demonstrate the utility of the Au(trz)Cl complexes as precatalysts in a reaction developed within one of our laboratories. We have previously shown that direct allylic etherification using unactivated allylic alcohols and alcohol nucleophiles is possible using gold catalysis (e.g., Scheme 2). 34 The original method requires excess (5 equiv) of the alcohol nucleophile (e.g., 13) for best results, using Au(PPh 3 )NTf 2 as the catalyst. 15e,35 An excess of 13 is to ensure that the allylic alcohol 12 does not react with itself and also to improve selectivity under these conditions. To our delight and surprise, using the new Au(trz)Cl complexes 7a−f as precatalysts allows not only for a significant reduction in the amount of alcohol nucleophile to 1.1 equiv but also for the reaction to be carried out at a much milder room temperature (vs. 50°C, Table 4), thus greatly improving on the original conditions shown in Scheme 2.
As shown in Table 4, all the Au(trz)Cl precatalysts 7a−f screened (entries 1−6) provide good yields of the desired product 14 in excellent regioselectivities (>20:1 of 14:15 vs 12:1 using the original conditions in Scheme 2) using only 1.1 equiv of the alcohol nucleophile 13 and a mild 25°C. Unlike the enyne skeletal rearrangement reaction shown in Table 3, the allylic etherification reaction is sensitive to electronic tuning on the trz ligand (entries 1−3). Changing from the parent precatalyst 7a (entry 1) to the more electron rich 7b (entry 2) gives a slightly improved yield and E:Z selectivity, while the more electron-withdrawing 7c shows a noticeably lower yield of 14 (entry 3). Tuning the sterics around the trz ligand (entries 4−6) does not really seem to affect the yield of 14 or the regioselectivity. Next, we were keen to see how Au(trz)Cl precatalysts 7a−f compare with other commonly used gold(I) catalysts (entries 7−10). Using the original catalyst Au(PPh 3 )-NTf 2 36 under these conditions results in incomplete conversions and poor selectivities, including at least 9% of selfreaction of 12 (entry 7). The commercially available NHC precatalysts Au(IPr)Cl and Au(IMes)Cl were also investigated for comparison purposes (entries 8 and 9). Au(IPr)SbF 6 , like Au(PPh 3 )NTf 2 , results in poor selectivities and conversions (entry 8). Au(IMes)SbF 6 , on the other hand, provides a good yield of 14 although the E:Z selectivity is poorer than with Au(trz)SbF 6 (entry 9). Finally, the phosphine counterpart Au(PPh 3 )SbF 6 provides poorer yields as well as selectivities (entry 10) than the Au(trz)SbF 6 catalysts.
The gold(I)-catalyzed reactions were carried out without the need for dry solvents or inert atmosphere. However, AgSbF 6 was stored and weighed out in a glovebox as a precaution to avoid hydrolysis to the corresponding Brønsted acid in our screening experiments.

Organometallics
Special Conditions/Variations. 7b: the benzyl group (C9−C15) appeared to be rotationally disordered about the C11/C14 phenyl ring axis and was modeled over three positions, refining to 0.5:0.25:0.25 occupancies. The disordered fragment was refined with the use of SAME, FLAT, EADP, and ISOR restraints/constraints. The largest residual electron density peaks are ca. 1 Å from the gold atom.
8d: the whole dibenzylmethyltriazolylidene ligand (C1−C17, N1− N3) appeared to be rotationally disordered about the Pd1−C1 bond (by ca. 180°for the triazolyl ring). The triazolylidene ligand and both bromide ligands were modeled over two positions, refining to 0.75:0.25 occupancies. The disordered triazolylidene ligand was refined with the use of SAME restraints across the two orientations, and a FLAT restraint and EADP constraints were applied to each ring. Author Contributions §