Balancing Bulkiness in Gold(I) Phosphino‐triazole Catalysis

The syntheses of a series of 1‐phenyl‐5‐phosphino 1,2,3‐triazoles are disclosed, within which, the phosphorus atom (at the 5‐position of a triazole) is appended by one, two or three triazole motifs, and the valency of the phosphorus(III) atom is completed by two, one or zero ancillary (phenyl or cyclohexyl) groups respectively. This series of phosphines was compared with tricyclohexylphosphine and triphenylphosphine to study the effect of increasing the number of triazoles appended to the central phosphorus atom from zero to three triazoles. Gold(I) chloride complexes of the synthesised ligands were prepared and analysed by techniques including single‐crystal X‐ray diffraction structure determination. Gold(I) complexes were also prepared from 1‐(2,6‐dimethoxy)‐phenyl‐5‐dicyclohexyl‐phosphino 1,2,3‐triazole and 1‐(2,6‐dimethoxy)‐phenyl‐5‐diphenyl‐phosphino 1,2,3‐triazole ligands. The crystal structures thus obtained were examined using the SambVca (2.0) web tool and percentage buried volumes determined. The effectiveness of these gold(I) chloride complexes to serve as precatalysts for alkyne hydration were assessed. Furthermore, the regioselectivity of hydration of but‐1‐yne‐1,4‐diyldibenzene was probed.


Introduction
Bulky phosphines offer significant and well-documented advantages as ligands in metal-catalysed reactions. The landmark contributions of Buchwald and co-workers have led to a deeper understanding of the synthetic chemistry enabled by such bulky ligands, and ligands such as S-Phos and X-Phos are an often-required component of the toolbox of today's synthetic chemists. [1] There is emerging interest in the importance of categorising and evaluating steric and electronic parameters [2] of these types of bulky phosphines [3] in order to correlate and ultimately predict their suitability for use in metal-mediated catalysis. [4] The Tolman cone angle has been used to describe the bulkiness of ligands, [5] this descriptor has been complemented by Nolan describing bulkiness in terms of a percentage buried volume (%V bur ). [6] The %V bur described by a ligand can be calculated by using Cavallo and co-workers' web tool SambVca (2.0). [7] Some authors of this report previously detailed the synthesis, and application to palladium-catalysed Suzuki-Miyaura crosscoupling reactions, of 1,2,3-triazole-containing phosphines including analogues of the aforementioned Buchwald-type ligands, such as 1a (Figure 1). [8] Analysis of the steric parameters (buried volume) [6,9] using the SambVca (2.0) web tool, [7] confirmed that the bulkier ligands of those tested were the most effective in said catalysis. [8] pared and analysed by techniques including single-crystal X-ray diffraction structure determination. Gold(I) complexes were also prepared from 1-(2,6-dimethoxy)-phenyl-5-dicyclohexyl-phosphino 1,2,3-triazole and 1-(2,6-dimethoxy)-phenyl-5-diphenylphosphino 1,2,3-triazole ligands. The crystal structures thus obtained were examined using the SambVca (2.0) web tool and percentage buried volumes determined. The effectiveness of these gold(I) chloride complexes to serve as precatalysts for alkyne hydration were assessed. Furthermore, the regioselectivity of hydration of but-1-yne-1,4-diyldibenzene was probed. Figure 1. Selected examples of a previously reported 1-aryl-5-phosphino triazoles (left) (1a-b) and single-crystal X-ray diffraction structure of 1a (right), as reported elsewhere, H-atoms omitted for clarity. [8] The 1-aryl-5-phosphino-triazole ligands, such as 1a, present their 1-aryl fragment in the same orientation as the phosphorus lone pair (of the free ligand) or in the direction of the metal (of a phosphorus-metal complex thereof ), thus significantly impacting the determined buried volume. [8] Noting that orthoaryl tris-phenylene phosphines (first reported, and somewhat overlooked, in 1940) [10] can impart favourable properties as ligands in catalysis, [11] it struck us that a bulky, and thus possibly superior, version of the triazole phosphine ligand is possible if the phosphorus atom were to be flanked by up to three, rather than one, triazole. A series of ligands ranging from zero to three triazoles, for comparison of the manifested bulk about the phosphorus centre by 1,5-disubstituted-triazole motif(s), was proposed. As such, a series of phosphines comprising of triphenylphosphine (2a) and tricyclohexylphosphine (2b) along with mono-triazole-appended phosphines 3a and 3b (available from a previous study), and the previously unprepared bis-triazoles 4a and 4b and 1-phenyl-5-phosphino-tris-triazole, 5 (Figure 2), was conceived. Figure 2. The series of phosphines conceived to probe the influence the number of 1-phenyl-5-phosphino triazole fragments about the phosphorus centre upon their "bulkiness" as ligands.

Results and Discussion
The substituents attached to phosphorus and the triazole nitrogen, could in principle be varied significantly. To provide a benchmark in this initial investigation cyclohexyl (a) and phenyl (b) groups attached to phosphorus were selected and the substituent attached at the 1-N-position of the triazole was restricted to phenyl in the first instance ( Figure 2). As such two series of aryl-and alkyl-substituted phosphines were planned.
The same protocol for selective deprotonation of 6 at the 5position, [12] was followed by addition of cyclohexyl-or phenylphosphorus dichloride leading to the formation and subsequent isolation of 4a and 4b in 71% and 54% yield respectively [Scheme 1 (iii)]. When a third of an equivalent of phosphorus trichloride was added to deprotonated 6, the expected tris-1phenyl 5-phosphino triazole 5, resulted and was subsequently isolated in 77% yield [Scheme 1 (iv)]. The proton and carbon NMR spectrums of 5 (in [D]chloroform at ambient temperature) displayed a single set of well-defined resonances, corresponding to the three equivalent triazole arms of the 5, suggesting a high degree of symmetry in solution on the NMR timescale.
Both 4a and 4b provided single crystals suitable for structural analysis by XRD ( Figure 3a and Figure 3b, respectively, and supplementary material). In both cases the 1-phenyl substituents of the triazole components point broadly in the same general direction as the phosphorus lone pair, thus boding well for a systematic study of the effect of modulating the steric crowding or bulkiness about a coordinated metal in a catalytically relevant complex. For compound 5, a single crystal XRD structure was obtained (Figure 4), the solid-state conformation contains two unique molecules in the unit cell, both of which show the aryl arms of the triazole are directed along the same orientation as the phosphorus lone pair, creating a bowl-shaped ligand.

Gold(I) Chloride Complex Synthesis and Structural Analysis
Complemented by commercially sourced tricyclohexylphosphine 2a and triphenylphosphine 2b, two ligand sets corresponding to Figure 2 (R = Cy or Ph) were therefore available for comparison in the complexation of a chosen metal. Owing to the importance of gold(I) phosphines in catalysis, [14] and that X-ray crystal structures of gold(I) chloride complexes 2a and 2b have been previously reported by others, [15] the gold(I) chloride complexes of the ligands in the series presented in Figure 2 were compared. Ligation of the triazole-phosphines (3-5) to gold(I), through phosphorus, was achieved by performing dimethyl sulfide-phosphine exchange reactions on chloro(dimethyl sulfide)gold(I) (Scheme 2). Isolated yields of the corresponding gold(I) chloride complexes ranged from 58 to 91%. Triphenylphosphine and tricyclohexylphosphine gold(I) chloride complexes (7a and 7b) were purchased from commercial suppliers.
The single-crystal X-ray structures of gold(I) chloride complexes 7a and 7b [gold(I) chloride complexes of tricyclohexylphosphine and triphenylphosphine] have been previously reported in the literature. [15] The deposited PDB files were used to render images [Ortep III for Windows and PovRay, Figure 5a  10 ( Figure 8) were obtained from the complexes synthesised herein, and similarly analysed to determine the corresponding buried volumes described by the ligands in their complexes with gold(I) chloride (presented in Figure 9). The crystal structures of the gold-phosphorus bond lengths used in the buried volume determinations were those obtained crystallographically.
Since previously prepared mono-triazole-containing phosphines 1a and 1b were readily available to this programme of study, and as 1a was shown previously to have been among the superior ligands of the mono-triazole set in earlier palladiummediated Suzuki-Miyaura catalysis, the synthesis of gold(I) chloride complexes thereof was also attempted. The gold(I) chloride complex of 1a (11) was prepared by the aforementioned dimethyl sulfide ligand exchange reaction, in good yield (95%, 0.25 mmol scale). A single crystal XRD structure of 11 was obtained (Figure 10a), and closely matched (by visual inspection) the structure of 9a, with a 46.0%V bur . Attempts to prepare a 1:1 complex of 1b and gold(I) chloride under the same conditions were inconclusive. However, a trigonal 2:1 ligand/gold(I) chloride complex 12 was identified by singlecrystal X-ray diffraction crystal structure determination (Figure 10b) from a mixture of otherwise unidentified products. A 76.6%V bur (Figure 10b) was determined from the obtained crys- tal structure. Attempts to prepare 12 by employing two equivalents of ligand 1b gave inconclusive results. It is noteworthy that this 2:1 trigonal gold(I) structure is similar to crystal structures reported for [(Ph 3 P) 2 Au (I) Cl] [16] and [(Ph 3 P) 2 Au (I) (SCN)]. [17] Furthermore a linear cationic gold(I) chloride complex [(Ph 3 P) 2 Au + ][Cl -] bearing two triphenyl phosphine ligands with a fully dissociated chloride counterion also been reported. [15b] Therefore, the crystal structure of 12 is included for completeness. Attempts to prepare two and three 2,6-dimethoxyl phenyl triazole-containing variants of 4(a or b) and 5 failed to deliver any desired products. [18]

Catalysis
With the set of gold chloride complexes of Figure 2 (7a-b, 8ab, 9a-b and 10) along with complex 11 in hand, their effectiveness as precatalysts for gold-catalysed alkyne hydration was probed (Table 1). [19] Catalytically active cationic gold(I) species may be generated by silver-mediated halide abstraction, and choice of appropriate counter-anion has been shown to modulate catalytic effects. [20] In this case, triflate was selected as a counter-anion across all cationic gold(I) catalytic systems stud-ied. Initially the gold-catalysed hydration of dec-1-yne (13a) was performed by in situ preactivation of 0.5 mol% of the corresponding gold complex through silver triflate-mediated halide abstraction, using an excess of silver triflate in methanol (or as explained below, dichloromethane in the case of 10), alkyne 13a was stirred in methanol to which the preactivated catalyst solution was added. Since complex 10 was not well solubilised in methanol a protocol of activation in dichloromethane and dilution in methanol was deployed. To facilitate comparison in the subsequent reactions with lower catalyst loadings (0.25 and 0.05 mol% of gold complex), the dichloromethane activation and methanol dilution protocol was used in those cases. Following the addition of water (2 equiv.) all reactions were heated at 80°C in a sealed tube for two hours and conversions to ketone 14a were determined by gas chromatography. At the highest catalyst loading of 0.5 mol% only the bulkiest complex (10) failed to give complete conversion to product 14a (Table 1, entry 10 vs. other entries in same table ). When lower catalyst loadings were probed 0.05 mol% of gold(I) complexes proved to be too low to achieve satisfactory conversion within two hours (the maximum conversion observed across the set was 10% under these conditions). Good gradation across the test series was seen at a 0.25 mol% catalyst loading (Table 1) and confirmed the superior ligands, for this gold-catalysed transformation, to be those with one triazole substituent. The dicyclohexylphosphine-containing complex 8a afforded quantitative conversion to 14a (Table 1, entry 2). The corresponding mono-triazole-diphenylphosphine-containing complex 8b gave the best conversion to 14a among the phenyl-appended phosphine series of 57% (Table 1, entry 6). The three triazole congener (10) ( Table 1, entry 4) gave only 11% conversion under the same conditions. Pleasingly, however, catalysts derived by hal-ide abstraction in the same manner from complex 11 gave quantitative conversion of 13a to 14a (Table 1, entry 8).
Synthetic modification of phosphorus-containing ligands can result in significant differences in the outcomes of reactions catalysed by their corresponding cationic gold(I) complexes. [4c,19a,21] The regioselectivity of hydration of unsymmetrical internal alkynes has been previously probed by Nolan and co-workers who identified anti-Markovnikov-selective gold(I)carbene complex-catalysed hydration. [22] As such, a model reaction, namely the hydration of unsymmetrical internal alkyne 15 was selected to probe selectivity that might arise from the structural modifications across the series of gold(I) complexes of phosphino-triazoles prepared in this study ( Table 2). The reaction of water at the benzylic position (15-a) represents the generally expected (Markovnikov) outcome, with selective reaction at the alternate alkyne position  being more challenging. Compound 15 was added to a mixture of catalyst precursor (1 mol%) (which had undergone silver triflate (2 mol%)-mediated in situ halide abstraction) in a degassed methanol/water mixture. [23] Consumption of 15 and the ratio of products 16a:16b was determined by 1 H NMR spectroscopy or gas chromatography. That quantitative conversion was observed in allbut-one case (tris-triazole complex 10, Table 2 entry 4), [19a] is in keeping with the results from the hydration of dec-1-yne 13a ( Table 1). All complexes showed a preference for Markovnikov addition at the benzylic position of the alkyne (position [a]). Whilst all cases gave product 16a in preference to 16b (ranging from 4.2:1 to 2.3:1 ratios of products 16a/16b), it is noteworthy that use of complexes 8a and 11 as catalyst precursors ( Table 2, entries 2 and 8 respectively) afforded a greater proportion of the challenging anti-Markovnikov product 16b. Table 3. Substrate scope survey for triazole-appended phosphine ligate gold(I) catalysed alkyne hydration.
[b] Significant Boc-deprotection observed, isolated yield of 14j given in parenthesis.

Conclusions
1,2,3-Triazoles bearing 1-aryl and 5-phosphino functionalities were prepared, wherein the phosphorus atom was appended to one, two or three such triazole motifs. In addition, the corresponding gold(I) chloride coordination complexes bearing phosphino triazoles [with ancillary cyclohexyl or phenyl groups on phosphorus to complete the valence requirement of phosphorus(III) where required] were also prepared and their relative bulkiness determined using the online tool SambVca (2.0). Two more gold(I) chloride complexes were prepared from ligands containing a phosphorus atom appended by one triazole of the same general formula, where the 1-aryl substituent on the triazole was 2,6-dimethoxy phenyl. These complexes, along with complexes of tricyclohexyl phosphine and triphenyl phosphine were in situ converted into their corresponding cationic gold(I) triflate phosphine ligated congeners and compared as alkyne hydration catalysts. Increasing the "triazole number" about the phosphorus atom confirmed that, among the triazole appended phosphines investigated, the monotriazole-containing phosphines were the most effective ligands for cationic gold(I)-catalysed hydration of alkynes.
In summary, the data derived from the above experimentation show triazole-phosphine ligand 1a to be the most promising of the ligands reported. The ease of synthesis and modification of the triazole ligand framework should prove useful in future ligand design, screening and optimisation campaigns.
Supporting Information (see footnote on the first page of this article): General and experimental procedures are available as supporting information, which also includes a summary of the crystallographic data collection and analysis, and further analysis of structural features of the crystal structures of the complexes discussed. All crystal structures disclosed in this report have had the corresponding data deposited at the CCDC with deposition numbers 1922101-1922111. Experimental procedures, including the synthesis of non-commercially available alkynes, characterisation data and spectroscopic information are available in supporting information as are selected, preliminary palladium-catalysis findings. Citations therein refer to methods, data and software used. [25] A pre-peerreview version of this manuscript has been submitted to a preprint repository. [26] CCDC 1922101 (for 4a), 1922102 (for 4b), 1922103 (for 5), 1922104 (for 8a), 1922105 (for 8b), 1922106 (for 9a), 1922107 (for 9b), 1922108 (for 10), 1922109 (for 11), 1922110 (for 12), and 1922111 (for S5 (see supplementary material)) contain the supplementary crystallographic data for this paper. These data can be obtained free of charge from The Cambridge Crystallographic Data Centre.

Author Contributions
All authors contributed in varying degrees to planning the experiments, evaluating results and writing of the manuscript, specific contributions in addition to this are listed for each co-author in alphabetical order: BRB helped direct aspects of the research and gave input and critical assessment throughout the progress of the project; PWD suggested experiments, supervised and provided critical assessment for aspects of the work; JSF led and co-conceived the project, providing critical assessment of data, day-to-day project management and oversight, directed most aspects throughout, supervised most of the experimental work and wrote the majority of the manuscript; FM and TJS synthesised some of the alkynes used in Table 3; HvN contributed to preliminary studies detailed in the supplementary material and some aspects of mass spectrometry of complexes; MGW advised on the preparation of gold complexes and conducted experiments of Table 2; YZ co-conceived aspects of the project, conducted all ligand synthesis and most of the reactions, drafted a proportion of the ESI, offered critical suggestions and conducted the XRD data collection and analysis herein. J003220/1). JSF and HvN. are thankful for the support of the NEURAM H2020 FET Open project 712821. Dr Chi Tsang, Alex S. Quy and Dr Peter Ashton are thanked for assistance with mass spectrometry. Dr Cécile S. Le Duff is thanked for advice on NMR spectroscopy. JSF and BRB acknowledge the support of a Wellcome Trust ISSF award within the University of Birmingham. A Royal Society Research Grant (2012/R1) underpinned initiation of this project.