Atropisomeric [(diphosphine)Au2Cl2] Complexes and their Catalytic Activity Towards Asymmetric Cycloisomerisation of 1,6-Enynes

X-ray crystal structures of two [(diphosphine)Au2Cl2] complexes (in which diphosphine=P-Phos and xylyl-P-Phos; P-Phos=[2,2′,6,6′-Tetramethoxy-4,4′-bis(diphenylphosphino)-3,3′-bipyridine]) were determined and compared to the reported structures of similar atropisomeric gold complexes. Correlations between the Au⋅⋅⋅Au distances and torsional angles for the biaryl series of ligands (MeOBIPHEP, SEGPhos, and P-Phos; BIPHEP=2,2′-bis(diphenylphosphino)-1,1′-biphenyl, SEGPhos=[(4,4′-bi-1,3-benzodioxole)-5,5′-diyl]bis[diphenylphosphine]) can be made; these measurements appear to be very dependent upon the phosphorous substituent. Conversely, the same effect was not observed for ligands based on the binaphthyl (BINAP) series. The catalytic activity of these complexes was subsequently assessed in the enantioselective cycloisomerisation of 1,6-enynes and revealed an over-riding electronic effect: more-electron-rich phosphines promote greater enantioselectivity. The possibility of silver acting as a (co-)catalyst was ruled out in these reactions.


Introduction
The recent upsurge of interest to explore the medicinal properties of three-dimensional molecular structures and fragments [1][2][3][4] has driven the need for synthetic methodologies that can generate complex scaffolds from acyclic precursors in a single step. An exemplar of such reactions is the cycloisomerisation of 1,6-enynes, which generates carbo-and heterobicyclo[4.1.0]heptene derivatives with up to three stereocentres, thus provides extremely valuable compounds for organic synthesis. [5][6][7] The reaction was first discovered in 1995 by Blum and coworkers by using a catalytic amount of PtCl 4 . [8] Since then, a number of other p-acidic transition-metal catalysts were reported, including Rh, [9,10] Ir, [11] Pt [12][13][14] and Au catalysts. [15,16] The first enantioselective examples were provided by Michelet and co-workers in 2009, by using a gold(I) complex of the chiral diphosphine ligand L1 (Scheme 1). [17,18] Subsequently, the enantiomeric excess (ee) values obtained have been superseded by other transition-metal catalysts (Pt, [19] Rh [20][21][22] and Ir [23] ), but gold is the only reported system that is catalytically active at room temperature.
The catalytic activity of cationic gold complexes continues to be a subject of considerable intrigue and debate. [24] Diphosphine ligands were frequently found to be effective for many asymmetric transformations catalysed by gold. More often than not, silver salts are required to activate [(diphosphine)Au 2 Cl 2 ] precursors towards catalysis. This is thought to gen-erate a cationic species I (Scheme 2) with sufficient p-Lewis acidity for catalysis, but the characterisation of such cationic gold complexes remains elusive; no X-ray crystal structure has been determined yet. Most diphosphine ligands do not form chelate rings with gold(I), instead linear geometries that may contain short gold-gold distances, indicative of aurophilic interactions, are favoured. [25] Thus, a significant gold-gold cooperative effect may be important for the catalytic activity. In reactions catalysed by [AuCl(PR 3 )]/AgX systems, chloro-bridged species II (Scheme 2) has been isolated and identified. Also, the participation of silver in the catalytic reaction (presumably via bimetallic complexes) cannot be ruled out. [26,27] In this paper, we report the results of a preliminary study conducted with different diphosphine-gold(I) complexes for the cycloisomerisation reactions of 1,6-enynes. We have attempted to correlate the structural parameters of the gold complexes to the observed enantioselectivity and to investigate the role of the silver salt on the outcome of these reactions.

Results and Discussion
The study began with a search in the CCDC database for reported structures of chiral atropisomeric biaryl [(diphosphine)Au 2 Cl 2 ] complexes, from which 14 discreet structures corresponding to 9 compounds can be found (diphosphine = L1-L9, Table 1). [28] The major conformational parameters that vary between the complexes are given in Table 1; namely, the torsional twist about the central aryl-aryl bond (q), the AuÀCl bond length and the intramolecular Au···Au separation.
Additionally, two complexes that contained P-Phos (L10) and xylyl-P-Phos (L11) were also prepared and their structures were determined by X-ray crystallography (Figures 1 and 2). Complex [(L10)Au 2 Cl 2 ] was found to crystallise with two independent molecules (A and B) in the asymmetric unit ( Figure 1 Table 1).
Manifestation of aurophilicity is generally indicated by Au···Au distances that are shorter than the sum of two van der Waals radii of 3.7 . [40] For the atropisomeric diphosphine ligands (MeOBIPHEP (L1-L3), BINAP (L7-L9), SEGPHOS (L4-L5) and P-Phos) it is expected that the Au···Au distance is dependent upon the degree of rotational freedom between the aromatic rings that defines the axial chirality, which is, in turn, determined by the nature of the biaryl that constitutes the scaffold, as well as the P substituents. With this in mind, a scatter plot of the torsional angle between the two central aryl rings against the Au···Au distance was produced ( Figure 3) and revealed some interesting trends.  Greater conformational flexibility was found in the biaryl series MeOBIPHEP, SEGPHOS and P-Phos, for which it is common to find more than one independent molecule for each complex. Nevertheless, the Au···Au distance tends to increase with a greater torsional angle between the two aromatic rings. Within the MeOBIPHEP series (L1-L3,^, Figure 3) the introduction of tert-butyl and methoxy substituents on the P-aryl group appear to augment both of these values. The difference between L1 and L3 is particularly notable: the presence of an additional methoxy substituent at the para position appears to lengthen the Au···Au distance quite significantly. The same trend can be observed in the P-Phos series (L12 and L13, &, Figure 3), in which the presence of electron-donating xylyl groups increased the Au···Au distance. For the SEGPHOS series (L4 and L5,~, Figure 3), substitution of PPh 2 by PCy 2 (Cy = cyclohexyl) groups also has a dramatic effect on the intermetallic distance. As a result, Au···Au distances of < 3.7 were only observed in four of these biaryl diphosphine ligated complexes, which all contained PPh 2 substituents. Thus, it appeared that intermetallic interactions are discouraged by more-electron-rich P donors.
In contrast, Au···Au distances in the gold complexes of the BINAP series (L7-L9) are essentially insensitive to the P substituent (*, Figure 3). This is attributed to the higher barrier of rotation for the binaphthyl ring. For this particular series of gold complexes, all Au···Au distances were found to be > 5.4 . These observations suggest that aurophilicity is relatively uncommon for these atropisomeric [(diphosphine)Au 2 Cl 2 ] complexes, inhibited by more-electron-rich P-donor groups and by restriction of the conformational flexibility of the aryl-aryl bond.
In the next phase of this study, the catalytic activity of selected complexes was tested in the cycloisomerisation of ether substrates 1 a-d (Scheme 3). Selected results are presented in Table 2.
The cyclisation of ether-tethered enyne substrate 1 a was examined initially as the model reaction, performed in toluene at room temperature. The product was somewhat unstable, therefore quick purification by flash chromatography was necessary to achieve good isolated yields. The benchmark for this   study was set by reproduction of the result reported by Michelet and co-workers by using ligand L1: [17,18] the product 2 a can be obtained in 87 % yield with 99 % ee ( Table 2, entry 1). Guided by the structural analysis (Table 1), two complexes from the BINAP series (L7 and L8) were tested as catalysts because they have similar Au···Au distances (> 5.3 ). In both cases, the product was isolated in good yields. By using the BINAP ligand L7, a modest 46 % ee was obtained, whereas tolyl-substituted L8 gave a much higher 75 % ee (Table 1,  entries 2 and 3). Similarly, two ligands from the P-Phos series (L10 and L11) were selected for their similar torsional angles to L1. In these cases, excellent yields of the product were obtained. Once again, the more-electron-rich P-xylyl-substituted complex L11 offered significantly better enantioselectivity than L10 (38 vs. 60 % ee; Table 1, entries 4 and 5).
Correlations between structural parameters obtained from solid-state structures and homogeneous catalytic reactions should always be exercised with caution. Nevertheless, these preliminary studies seem to suggest that the enantioselectivity of atropisomeric biaryl diphosphine ligands can be enhanced by using more-electron-rich P substituents.
Next, we examined the importance of the silver salt in these reactions. In previous studies, the choice of silver salt/counteranion often had a profound effect on the catalytic activity and it has been suggested that the participation of silver catalysis cannot be discounted in such reactions. [26] In this case, a control reaction showed that the linear substrate 1 a remained unchanged in the presence of AgBF 4 alone at ambient tempera-ture (Table 2, entry 7), that is, any background racemic process due to the silver salt can be categorically ruled out. Subsequently, by using the L11-ligated complex, the cycloisomerisations of three other substrates 1 b-d were performed in the presence of various amounts of silver salt, with the stoichiometric ratio of Au/Ag either 1:1 or 2:1 ( Table 2, entries [8][9][10][11][12]. In each case, the product yield was lower when silver salt (0.5 equiv) was employed but the level of enantioselectivity remained invariant. This is a significant observation. Considering the different cationic complexes that can be generated in each case (Scheme 4), the involvement of a bimetallic Au/Ag complex as the active catalyst can be ruled out because this can only be generated in a significant quantity when a sub-stoichiometric amount of silver salt is used. [41] Furthermore, the result also strongly suggests that each metal centre operates independently as an active catalytic site, that is, the presence of any cooperative effect between the two proximal Au metal centres may be discounted. This is commensurate with the previous study, in which greater enantioselectivity was observed with complexes that had larger intermetallic distance/less propensity to form Au···Au interactions for steric (L1) and/or electronic (L8, L11) reasons.
Finally, the active catalyst was also generated in situ by premixing the gold-chloride complex with the silver additive prior to addition of the substrate, or by introducing an intermediate operation, in which the AgCl formed was removed by filtration through a bed of Celite before the introduction of the substrate. These procedures had no effect on the results. Hence, the involvement of [LÀAuÀClÀAg] + is also highly unlikely.

Conclusion
We have conducted a structural study on a series of [(diphosphine)Au 2 Cl 2 ] complexes. Even within the small selection of examples, some trends can be observed within the diaryl diphosphine complexes, for instance an increase in the electron-donating character of the P-Ar substituent increases the Au···Au distance. This effect is not observed, however, in the BINAP series of ligands, which is attributed to the greater barrier to rotation/more-rigid structure. The catalytic activity of selected complexes was tested in the cycloisomerisation of 1,6-enynes; the incorporation of electron-rich P donors enhanced the stereoselectivity. Critically, the enantioselectivity of the process was not affected by using different gold/silver ratios or by removal of the silver by-product. This suggests that each metal Scheme 3. Au-catalysed cycloisomerisation of 1,6-enynes 1 a-d. centre is likely to operate independently and that silver does not participate in the process. These observations are useful for the future design of better catalysts for these reactions.