Chirality Transfer in Gold(I)‐Catalysed Hydroalkoxylation of 1,3‐Disubstituted Allenes

Abstract Gold(I)‐catalysed intermolecular hydroalkoxylation of enantioenriched 1,3‐disubstituted allenes was previously reported to occur with poor chirality transfer due to rapid allene racemisation. The first intermolecular hydroalkoxylation of allenes with efficient chirality transfer is reported here, exploiting conditions that suppress allene racemisation. A full substrate scope study reveals that excellent regio‐ and stereoselectivities are achieved when a σ‐withdrawing substituent is present.

Meanwhile, we recently carriedo ut af ull investigation into chirality transfer in gold-catalysed allylic etherification reactions;t his reactioni sr elated to the title reaction by virtue of forming similar allylic ether products 6,t hough with different substrate and mechanism.S urprisingly,w ed iscovered that the addition of molecular sieves (MS) wasc rucial for both regioselectivity and efficient chirality transfer (Scheme 4). [21,22] In the absence of molecular sieves, only racemic allylic ether products 6 were observed. Inspired by this discovery,w ed ecided to revisit the gold-catalysed hydroalkoxylation of allenes (Scheme 2 and Scheme 3) in order to investigate whetherasimilar ap-proach would allow us to finally realise efficient chirality transfers. In this full article, we describe our endeavours and disclose for the first time amethod for intermolecular hydroalkoxylationo fa llenes with high degree of chirality transfer,a long with full substrate scope studies.

Results and Discussion
We initiated our investigations using the allene 4b,since regioselectivity is not an issue with this substrate [7] (see below), and also in order to readily compareour results with previousstudies (Scheme 3). Working on the assumption that any erosion of chirality transfer originates from rapid racemisation of the allene substrate, as suggested by Widenhoefer and Yamamoto, while not ruling out racemisation of the allylic ether product through isomerisation (see reaction (2), Scheme1), it is clear in either case that the conditions need to be modified in ordert o retard the racemisation step while stilla ctively catalysing the desired hydroalkoxylation reaction. Based on our previous related studies mentioned above,w eh ypothesised that our two previouss ets of conditions could potentially be solutions: 1) conditions in reaction (3), Scheme 1( DMF,08C, excess alcohol), which was found to suppress any further isomerisation of the allylic ether products, and 2) conditions using molecular sieves, as shown in Scheme 4, which was also shown to favour the kinetic products. [23] Pleasingly,o ur initial resultswere promising and are summarised in Scheme 5. Addition of molecular sieves did indeedr esult in higher chirality transfer (97:3 e.r.), however,a tt he cost of ad rop in yield (30 %, reaction(1), Scheme 5). Unfortunately,f urther attempts at optimisation by increasing the time, temperature and catalystl oading did not significantly improve the yield without as ubsequent drop in e.r.T herefore, we turned to our previous conditions shown in reaction( 3), Scheme 1i nstead:( IPr)AuCl/AgOTf( Tf = triflate) pre-catalyst at 0 8Ci nD MF,w ith excess alcohol. To our delight, high chirality transfer (98:2 e.r.) is observeda nd in am uch better6 5% yield of 6bb (reaction(2), Scheme5).
Scheme5.Initiala ttempts at increasing the levels of chirality transfer. Before proceeding to the substrate scope, we carried out control reactions to ascertain whether erosioni ne nantiomeric excess occurs through racemisation of the allylic ether product 6 or racemisation of the allene substrate 4,o rb oth. To wards this end, the allylic ether products 6bb and 6cb were subjected to two reactionc onditions:t he originalc onditions used by Widenhoefer in reaction(2) Scheme3 (hereafter referred to as ConditionsA)a nd also our conditions shown in reaction (2), Scheme 5( hereafter referred to as Conditions B). As shown in Scheme 6, there was either no or very little erosion of enantiomeric excess, thereby suggesting that isomerisation of products 6 is not the major cause of ee erosion in the gold-catalysed hydroalkoxylation reaction.
Next, the allene substrate 4b was subjected to similarc ontrols (Table 1). This time, however, sterically hindered tBuOH (which is as luggish nucleophile under these conditions, see below) was added to these control reactions in order to replicate the reaction conditions whilst avoiding hydroalkoxylation. [24] As shown in Ta ble 1, ConditionsA clearly result in much faster racemisation of the allene 4b compared to Conditions B.
Having confirmed that the racemisation of the allene substrate is indeed the major culprit for ee erosionu nder Conditions A, we then proceeded to ascertain whether it is the solvent (DMF versus toluene), temperature (0 8Cv ersus RT) or alcohol concentration that is causing the stark differencei nr esults using Conditions Av ersus Ba ss hown in Ta ble 1. To wards this end, the racemisationc ontrol under ConditionsBw as investigated with toluene insteado fD MF as the solvent (Scheme7). The resulting enantiomeric ratios are very similar (withine rror): 96.5:3.5 e.r.i nD MF and 97.5:2.5 e.r.i nt oluene, which initially suggests that the solvent difference between ConditionsA and Bi sn ot the most crucial change with this substrate. The temperature, however,i sc rucial (Scheme 7); when the same toluenee xperiment is repeated at room temperature,the e.r.o f4b plummets to 81:19.
With these observations in hand, we then proceeded to study the effect of the solvent in the actual gold-catalysed hydroalkoxylation reactionw ith allene 4b (Table 2). Indeed, the change of solvent from DMF and tolueneh as only as mall effect on resulting enantiomeric ratios for substrate 4b (98:2 vs. 97:3), but the yield is slightly better with toluene as solvent (65 vs. 81 %, entries 1a nd 2). Sticking with toluene, the alcohol equivalents was investigated next (entries 3-6). The enantiomeric ratio appears to drop slightly as the alcohol equivalents is decreased from 4t o2to 1.1 equivalents (entries3-6). As ar esult of this screen,w ei nitially decided to use the conditions shown in entry 4( hereafter referred to as Conditions C) Scheme6.Controlr eaction to test whether the products 6 racemise under the reaction conditions.
We therefore conclude that, with the exception of the initially studied allene 4b,i ng eneral, the solvent, temperature and alcohol equivalents all cumulatively affect the efficiency of the chirality transfer reactionw hen comparing Conditions Bt ot he previously reported ConditionsA.A lthough the yield is lower using DMF for this particularc ombination of reactants (4c with BnOH), the yields were pleasingly all good to excellent when ConditionsB werea pplied to other substrates (see later, Ta ble 4a nd Table 5).
Having ascertained that Conditions Bi st he most general, we proceeded to investigate the allene substrate scope (Table 4). [25] Suspecting that the OBz substituent plays am ajor role in the excellent regioselectivity observed, we first investigated the effect of variousd ifferent protecting groups on the oxygen (entries 1-4). To our delight,r emoving the carbonyl (4c,e ntry 2) or Ph (4d, 4e,e ntries 3a nd 4) does not seem to significantly affect enantiomeric ratios and ah ighd egree of chirality transfer occurs in all cases. Replacing the Me in 4b with al onger npentylin4falso results in high yield and chirality transfer (entry 5). Inserting an extra CH 2 to place the OBz group further from the allene in 4g,h owever,d oes cause am ore noticeable drop in chirality transfer,a lthough 6gb is still formed in ag ood 79 %yield and 90:10 e.r.(entry 6).  Next, we proceeded to replacet he Oi n4b with N( 4h)a nd pleasingly,t his still gives ah igh8 1% yield and 97:3 e.r.o f6hb ( Table 4, entry 7). Once again, placing the NPhtalate functionality further from the allene in 4i results in am ore noticeable drop in enantiomeric ratio (81:19e .r., entry 8). While all previous examples (entries 1-7) provided exclusively one regioisomer,t he regioselectivity with 4i is lower,a lbeit as till very good 9:1( entry 8), suggesting that the functionality on the substituent is indeedr esponsible for the excellent regioselectivities observed thus far.N evertheless,w hen the silver free catalyst( IPr)AuNTf 2 is used, the regioselectivity is restored to > 20:1 (82:18 e.r.).
Having successfully demonstratedg ood regioselectivities and enantiomeric ratios with aw ider ange of functionalised allenes, we next turned our attention to unfunctionalised ones. Aryl substituted allene 4a,o riginally investigated by Yamamoto (reaction (1), Scheme 3), gave ag ood 10:1 regioselectivity but still racemised under these conditions (Table 4, entry 14). It is likelyt hat the aryl substituent renders the allene isomerisation too rapid for successful chirality transfer under gold-catalysed hydroalkoxylation conditions. [20b] Next, we investigated whether steric differentiation in ad ialkyl 1,3-substituted allene (4n)c ould result in good regioselectivity.D isappointingly,t his is not the case and 6nb and 6nb' is formed as an inseparable regioisomeric mixture (entry 15). [26] The poor regioselectivity with double alkyl substituents (4n)i sn ot necessarily am ajor drawback for synthetic purposes, as functionalised substituents are much more useful as ah andle for subsequente laboration in synthesis.
In order to investigate the minimum amount of functionality required to achieve good regioselectivity,t he ether allene 4o was investigated next (entry 16). Pleasingly,t he reactioni sr egioselective, and ad ecent 87:13 e.r.i so bserved for product 6oc.I ti sc lear that some functionality on one substituent is required for good regio-and stereoselectivities, andt he heteroatom (4o)a nd carbonyls (4b, 4d-l)a ll seem to play ar ole in the observeds electivity.
Finally,i no rder to ascertain whether the heteroatom and carbonyl groups in 4b-l and 4o impart excellentr egioselectivity through achelatione ffect or asimple inductive withdrawing effect, the reactionw ith allene 4p was investigated (Scheme 8). Allene 4p contains an on-chelating electron-withdrawing substituent (CF 3 )i np lace of the O, No rC =Ow ithdrawingg roupsin4b-4l, 4o,b ut still produces ah ighly regioselective reaction( Scheme 8). This suggests that the preference for reaction at a is purely due to electronics:T he inductive withdrawing effect of the functional group (CF 3 in this case) results in electronically differentiated p-bonds a and b, with the LAu + catalyst preferring to coordinate to the more electron-rich p-bond at position a.I ts hould be noted though that allene 4m,w ith electronicallyd ifferentiated p-bonds through ac onjugated rather than inductive withdrawing group,p erformsv ery poorly in the reaction (entry 13, Table 4). Furthermore, because the unwanted and competing gold-catalysed allene racemisation is thought to occur througha chiral h 1 -allylic cation intermediates, [20] it is possible that allenesw ith substituents that stabilise these intermediates (such as Ph in 4a)u ndergo fast racemisation, and therefore provide poor chirality transfer in the reaction,w hereas inductive withdraw-ing substituents have the opposite effect and allow for excellent transfer of chirality.
It should be noted that the reaction is highly stereoselective both in terms of chirality transfer as well as E selectivity (> 20:1 E:Z by 1 HNMR analysis). Ap lausible reason for the observed selectivity is shown in Scheme 9. The gold catalyst can coordinate to either face of the allene (I and I'), with al ow barrier of interconversion between I and I'. [27] Because nucleophiles typically approach anti to the Au I , [1] the approacho ft he alcohol onto I shouldh avet he lower energy barrier as it approaches from the less hindered face of the allene.T his leads to intermediate II and the observed (R,E)-6 product upon protodeauration. Conversely,n ucleophilic attack onto I' is predicted to be kinetically unfavourable, and indeed (S,Z)-6 is never observed experimentally. [28] Furthermore, subjecting am ixture of E:Z isomerso fa na llylic ether 6 to the gold-catalysed hydroalkoxylation conditions (see Supporting Information) resultsi n no change to the E:Z ratio, suggesting that the E selectivity is not due to thermodynamic control.

Conclusion
Gold(I)-catalysed intermolecularh ydroalkoxylation of enantioenriched 1,3-disubstituted allenesw as previously reported to occur with poor chirality transfer due to rapid racemisation of the allene substrate. We have developed conditions to overcome this limitation and to successfully carry out gold(I)-catalysed intermolecularh ydroalkoxylation of allenes with high degree of chirality transfer (up to 99:1 e.r.), excellent E selectivity and good substrates cope. The combined use of the coordinating solventD MF,l ower temperatures and highere quivalents of alcohol nucleophile suppresses allene racemisation and thereby allows for the asymmetric hydroalkoxylation using aw ide range of alcohol nuceophiles,i ncluding sterically hindered and acid sensitive ones. Av ariety of functional groups are tolerated on the 1,3-disubstituted allene substrate, and the former are in fact necessary for excellentr egioselectivities. Controle xperiments suggest that the excellent regioselectivity is determined by inductive withdrawing effects rather than chelation controlo rs teric differentiation.

Experimental Section
General procedure (IPr)AuCl (10 mol %), followed by AgOTf (10 mol %) were added to as olution of allene 4 (0.14 mmol, 1equiv), alcohol nucleophile (1.4 mmol, 10 equiv) and DMF (0.14 mL) at 0 8C. The resulting reaction mixture was allowed to stir at 0 8Cf or 24 h. The crude mixture was then filtered through two plugs of silica, washing with Et 2 O. The filtrate was washed with water and brine, and the resulting organic layer was dried (MgSO 4 )a nd concentrated in vacuo. The crude product was purified by column chromatography to yield products 6.F ull experimental procedures, characterisation for all new compounds and copies of 1 Ha nd 13 CNMR spectra are provided in the Supporting Information.