Asymmetric transfer hydrogenation of unsaturated ketones; factors in ﬂ uencing 1,4- vs 1,2- regio- and enantioselectivity, and alkene vs alkyne directing effects

A detailed study has been completed on the asymmetric transfer hydrogenation (ATH) of a series of enones using Ru(II) catalysts. Electron-rich rings adjacent to the C ] O group reduce the level of C ] O reduction compared to C ] C. The ATH reaction can readily discriminate between double and triple bonds adjacent to ketones, reducing the double bond but leaving a triple bond intact in the major product.

ATH, using the Ru(II) catalysts 1e5, of a,b-unsaturated ketones reveals a rather more complex pattern of selectivities. Deng et al. [7] described the reductions of ketones with catalyst 1 to the corresponding allylic alcohols, using a combination of formic acid and trimethylamine (FA/TEA) as solvent and reducing agent, in high yield; enantioselectivities depended on the substitution pattern on the alkene (Fig. 3a). Using chalcone as the substrate for TH with the achiral ligand N-tosylethyenediamine (TsEN), the reduction product was a ca. 3:1 mixture of saturated ketone and saturated alcohol [7].
Increasing the steric bulk of the alkyl substituent on benzylidineacetones has a complicated effect on the reduction selectivity using catalyst 1 (Fig. 3b) [8]. The ethyl substituent increased the 1,2 selectivity whilst the larger isopropyl group gave equal proportions of 1,2-and 1,4-reduction products in lower conversion. The tertbutyl substituent strongly disfavoured all ketone reduction. Noyori catalyst 1 was applied to the ATH of b-alkyl b-trifluoromethyl a,bunsaturated ketones to yield 1,2-reduction products selectively [9]. Aryl-ketone substrates were reduced in high ee as expected, while a methyl ketone gave a product in poor enantioselectivity (Fig. 3c).
The (non-asymmetric) TH of C]C bonds in enones, using TsEN as a ligand in the Ru(II) complex, can be promoted by incorporating an electron withdrawing groups into the substrate [7], including nitro, ester, nitrile and carboxylic acids. Some asymmetric examples of C]C reduction have been reported [7]. A series of asubstituted cyclic a,b-unsaturated ketones were reduced selectively with the Noyori-Ikariya complex 1 to the corresponding cycloalkenols 8e10, although the carbamate substrate also yielded a small amount of the 1,4-reduction product 11 (Fig. 4) [10]. The acyclic analogue of 10 in contrast was reduced with complete 1,4alkene selectivity, yielding a mixture of 33% saturated ketone and 67% saturated alcohol.
In some cases an enone can be formed in situ and directly reduced. Adolfsson's a-amino acid hydroxyamide ligand 12 has been applied to ATH of allylic alcohols by oxidation to the corresponding enone, followed by complete reduction of alkene and carbonyl functionalities (Fig. 5) [11]. Use of the stronger base potassium tert-butoxide was important for the success of this reduction.
Kosmalski applied the Noyori catalyst 1 to the reduction of bdimethylamino-acetophenone and found that the main product was the partially reduced elimination product [12]. Therefore there is still scope for increased understanding of the subtle effects of substrate structure on the regioselectivity of enone ATH. In this paper, we report our results from our investigation into this area.

Results and discussion
We first examined the ATH of b-chloropropiophenone 13 [13].
Reduction using catalyst (S,S)-2 in FA/TEA at 60 C gave complete conversion to 1-phenylpropan-1-ol 14, in high enantioselectivity (Fig. 6). Subjecting 15 to the same ATH conditions gave 14, in identical ee as obtained previously, as did the reduction of 16, indicating that both were likely to be common intermediates in the formation of 14. Vinyl alcohol 17 was inert under the same conditions in contrast to the result in Fig. 5 [11] and no 17 was observed in the reduction of 15 or 16. Initial 1,4-reduction of 15 is expected to be favourable due to the high reactivity of the unhindered mono-substituted vinyl group. trans-Benzylideneacetophenone 18 (chalcone) was reduced using both catalyst (S,S)-2 and the methoxy analogue 3 under a variety of conditions (Table 1).
With both catalysts, 1,4-reduction was favoured. The substrate was fully consumed but some saturated ketone 20 was observed. Alcohol 19 was produced with consistently good ee of ca. 95e98% in FA/TEA for both catalysts. Compound 21 was formed with a lower ee (73e85%), which is consistent with the presence of two p systems that could compete as directing groups for reduction (Fig. 2). Catalyst (R,R)-3 delivered products in slightly higher ee than (S,S)-2 under the same conditions. Racemic standards were prepared using sodium borohydride; alcohol 21 was prepared by Luche reduction [12], while a sample of 20 was produced by a one pot reduction in the presence of palladium on carbon, acetic acid, isopropanol, and sodium borohydride [13]. The products ees were measured by   chiral HPLC, and the product ratio was determined by NMR spectroscopy to ensure that the measurement was quantitative. Cerium trichloride [12] was tested as an additive (Table 1, entry 5), but it had only a marginal effect. However the additional methanol co-solvent was advantageous, as substrate 18 was poorly soluble in FA/TEA. Further reactions using equal quantities of FA/ TEA and MeOH at lower concentration (0.5 M instead of 2 M) gave full conversion to product (Table 1, entries 3 and 4). Using H 2 O/ MeOH as the solvent system and sodium formate as hydrogen donor (Table 1, entry 6) [14], with (R,R)-3 as the catalyst, the reduction was slower, with incomplete conversion after 45 h at 60 C and the ee of alcohol 19 reduced to 86%. Increasing the FA/ TEA/MeOH reaction temperature to 60 C or decreasing it to 25 C had a marginal effect on the selectivities (entries 7 and 8). Screening of alternative co-solvents in the reduction of 18 was undertaken (Supporting Information). Aprotic solvents tested performed similarly, giving similar or slightly improved 1,4-selectivity and ee compared to reactions with MeOH. Water was also tested as a co-solvent with FA/TEA, however the solubility of 18 in the aqueous FA/TEA mixture was poor, and the enantioselectivty of both products lower than for the other solvents tested.
It was expected that the configurations of both 19 and 21 were the same and this was confirmed by hydrogenating the product mixture from Table 1 entry 4 using Pt 2 O as a catalyst. Given the ratio of alcohols 19 and 21, the predicted ee of 19 after alkene hydrogenation is 97% if the configuration of both alcohols is the same, and 91% if it is different (Supporting Information). The experimental measurement of ee after hydrogenation was 97%, confirming that 19 and 21 must have the same configuration. The electronic nature of the aromatic ring adjacent to the ketone could influence the 1,4-vs 1,2-selectivity of reduction [15]. To test this, chalcone  [7] b, Variation between 1,2-and 1,4-reduction of alkyl-benzylidineacetone derivatives (the ee of C was not determined) [8]. c . Products of selective 1,2-reduction of b-trifluoromethyl enones using catalyst 1 [9].    derivatives containing p-Cl, p-OMe and p-NMe 2 were reduced with catalyst (R,R)-3 to products 22e24 (Fig. 7). Electron donating substituents slow down the rate of reduction, and increase the 1,4selectivity. The proportion of 1,2-product was so low for the pmethoxy and p-dimethylamino products 23 and 24 respectively, that the ee of the unsaturated product could not be determined.
In order to establish the importance of each aromatic ring in the chalcone derivatives, the corresponding cyclohexyl substituted substrates were reduced by ATH (Fig. 7). The ketone with a cyclohexyl adjacent to the alkene reacted with similar selectivity to chalcone, to give 25 with a slight increase in 1,4-selectivity (97% ee). Reduction of the ketone with the cyclohexyl adjacent to the ketone, in contrast, gave different products 26 depending on which catalyst was used. In general it gave a much higher proportion of 1,2reduction product, although 1,4-selectivity was highest under aqueous conditions using sodium formate as the hydrogen source (Fig. 7). The ee of reduction was also poor. This demonstrates the importance of the aromatic ring adjacent to the ketone for the control of enantioselectivity, but that the aromatic ring on the alkene is of secondary importance. ATH of a substrate with a b,bdisubstitution gave a product in high 1,4-selectivity, with predominant formation of the saturated alcohol 27 over the equivalent allylic alcohol although in only 55% ee.
It is known in the literature that alkynes are generally inert under ATH conditions, and are capable of acting as directing groups ( Fig. 2) [3d, 4,6]. It was therefore of interest to establish the outcome of the ATH of substrates containing both an alkyne and an alkene flanking the central ketone (Fig. 8).
The precursor ketones were prepared by reaction of the required lithiated acetylene with the Weinreb amide of cinnamic acid. Racemic standards were obtained for all the products and HPLC was used to directly assess the regio-and stereoselectivity of the reactions. Product ratios were also determined by 1 H NMR data. In the ATH of the diphenyl substrate, using (R,R)-2, the saturated and unsaturated products 28 and 29 respectively were formed in an 83:17 ratio (Fig. 8). The ee of 28 was highest, presumably because the alkene is reduced first and the resulting propargylic ketone is selectively reduced following existing precedent [1,3d,4c]. The 1,2reduction product 29 was formed in lower ee likely due to competing electron-rich unstaturated bonds in the CH/p of the reduction transition state (Fig. 2). The absolute configuration of the products was not unambiguously determined. However it is likely that the R-configuration products will be formed using the (R,R)configuration catalyst, based on the precedent for this class of reductions. A reaction/time study was completed to investigate the ATH using HPLC (Supporting Information). From an early stage in the reaction, the formation of the intermediate saturated ketone was essentially instantaneous, and a small amount of unsaturated product was also observed. As time increases, starting material and intermediate ketone disappear and the two alcohol products are formed. The effect of solvent on the reaction was also investigated and a time study for each solvent was undertaken to explore the relative rate of formation of each of the intermediate and product species over time (Supporting Information). DFT studies have indicated that using MeOH engages in hydrogen bonding interactions to the ketone during the reduction [2a,2e] and in some cases different solvents have been demonstrated to reverse the enantioselectivity [16]. However the results, whilst similar to those in Fig. 8, were inferior with respect to product selectivity, enantioselectivity and conversion. Four further ATH catalysts, 1, 3, '4Ctethered' 6 and the 'benzyl-bridged' 7 were also used (Fig. 8). Time studies were also conducted to track the formation of products and intermediates in each case (Supporting Information). Catalyst 6 produced similar ratios of alcohols and ee values as seen with 2, however, it gave lower conversion. Catalyst 3 gave a similar result to that of 6. ATH with 3 at the lower temperature of 25 C gave a more selective product ratio (79:21 28:29) with high ee's of 98% and 87% for the saturated OH and unsaturated OH respectively. However the conversion was slightly lower at 85%. Catalyst 7 gave approximately a 1:1 ratio between the saturated and unsaturated products which could be due to the hindered nature of the catalyst.
The ATH (1 mol% catalyst (R,R)-2, MeOH co-solvent at 40 C) was tested with other substrates and in all cases, standards of the reduction products were prepared for HPLC analysis. Attempts were made at ATH of the TMS-substituted alkyne but decomposition was observed. However the ATH of the TIPS-enynone was successful. Products 30 and 31 were isolated as a mixture, in a ratio of 69:31 and in a conversion of just 81%. The ATH was less selective for the 1,4-reduction pathway and hence, produced more of unsaturated alcohol 31. The enantioselectivity was high for both alcoholic products however. The products of ATH of the pmethoxyphenyl derivative were obtained as an 86:14 mixture of 32 and 33. Hence the electron-donating group does not appear to significantly affect the reaction mechanism or product distribution. The p-chlorobenzene (PCB) derivative gave a product as a mixture of 34 and 35, with slightly lower enantioselectivity. An impurity was found in the 1 H NMR spectrum and it was hypothesised that this was the alkyne reduction production 36, based on the observation of a further ABX system in the 1 H NMR spectrum (see the Supporting Information).

Conclusions
Investigation of the ATH of a range of enones showed that predominant 1,4-reactivity is favoured and the majority of aromatic-ketone substrates were reduced to their saturated alcohols with high ee. Electron-donating para substituents on the ketone favoured 1,4-reduction further. The ATH of alkene/alkyne ketones leads to a mixture of chiral alcohols via 1,4 and 1,2reduction pathways, the 1,4-pathway being predominant in the reaction. A time study of the reaction confirmed that the C]C bond is rapidly reduced early in the reduction process, and the saturated products were formed in higher ee than the unsaturated ones.

General experimental
All reagents and solvents were used as purchased and without further purification, with the exception of cyclohexane carboxaldehyde which was redistilled for storage.
All reactions were carried out under a nitrogen atmosphere unless otherwise specified. Reactions at elevated temperature were maintained by thermostatically controlled oil-baths or aluminium heating blocks. A temperature of 0 C refers to an ice slush bath, À78 C to a dry ice acetone bath.
NMR spectra were recorded on a Bruker AV (250 MHz), Bruker DPX (300 or 400 MHz), Bruker DRX (500 MHz) or Bruker AV-II. (700 MHz). All chemical shifts are rounded to the nearest 0.01 ppm for 1 H spectra and the nearest 0.1 ppm for 13 C spectra, and are referenced to the solvent chemical shift. Coupling constants are rounded to the nearest 0.1 Hz. Mass spectra were recorded on an Esquire 2000 and high resolution mass spectra were recorded on a Bruker Micro ToF or MaXis. IR spectra were recorded on a Perki-nElmer spectrum 100 and peaks are reported in wavenumbers. Optical rotations were measured on an Optical Activity Ltd. AA-1000 Polarimeter and are reported in deg dm À1 cm 3 g À1 .
The chiral GC measurements were performed using a Perki-nElmer 8500 or Hewlett-Packard 1050 instrument linked to a PC running DataApex Clarity software. HPLC measurements were performed out using a Hewlett Packard 1050 Series with a quaternary pump, autosampler and variable wavelength detector linked to a PC running DataApex Clarity software.
Melting points were determined on a Stuart scientific melting point apparatus and are uncorrected. Flash column chromatography was performed using silica gel of mesh size 230e400, Thin layer chromatography was carried out on aluminium backed silica gel 60(F254) plates, visualised using 254 nm UV light, potassium permanganate, iodine stains or cerium ammonium molybdate (CAM) as appropriate. Column chromatography was performed either by gradient elution (reported as a range, eg EtOAc/Petroleum ether (2e12%), or by isocratic elution.

rac-1-Phenylpropan-1-ol 14
This compound is known [17]. To a solution of propiophenone 16 (66 mg, 0.49 mmol, 1 eq) in methanol (0.9 mL) and water (0.1 mL) was added sodium borohydride (41 mg, 1.08 mmol, 2 eq) as a solid in one portion. The reaction was monitored by TLC. After stirring for 6 h, the reaction mixture was concentrated under vacuum, the residue suspended in water (1 mL) and extracted with Et 2 O (3 mL total). The organic layer was dried (Na 2 SO 4 ) and concentrated to give the product 35 as a clear oil (35 mg, 0.26 mmol, 52%). The spectroscopic data were consistent with those observed for the asymmetric product below.

(E)-3-Cyclohexyl-1-phenylprop-2-en-1-one
This compound is known [31]. To a suspension of sodium hydride (60 wt% dispersion in mineral oil, 0.20 g, 5.0 mmol, 1.0 eq) in THF (5 mL) at 0 C was added dropwise a solution of the diethyl (2oxo-2-phenylethyl)phosphonate [30] (1.25 g, 4.9 mmol, 1.0 eq) in THF (5 mL) and the resulting clear solution was stirred for 30 min at room temperature. Cyclohexanecarboxaldehyde (0.57 g, 5.1 mmol, 1.0 eq) was added neat and the reaction mixture stirred at room temperature overnight. The reaction was quenched with NH 4 Cl (half saturated, 30 mL) and extracted with ethyl acetate (3 Â 15 mL), the organic extracts washed with brine (25 mL), dried over Na 2 SO 4 and concentrated to give the crude product as a clear oil (1.13 g). The crude was taken up in methanol (50 mL) and cooled to À72 C. The resulting white precipitate was filtered and dried to give the pure product as a white solid (492 mg, 2.30 mmol, 45%). Mp

(E)-1-Cyclohexyl-3-phenylprop-2-en-1-one
This compound is known [33]. Sodium methoxide solution (25% w/w, 5.96 g, 27.6 mmol, 1 eq) was diluted to 50 mL with methanol and added to cyclohexylmethyl ketone (3.33 g, 26.4 mmol, 1 eq). The mixture was cooled to 0 C and a solution of benzaldehyde (2.81 g, 26,5 mmol, 1 eq) in methanol (15 mL) was added. The reaction mixture was warmed to 40 C and stirred for 3 days. The reaction was quenched with 0.25 M HCl (100 mL) and extracted with diethyl ether (4 Â 100 mL), the organic layers were dried and concentrated to give the crude product as a yellow oil that solidifies slowly on standing. The oil was dissolved in~150 mL of methanol and cooled to À78 C, the resulting precipitate was filtered and washed once with cold methanol and dried to give the purified product as a white solid (2.78 g, 13.0 mmol, 49%). Mp 54e58 C; d H

rac-1-Cyclohexyl-3-phenylpropan-1-ol 26
This compound is known [32]. To a solution of cyclohexane carboxaldehyde (128 mg, 1.14 mmol, 1 eq) in THF (1 mL) was added phenethyl magnesium chloride (1 M in THF, 1 mL, 1.0 mmol, 1 eq) at À78 C. The reaction was stirred for 2.75 h while gradually warming to~0 C, then quenched with NH 4 Cl (sat. soln, 2 mL) and water (1 mL). The suspension was extracted with Et 2 O (2 Â 2.5 mL), the organic layers dried over MgSO 4 and concentrated to give the crude product as a white solid. The crude was purified by column chromatography (10% EtOAc in petroleum ether) to give the pure product as a white powder (110 mg, 0.51 mmol, 45%

2-(1-Hydroxycyclohexyl)-1-phenylethan-1-one
This compound is known [35]. TiCl 4 (1 M in DCM, 12 mL, 12 mmol, 1.2 eq) was added dropwise at 0 C to a solution of cyclohexanone (1.23 g, 12.5 mmol, 1.25 eq) in DCM (20 mL) and stirred for 25 min. To the resulting yellow suspension was added dropwise 1-phenyl-1-(trimethylsiloxy)ethylene (1.94 g, 10 mmol, 1.0 eq). The resulting orange suspension was allowed to warm to rt and stirred for 24 h before being quenched with water (35 mL). The mixture was extracted with DCM (2 Â 20 mL), washed with brine (10 mL) and filtered through a plug of silica gel (~4 g) with DCM to give the crude product as a thick yellow oil that crystallises on standing (2.69 g).The crude was dissolved in hot methanol, concentrated to a thick oil and then crystallised by addition of hexane (~10 mL) to give the pure product as a white crystalline solid (0.95 g, 4.3 mmol, 43%). A second crop was isolated by concentration of the mother liquors and addition of hexane to give white plates (0. 20

rac-2-Cyclohexylidene-1-phenylethan-1-ol
This compound has been reported as part of a mixture of isomers but has not been fully characterised [38]. To a suspension of 2cyclohexylidene-1-phenylethan-1-one (101 mg, 0.5 mmol, 1 eq) and cerium trichloride heptahydrate (185 mg, 0.5 mmol, 1 eq) in methanol (1 mL) was added sodium borohydride (29 mg, 0.8 mmol, 1.5 eq) at 0 C. The reaction was stirred for 2 h and quenched with NH 4 Cl (sat., 0.5 mL), diluted with water (0.5 mL) and extracted with diethyl ether (3 Â 2 mL). The organic extracts were dried over Na 2 SO 4 and concentrated to give the product as a clear oil (103 mg, 0.50 mmol, 100%). The crude product was purified by column chromatography on silica gel (2.6 g) with 10% diethyl ether in petroleum ether as eluent, to yield the pure product as a clear oil

(S)-2-Cyclohexyl-1-phenylethan-1-ol 27 and (S)-2cyclohexylidene-1-phenylethan-1-ol
A suspension of 2-cyclohexylidene-1-phenylethan-1-one (95 mg, 0.47 mmol, 1 eq) and catalyst (S,S)-2 (3.1 mg, 0.005 mmol, 1%) in FA/TEA (5:2, 0.5 mL) and MeOH (0.5 mL) was stirred at 40 C for 22.5 h. The mixture was diluted with diethyl ether (2 mL) and quenched with NaHCO 3 (sat., 2 mL), the aqueous layer was extracted further with ether (2 Â 2 mL) and the organic extracts dried over Na 2 SO 4 and passed through a silica plug to yield the crude product as an off white solid (86 mg, 0.42 mmol, 89%). The product was obtained in full conversion as a mixture of saturated and unsaturated alcohols, ratio 94:6 by 1 H NMR. Major product 55% ee as calculated by HPLC. Purification by chromatography on silica (8% Et 2 O/Petroleum ether) separated the unsaturated alcohol and gave the purified product as a white solid (60 mg, 0.29 mmol, 62%). Spectroscopic data for asymmetric product is consistent with the prepared standards. The reaction was monitored by TLC and/or HPLC. Once completed, the mixture was quenched with NaHCO 3 (sat. soln.) and extracted with EtOAc. The organic layers were combined, dried with MgSO 4 and concentrated in vacuo to give the crude product. Where appropriate further purified was undertaken.
4.1.28.2. General n-BuLi procedure for racemic alcohol synthesis (procedure 2). A degassed solution of acetylene in THF (anhyd.) was cooled to À78 C. Once cooled, n-butylithium was added dropwise, the reaction mixture was left to stir at À78 C for 30 min. Aldehyde was added dropwise and the reaction mixture was left to stir at À78 C. After 1 h, the reaction mixture was allowed to warm to r.t. The reaction was monitored by TLC. Once completed, the reaction was quenched with NH 4 Cl (sat. soln) and extracted with EtOAc. The organic layers were combined, dried with MgSO 4 and concentrated in vacuo to give the crude product. The crude product was further purified using column chromatography to give the desired product.
4.1.28.3. General procedure for MnO 2 oxidation for ketone synthesis (procedure 3). Alcohol and activated MnO 2 were dissolved in DCM (anhyd.) and stirred at r.t. The reaction was monitored using TLC. When completed, the reaction mixture was diluted with DCM and filtered through a Celite pad and concentrated in vacuo to give the crude product.

Data sharing statement
The research data (and/or materials) supporting this publication can be accessed at http://wrap.warwick.ac.uk/.

Declaration of competing interest
The authors declare no conflicting interests.