Visible‐Light‐Driven Catalytic Deracemization of Secondary Alcohols

Abstract Deracemization of racemic chiral compounds is an attractive approach in asymmetric synthesis, but its development has been hindered by energetic and kinetic challenges. Here we describe a catalytic deracemization method for secondary benzylic alcohols which are important synthetic intermediates and end products for many industries. Driven by visible light only, this method is based on sequential photochemical dehydrogenation followed by enantioselective thermal hydrogenation. The combination of a heterogeneous dehydrogenation photocatalyst and a chiral molecular hydrogenation catalyst is essential to ensure two distinct pathways for the forward and reverse reactions. These reactions convert a large number of racemic aryl alkyl alcohols into their enantiomerically enriched forms in good yields and enantioselectivities.


Introduction
Form ost bioactive chiral compounds,o nly one enantiomer has the desired activity whereas the other enantiomer is inactive or even detrimental. Thus,enantioselective synthesis of chiral compounds is essential for the pharmaceutical and chemical industries.C onverting readily available racemates into single enantiomers is an attractive strategy in enantioselective synthesis. [1][2][3] Kinetic resolution and dynamic kinetic resolution are commonly used to prepare enantiomerically enriched chiral compounds from racemates. [2b, 3] However, kinetic resolution has amaximum yield of 50 %, and dynamic kinetic resolution requires one or more steps to convert the chemically modified products back to the original compounds.While potentially efficient, direct deracemization [1] of chiral compounds is challenging to achieve due to (i) unfavorable thermodynamics (a positive DG q of about 0.4 kcal mol À1 at 298 K) and (ii)t he principle of microscopic reversibility in thermal reactions which interconverts the two enantiomers in asingle potential energy surface ( Figure 1a). [4] There are limited examples of single-pot thermal deracemization where compatible (or compartmentalized) oxidants and reductants are used to drive the forward and reverse steps via distinct mechanisms. [2c,5] Photocatalysis is ideally suited for deracemization [1] because light can provide the energy input and enable two distinct potential energy surfaces for the forward and reverse reactions,o vercoming the microscopic reversibility. [4] Only photons are consumed in the process.Despite these appealing features,p hotocatalytic deracemization reactions generally had low enantioselectivity [6] until recently (Figure 1b). [7,8] In their breakthrough work, Bach and co-workers reported light-driven deracemization of axially chiral allenes,w here high enantioselectivity was achieved via selective energy transfer from ac hiral photosensitizer to one enantiomer of as ubstrate. [7a] In another remarkable development, the groups of Miller and Knowles reported photochemical deracemization of cyclic ureas via excited-state electron transfer. [8] Favorable sequential electron, proton, and hydrogen-atom transfer (HAT) steps orchestrated by an achiral photoredox catalyst and chiral phosphate base and apeptide thiol catalyst result in ah ighly enantioselective process. Despite the conceptual advances,the scope of the reactions in these two reports is limited to substrates such as allenes and cyclic ureas.
Here we describe acomplementary approach based on an irreversible photochemical oxidation followed by its thermochemical reverse reaction. Thet hermal reaction is enantioselective by virtue of ac hiral catalyst, which leads to overall deracemization. We demonstrate this approach for one-pot deracemization of secondary alcohols,using aheterogeneous dehydrogenation photocatalyst and ac hiral homogeneous hydrogenation catalyst (Figure 1c). Enantiomerically pure alcohols are ubiquitous synthetic intermediates and end products in the pharmaceuticals,a grochemicals,a nd food industries.A lthough methods based on sequential or cyclic oxidation and reduction have been developed for deracemization of secondary alcohols, [2,5c, 9] over-stoichiometric oxidants and reductants are used to provide the driving forces in these methods,a nd strategies have to be developed to avoid the self-quenching of the redox reagents.Our photochemical method uses only light as the energy input and can be conducted in asimple reaction vessel. As such, we avoid sacrificial chemical redox reagents and complicate reaction systems employed in previous methods,making our process more environmentally friendly and more convenient to conduct.

Results and Discussion
Our design relies on the cyclic action of an achiral photochemical dehydrogenation catalyst and achiral thermal hydrogenation catalyst. Forthe photocatalyst, we first considered heterogeneous semiconductors in view of their high performance. [10] In particular we noticed ar ecent report of photochemical dehydrogenation of alcohols catalyzed by Ni-modified cadmium sulphide (Ni-CdS). [11] Fort he hydrogenation catalyst we chose Noyori catalysts which are efficient and commercially available. [12] Protic solvents are crucial for the reactivity and enantioselectivity of Noyori catalysts,t hus,w e screened for aprotic solvent for the dehydrogenation step.Commonly used protic solvents such as methanol, ethanol, and isopropanol were ruled out because they were susceptible to dehydrogenation. Instead, we checked water, which is hard to be oxidized by visible light (See Table S1 in the Supporting Information), and alcohols without a a-H (See Table S1 in the Supporting Information) as solvents for the dehydrogenation of racemic 1-phenethylalcohol (rac-1a)( Table 1, entries [1][2][3][4][5][6][7][8][9][10][11][12]. We found that dehydrogenation occurred in modest to high yields in acetonitrile (CH 3 CN), water, and the mixtures of water with CH 3 CN,d imethylacetamide (DMA), or tert-butyl alcohol ( t BuOH) ( Table 1, entries 1, 8-10 and 12). We then tested these solvents for asymmetric hydrogenation of acetophenone (2a)with aNoyori catalyst (Ru*-1)( Table 1, entries [13][14][15][16]. Good yields and enantioselectivities were achieved in water alone as well in amixture of water and t BuOH ( Table 1, entries 13 and 16). Considering the good solubility of many alcohols in t BuOH, we decided to use amixture of water and t BuOH as the solvent.
Asymmetric hydrogenation is typically conducted at ahigh pressure (e.g.,10atm or above). [12c] In the small-scale screening tests (e.g., 0.2 mmol of substrate), the H 2 generated from dehydrogenation provides al ow pressure of H 2 for the following hydrogenation step,w hich makes the deracemization process slow (Scheme S2, Supporting Information). Thus, we decided to introduce a10atm pressure of H 2 in the system to accelerate the hydrogenation step.T here is no net 1d ehydrogenation 1a [b] Yields were measured by aGCwith an FID detector; n-dodecane was used as the internal standard.
consumption of H 2 though. Thep hotochemical deracemization of racemic 1-phenethylalcohol (rac-1a)w as tested using several variants of Ni-CdS and Noyori catalysts ( Table 2). We found that using Ni-CdS generated in situ from CdS nanoparticles (CdS NPs) and NiCl 2 ·6H 2 Oand Ru*-1,noderacemization occurred ( Table 2, entries 1-3). When the hydrogenation of acetophenone catalyzed by Ru*-3 was conducted in the presence of 5mol %N iCl 2 ,t he reaction was almost completely inhibited (< 5% yield;S cheme S3, Supporting Information). We reasoned that residual Ni salts in the system would coordinate to the ligands in the Ru hydrogenation catalyst and destroy it;m eanwhile,t he lack of Ni species on CdS was detrimental to the dehydrogenation as well. We then used pre-formed Ni-CdS [11] in combination with Ru*-1,a nd obtained an encouraging 26 % ee in the deracemization ( Table 2, entry 4). We screened several other Noyori catalysts ( Table 2, entry 5-7) and found Ru*-3 to be best. [12c] Hydrogenation appeared to be the slow step in the process,s ot o ensure ah igh yield the reaction was continued for 12 ha fter illumination. Increasing reaction time increased the enantioselectivity ( Table 2, entries 8-10).
We further optimized the deracemization method by varying the catalyst loading, solvent ratio,a nd reaction time (Table 2, entries [11][12][13][14][15][16]. Small improvements in yield and enantioselectivity were obtained using ahigher catalyst loading or longer reaction time (Table 2, entries [11][12][13]. Ther atio of solvent components influenced the reaction rate.I ncreasing the ratio of water to t BuOH from 2.5/2.0 to 3.0/1.5 accelerated the deracemization ( Table 2, entry 14), but further increasing it to 3.0/1.0 decreased the enantioselectivity ( Table 2, entry 15), probably due to al ower solubility of rac-1a in the solvent mixture.I ncreasing the dark reaction time after illumination from 3h to 4h further improved the reaction, and a9 9% isolated yield with 98 % ee was achieved (Table 2, entry 16). (S)-RUCY -Xyl-BINAP (Ru*-5), [13] which bears au nique ruthenabicyclic structure, also gave ah igh yield and ee ( Table 2, entry 17). We tested other semiconductors and thermal dehydrogenation catalysts,s uch as Cu 2 O, Ni-NCN CNx, Ni-g-C 3 N 4 ,a nd some molecular complexes as dehydrogenation catalysts, [14] but none of them showed activities (See Table S4 and S5 in the Supporting Information). Although several homogeneous dehydrogenation photocatalysts have been reported, [15] we could not find proper solvents to integrate them with Ru * -3 to facilitate deracemization (See Table S6 in the Supporting Information).
We tested the scope of the deracemization with al arge array of aryl alkyl alcohols.S ubstrates with an electron donating group at the 4-or 3-position of the aryl group were deracemated in high yields and ees( table 3, 1a-1e, 1g). Different substrates can have different reaction rates so the optimized reaction time and solvent ratios vary for some substrates (footnote of Table 3a nd Section 5inthe Supporting Information). Like rac-1a,(R)-1awas converted to (S)-1a in high yield and ee. Thealcohol 1fwith amorpholine group at the 4-psotion was deracemated in ay ield of only 65 %, possibly due to the instability of this group.S ubstrates with two and more substituents on the aryl groups were deracemated in high ees ( Table 3, 1h-1k). Alcohols with an electron-withdrawing substituent in the aryl group were suitable substrates as well (Table 3, 1l-1p). Thel ower eeso f 1l and 1p (84 %a nd 77 %, respectively) were likely due to as low dehydrogenation as ar esult of having as trongly electron-withdrawing substituent. Naphthyl alcohols 1q and 1r have low solubility in the solvent mixture.Byusing alarge   Table 3a nd Section 5i nt he Supporting Information), decent yields and eesw ere obtained. Alcohols with as ubstituent at the 2-position of the aryl group were deracemated in lower ees( 76-88 % ee, Table 3, 1r-1t), likely due to the steric hindrance of the substrate,w hich slow down the reaction. This steric effect might explain the eesofproducts where the alkyl group vary from ethyl (Table 3, 1u)t oc yclopropyl ( Table 3, 1v)a nd to isopropyl (Table 3, 1w). The eesdecreased from 97 %to80% and to 64 %. Thereaction of abiaryl alcohol 1x had alow ee (47 %) suffering from both electronic and steric effects. Alcohols with ah eteroaryl group were deracemated in high yields and ees ( Table 3, 1y-1ac). We found that the Ni-CdS photocatalyst could be reused while keeping the same reaction efficiency.X RD patterns and UV/Vis spectra of Ni-CdS after the frist and second uses are nearly identical to those of the pristine catalyst, indicating the stability of the catalyst during deracemization (See Section 4.3 and 4.4 in the Supporting Information).
In the above reactions,aH 2 pressure was used to accelerate the hydrogenation step.W er easoned that in al arger reaction scale and lower head space of the reaction vessel, the H 2 generated from the dehydrogenation step would yield an enough pressure that the hydrogenation step could proceed in ar easonable rate without an external H 2 pressure.I ndeed we found that by increasing the scale from 0.2 to 5.0 mmol and using an early filled sealed vessel, deracemization of 1a occurred smoothly in 97 %y ield and 96 % ee (Table 4, 1a). Doing reaction in this scale also allowed the lowering of the catalyst loading:1 50 mg Ni-CdS (20 mol %C dS and 1.7 mol %N i) and 0.67 mol %o fRu*-3 were sufficient. Thee xternal H 2 -free protocol could be applied for the deracemization of many other alcohols (Table 4, 1b, 1c, 1f, 1g, 1n, 1z and 1ac). The eeso f electron-rich aryl alcohols (Table 4, 1b,1c, 1f and 1g)a re slightly higher than that of an electron-poor aryl alcohol (Table 4, 1n). These results confirm the catalytic deracemization using light as the only energy input. They suggest that deracemization of al arge number of alcohols could become possible under external H 2 -free conditions with ap roper design of reaction system.
We monitored the eeso f1a during deracemization and found it to increase gradually during the reaction time ( Figure 2a). This result is consistent with the sequence of chirality-removing dehydrogenation and enantioselective hydrogenation steps.The dehydrogenation occurs at the excited state of Ni-CdS.U pon illumination, electron-hole pairs are [a] Reaction conditions:0.2 mmol alcohol 1,18mgNi-CdS photocatalyst, 2.0 mol %Rucat.,20mLKOH (1.0 M), t BuOH/H 2 O = 1.0/3.0 or 1.0/2.0 or 2.0/4.0 mL, 10 atm H 2 ,two reaction tubes shared two blue LEDs, reaction time:24-96 h; yields are isolated yields, eeswere determined by an HPLC with chiral columns.For 1a-1c, 1g and 1n,the reaction time for light + dark is 24 h + 4h;for 1k,the reaction time for light + dark is 48 h + 12 h; for 1p-1r,the reaction time for light + dark is 96 h + 12 h; the reaction time for light + dark for the other substrates is 72 h + 12 h. For 1a-1c, 1g, 1k and 1n,t he solvent ratio of t BuOH/H 2 Ois1 .5/3.0, for 1d, 1f, 1i, 1j, 1m, 1o and 1i,the solvent ratio of t BuOH/H 2 Ois1 .5/2.5, for 1q, 1r and 1x,the solvent ratio of t BuOH/H 2 Ois2.0/4.0, the solvent ratio of t BuOH/H 2 Of or the other substrate is 1.0/2.0. LED = light-emitting diode, r.t. = room temperature. generated in CdS.T he holes oxidizes an alcohol to give ak etone and adsorbed protons,p robably via ap rotoncoupled electron transfer, although the detailed mechanism remains unclear. [10f, 11, 16] Theelectrons then reduce the protons to dihydrogen, which is known to occur at Ni sites. [16,17] Dehydrogenation of an alcohol is endothermic,but with light excitation, it becomes favorable (Figure 2b). Ther everse reaction, the hydrogenation of the resulting ketone,i s exothermic but does not occur on Ni-CdS due to ah igh kinetic barrier (Figure 2b). In the presence of achiral hydrogenation catalyst, the hydrogenation then proceeds to give the enantiomerically enriched alcohol. Thec ombination of Ni-CdS and Ru*-3 seems to fulfill all the kinetic requirements for an efficient deracemization system.

Conclusion
We developed ao ne-pot deracemization method for secondary alcohols.B yatandem action of ah eterogeneous dehydrogenation photocatalyst and ac hiral homogeneous hydrogenation catalyst, aw ide range of enantiomerically enriched alcohols can be produced from their racemates using light as the only energy input. Neither stoichiometric oxidants and reductants,n or special compartmentation, was required for the process.A lthough at the first stage of development, the scope and efficiencyo ft his deracemization method are inferior to the well-established asymmetric hydrogenation of ketones,i t represents anew approach for deracemization of alcohols, which employs the readily available racemic alcohols as starting reagents.I nt he current system the photocatalyst is not enantioselective,s o each molecule of alcohol likely goes through multiple oxidation/reduction steps in the process.T his scenario increases the energy consumption, although the energy comes from light which is less costly than chemical reagents.T he energy efficiency can be improved if an enantioselective photocatalyst for dehydrogenation can be found, which is ac hallenging future subject of research. Further development of our strategy will potentially yield efficient deracemization methods for other types of substrates,l eading to promising applications in asymmetric synthesis.