Reductive deoxygenation of alcohols is an essential step in the production of platform chemicals from biomass, because biomass-derived resources are generally oxygen-rich and removing oxygen atoms from such chemicals allows production of more easily handled, carbon/hydrogen-rich chemicals with higher solubility in organic solvents and lower boiling point , . Direct cleavage of the C–O bonds of allylic alcohols to alkenes is a representative reductive deoxygenation that is important not only in the conversion of biomass-derived allylic alcohols, but also in the synthesis of complex organic molecules with olefin functionalities . Because of its synthetic importance, hydrogenolysis of allylic alcohols to alkenes has been studied for decades, and catalytic methods for this reaction using cobalt complexes , , heteropolyacid (H3[PW12O40]·nH2O)  or bismuth triflate , and stoichiometric methods using titanocene , , ,  or NaBH3CN/BF3·OEt2  are available without significant loss of C=C π-bonds. Those methods, however, require the use of stoichiometric salt additives, silicon, titanium, or boron reagents (Scheme 1). A greener method for directed cleavage of the C–O bonds of allylic alcohols to alkenes remains elusive.
In the course of our studies to explore selective synthetic methods using metal-loaded semiconductor photocatalysts , , , we have recently found that palladium-loaded titanium(IV) oxide (Pd/TiO2) photocatalysts effectively promote hydrogenolysis of allylic alcohols to alkenes in methanol at room temperature (rt) under light irradiation (Scheme 1) . As hydrogen atom(s) from methanol is incorporated in the alkene product in the photocatalytic process, we characterized this reaction as photocatalytic transfer hydrogenolysis (PcTH) , . The PcTH of allyl alcohols is highly chemo- and redox-selective: hydrogenolysis of the C–O σ-bond predominates over reduction of the C=C π-bond and oxidation of the HC–OH bond. The regio- and stereoselectivity in the PcTH of substituted allylic alcohols are readily predictable. In this paper, we focus on our recent efforts in developing the PcTH of allylic alcohols in the context of production of platform chemicals and fine chemicals.
Results and discussion
Preparation of palladium-loaded titanium oxide
We chose palladium-loaded titanium oxide (Pd/TiO2) photocatalyst for the PcTH of allylic alcohols because palladium nanoparticles and TiO2 are privileged tools for catalytic conversion of allylic compounds  and photocatalytic hydrogen evolution from alcohols , , respectively. First, Pd/TiO2 was prepared by impregnating TiO2 [from Sigma-Aldrich, anatase–rutile mixture, <100 nm particle size (BET)] with an aqueous solution of PdCl2, followed by drying under vacuum and reduction with NaBH4 , , . Inductively coupled plasma optical emission spectroscopy (ICP-OES) of digested samples of the Pd/TiO2 showed that the palladium content was 4.78±0.06 wt%. The presence of Pd nanoparticles loaded on TiO2 (diameter: 4–7 nm) was confirmed by high-resolution transmission electron microscopy (HRTEM) and energy-dispersive X-ray spectrometry (EDX) (Fig. 1a and b). X-Ray photoelectron spectroscopy (XPS) analysis indicated that the valence number of the Pd nanoparticles was mainly Pd0 (Fig. 1c) .
PcTH of allyl alcohol using Pd/TiO2
Scheme 2 shows the result of our initial examination of PcTH of allyl alcohol (1a) using Pd/TiO2. 1a was smoothly converted to 2a in 70% yield with Pd/TiO2 [1a/Pd (mol/mol)=300] in CH3OH in the presence of TsOH·H2O as an additive under near-UV–vis light irradiation (λ>365 nm) at 25°C (Scheme 2). GC/MS analysis of the reaction mixture indicated concomitant formation of methylal and methoxymethanol. In the absence of any acidic additive, PcTH of 1a gave 2a in a lower yield. Under slightly acidic conditions with TsOH, the OH group of 1a may be more readily adsorbed on the photocatalyst covered by a higher concentration of H+, which could also activate the OH group more effectively, giving protonated 1a. Pd, TiO2, and near-UV light irradiation were all essential for PcTH to proceed.
GC monitoring of the product distribution revealed high selectivity of the PcTH for propylene generation (Fig. 2). PcTH of 1a afforded 2a in 94% yield at t=9 h. Hydrogen gas (H2) generation was negligible before the complete consumption of 1a. After consumption of 1a, evolution of H2 and reduction of 2a to propane (3a) started, implying that the presence of 1a kinetically suppresses (1) generation of H2 gas; and (2) hydrogenation of the alkene of 1a and 2a on the Pd surface, both of which are otherwise favorable processes. This chemoselectivity is completely opposite to that observed in thermal catalysis over Pd nanoparticles under a hydrogen atmosphere .
Catalyst recycling experiment
Catalyst recycling experiments (Fig. 3) indicated that (1) PcTH of 1a using recovered photocatalyst gave 2a without any appreciable decrease in yield; (2) the catalyst could be used at least five times; (3) throughout the recycling experiments, selectivity for producing 2a over 3a was consistently >99:1. The total turnover number through five runs calculated based on the amount of Pd loaded on TiO2 was ca. 900 (mol/mol).
PcTH of γ-substituted allylic alcohols
In contrast to non-substituted allylic alcohol 1a, PcTH of γ-substituted allylic alcohols raises issues of regioselectivity and stereoselectivity. For instance, hydrogenolysis of cinnamyl alcohol (1b) can potentially give stereoisomers of (E/Z)-β-methylstyrene [(E)-2b and (Z)-2b] and a regioisomer, allylbenzene (iso-2b). To our delight, however, after optimization of the reaction conditions, we found that PcTH of 1b using Pd/TiO2 and an acidic additive (0.5 mol%) selectively gave E-2b (93% yield) together with a small amount of over-reduced product 3b (7%, Scheme 3). Amounts of other isomers (Z-2b and iso-2b) were below the detection limit based on the GC calibration line (<2% yield). In this experiment, we used Pd/TiO2 prepared by impregnating TiO2 [Degussa Aeroxide® P25, anatase-rutile mixture, 21 nm particle size, 35–65 m2/g surface area (BET)] with PdCl2(CH3CN)2, without pre-reduction with NaBH4 . Again, we found that high chemoselectivity for C–O bond cleavage over C=C π-bond reduction is characteristic of this photocatalytic system: when the reaction mixture was left under a hydrogen atmosphere in the dark, hydrogenation of the C=C π-bond of 1b afforded 3-phenylpropan-1-ol (86% yield) and scant hydrogenolysis of the C–O bond of 1b took place.
Deuterium labeling experiment
To explore the characteristics of the PcTH, deuterium-labeling experiments were conducted using deuterated methanol (CD3OD) instead of CH3OH (Scheme 4, above). 1H and 13C NMR analyses of the reaction mixture indicated that 1b was converted to mainly E-2b-d1 [E-2b-d1 (64% D):1b:3b-d3=3:<0.02:<1]. Tri-deuterated 3b (3b-d3) was generated through over-reduction of E-2b-d1. Since E-2b-d1′ and iso-2b-d1 were not formed, the PcTH of 1b appears to proceed via the SN2 pathway, affording E-2b-d1, rather than the SN2′ pathway followed by migration of the C=C double bond. In contrast, PcTH of 1-phenylallyl alcohol (iso-1b) using CH3OD gave only E-2b-d1 with high deuterium content (Scheme 4, below). This result suggests that iso-1b was reduced in the SN2′ manner and that protic deuterium of methanol-d1 is predominantly used for reduction of iso-1b rather than C–H hydrogen at the methyl group of methanol-d1. In other words, these results indicate that the protic hydrogen (or deuterium) is regioselectively incorporated into sterically less congested carbon, irrespective of the position of the OH group in the starting allylic alcohol. This predictability of regioselectivity in the hydrogenolysis of allylic alcohols is unprecedented .
With the optimized conditions in hand, the substrate scope of the PcTH was investigated. A variety of allylic alcohols with aliphatic or aryl substituents underwent hydrogenolysis (selected examples are shown in Scheme 5). A simple aliphatic allylic alcohol E-1c was selectively converted to the corresponding E-2c (Scheme 5). PcTH of geraniol (E-1d) and nerol (Z-1d) stereospecifically gave the corresponding alkenes, E-2d and Z-2d, respectively (Scheme 5). In these cases, SN2′ reaction leading to iso-2d (15–18%) occurred to a greater degree, compared with the cases of 1b and E-1c. Similar to iso-1b, PcTH of secondary alcohol iso-1d proceeded via apparent SN2′ reaction to give a stereoisomeric mixture of two major products [E-2d and Z-2d (51:25)] (Scheme 5). The reactivity is significantly higher than that in the reported cobalt catalysis, which was not effective for the reduction of E-1d to the corresponding alkene . PcTH of (S)-perillyl alcohol (1e) for 10 h gave (S)-limonene (2e) in 80% yield (Scheme 5). No racemization at the chiral carbon center of 1e took place, and the presence of a C=C bond located distal from the allylic alcohol fragments of Z-1d, iso-1d and 1e was fully tolerated.
Derivation of ene products
Alkene products obtained by the PcTH of allylic alcohols could be directly derivatized to more highly functionalized products by bromination and epoxidation (Scheme 6). PcTH of 1b with Pd/TiO2–methanol followed by bromination with Br2 selectively afforded trans-4b, together with a small amount of cis-4b. Treatment with m-chloroperbenzoic acid (MCPBA) instead of Br2 gave trans-5b stereospecifically.
Application to (S)-(+)-lavandulol synthesis
The high compatibility of PcTH with olefin functionalities encouraged us to attempt a short-step synthesis of (S)-(+)-lavandulol (S-8) via PcTH. Lavandulol is found in several kinds of essential oils and is a sex pheromone of mealybugs , . It is used industrially as an additive for perfumes and flavors . Koo et al.  recently reported a stereoselective synthesis of S-8 from (R)-(–)-carvone (R-6) involving chemoselective but indirect deoxygenation of allylic alcohol S-7 with retention of the homoallylic hydroxyl group (Scheme 7). Their method is superior in terms of ready availability of the starting material to other reported examples of racemic and stereoselective synthesis of lavandulol , , , , , , . However, the key deoxygenation of S-7 to S-8 requires a tedious multi-step protection/deprotection sequence and stoichiometrically generates salt as a waste product .
We expected that our PcTH would cleave the allylic hydroxyl group rather than the homoallylic one of diol S-7 (Scheme 7). Indeed, PcTH of S-7 chemo- and regioselectively gave S-8 as a major product with retention of the homoallylic hydroxyl group, C=C double bonds, and enantiomeric purity of the chiral center. The present single-step photocatalytic transfer hydrogenolysis thus makes it possible to skip the critical but tedious protection/deprotection steps in the previous route from S-7 to S-8 .
We describe a selective photocatalytic transfer hydrogenolysis (PcTH) reaction of allylic alcohols to afford alkenes, promoted by palladium-loaded TiO2 (Pd/TiO2) in CH3OH under near-UV–vis light irradiation at room temperature. The photocatalyst is easily preparable and reusable. The hydrogen source for this reaction is methanol, and consequently, essentially no salt is generated as a wasteful product. PcTH of allyl alcohol opens up a straightforward synthetic route from glycerol to an important platform chemical, propylene. Alkene products of PcTH could be directly derivatized to more highly functionalized, synthetically useful compounds such as epoxides and dibromides. PcTH proceeded in either an SN2 or SN2′ manner, in which a hydrogen atom was preferentially incorporated into the sterically less congested carbon of the allylic alcohol. This predictable regioselectivity is unprecedented among reported hydrogenolysis reactions of allylic alcohols . Further, C=C double bonds and hydroxyl groups at non-allylic positions of substrates were fully tolerated under the reaction conditions. Such unique chemoselectivity should be useful for fine chemical synthesis, as illustrated by a short-step synthesis of (S)-(+)-lavandulol without the need for protection/deprotection steps. The PcTH strategy is expected to be useful for selective, direct and green access to a variety of platform chemicals and fine chemicals from renewable biomass resources and light energy.
Y.T. is grateful for a JSPS Research Fellowship for Young Scientists. This research was supported by “Advanced Catalytic Transformation program for Carbon utilization (ACT-C, Grant# JPMJCR12YJ)”, JST, Japan (to S.S.), Asashi Glass Foundation (to S.S.), MEXT (a Grant-in-Aid for Scientific Research C, General, 26410115, to H.N.) and Tobe Maki Scholarship Foundation (to H.N.). The authors wish to thank Profs. Ryo Kitaura (Nagoya University) and Akihiko Kudo (Tokyo University of Science) for HR-TEM/EDX and XPS measurement, respectively.
M. M. Smith. March’s Advanced Organic Chemistry 7th ed. John Wiley & Sons, NJ (2013). Google Scholar
T. Funabiki, Y. Yamazaki, K. Tarama. J. Chem. Soc. Chem. Commun. 63 (1978). Google Scholar
Reduction method adapted from a standard method to reduce the Ni-Zn/TiO2 catalyst: S. Myagmarjav, H. Takahashi, Y. Sunagawa, K. Yamamoto, N. Sato, E. Matsubara, A. Muramatsu. Mater. Trans. JIM 45, 2035 (2004). CrossrefGoogle Scholar
Y. Ueno, S. Aoki, M. Okawara. J. Chem. Soc. Chem. Commun. 683 (1980). Google Scholar
S. Araki, S. Kambe, K. Kameda, T. Hirashita. Synthesis 751 (2003). Google Scholar
About the article
Published Online: 2017-09-22
Published in Print: 2018-01-26
Citation Information: Pure and Applied Chemistry, Volume 90, Issue 1, Pages 167–174, ISSN (Online) 1365-3075, ISSN (Print) 0033-4545, DOI: https://doi.org/10.1515/pac-2017-0501.http://creativecommons.org/licenses/by-nc-nd/4.0/.