Gold Nanoparticle-Catalyzed Environmentally Benign Deoxygenation of Epoxides to Alkenes

We have developed a highly efficient and green catalytic deoxygenation of epoxides to alkenes using gold nanoparticles (NPs) supported on hydrotalcite [HT: Mg6Al2CO3(OH)16] (Au/HT) with alcohols, CO/H2O or H2 as the reducing reagent. Various epoxides were selectively converted to the corresponding alkenes. Among the novel metal NPs on HT, Au/HT was found to exhibit outstanding catalytic activity for the deoxygenation reaction. Moreover, Au/HT can be separated from the reaction mixture and reused with retention of its catalytic activity and selectivity. The high catalytic performance of Au/HT was attributed to the selective formation of Au-hydride species by the cooperative effect between Au NPs and HT.


Introduction
The chemoselective reduction of organic compounds is one of the most fundamental reactions in organic chemistry. Among these reductive reactions, the deoxygenation of epoxides into alkenes (Scheme 1) has attracted much attention in both organic synthesis and biological chemistry where, for OPEN ACCESS example, it occurs in the protection-deprotection cycle of carbon-carbon double bonds [1][2][3] and in the production of vitamin K in the human body [4,5]. Scheme 1. Deoxygenation of epoxides to the corresponding alkenes.
Gold nanoparticles (Au NPs) have received much attention in the catalysis field due to their unique and high oxidation ability in various reactions such as oxidation of CO [29] and alcohols [30,31], and epoxidation of alkenes [32,33]. On the other hand, the reduction ability of the Au NP catalysts has not yet been widely studied despite their remarkable activities [34,35]. We have focused on exploring the novel catalytic reactions of Au NPs, and found that supported Au NPs showed unprecedented reduction activities in the deoxygenation of epoxides [36][37][38], N-oxides, sulfoxides, and amides [39].
In this account, we review our recent progress on the unique catalysis of Au NPs supported on hydrotalcite [HT: Mg 6 Al 2 CO 3 (OH) 16 ] (Au/HT) for deoxygenation of epoxides to the corresponding alkenes using alcohols [36] or CO/H 2 O [37] as reductant (Scheme 2). A wide range of epoxides were deoxygenated to the corresponding alkenes with over 99% selectivities. The C=C bonds of the products were not hydrogenated at all. After the reaction, the solid Au/HT catalyst could be easily separated and reused without loss of its activity or selectivity. Furthermore, Au/HT was successfully applicable to an ideal deoxygenation, i.e., the H 2 -mediated deoxygenation of epoxides where only water is formed as a by-product [38].

Characterization of Au/HT
Au/HT was prepared by the deposition precipitation method (see Experimental). From atomic force microscopy (AFM) and transmission electron microscopy (TEM) analyses, the mean diameter (d) of Au NPs on the surface of HT was 2.7 nm with a standard deviation (σ) of 0.7 nm (Figure 1).

Au/HT-Catalyzed Deoxygenation of Epoxides Using Alcohols as a Reductant
Recently, we have found that Au/HT could catalyze the highly efficient aerobic oxidation of alcohols [40] and the lactonization of diols [41]. These results allowed us to predict that if epoxides could be employed as hydrogen acceptors in place of molecular oxygen under the alcohol oxidation conditions, a green catalytic deoxygenation of epoxides with alcohols could be developed (Scheme 3) [36]. Scheme 3. The oxidation of alcohols using O 2 vs. deoxygenation of epoxides using alcohols.

Epoxides Alkenes
To demonstrate the above hypothesis, we carried out the deoxygenation of trans-stilbene oxide (1) using Au/HT with 2-propanol in toluene as the solvent at 110 °C under Ar atmosphere for 4 h. 1 was successfully deoxygenated to give the corresponding alkene trans-stilbene (2) in quantitative yield. Notably, no by-products such as 1,2-diphenylethanol or 1,2-diphenylethane resulting from the hydrogenation of 1 or 2 were formed ( Table 1, Entry 1). During the deoxygenation of 1, the amounts of acetone and water generated were almost equivalent to that of 2. Among the alcohols tested, 1-phenylethanol and benzyl alcohol could also function as reductants (Entries 2 and 3), while the use of an aliphatic primary alcohol such as 1-octanol resulted in lower yield (Entry 4). Next, the effects of inorganic supports of Au NPs were investigated. Au NPs on basic supports of Al 2 O 3 and MgO afforded good to moderate yields of 2 (Entries 5 and 6), whereas non-basic supports like TiO 2 and SiO 2 were not effective (Entries 7 and 8). Other Au compounds like HAuCl 4 , Au 2 O 3 and bulk Au metal did not promote the deoxygenation (Entries 9-11). These results indicate that the combination of Au NPs and a basic support is necessary to achieve the high catalytic activity for the deoxygenation. Other metal NPs on HT were examined for the deoxygenation of 1 ( Figure 2). Among the catalysts tested, Ag/HT also showed excellent catalytic activity for deoxygenation, while other metal NPs did not function as catalysts. To investigate the possibility of the leaching of active metal species from Au/HT into the reaction mixtures, Au/HT was filtered from the reaction mixture at 50% conversion of 1, and treatment of the filtrate with additional stirring under similar conditions did not give any product. Furthermore, inductively coupled plasma atomic emission spectral (ICP-AES) analysis revealed no Au species in the filtrates (detection limit: 0.1 ppm). These results clearly proved that no leaching occurred and the deoxygenation proceeded on the Au NPs on HT.
The outstanding catalytic activity of Au NPs encouraged us to investigate the scope of epoxides in the deoxygenation (Table 2). Various epoxides were efficiently converted into the corresponding alkenes with over 99% selectivity. Both aromatic and aliphatic epoxides could be deoxygenated. Epoxides having ether and hydroxyl groups were also successfully employed as substrates (Entries 9 and 16). Notably, the reducible C=O bonds of epoxyketones were tolerated in the deoxygenation (Entries 14 and 15). cis-Stilbene oxide and cis-2,3-epoxyoctane gave (Z)/(E)-alkene stereoisomers. The selectivities for cis-alkenes were 60% and 50%, respectively (Entries 4 and 13).  After the deoxygenation of 1, the solid Au/HT catalyst could be easily separated from the reaction mixtures and reused with retention of its performance (Entries 2 and 3). TEM images showed that the Au NPs on HT after reuse were similar to fresh Au/HT in average diameter and size distribution and no aggregation of the used Au NPs was observed ( Figure 3). Atomic-scale analysis using Au L-edge EXAFS of Au/HT showed that the intensity of the FT peak derived from the Au-Au shell at 2.8 Å was not changed, supporting the observation that the Au NPs after reuse were of the same size as the originals. These results are consistent with the high durability of Au/HT in the recycling experiments.
Au/HT was also applicable in a preparative scale reaction (Scheme 4). Thus, the deoxygenation of 20 mmol of 1 successfully proceeded to afford 2 with 95% isolated yield in 2 In separate experiments, the use of d-benzhydrol [C 6 H 5 CD(OD)C 6 H 5 ] as a reductant for the deoxygenation of 1 afforded 2 and D 2 O with all hydrogen atoms in the alkene product retained. From both these results and the positive effect of basic supports as shown in Table 1 (Entries 1, 5 and 6), we propose the following mechanism as shown in Scheme 5.

Scheme 5.
A plausible reaction mechanism for the Au/HT-catalyzed deoxygenation of an epoxide through the cooperation of the Au NPs with basic sites (BS) of HT.
A basic site (denoted as BS) of HT abstracts the H + from the hydroxyl group of the alcohol to promote the formation of an Au-alcoholate species which subsequently forms an Au-hydride species and the corresponding carbonyl compound through β-hydride elimination [42]. The Au-hydride species and H + attack an epoxide, providing an alkene and H 2 O. The distinguished deoxygenation activities of Au NPs from those of other metal NPs should be attributed to the reactivity toward an epoxides (III→I) because the other metal NPs of Cu, Ru, Pd and Rh, which can form metal-hydride species from the reaction with the alcohol (I→III), do not deoxygenate the epoxide.

Deoxygenation of Epoxides with CO/H 2 O
We have previously reported that Rh carbonyl species can deoxygenate nitro compounds in the presence of amines as bases under water-gas shift reaction conditions (CO + H 2 O → CO 2 + H 2 ) [43]. In this reaction, CO and H 2 O react with the Rh carbonyl species and an amine to form a Rh-hydride species that is active for the reduction of nitro compounds. The formation of the Rh-hydride species in cooperation with amines under water-gas shift reaction conditions inspired us to develop an alternative catalytic deoxygenation system using Au/HT. Namely, we proposed that an active Au-hydride species for the deoxygenation of epoxides can be formed through the cooperative effect of HT as a base under water-gas shift reaction conditions (Scheme 6). Thus, the attack of H 2 O on the basic sites of HT to CO adsorbed on Au NPs generates [Au-COOH] − , followed by the elimination of CO 2 to give the Au-hydride species and H + , which then act in concert to deoxygenate epoxides to alkenes [37].  Based on the above hypothesis, we carried out the deoxygenation of styrene oxide (1a) under watergas shift reaction conditions in the presence of Au/HT. Styrene (2a) was quantitatively obtained as the sole product under atmospheric pressure CO in water at room temperature ( Table 3, Entry 1). Various epoxides tested in the Au/HT-2-propanol system were also reactive under water-gas shift conditions (Table 4). Compared with the Au/HT-2-propanol system, this Au/HT-CO/H 2 O system can promote the deoxygenation of epoxides under mild and convenient reaction conditions, e.g., at room temperature, 1 atm of CO, and in the absence of organic solvents.   The solvent effect on the deoxygenation is shown in Scheme 7. Notably, water was found to provide the highest yield among all the solvents tested despite the water-insoluble nature of 1a. In the case of the epoxy alcohol 2,3-epoxy-3-phenyl-1-propanol (1b), the highest yield and selectivity of cinnamyl alcohol (2b) were obtained in water. After the deoxygenation of 1a, 2a was easily extracted from the reaction mixture by n-hexane and the recovered aqueous phase containing solid Au/HT could be recycled with no decrease in catalytic activity (Table 3, Entries 2 and 3). To gain more insight into the deoxygenation under the water-gas shift conditions, the following control experments were carried out. When the reaction was conducted in the absence of 1a under otherwise identical conditions, H 2 was not detected. The use of D 2 O in place of H 2 O significantly affected the reaction rate for the deoxygenation of 1a with a k H /k D value of 3.9. These results rule out the participation of H 2 in the Au/HT-catalyzed deoxygenation reaction, while indicating that not only CO functions as a sole reductant, but also water takes part in the deoxygenation. An additional experiment using trans-2-octenal in place of 1a revealed that chemoselective reduction occurred to give trans-2-octen-1-ol as the sole product while retaining the C=C double bond of the starting material. From the above results, we believe that an active Au-hydride species is generated in situ from the reaction of H 2 O with CO during the deoxygenation of epoxides. According to the proposed reaction mechanism (Scheme 6), a basic site of HT facilitates the formation of the Au-hydride species through the nucleophilic attack of OH − on the Au-CO species followed by a decarboxylation, which is well evidenced by the positive effect of the additional base of Na 2 CO 3 to the Au NPs on the non-basic material of TiO 2 (  (Figure 4). Next, this treated Au/HT was exposed to the vapor of 1a, and the band attributed to the Au-H species gradually disappeared. The detection of the Au-H species agreed with recent IR and theoretical studies on Au-H species that predicted a band around 1800 cm −1 [44,45]. These above control experiments are consistent with the proposed reaction mechanism as shown in Scheme 6. The heterolytic H + and Au-hydride species generated in situ on Au/HT deoxygenate the epoxide to form the corresponding alkene and water. After the publication of our Au/HT-CO/H 2 O catalyst system, Cao et al. reported a deoxygenation method using Au/TiO 2 -VS (very small gold NPs on TiO 2 ) under water-gas shift reaction conditions [46]. Au/TiO 2 -VS showed high catalytic activity (TON = 9,600, TOF = 400 h −1 ) in the deoxygenation of styrene oxide in the mixed solvent of acetone with H 2 O under a high pressure CO atmosphere (10 atm).

Selective Deoxygenation of Epoxides Using Molecular Hydrogen
The ideal green methodology for the catalytic deoxygenation of epoxides is to utilize molecular hydrogen (H 2 ) as a reductant due to the formation of non-toxic water as a by-product. However, the use of H 2 often causes non-selective reduction of epoxides to yield alcohols and alkanes as byproducts through hydrogenation of the epoxides and overhydrogenation of the desired alkenes, respectively. Although there are a few successful reports on the selective deoxygenation of epoxides using H 2 , high selectivity for alkenes is restricted to low conversion levels [46] or a limited range of substrates [47]. Therefore, the development of an efficient catalytic system for the selective deoxygenation of epoxides to the corresponding alkenes using H 2 is a challenging task.
With supported Au NPs in hand, the deoxygenation conditions were optimized [38]. When the deoxygenation of 1a using Au/HT was carried out at 80 °C under 1 atm of H 2 for 6 h, 1a was converted to 2a in 97% yield accompanied by the formation of ethylbenzene (3a) as a byproduct through the hydrogenation of the desired product 2a (Table 5, Entry 1). Next, Au NPs on other supports were investigated in the deoxygenation of 1a under similar reaction conditions. Au/CeO 2 and Au/Al 2 O 3 had lower selectivities for 2a, which caused hydrogenation of 2a (Entries 4 and 5) [48]. Interestingly, Au/TiO 2 gave 2a with over 99% selectivity, though the conversion of 1a was much lower than that of Au/HT (Entry 6). Au/SiO 2 did not exhibit any catalytic activity toward this reaction (Entry 7). Remarkably, when the reaction was carried out at 60 °C for 8 h, Au/HT produced 2a in over 99% yield without formation of any side products (Entry 2). Moreover, the C=C bond of 2a was completely intact when the reaction time was prolonged (Entry 3). The Au NP catalysis exhibited different activity from other metal NPs. Ag/HT, Rh/HT, Ru/HT, and Cu/HT did not function as catalysts for this reaction (Entries 10-13). On the other hand, Pd/HT and Pt/HT afforded undesired 4a with over 99% selectivity through the hydrogenation of 1a (Entries 8 and 9).  Next, we conducted further studies on Au/HT and Au/TiO 2 in the hydrogenation of 2a in the presence or absence of p-methylstyrene oxide (1c) (Scheme 8). In the absence of 1c, Au/TiO 2 hydrogenated 2a into 3a. Surprisingly, Au/HT did not show any activity toward the hydrogenation of 2a. Neither Au/HT nor Au/TiO 2 hydrogenated 2a in the presence of 1c. These sharply contrasting results indicate that the hydrogen species generated on Au/HT are active for the deoxygenation of epoxides, but are completely inactive for the hydrogenation of C=C bonds. On the other hand, the high selectivity of Au/TiO 2 for alkenes in the deoxygenation of epoxides (Table 5, Entry 6) is attributable to the preferential adsorption of epoxides over alkenes, which is a similar phenomenon to the previous report that the nitro group in 3-nitrostyrene was reduced by Au/TiO 2 catalyst while retaining C=C bonds [49]. Various Au/HTs with different mean diameters of Au NPs were tested for the deoxygenation of 1a with H 2 ( Figure 5). Different sized Au/HTs could be prepared by varying the concentration of HAuCl 4 solution [38]. Interestingly, larger Au NPs (>3 nm) showed lower catalytic activity and selectivity for the deoxygenation due to the hydrogenation of the C=C bond. The selectivity and yield of 2a increased with decreasing mean diameter of supported Au NPs. From these results, it can be said that immobilizing small Au NPs (<3 nm) is the key to promoting the selective deoxygenation of epoxides to alkenes. The lower selectivity of larger Au NPs indicates that non-polar hydrogen species active for the hydrogenation of C=C bond were easily generated on the surface of larger Au NPs through the homolytic dissociation of H 2 . Au/HT with a mean diameter of 2.7 nm showed high catalytic activity and selectivity for the H 2 -driven deoxygenation of both aromatic and aliphatic epoxides to alkenes ( Table 6). After the reaction, Au/HT could be reused with no loss of its catalytic efficiency (Entries 2 and 3).   Bearing in mind that basic ligands of transition-metal complexes promote heterolytic cleavage of H 2 to give metal-hydride species, we propose a concerted effect between the basic sites of HT and Au NPs. TiO 2 (JRC-TIO-4) were supplied by the Catalysis Society of Japan (Tokyo, Japan). Inductively coupled plasma measurements were performed by SII Nano Technology SPS7800. 1 H and 13 C-NMR spectra were recorded on JEOL JNM-AL400 and JNM-GSX270 spectrometers, respectively. GC (Shimadzu GC-2014) analysis was carried out with a KOCL-3000T and column. High-performance liquid chromatography (HPLC) was performed on a Shimadzu LC-10ADvp: STR ODS-IV. Atomic force microscopy (AFM) and transmission electron microscopy (TEM) micrographs were obtained with a Shimadzu SPM-9700 and Hitachi HF-2000 type microscope, respectively. Au L-edge X-ray absorption spectra were collected in the quick mode and recorded at room temperature in transmission mode at the facilities installed on the BL-01B1 line attached with a Si (311) monochromator at SPring-8, Japan Atomic Energy Research Institute (JASRI), Harima, Japan. Data analysis was performed using the REX 2000 program, ver. 2.5.7 (Rigaku). Fourier transformation (FT) of k 3 -weighted extended X-ray absorption fine structure (EXAFS) data was performed to obtain the radial structural function. FT-IR data were collected in a JASCO FT-IR 410 spectrometer equipped with a MCT detector. Self-supporting pellets were prepared from the sample powders and treated directly in the IR cell allowing thermal treatments under a controlled atmosphere.

Preparation of Au/HT
HT was prepared by the previously reported method [40]. Au/HT was synthesized as follows: HT (1.0 g) was added to 50 mL of an aqueous HAuCl 4 solution (2 mM). After stirring for 2 min, 0.09 mL of aqueous NH 3 solution (10%) was added. The mixture was further stirred at room temperature for 12 h. The obtained slurry was filtered and washed with deionized water, and dried in vacuo to afford HT-supported Au(III) species [Au(III)/HT] as a pale yellow powder. Au(III)/HT (0.9 g) was subsequently stirred in 50 mL of KBH 4 solution (18 mM) under Ar atmosphere at room temperature for 1 h. The solid was filtered and washed with deionized water to give Au/HT as a reddish purple powder.

Conclusions
We discovered that Au/HT catalyzed the highly efficient deoxygenation of epoxides to the corresponding alkenes using various reductants. The obtained TON in the Au/HT-alcohol system was three orders of magnitude greater than that of previous reports. An alternative catalytic deoxygenation system was developed using CO/H 2 O as a reductant. The Au/HT-CO/H 2 O system could promote the deoxygenation of epoxides under mild reaction conditions (water, at room temperature, under 1 atm of CO). Under the water-gas shift reaction conditions, IR experiments revealed the in situ generation of the Au-hydride species. Finally, Au/HT realized an ideal green deoxygenation of epoxides using H 2 as a reductant with water as the sole by-product. During these deoxygenations, no leaching of Au NPs from Au/HT to the reaction mixture occurred and Au/HT could be reused with retention of its high catalytic activity and selectivity. It is notable that the non-polar C=C bonds of products remained intact during these deoxygenations. The key to the above successful deoxygenation is the in situ generated Au-hydride and H + species obtained through the concerted effect of the interface between Au NPs and basic sites of HT. We believe that this Au NP catalysis can be applied to other chemoselective reductions.