HMPA-Catalyzed Transfer Hydrogenation of 3-Carbonyl Pyridines and Other N-Heteroarenes with Trichlorosilane

A method for the HMPA (hexamethylphosphoric triamide)-catalyzed metal-free transfer hydrogenation of pyridines has been developed. The functional group tolerance of the existing reaction conditions provides easy access to various piperidines with ester or ketone groups at the C-3 site. The suitability of this method for the reduction of other N-heteroarenes has also been demonstrated. Thirty-three examples of different substrates have been reduced to designed products with 45–96% yields.

In recent years, frustrated Lewis base pairs have been proven as efficient catalyst systems for the regio-and chemoselective reduction of pyridines with various reducing agents. In particular, Stephan [25] and Du [26], respectively, reported the metal-free organoborane-catalyzed hydrogenation of pyridines with H 2 . Later on, Du [27] developed a method for the metal-free organoborane catalyzed transfer hydrogenation of pyridines with ammonia boranes. Chang [28] and Wang [29] reported the B(C 6 F 5 ) 3 catalyzed reduction of pyridines with Et 2 SiH 2 and PhMe 2 SiH, respectively ( Figure 1). For the reduction of the 3-carbonyl pyridines, Rueping reported the first example of organocatalytic transfer hydrogenation of 3-carbonyl pyridines with Hantzsch ester for the preparation of chiral 1,4-dihydropyridine (DHP) derivatives [30]. Although a variety of 3-carbonyl piperidines derivatives could be prepared with these methods, there are still some drawbacks: (1) For most of these reactions, a high temperature and (2) high pressure of H 2 was required. Therefore, the search for new methods for the reduction of 3-carbonyl pyridines still remains a challenging task. Trichlorosilane with a Lewis base as an activator is a well-known unsaturated double bond reduction method. Its strength has been well demonstrated by others and ourselves in terms of asymmetric reduction of double bonds [31][32][33][34][35][36][37][38]. However, to the best of our knowledge, the reduction of pyridines by use of this system still presents a great challenge, and no successful protocol has been reported. In our previous study, we found trichlorosilane could activate the imine substrate through coordination of the nitrogen atom. A similar coordination was also found when trichlorosilane and pyridines were added together. Thus, we envisioned that pyridines would be reduced by trichlorosilane with a proper Lewis base activator. Here in, we wish to communicate the results of our study and present a highly effective new method to reduce 3-carbonyl pyridines under an organic Lewis base activated trichlorosilane system.

Results and Discussion
To implement our design, phenyl(pyridin-3-yl)methanone 1a was used as a model substrate to test the catalytic activity of various commercially available Lewis bases. We found that the reduced product could be obtained with 49% and 37%, respectively, when 0.2 equivalent of HMPA (hexamethylphosphoric triamide) or POPh3 were used. However, only a trace amount of reduced product could be detected when DMF (N,N-dimethylformamide) was used, which has been proven as an efficient catalyst for the reduction of C=O and C=N bonds with trichlorosilane. Dichloromethane was the most suitable solvent for this reaction. An 86% yield could be obtained when six equivalents of trichlorosilane were added as a reducing agent. The yield could be increased further to 96% when the reaction was stirred at 25 °C for 24 h. Decreasing the amount of HMPA to 10 mol% and 5 mol%, respectively, both caused a clear drop in the yield. After a careful investigation, we identified the best reaction conditions in which the substrate 1a was reduced with trichlorosilane (6.0 equivalent) under the catalysis of HMPA (20 mol%) in DCM (dichloromethane) at 25 °C for 24 h (Table 1).  Trichlorosilane with a Lewis base as an activator is a well-known unsaturated double bond reduction method. Its strength has been well demonstrated by others and ourselves in terms of asymmetric reduction of double bonds [31][32][33][34][35][36][37][38]. However, to the best of our knowledge, the reduction of pyridines by use of this system still presents a great challenge, and no successful protocol has been reported. In our previous study, we found trichlorosilane could activate the imine substrate through coordination of the nitrogen atom. A similar coordination was also found when trichlorosilane and pyridines were added together. Thus, we envisioned that pyridines would be reduced by trichlorosilane with a proper Lewis base activator. Here in, we wish to communicate the results of our study and present a highly effective new method to reduce 3-carbonyl pyridines under an organic Lewis base activated trichlorosilane system.

Results and Discussion
To implement our design, phenyl(pyridin-3-yl)methanone 1a was used as a model substrate to test the catalytic activity of various commercially available Lewis bases. We found that the reduced product could be obtained with 49% and 37%, respectively, when 0.2 equivalent of HMPA (hexamethylphosphoric triamide) or POPh 3 were used. However, only a trace amount of reduced product could be detected when DMF (N,N-dimethylformamide) was used, which has been proven as an efficient catalyst for the reduction of C=O and C=N bonds with trichlorosilane. Dichloromethane was the most suitable solvent for this reaction. An 86% yield could be obtained when six equivalents of trichlorosilane were added as a reducing agent. The yield could be increased further to 96% when the reaction was stirred at 25 • C for 24 h. Decreasing the amount of HMPA to 10 mol% and 5 mol%, respectively, both caused a clear drop in the yield. After a careful investigation, we identified the best reaction conditions in which the substrate 1a was reduced with trichlorosilane (6.0 equivalent) under the catalysis of HMPA (20 mol%) in DCM (dichloromethane) at 25 • C for 24 h (Table 1). of trichlorosilane were added as a reducing agent. The yield could be increased further to 96% when the reaction was stirred at 25 °C for 24 h. Decreasing the amount of HMPA to 10 mol% and 5 mol%, respectively, both caused a clear drop in the yield. After a careful investigation, we identified the best reaction conditions in which the substrate 1a was reduced with trichlorosilane (6.0 equivalent) under the catalysis of HMPA (20 mol%) in DCM (dichloromethane) at 25 °C for 24 h (Table 1). With the optimized reaction conditions in hand, the scope and limitations for the substrates were investigated. We found a series of phenyl(pyridin-3-yl)methanone derivatives could be reduced under the existing reaction conditions to get the desired product with good yields ( Figure 2). The desired products 2b-2h were obtained with 62-91% yields when the phenyl group of phenyl(pyridin-3-yl)methanone was replaced with other aryl and alkyl substituents. The 3,5-disubstituted pyridines could also be reduced under the existing reaction conditions. The substituents of the 5-position of the pyridine ring could be aryl and alkyl groups. The 5-phenethyl and 5-methyl substituted substrates could be reduced to the desired products 2i and 2j with 61% and 77% yields, respectively. When the 5-position substituent groups of pyridines were Ph, 4-MePh, 4-FPh and 1-napthyl, these pyridines could be reduced to 3,5-disubsitituted piperidines with 82-88% yields. The substrates with a hetero aromatic group at the 5-position of the pyridines are also tolerated. The desired products 2o and 2p were obtained with 73% and 82% yields, respectively, when the thiopen-2-yl and thiopen-3-yl substituted substrates were reduced under the existing reaction conditions. Next, we found that the 3,6-disubstituted substrates could be reduced with moderate yields. The desired reducing products 2q-2s could be obtained with 40-53% yields.
Ethyl nicotinate and its 5-position substituted derivatives are tolerated under the existing reaction conditions. Ethyl nicotinate was reduced to the corresponding product 2t with a 75% yield. The 5-methly substituted ethyl nicotinate was reduced with a 41% yield, and the 5-Ph, 4-MeOPh and 4-FPh substituted ethyl nicotinate were reduced to their corresponding products 2v-2x with 75-76% yields. The 5-BnO substituted ethyl nicotinate could also be reduced to the corresponding product 2y with a 45% yield. In order to confirm the relative configuration of the main product, compounds 5 and 6 were synthesized according the literature, and the trans product was confirmed to be the main product [39]. Pyridine derivatives such as 3-Br, 3-CF3, 3-NO2, and 3-CN substituted pyridines could not be reduced under the existing reaction conditions. However, other N-heteroarenes such as quinoxaline (3a) and 2-phenylquinoxaline (3b) could be reduced to their tetrahydroquinoxaline derivatives 4a [29] and 4b [40] with high yields. The substrates 3c and 3d were partially reduced to the products 4c and 4d. Quinolone (3e) and isoquinoline (3f) were reduced to products 4e [41] and 4f [41], respectively, with moderate yields. 1,5-Naphthyridine (3g) and 1,10-phenanthroline (3h) could only be partially reduced to the products 4g [42] and 4h [43] (Figure 3). All attempts to achieve the fully reduced products of 3c, 3d, 3g and 3h have failed. Pyridine derivatives such as 3-Br, 3-CF 3 , 3-NO 2 , and 3-CN substituted pyridines could not be reduced under the existing reaction conditions. However, other N-heteroarenes such as quinoxaline (3a) and 2-phenylquinoxaline (3b) could be reduced to their tetrahydroquinoxaline derivatives 4a [29] and 4b [40] with high yields. The substrates 3c and 3d were partially reduced to the products 4c and 4d. Quinolone (3e) and isoquinoline (3f) were reduced to products 4e [41] and 4f [41], respectively, with moderate yields. 1,5-Naphthyridine (3g) and 1,10-phenanthroline (3h) could only be partially reduced to the products 4g [42] and 4h [43] (Figure 3). All attempts to achieve the fully reduced products of 3c, 3d, 3g and 3h have failed.  In order to shed light on the mechanistic pathway, in situ NMR analysis was performed. We first found that ethyl nicotinate could form a complex with HSiCl3 when the reaction was set in CDCl3 under the otherwise identical reaction conditions. An obvious chemical shift in the aromatic region could be detected when HSiCl3 was added to the solution of 1t. Besides the signals of the designed products, a group of peaks that matched the intermediate C were also detected at the beginning. The intensity of these peaks would decrease with the addition of water. At the end of the reaction, only the peaks of designed product and HMPA could be detected (Figure 4). In order to shed light on the mechanistic pathway, in situ NMR analysis was performed. We first found that ethyl nicotinate could form a complex with HSiCl 3 when the reaction was set in CDCl 3 under the otherwise identical reaction conditions. An obvious chemical shift in the aromatic region could be detected when HSiCl 3 was added to the solution of 1t. Besides the signals of the designed products, a group of peaks that matched the intermediate C were also detected at the beginning. The intensity of these peaks would decrease with the addition of water. At the end of the reaction, only the peaks of designed product and HMPA could be detected (Figure 4). Molecules 2019, 24, x 6 of 15 Based on the above observations and precedents indicating a stepwise process in the reduction of unsaturated pyridines [27], we proposed a possible mechanistic pathway for the present HMPAcatalyzed reduction of pyridines ( Figure 5). The first step was assumed to be the formation of a HSiCl3 and substrate complex, followed by the hydride attack at the C-4 position to produce the 1,4dihydropyidine intermediate, A, which will transfer to B in the presence of a proton, and then be reduced to D with HSiCl3 under the catalysis of HMPA. The D to E step is the rate-determining step, since only the intermediate D could be detected and isolated from the reaction. The proton which is coming from the hydrochloride that is formed by the hydrolysis of HSiCl3 is important for the existing reaction. Based on the above observations and precedents indicating a stepwise process in the reduction of unsaturated pyridines [27], we proposed a possible mechanistic pathway for the present HMPA-catalyzed reduction of pyridines ( Figure 5). The first step was assumed to be the formation of a HSiCl 3 and substrate complex, followed by the hydride attack at the C-4 position to produce the 1,4-dihydropyidine intermediate, A, which will transfer to B in the presence of a proton, and then be reduced to D with HSiCl 3 under the catalysis of HMPA. The D to E step is the rate-determining step, since only the intermediate D could be detected and isolated from the reaction. The proton which is coming from the hydrochloride that is formed by the hydrolysis of HSiCl 3 is important for the existing reaction. In order to illustrate the synthetic potential of these methodologies, a gram-scale reaction was carried out using 1t as the substrate. Fortunately, the desired product, 2t, was obtained in a yield of 69% ( Figure 6).

Materials and Methods
All solvents used in the reactions were distilled from the appropriate drying agents prior to use. All substrates were analogously prepared and characterized as reported in the Supplementary Materials. Reactions were monitored by thin layer chromatography, using silica gel HSGF254 plates. Flash chromatography was performed using silica gel HG/T2354-92. 1 H-and 13 C-NMR (400 and 100 MHz, respectively) spectra were recorded in CDCl3. The 1 H-NMR chemical shifts are reported in ppm (δ) relative to tetramethylsilane (TMS), with the solvent resonance employed as the internal standard (CDCl3, δ 7.26 ppm). Data are reported as follows: Chemical shift, multiplicity (s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet, br = broad, dd = double doublet), coupling constants (Hz) and integration. 13   In order to illustrate the synthetic potential of these methodologies, a gram-scale reaction was carried out using 1t as the substrate. Fortunately, the desired product, 2t, was obtained in a yield of 69% ( Figure 6). In order to illustrate the synthetic potential of these methodologies, a gram-scale reaction was carried out using 1t as the substrate. Fortunately, the desired product, 2t, was obtained in a yield of 69% ( Figure 6).

Materials and Methods
All solvents used in the reactions were distilled from the appropriate drying agents prior to use. All substrates were analogously prepared and characterized as reported in the Supplementary Materials. Reactions were monitored by thin layer chromatography, using silica gel HSGF254 plates. Flash chromatography was performed using silica gel HG/T2354-92. 1 H-and 13 C-NMR (400 and 100 MHz, respectively) spectra were recorded in CDCl3. The 1 H-NMR chemical shifts are reported in ppm (δ) relative to tetramethylsilane (TMS), with the solvent resonance employed as the internal standard (CDCl3, δ 7.26 ppm). Data are reported as follows: Chemical shift, multiplicity (s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet, br = broad, dd = double doublet), coupling constants (Hz) and integration. 13

Materials and Methods
All solvents used in the reactions were distilled from the appropriate drying agents prior to use. All substrates were analogously prepared and characterized as reported in the Supplementary Materials. Reactions were monitored by thin layer chromatography, using silica gel HSGF254 plates. Flash chromatography was performed using silica gel HG/T2354-92. 1 H-and 13 C-NMR (400 and 100 MHz, respectively) spectra were recorded in CDCl 3 . The 1 H-NMR chemical shifts are reported in ppm (δ) relative to tetramethylsilane (TMS), with the solvent resonance employed as the internal standard (CDCl 3 , δ 7.26 ppm). Data are reported as follows: Chemical shift, multiplicity (s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet, br = broad, dd = double doublet), coupling constants (Hz) and integration. 13 C-NMR chemical shifts are reported in ppm from tetramethylsilane (TMS) with the solvent resonance as the internal standard (CDCl 3 , δ 77.0 ppm). ESIMS (Electron Spray Ionization Mass Spectrometry) spectra were recorded on BioTOF Q (Bruker, Billerica, MA, USA).

Conclusions
In conclusion, we have developed a HMPA-catalyzed metal-free transfer hydrogenation method for the reduction of pyridines. The functional group tolerance of this method provides an easy access method to various piperidines with ester or ketone groups at the C-3 position. The suitability of the method for the reduction of other N-heteroarenes has also been demonstrated. Efforts to extend the application of chiral HMPA derivatives in metal free pyridine reduction with HSiCl 3 are currently underway.
Supplementary Materials: The following are available online: General experimental procedures, substrates, product characterization data, 1 H-and 13 C-NMR spectra substrates and NMR spectra.

Conflicts of Interest:
The authors declare no conflict of interest.