Sydney
Williams
a,
Leiming
Qi
a,
Robert J.
Cox
b,
Prashant
Kumar
a and
Jianliang
Xiao
*a
aDepartment of Chemistry, University of Liverpool, Crown Street, L69 7ZD, Liverpool, UK. E-mail: jxiao@liverpool.ac.uk
bChemical Development, AstraZeneca, Silk Road Business Park, SK10 2NA, Macclesfield, UK
First published on 3rd January 2024
Piperidines are one of the most widely used building blocks in the synthesis of pharmaceutical and agrochemical compounds. The hydrogenation of pyridines is a convenient method to synthesise such compounds as it only requires reactant, catalyst, and a hydrogen source. However, this reaction still remains difficult for the reduction of functionalised and multi-substituted pyridines. Here we report the use of a stable, commercially available rhodium compound, Rh2O3, for the reduction of various unprotected pyridines. The reaction only requires mild conditions, and the substrate scope is broad, making it practically useful.
There are multiple approaches for the synthesis of piperidines. Examples include cycloadditions, reductive amination, nucleophilic substitutions, radical cyclisation and reduction of pyridines.8–21 In principle, hydrogenation is a favourable method as it only requires a pyridine substrate, a catalyst and a hydrogen source. However, this remains a challenging reaction and the current reported catalytic methods tend to require harsh conditions and are generally substrate specific.17–19
Hydrogenation of N-heterocycles has been extensively studied using a variety of hydrogen sources and catalysts.18–21 Although using molecular hydrogen carries risks with the use of specialised high pressure equipment, previous research has shown that this type of hydrogenation can effectively reduce activated N-heterocycles with both high yield and high stereoselectivity.21 Issues can arise when reducing unactivated or neutral N-heterocycles, such as pyridines, as the nitrogen can bind to the metal centre and poison the catalyst in addition to the reduced reactivity of the ring toward metal hydride.16 These issues can be alleviated by creating a quarternised pyridinium salt.22,23 Although the hydrogenation of unprotected pyridines is less common with homogeneous catalysts, the use of heterogenous catalysts have been reported.10–12,18,24–29 Some recent examples are shown in Fig. 2. The most commonly used catalysts include palladium on carbon, rhodium on carbon and platinum oxide.12,27–29 However, these reactions tend to require harsh conditions, high catalyst loading and/or long reaction times with varying degrees of success. The substrate scope also tends to be narrow, featuring few functional groups.
Herein we report the use of rhodium(III) oxide (Rh2O3) for the hydrogenation of a wide range of unprotected pyridines with H2 under mild conditions using low catalyst loading. Rh2O3 is one of the commercially most easily available rhodium compounds, and it is stable and easy to handle and store. Whilst rhodium compounds are one of the most widely used catalysts in hydrogenation, it is surprising somewhat that Rh2O3 has rarely been explored in any catalytic reactions, and our literature search found no reported use of Rh2O3 for the hydrogenation of pyridines. Only a few studies of using rhodium oxide as catalysts are available, e.g. for the reduction of carbon monoxide and the hydroformylation of alkenes.30–32
Entry | Catalyst | Solvent | H2 pressure (bar) | Temperature (°C) | Conversionb (%) | cis:trans |
---|---|---|---|---|---|---|
a Reaction conditions (unless otherwise stated): 2,6-lutidine (0.79 mmol.) and catalyst in solvent (1 mL). b Conversion determined by 1H NMR spectroscopy using an internal standard (1,3,5-trimethoxybenzene). c Reaction time 1.5 hours. | ||||||
1 | Rh/C (2 mol%) | HFIP | 50 | 45 | 100 | 96:4 |
2 | Rh/Al2O3 (2 mol%) | HFIP | 50 | 45 | 100 | 96:4 |
3 | Rh2O3 (1 mol%) | HFIP | 50 | 45 | 100 | 97:3 |
4 | Rh2O3 (1 mol%) | HFIP | 10 | 45 | 100 | 95:5 |
5 | Rh2O3 (1 mol%) | H2O | 10 | 45 | 0 | N/A |
6 | Rh2O3 (1 mol%) | THF | 10 | 45 | 0 | N/A |
7 | Rh2O3 (1 mol%) | DCE | 10 | 45 | 17 | N.D |
7 | Rh2O3 (1 mol%) | 1,4-Dioxane | 10 | 45 | 0 | N/A |
9 | Rh2O3 (1 mol%) | Cyclohexane | 10 | 45 | 7 | N.D |
10 | Rh2O3 (1 mol%) | MeOH | 10 | 45 | 100 | 94:6 |
11 | Rh2O3 (1 mol%) | TFE | 10 | 45 | 100 | 96:4 |
12 | Rh2O3 (1 mol%) | TFE | 5 | 45 | 100 | 94:6 |
13 | Rh2O3 (1 mol%) | MeOH | 5 | 45 | 14 | N.D |
14 | Rh2O3 (1 mol%) | TFE | 1 | 45 | 6 | N.D |
15 | Rh2O3 (1 mol%) | TFE | 5 | 40 | 100 | 97:3 |
16 | Rh2O3 (0.5 mol%) | TFE | 5 | 40 | 100 | 97:3 |
17 | Rh2O3 (0.5 mol%)c | TFE | 5 | 40 | 60 | 97:3 |
18 | Nishimura‘s catalyst (1 mol% Rh)c | TFE | 5 | 40 | 54 | 98:2 |
Results from the optimisation are also shown in Table 1, testing a variety of parameters, such as solvent, hydrogen pressure, and temperature. Reducing the pressure from 50 bar to 10 bar did not affect the reaction yield (entry 4). A number of other solvents were screened (entries 5–11), among which trifluoroethanol (TFE) showed the best result in terms of activity and selectivity (entry 11). Reducing the pressure to 5 bar did not affect the activity but there was a slight reduction in the selectivity between cis and trans isomers (entry 12). Although the reaction in methanol also showed excellent activity at 10 bar (entry 10), the drop to 5 bar dramatically reduced the activity (entry 13). It was also possible to reduce the temperature and catalyst loading to 40 °C and 0.5 mol%, respectively (entry 16). We compared Rh2O3 to the commercially available Nishimura‘s Catalyst, which is a combination of Rh2O3 and PtO2 and has been reported for the hydrogenation of aromatic compounds.34–37 Interestingly, when using the same amount of rhodium, Rh2O3 outperformed the Nishimura‘s Catalyst in activity (entries 17 and 18).34 The conditions in entry 16 represent the now optimised reaction conditions for the substrate scope investigations, i.e. 5 bar of H2 with 0.5 mol% of Rh2O3 at 40 °C in TFE.
We went on to examine the substrate scope of the hydrogenation. Although the reduction of 2,6-lutidine (1e) showed 100% conversion within 3 hours, some substrates with substituents at the 3 and 4 positions required a longer time. It was therefore decided that it was more efficient to expand the substrate scope using an overnight reaction. Due to the volatility of most piperidines, we used 1H NMR in the presence of an internal standard to determine the product yield. However, a few compounds were isolated, aimed to validate the NMR yield. The optimal conditions were highly successful in reducing a variety of alkyl pyridines, as shown in Fig. 3. Notably, the reaction was also successful in reducing pyridines with bulky groups that sterically hinder the nitrogen, e.g. 2,4,6-trimethylpyridine (1k) and 2,6-di-tert-butylpyridine (1m). The majority of multisubstituted pyridines were reduced to give mainly the cis product, which is expected for heterogeneous arene hydrogenation (see ESI† for details).38 The diastereoselectivity ratio is included in Fig. 3 for the multisubstituted piperidines that exhibit both cis and trans isomers. The trans isomer would be expected to be the thermodynamically favoured configuration for 2,3- and 2,5-dimethylpiperidine (2f, h) when considering the possible chair conformations of disubstituted piperidines. This may explain why the selectivity is lower when compared to 2,6- (2e) and 2,4-dimethylpiperidine (2g) and indicates that thermodynamics plays a role in determining the product selectivity.39 Although a variety of multi-substituted alkyl pyridines were fully reduced, a lower yield was encountered when the fourth position is sterically hindered, as observed in the case of 3,4- and 3,5-lutidine (2i–j) and 4-tert-butylpyridine (c.f.2r and 2q). This provides some insight into the potential mechanism of the reaction, which was investigated further (vide infra). An issue of chemoselectivity was observed when reducing 2-vinylpyridine (1p) and 4-vinylpyridine (1q), where the alkene was also reduced.
The hydrogenation of pyridines with alcohol groups was also achieved, shown in Fig. 4. The optimised conditions were effective in reducing pyridines with alcohol groups attached directly to the ring but also separated by a carbon chain. It was found that the hydrogenation of 2-hydroxypyridines (3a–c) formed δ-lactams rather than 2-hydroxypiperidines. This was expected as similar occurrences have been reported in the reduction of 2-hydroxypyridine as a result of amide-iminol tautomerization.27,40,41 The reduction of 2-pyridinemethanol (3e) was achieved with a lower yield (76%). This is believed to be caused by possible coordination of the pyridine or the product to the rhodium metal to give 5-membered ring metal species, poisoning the catalyst. An increase in time from 16 to 24 hours did not improve the conversion, neither did an increase in pressure to 10 bar. Further supporting our surmise, 3-pyridinemethanol gave 86% conversion (4f), with the rest of the product subject to an elimination reaction to afford 3-methylpiperidine (2c), and pyridines 3g and 3h were fully reduced during the reaction to give the desired products 4g and 4h. These results indicate that it is not the alcohol functional group itself that affects the hydrogenation.
It was also possible to reduce pyridines with primary and tertiary amines at the 2-position (Fig. 5). For the hydrogenation of substrates with an amino unit at the 2-position, 1H NMR indicated the formation of 3,4,5,6-tetrahydopyridin-2-amines, rather than 2-aminopiperidines. This observation for 2-aminopyridine (5a) is expected as the substrate exhibits tautomerism and the resulting 3,4,5,6-tetrahydopyridin-2-amine (6a) product shows some resistance towards further reduction under the slightly acidic conditions (pKa of TFE: 12.5).42,43 The same is true with pyridines 5b and 5c. These observations are also reported within the literature regarding tertiary amine at the second position of the pyridine,44,45 and are reminiscent of that observed with 3a. However, an elimination side reaction was observed when there was an amine attached to the pyridine ring (e.g.6a and 6b);46 improved yields were obtained at a shorter reaction time. In the case of 2-picolylamine (5d) the conversion is low. This is similar to the reduction of 2-pyridine methanol (3e) and the lower conversion can be explained by the stronger coordinating capability of a free amine. In line with this, complete reduction of Boc-protected 2-picolylamine (5e) was achieved. Unfortunately, 2-nicotinonitrile was not tolerated and the 4-dimethylamino-substituted pyridine (5f) afforded only a low yield, reminiscent of 1r.
The reaction was successful in reducing the pyridine ring when a carboxylic acid, ester or amide is directly attached to the ring (Fig. 6). Both 2- and 3-substituted carbonyl compounds were well tolerated, and di-substituted carbonyl compounds also showed excellent activity to give the corresponding cis product as the major isomer. Pipecolic acid (8a), piperidine-3-carboxamide (8g), 5-(methoxycarbonyl)piperidine-2-carboxylic acid (8h) and dimethyl-piperidine-2,6-dicarboxylate (8j) were isolated with excellent yield, supporting the use of an NMR internal standard to determine yield. There was an issue with the chemoselectivity when reducing 2-acetylpyridine (7d), in which the ketone was also reduced along with the pyridine ring to give 8d.
Poor reduction of phenyl pyridines was observed when the benzene ring is directly attached to the pyridine (Fig. 7). This has been previously seen in the literature.47,48 We were unsure whether this was caused by issues with the individual substrate or catalyst poisoning. As there is low conversion for both 2-phenylpyridine (9a) and 3-phenylpyridine (9b), the issue does not appear to be caused by potential coordination to the rhodium via e.g. C–H activation. The issue could be due to interactions of the aromatic rings with the rhodium atoms, as 2- and 3-phenylpyridines are planar molecules. This is to some degree supported by the complete reduction of 2-benzylpyridine 9d and 2-benzoylpyridine 9c. Although 9c is also a planar molecule, the molecule would be non-planar if the ketone is reduced first. We also attempted the hydrogenation of halogenated pyridines. Unfortunately, the hydrogenation conditions employed resulted in dehalogenation, regardless of halogen or position. This has also been observed within the literature.40,44,45,49
The reduction of pyridines where the 4th position was sterically hindered was difficult, evident from the low yields of 2i, 2j and 2r. This is an indication that the reduction may proceed via hydrogen addition at the 4th position. To shed more light on this conjecture, a few 4-substituted pyridines with different electronic effects were examined in a short reaction time (Fig. 8). As expected, the reduction of 4-methoxypyridine (11a) gave the desired product (12a) in a low yield, presumably because the 4th position has a strong electron donor which would disfavour the hydride addition. When a less electron-donating methyl group was at the same position, a higher product yield was observed (2d), and more remarkably, the electron withdrawing trifluoromethyl analogue afforded full conversion (12b), although sterically 11a is less demanding. Together with the observations made with 1m and 1r, these results suggest that the hydrogenation starts with hydrogen transfer to the 4th position of pyridines. This is common with the reduction of pyridines, pyridinium salts and quinolines under the conditions of homogeneous catalysis.21
Footnote |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d3ob01860a |
This journal is © The Royal Society of Chemistry 2024 |