Hydrogenation of unsaturated bonds is an important approach for molecular functionalization [1-5]. The hydrogenation of polar unsaturated bonds is often carried out using homogeneous catalysts; this process generates metal hydride (M–H) species in situ through the heterolysis of hydrogen molecules (H₂) [6, 7]. Heterogeneous catalysts have been widely used in the industry because of their stability, reusability, and easy separation. The hydrogenation of nonpolar unsaturated bonds can be easily achieved on the surface of heterogeneous catalysts and involves the homolysis of H₂ [8]. However, the hydrogenation of polar unsaturated bonds, which proceeds at mild conditions (low H₂ pressure and temperature) using a heterogeneous catalyst, remains challenging. Compared with heterolytic splitting, homolytic H₂ splitting can occur easily on the surface of heterogeneous catalysts.

The chemoselective hydrogenation of quinolines to obtain 1, 2, 3, 4-tetrahydroquinolines (py-THQs), which is an important framework in numerous pharmaceuticals, agrochemicals, dyes, and biologically active natural products [9], has been demonstrated. This process is the most convenient and promising technique to produce py-THQs, because of its high atom utilization and the easy availability of the raw material [10]. Many types of homogeneous [11-19] and heterogeneous [20-35] catalyst systems have been developed for the chemoselective hydrogenation of quinolines. Although heterogeneous catalyst systems have recently attracted considerable attention and are successfully employed in the chemoselective hydrogenation of quinolines (Scheme 1), relatively high H₂ pressures (10–50 atm) and high reaction temperatures (60–150 ℃) are required. Therefore, hydrogenation under much milder conditions is highly desirable. In addition, no evidence was found for the heterolytic H₂ splitting on the surface of heterogeneous catalysts during the hydrogenation of quinolines.

Catalysts based on unsupported nanoporous metal materials are attracting considerable interest because of their "green" potential and sustainable catalytic properties [36, 37]. Nanoporous metals are attractive as heterogeneous catalysts because of their nontoxicity, robustness, high recyclability, and easy recovery. We have previously demonstrated that nanoporous metals are promising green heterogeneous catalysts for liquid-phase organic synthesis [38-41]. In the present work, we investigated the hydrogenation of quinolines using PdNPore as a catalyst. The chemoselective hydrogenation of quinolines proceeded smoothly under relatively low H₂ pressures (2–5 atm) and temperatures (room temperature to 50 ℃), to give py-THQs in satisfactory to excellent yields without leaching of palladium (Scheme 1). The results of a mechanistic analysis show that the chemoselective hydrogenation of quinolines involves heterolysis of H₂.

The starting materials were purchased from Energy Chemicals Co., Ltd. Solvents were purified by standard techniques without special instructions. All other reagents were used as received. ¹H and ¹³C NMR spectra were recorded on either a Varian Inova-400 (400 MHz for ¹H, 100 MHz for ¹³C) or a Bruker Avance II-400 (400 MHz for ¹H, 100 MHz for ¹³C) spectrometer; CDCl₃ was used as a solvent, while tetramethylsilane (TMS) was used as an internal standard. In the following, chemical shifts are reported in ppm downfield (δ) from TMS, whereas the coupling constants J are expressed in Hz. The peak patterns are labeled as follows: s, singlet; d, doublet; t, triplet; q, quartet; m, multiplet. Thin-layer chromatography (TLC) was carried out on SiO₂ (silica gel 60 F254, Merck), and the spots were located with UV light, iodoplatinate reagent, or 1% aqueous KMnO₄. Flash chromatography was carried out on SiO₂ (silica gel 60, 200–300 mesh).

The hydrogenation of quinoline (1a, Table 1) in the presence of PdNPore (5 mol%) was chosen as a model in preliminary tests aimed to optimize the reaction conditions. The results are shown in Table 1. It was expected that a polar and Lewis-basic solvent could assist the heterolytic cleavage of H–H bonds, facilitating the chemoselective hydrogenation of quinoline to py-THQ [24, 42]. Accordingly, the reduction was initially tested in polar and Lewis-basic solvents, dimethylformamide (DMF), and 1, 4-dioxane (Dioxane), under 5 atm H₂ at room temperature. As expected, 2a was obtained as the sole product in moderate yields (entries 1 and 2, 55% and 51%, respectively). The yield of 2a decreased when toluene, a nonpolar solvent, was used (entry 3, 30%). In contrast, the yield of 2a increased to 72% by adding triethylamine (Et₃N), a polar and Lewis-basic additive, to toluene (entry 4). Thus, Et₃N was used as the solvent, and the yield of 2a further increased (entry 5, 80%). The desired product 2a was finally obtained in excellent yield (93%) when acetonitrile (CH₃CN) was used as a solvent (entry 6); in this case, ethylamine (1.5 mmol) was also obtained. These results indicate that CH₃CN was hydrogenated to the corresponding amine under the reaction conditions. The obtained ethylamine may act as a polar and Lewis-basic additive, in addition to CH₃CN itself. The yield decreased when the H₂ pressure and catalyst loading were decreased (entries 7 and 8). The yield of the 2a product also decreased with a shorter reaction time (entry 9). No reaction was observed in the absence of PdNPore catalyst or when the precursor of PdNPore, the Pd₂₀Al₈₀ alloy, was used in the hydrogenation of 1a (entries 10 and 11). Accordingly, we set the standard conditions as follows: PdNPore (5 mol%), 5 atm H₂, and room temperature in CH₃CN. The desired 2a product was separated in low yield, along with a N-ethyl-1, 2, 3, 4-tetrahydroquinoline byproduct, when the palladium/carbon and the Lindlar palladium catalysts were used instead of the PdNPore catalyst (entries 12 and 13). These results clearly indicate that the activity and selectivity of the PdNPore catalyst are higher than those of the two commercially available catalysts.

The catalytic activity of PdNPore was further examined by using it for the hydrogenation of various substituted quinolines under the same conditions described above (Table 2). Monomethyl-substituted quinolines were chemoselectively hydrogenated to give the corresponding py-THQs in good to excellent yields, regardless of the position of the methyl substituent on the benzene or pyridine ring (entries 2–6, 2b–2f, 81%–95%). Compared with the quinolines 1a–1f, the substrate 1g, bearing two methyl groups on the 2- and 6-positions, exhibited relatively low reactivity and required slight heating to undergo the hydrogenation reaction. Product 2g was obtained in 62% yield (entry 7). Similarly, the quinolines 1h and 1i, bearing a strong electron-donating methoxy (MeO) group on the 5- or 6-position, also required slight heating to complete the hydrogenation reaction. The corresponding products 2h and 2i were obtained in good yields (84% and 80%, respectively; entries 8 and 9). Interestingly, 8-hydroxyquinoline (1j) smoothly underwent chemoselective hydrogenation at room temperature to yield the desired product 2j in 83% yield, despite bearing a strong electron-donating hydroxyl (OH) group on the 8-position [43]. This behavior could be attributed to the formation of intramolecular hydrogen bonds in substrate 1j (entry 10). The substrate 1k, an acylated derivative of 1j, exhibited higher reactivity than its precursor 1j, giving the product 2k in excellent yield (entry 11, 92%). Remarkably, the fluoro- and chloro-substituted quinolines 1l–1n smoothly underwent the desired chemoselective hydrogenation reaction under reduced H₂ pressure, to afford the corresponding products 2l–2n in satisfactory to good yields without generating the dehalogenated product (entries 12–14, 2 atm of H₂, 72%–83%). However, debromination was observed when the bromine-containing substrate bromoquinoline (1o) was employed under optimal conditions. The desired hydrogenation product 2o was obtained in 71% yield, along with the debromination product 2a in 8% yield (entry 15). Finally, the 1p–1r substrates, bearing electron-withdrawing and reducible groups such as formyl, ethoxycarbonyl, and aminocarbonyl, were tested. These substrates exhibited higher reactivity in the chemoselective hydrogenation reaction than those bearing electron-donating groups. The products 2p–2r were obtained in 80%–85% yields under reduced H₂ pressure. No reduction of the functional groups was observed, suggesting that further manipulation may produce more useful compounds (entries 16–18).

Several nitrogen-containing heteroaromatic compounds were examined under optimized or slightly modified reaction conditions, to further explore the scope and limitations of this kind of hydrogenation reaction. The results are shown in Scheme 2. The hydrogenation of quinoline N-oxide (1s) proceeded smoothly under the optimized reaction conditions, to yield the corresponding product 2a in excellent yield (Eq. (1), 93%). The hydrogenations of phenanthroline (1t) and phenanthridine (1u) were completed at 50 ℃ and produced 1, 2, 3, 4-tetrahydrophenanthroline (2t) and 5, 6-dihydrophenanthridine (2u) in moderate to good yields (65% and 76%, respectively; Eqs. (2) and (3)).

Leaching experiments were conducted to determine whether or not the PdNPore catalyst leached into the reaction mixture (Scheme 3). After the hydrogenation of quinoline 1a under standard conditions for 8 h, the yield of product 2a was determined to be 20%. Then, half of the reaction solution was transferred to another reaction vessel and continuously stirred for 16 h. No change in the yield of 2a was observed in the absence of a solid catalyst. In contrast, a 91% yield of 2a was obtained in the reaction of the residual solution containing the PdNPore catalyst. This result indicates that no Pd atoms were leached into the reaction solution, therefore confirming that the catalysis proceeded heterogeneously. Inductively coupled plasma-mass spectrometry (ICP-MS) measurements were used to further confirm that no Pd leached from the PdNPore catalyst (detection limit of 0.01 × 10^-6).

To further explore the practical applicability of our method, the chemoselective hydrogenation of quinoline was scaled up to the gram scale, and the result is shown in Scheme 4. When 1.29 g of 1a was treated under standard conditions, 1.16 g of py-THQ product 2a was obtained with 87% yield; this yield was slightly lower than that obtained in the small-scale experiments.

The stability of the PdNPore catalyst was further investigated by recycling tests, and the results are summarized in Table 3. Almost the same yields were observed in the hydrogenation of quinoline (1a) under standard conditions when the PdNPore catalyst was reused for six times. These results clearly indicate that PdNPore is a robust and recyclable catalyst.

Kinetic tests were performed to investigate the effect of ethylamine and py-THQ in the hydrogenation of quinolines. The results are shown in Fig. 1. The yield of the py-THQ 2a product was determined to be 30% in the first 12 h, and the decrease in the starting material 1a showed a very good correlation with the formation of 2a. However, the yield of 2a increased to 93% in the second 12 h, suggesting that the reaction rate increased as the reaction progressed. This result suggests that not only the CH₃CN solvent but also ethylamine and py-THQ, which were produced as the reaction progressed, would act as polar and Lewis-basic additives in this hydrogenation reaction.

The deuterium-hydrogen exchange reaction between H₂ and D₂O was performed under standard conditions to clarify whether or not the hydrogenation involves Pd–H species generated through the heterolytic cleavage of H₂ gas on PdNPore assisted by a polar and Lewis-basic additive, which might act as a weak nucleophile (Scheme 5). The formation of D₂ gas was detected by gas chromatography-mass spectrometry (GC-MS) (Fig. S5); D₂ is considered to be produced through the formation of DH. As shown in Scheme 6, adsorption of H₂ would occur on the surface of PdNPore, and then a base would capture H⁺ ions after the heterolytic splitting of H–H bonds (A), producing [H-base]⁺ and the Pd–H^- intermediate B. If D₂O is present in the reaction medium, the Pd–H^- intermediate B would react with D₂O to release DH gas. Then, the in situ-generated DH would further undergo heterolytic cleavage to generate the Pd–D^- intermediate C, which would subsequently react with D₂O to release D₂ gas.

Deuterium-labeling experiments were performed to further explore the mechanism behind the excellent activity and chemoselectivity in the quinoline hydrogenation provided by the PdNPore catalyst. The deuterium-labeled product 2a-d₃ was obtained in 85% yield, with 100% deuterium incorporation at the 2-, 3-, and 4- three positions, when the hydrogenation of quinoline 1a was performed under D₂ atmosphere under standard conditions (Scheme 7, Eq. (1)). The deuterium-labeled product 2a-d₃ was obtained in 82% yield, with 62% deuterium incorporation at all the 2-, 3-, and 4- positions, when the deuterium-labeling experiment was conducted in the presence of two equivalents of H₂O (Scheme 7, Eq. (2)). Non-deuterated 2a was obtained when a large amount of H₂O (20 equiv.) was added to the reaction mixture (Scheme 7, Eq. (3)). These results indicate that the D-H exchange reaction between D₂ and H₂O proceeded very rapidly to produce H₂ gas; therefore, the hydrogenation of quinoline 1a took place in the presence of both D₂ and H₂. Large amounts of H₂ were generated when a large amount of H₂O (20 equiv.) was added to the reaction mixture, leading to the formation of the non-deuterated product 2a. These results provide clear evidence for the heterolytic H₂ splitting on the surface of PdNPore in the hydrogenation of quinolines.

In conclusion, we have demonstrated for the first time that the unsupported PdNPore is an efficient heterogeneous catalyst for the chemoselective hydrogenation of quinolines. Good to excellent yields of py-THQs were obtained, with excellent chemoselectivities under low H₂ pressures and temperatures. The PdNPore catalyst could be easily recovered and reused several times without any loss of activity. To the best of our knowledge, this is the first report of heterolytic H₂ splitting on the surface of heterogeneous catalysts in the hydrogenation of quinolines. Mechanistic analyses revealed that the heterolytic cleavage of the H–H bonds on PdNPore and the in situ-generated Pd–H species play an important role in the very high chemoselectivity achieved with the present catalyst. The unique features of the PdNPore catalyst (low H₂ pressures and temperatures, high chemoselectivity, good functional group tolerance, high stability, and lack of leaching) make this material more useful for laboratory and industrial applications.