Important role of H2 spillover in asymmetric hydrogenation of quinolines in hybrid systems

Synergy effect of hybrid system in asymmetric hydrogenation

TsDPEN-Rh-Cp*-Cl is an efficient catalyst for the asymmetric transfer hydrogenation (ATH) of quinolines^{23,24,25,26,27,28,29,30,31}, but it is not an efficient hydrogenation catalyst due to the strong coordination ability of Cl^–22. TsDPEN-Rh-Cp*-Cl only showed 42% conversion with 79% ee in the asymmetric hydrogenation (AH) of quinaldine. The addition of Ni/TiO₂ sharply increases the conversion from 42% to 68% and the ee value slightly decreases to 71%, demonstrating the synergy effect of Ni/TiO₂ and TsDPEN-Rh-Cp*-Cl (Fig. 2a). A control experiment was performed with TiO₂ instead of Ni/TiO₂ (Table S1). The ee value remains almost the same before and after the addition of TiO₂, but the presence of TiO₂ causes a decrease in conversion. When the TsDPEN ligand was used to form the hybrid system with Ni/TiO₂ (entry 7, Table S1), only 0.3% conversion was detected, which is consistent with the blanket experiment of Ni/TiO₂ (Table S2). The above results imply that the conversion enhancement in the hybrid system is mainly atrributed to the Ni NPs and Rh complex. With the Ni/Rh ratio in the hybrid system increasing from 1 to 10, the conversion obviously increases from 48% to 92% (Table S1). But the ee value of the product is only 37% with Ni/Rh ratio of 10. Therefore, 5 mol% Ni (Ni/Rh molar ratio of 5) was used in the following discussion.

Comparison of the catalytic performance of Ni/TiO₂, TsDPEN-Rh-Cp*-Cl and the hybrid system in (a) 1 M NaOAc-HOAc buffer at pH=5 and (b) H₂O. (c) Reaction profiles and (d) change of pH for TsDPEN-Rh-Cp*-Cl and the hybrid system. Reaction conditions: 0.5 mmol quinaldine, 25 μmol Ni, 5 μmol Rh, 2 MPa H₂, 0.2 mL MeOH, 40 ^oC, 12 h, 1000 rpm, 1 mL 1 M NaOAc-HOAc buffer at pH=5.

Considering that the support modulates the properties of metal NPs in hydrogenation reactions, different Ni catalysts were explored in the hybrid system (detailed characterizations see Fig. S1, S2 and Table S3). The coupling of Ni/Al₂O₃ and Ni/SiO₂ with TsDPEN-Rh-Cp*-Cl significantly increases the conversion, but only 44% and 60% ee were obtained respectively with Ni/Al₂O₃ and Ni/SiO₂ (Table S2). Under similar reaction conditions, the activity of supported Ni catalysts in quinaldine hydrogenation follows the order of Ni/TiO₂ (0.3% conversion) 2 (1% conversion) 2O₃ (6% conversion) (Table S2). The decreasing sequence of ee value in a hybrid system is opposite to the increasing sequence of activity of supported Ni catalysts. The above results suggest that two essential prerequisites for metal NPs should be satisfied for the hybrid system. Firstly, the hydrogenation of quinolines on metal NPs should be eliminated so as not to affect the selectivity of enantiomers. Secondly, the metal NPs should be capable of hydrogen dissociation.

The catalytic performance of TsDPEN-Rh-Cp*-Cl and the hybrid system as a function of reaction time were investigated under the optimized reaction conditions. The reaction profile of TsDPEN-Rh-Cp*-Cl showed that the conversion increased rapidly in the first 4 h, and then the reaction rate slowed down (Fig. 2c). No obvious increase in conversion was observed after 9 h, which is a common phenomenon in the asymmetric hydrogenation with TsDPEN-M-Cl as catalysts²². This is possibly due to the deactivation of molecular catalysts during the reaction process. As for the hybrid system, the conversion increased gradually to 93% with the extension of reaction time to 24 h and a zero-order kinetic curve appeared from 6 h, which was obviously different from the reaction curve of a typical organometallic complex. No deactivation of the catalyst was observed for the hybrid system, showing the superiority of the synergy between TsDPEN-Rh-Cp*-Cl and Ni/TiO₂.

The influence of the reaction medium on the catalytic performance of the hybrid system was investigated. The coupling of Ni/TiO₂ and TsDPEN-Rh-Cp*-Cl did not notably increase the conversion in AH of quinaldine with H₂O as solvent, but an obvious enhancement in conversion was observed with NaOAc-HOAc buffer as solvent (Fig. 2a and b). Ni/TiO₂ only showed less than 0.3% conversion in the hydrogenation of quinaldine using either H₂O or NaOAc-HOAc buffer as solvent (Table S4). The conversion enhancement in buffer solution suggests that the acidic medium is important for the generation of the synergy effect. As compared with TsDPEN-Rh-Cp*-Cl, the ee value of the hybrid system is slightly reduced in NaOAc-HOAc buffer, but a very low ee value was observed in H₂O (Fig. 2a, b). These results suggest that the interaction between quinaldine and Ni NPs may alter the coordinate configuration of quinaldine and TsDPEN-Rh-Cp*-Cl and that this interaction can be weakened by substrate acidification in weakly acidic medium³². However, the enantioselectivity and activity of both TsDPEN-Rh-Cp*-Cl and the hybrid system diminished at pH of 4 as compared with pH of 5 and 6 in 1 M NaOAc-HOAc (Figure S3a). This is consistent with a previous report that the dissociation of Rh from the chiral ligands at low pH which deteriorates the activity and enantioselectivity²⁵.

The concentration of the buffer solution significantly affects the catalytic performance of the hybrid system (Fig. S3b). By increasing the concentration of NaOAc-HOAc (pH of 5.0) from 0.5 M to 4.0 M, the higher the concentration of NaOAc-HOAc, the more obvious activity enhancement of Ni/TiO₂. A volcano curve was observed in the enantioselectivity of the hybrid system with the maximum of 71% at 1 M NaOAc-HOAc. Both the hybrid system and TsDPEN-Rh-Cp*-Cl exhibited low catalytic activity in both KH₂PO₄-K₂HPO₄ and citric acid-sodium citrate buffer (1 M, pH = 5), accompanied by a drop in ee value (Table S5). The inferior reaction performance could be attributed to the anion effect in the Rh complex catalyzed quinaldine hydrogenation reaction. The similar influence of the component, buffer concentration and pH on the catalytic performance of the hybrid system and TsDPEN-Rh-Cp*-Cl suggests that the AH of quinaldine mainly takes place on TsDPEN-Rh-Cp*-Cl in the hybrid system.

To further verify the synergy effect, Ni/TiO₂ was added to the reaction system after 9 h of the AH of quinaldine catalyzed by TsDPEN-Rh-Cp*-Cl, and then the reaction was carried out for additional 12 h (Fig. 2c, green dots). Surprisingly, the conversion rate increased linearly with the extension of reaction time, and the reaction rate was almost the same as the hybrid system, further confirming the activity promotion effect of Ni/TiO₂. Based on the above observations, we propose that the deactivation of molecular catalyst is possibly due to the loss of H₂ dissociation ability of TsDPEN-Rh-Cp*-Cl, which could be reactivated by the hydrogen species dissociated over Ni/TiO₂. Both hybrid system and TsDPEN-Rh-Cp*-Cl show a slow decline in enantioselectivity as the reaction proceeds, which implies the structure/composition change of TsDPEN-Rh-Cp*-Cl during the hydrogenation process.

Another significant difference between TsDPEN-Rh-Cp*-Cl and the hybrid system lies in the fluctuation of the pH of reaction media during the hydrogenation process (Fig. 2d). The pH value for the two catalytic systems has a common trend in the initial stage, which decreases first and then increases and then remains almost the same. In the initial stage, the solution pH with either TsDPEN-Rh-Cp*-Cl or the hybrid system is almost the same, but the pH of the reaction solution with TsDPEN-Rh-Cp*-Cl is higher than that with the hybrid system after 4 h. The slightly different pH of the reaction solution is not the main reason for the different catalytic systems due to the fact that the hybrid system at pH of 6 is more active than TsDPEN-Rh-Cp*-Cl at pH of 5 (Table S6). Additionally, the addition of Ni/TiO₂ into the TsDPEN-Rh-Cp*-Cl catalyzed system leads to a slight decrease in pH value of the reaction solution (Fig. 2d, green dots), showing the release of proton to the reaction system with the addition of Ni/TiO₂. Based on the above results, it can be speculated that the hydrogen spillover from Ni/TiO₂ to TsDPEN-Rh-Cp*-Cl may take place through the proton-coupled electron transfer pathway¹.

Hydride transfer in hybrid system

To demonstrate the migration of dissociated H* from Ni NPs to TsDPEN-Rh-Cp*-Cl, Ni/TiO₂ was treated with hydrogen under the reaction conditions for 12 h, and TsDPEN-Rh-Cp*-Cl and quinaldine were added after removal of H₂. The reaction mixture was stirred without an extra H₂ supply. In 12 h, 8 μmol of 1,2,3,4-tetrahydroquinaldine was produced with an ee value of 35% (Fig. 3). Under similar conditions, no product was detected after H₂ treatment either in the absence of Ni/TiO₂ or TsDPEN-Rh-Cp*-Cl under N₂ (Fig. 3), confirming the transfer of dissociated H* from Ni NPs to TsDPEN-Rh-Cp*-Cl. It is well acknowledged that the dissociated H* can be stored in the reducible support through hydrogen spillover to further participate in the reduction reaction. Thus, the above results strongly illustrate the activated hydrogen transfer from Ni/TiO₂ to TsDPEN-Rh-Cp*-Cl via hydrogen spillover for quinaldine hydrogenation. Because there is no further H₂ supply, the adsorption of TsDPEN-Rh-Cp*-Cl on Ni/TiO₂ is required to transfer the dissociated H*, which may change the chiral environment of TsDPEN-Rh-Cp*-Cl and reduce the ee value of the hybrid system.

Ni/TiO₂ was treated with H₂ for 12 h, followed by the addition of quinaldine and TsDPEN-Rh-Cp*-Cl after H₂ removal, and the reaction was performed for another 12 h without H₂ supply. Reaction conditions: 0.2 mmol quinaldine, 1.0 eq. Ni, 1.0 eq. Rh, 2 MPa H₂, 0.2 mL MeOH, 40 ^oC, 1000 rpm, 1 mL 1 M NaOAc-HOAc buffer at pH = 5.

To further understand the reaction mechanism, the chemical environment of Ni/TiO₂ and TsDPEN-Rh-Cp*-Cl was further investigated with X-ray photoelectron spectroscopy (XPS) analysis (Fig. 4). The Ni 2p_3/2 spectrum of Ni/TiO₂ treated in NaOAc-HOAc exhibited fitting peaks at 852.1 eV, 855.6 eV, and 860.9 eV corresponding respectively to Ni⁰, Ni^δ+, and shake-up satellite signals^33,34 (Fig. 4a). The treatment of Ni/TiO₂ with H₂ in NaOAc-HOAc buffer has minimal impact on the Ni 2p BE, with only a 0.2 eV and 0.1 eV shift observed for the fitting peaks of Ni⁰ and Ni^δ+, respectively. The Ni⁰/Ni^δ+ ratio before and after the H₂ treatment (0.14 versus 0.12) is almost identical. Additionally, the Rh 3d_5/2 spectrum of TsDPEN-Rh-Cp*-Cl after treatment with NaOAc-HOAc buffer solution showed a fitting peak at 308.1 eV assigned to Rh(III)³⁵, which is slightly lowered than the reported BE of TsDPEN-Rh-Cp*-Cl (309.0 eV)³⁶ (Fig. 4b). This discrepancy may be attributed to the high concentration of OAc^– in the reaction media, which can potentially alter the chemical environment of Rh. Furthermore, a new peak at ~309 eV arose in the Rh 3d_5/2 spectrum of hybrid system and TsDPEN-Rh-Cp*-Cl after H₂ treatment, which implies the formation of Rh(III)-H species, indicating that the two catalytic systems have a similar reaction routine. No change in the valent state of Rh during H₂ treatment process suggests that TsDPEN-Rh-Cp*-Cl catalyzes the reaction through a stepwise H⁺/H^– transfer process outside the coordination sphere with M-H species served as reactive species involving H^– transfer. In this reaction pathway, the activation of quinoline with Brønsted acid was crucial to trigger the hydrogenation.

a Ni 2p spectra of Ni/TiO₂ (a) before and (b) after H₂ treatment in NaOAc-HOAc buffer. b Rh 3 d spectra of TsDPEN-Rh-Cp*-Cl treated (a) with H₂ in NaOAc-HOAc buffer, (b) with H₂ in NaOAc-HOAc in the presence of Ni/TiO₂ and (c) in NaOAc-HOAc buffer. c O 1 s XPS spectra of Ni/TiO₂ in the hybrid system in NaOAc-HOAc buffer (a) after and (b) before H₂ treatment and d Ti 2p XPS spectra of (a) Ni/TiO₂, Ni/TiO₂ in NaOAc-HOAc buffer (b) after and (c) before H₂ treatment.

To investigate whether the dissociated H species spillover onto the TiO₂ support, O 1s and Ti 2p XPS spectra of Ni/TiO₂ in hybrid system were further studied (Fig. 4c, d). The O 1s spectra of Ni/TiO₂ present two peaks with the BEs energies at ~531.5 eV and 529.5 eV³⁷. The lower BE peak could be assigned to the lattice oxygen species (O^2-), and the higher BE peak may be ascribed to the hydroxyls species (OH^–)³⁸. After the H₂ treatment, the ratio of OH^–/O^2- increased from 60:40 to 82:18. The increase in hydroxyl species may be attributed to the dissociation of hydrogen on the surface of Ni NPs, and H⁺ migrates to the TiO₂ support to form a hydroxyl group with the surface oxygen anion³⁹. The Ti 2p_3/2 binding energy (BE) of Ni/TiO₂ appears at 458.7 eV (Fig. 4d (a)), revealing a prominent Ti⁴⁺ species. According to the literature^37,40, the Ti 2p_3/2 BE of H₂-treated M/TiO₂ (M= metal nanoparticles) was observed at 458.3 eV due to the H insertion caused by the hydrogen spillover for TiO₂ supported catalyst. The BE of Ti 2p_3/2 was observed at 458.4 eV for Ni/TiO₂ in buffer solution with and without H₂ (Fig. 4d (b), (c)). The change of Ti 2p_3/2 BE of Ni/TiO₂ in buffer solution is possibly due to the coordination of OAc^{– 40}. Considering the similar BE of Ni/TiO₂ in buffer solution before and after H₂ treatment, it is hard to judge the formation of Ti³⁺ based on the BE of Ti 2p_3/2. Moreover, the results of Bader charge calculation showed that the H bonded with Ti and Ni respectively contains 1.413 electrons and 1.248 electrons, indicating that the H bonded with TiO₂ support carries more electrons, making it more reductive (Fig. S4)^41,42. Hence, the H species is more likely to spillover onto the support and then transfer to the metal complex which catalyzed the AH of the quinaldines. To further prove the hydrogen spillover in the hybrid system, WO₃ was used to diagnose the hydrogen transfer^16,43,44. As shown in Fig. S5, the mixture of Ni/TiO₂ and WO₃ in NaOAc-HOAc buffer turned dark blue after exposure to H₂, indicating the dissociation of H₂ and hydrogen transfer occurred in the buffer solution. The colour changing experiment of WO₃ provides strong evidence for the existence of hydrogen spillover in the hybrid system under reaction conditions.

Mechanism studies to uncover the reaction pathway

To probe the mechanism of quinaldine AH, labeling experiments were performed (Table 1). Using TsDPEN-Rh-Cp*-Cl or hybrid system, deuterated 1,2,3,4-tetrahydroquinaldine was detected with both D₂/solvent and H₂/D-solvent, implying that H₂O was involved in the reaction. The deuterated product distribution is quite different with D₂/solvent and H₂/D-solvent. For TsDPEN-Rh-Cp*-Cl, the major deuterated product is 2-position (63%) together with 3-position (4%), 4-position (40%) and 2-Me position (1%) with D₂/H₂O. When the reaction was carried out in H₂/D-solvent, deuterium incorporation mainly takes place in 3-position (95%) and 2-Me (84%) together with 2-position (5%) and 4-positions (6%). The comparison between Eqs. 1 and 2 suggests that hydride generated with H₂/D₂ is transferred to 2- and 4- position and the protons from the solvent to 3-position. The deviations from theoretical deuterium content suggest the exchange of Rh-H with protons in solution. Notably, the high deuteration at the 2-Me (84%) and 3-position (95%) and low deuteration at 2-position (5%) indicate that H⁺ causes multiple rearrangements in H₂/D-solvent. The deuterated product distributions of the hybrid system and TsDPEN-Rh-Cp*-Cl were similar in D₂/solvent, but the hybrid system showed a slightly lower deuteration at 2- and 4-positions and slightly higher deuteration at 3-position and the 2-Me group (Eq. 1 vs. Eq. 3). The slightly increased deuteration at 3-position suggests the higher amounts of proton generated from reducing atmosphere in the presence of Ni/TiO₂. Similar deuterium products were observed for TsDPEN-Rh-Cp*-Cl and the hybrid system in H₂/D-solvent (Eqs. 2, 4), however, the presence of Ni NPs leads to lower deuteration at 3-position, further confirming that more protons are generated in the presence of Ni/TiO₂. The labeling experiments confirm that the H* species relay from Ni/TiO₂ to quinaldine involves a proton-coupled electron transfer pathway, which was promoted with the assist of water¹.

Based on the above results, as well as the previous reports^45,46,47, the transfer of the dissociated H species from the supported Ni catalyst to TsDPEN-Rh-Cp*-Cl may take place in mixed pathways. The migration of H species from Ni NPs to TiO₂ and then to TsDPEN-Rh-Cp*-Cl was possible based on the combined results of the successful formation of the product by mixing H₂-treated Ni/TiO₂ with reactant under N₂ atmosphere, the XPS analysis and density functional theory (DFT) calculations. The synergistic effect can also be observed in the hybrid system constructed with Ni supported on hydrogen spillover inert supports, like Ni/SiO₂ and Ni/Al₂O₃. These results manifest that the spillover of the H species can also take place without the assistance of support. According to the results of the isotope labeling experiment of the hybrid system, the exchanges between the protons in the solution and the H species occurred (Table 1), demonstrating that the protons in the solution are also involved in the process of hydrogen spillover in the hybrid system. Therefore, the spillover of the dissociated hydrogen from the supported Ni catalyst to the metal complex may also take place with the assistance of H₂O.

A reaction mechanism was proposed for the hybrid system (Fig. 5). The H₂ is dissociated on Ni NPs and the activated H* species transfers to TsDPEN-Rh-Cp*-Cl through a proton-coupled electron transfer pathway with the assistance of TiO₂ and water, and then the ionized Rh complex (Rh³⁺) is reduced to form the active Rh-H species. The quinaldine is activated by the proton in solvent, which subsequently undergoes a 1,4-hydride transfer to afford enamine intermediate. The enamine intermediate isomerizes to yield iminium cation catalyzed by proton in the reaction media, accompanied with multiple enamine isomerization. Following a 1,2-hydride transfer step, the imine cation obtains another hydride from Rh-H species to give the chiral product.

Proposed mechanism for the AH of quinaldine in a hybrid system (ethylenediamine ligands and anions are omitted for clarity).

Substrate scope

The substrate scope of the hybrid system was investigated (Table 2). Regardless of the electronic property of the substituent group at the 6-position, the conversion of the hybrid system is significantly higher than that of TsDPEN-Rh-Cp*-Cl in a series of AH reactions of quinaldine derivatives, indicating that the activity promotion of TsDPEN-Rh-Cp*-Cl by Ni/TiO₂ is a common phenomenon. Although the enantioselectivity of the hybrid system is lower than that of TsDPEN-Rh-Cp*-Cl, the ee value of ~50% can still be obtained.

In summary, the presence of Ni/TiO₂ greatly promoted the activity of TsDPEN-Rh-Cp*-Cl in the asymmetric hydrogenation of quinolines, which is attributed to the H₂ spillover from Ni/TiO₂ to TsDPEN-Rh-Cp*-Cl. It was found that the H* species relay is strongly related with the pH of the reaction medium. Mechanism studies indicate that both the hybrid system and TsDPEN-Rh-Cp*-Cl follow a similar reaction pathway involving a stepwise H⁺/H^– transfer process outside the coordination sphere with M-H species serving as reactive species for H^– transfer. The labeling experiments confirm that the H* species relay from Ni/TiO₂ to TsDPEN-Rh-Cp*-Cl and then to quinaldine involves a proton-coupled electron transfer pathway, which was promoted with the assist of TiO₂ and water. The synergy of heterogeneous catalyst and chiral organometallic complex can combine their respective advantages and provide an efficient approach for asymmetric catalytic chemical synthesis.