Introduction

Optically pure α-substituted propionic acids are widely used as biologically active compounds and important organic intermediates1,2,3,4,5,6 (Fig. 1a). For instance, chiral α-aryl substituted propionic acids represent a very important class of anti-inflammatory and analgesic agents, such as (S)-Ibuprofen, (S)-Naproxen, (S)-Flurbiprofen, and (S)-Ketoprofen, because they are easily absorbed by organisms and have the effect of inhibiting the synthesis of prostaglandins1. Also, there are many other useful molecules, such as chiral α-aryl substituted propionate esters, which have been developed through simple derivations of chiral α-aryl substituted propionic acids. For example, chiral α-aryl substituted propionate esters can be used as an anti-tumor candidate drug2 and potent inhibitors against the inflammatory phenotype of cystic fibrosis3. Additionally, the chiral α-alkyl substituted propionic acid derivatives can also serve as important drugs and synthetic intermediates. Two representative examples are the world-renowned antimalarial drug Artemisinin4 and the widely used chiral intermediate, (S)-Roche ester5. Given the very important applications of these compounds, their asymmetric synthesis has long been a prominent research topic for chemists. Among the reported methodologies, asymmetric hydrogenation, as an efficient and easy-to-industrialize preparation method, is one of the most attractive approaches7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35 (Fig. 1b).

Fig. 1: The study on asymmetric hydrogenation of α-substituted acrylic acids.
figure 1

a Representative molecules bearing chiral α-substituted propionic acid scaffolds. b Transition metal-catalyzed asymmetric hydrogenation of α-substituted acrylic acids.

In the pioneering work on asymmetric hydrogenation, Prof. Knowles discovered that Rh complexes can catalyze the asymmetric hydrogenation of α-substituted acrylic acids20. Although the corresponding product was obtained in only 15% ee, this work opened the door to the study of transition metal-catalyzed asymmetric hydrogenation. Afterwards, through the efforts of Profs. Mathey21, Ding22, Zhang23,24, significant progress has been made in the asymmetric hydrogenation of these types of substrates catalyzed by rhodium complexes, affording the desired products in up to 99% ee and 20000 S/C23. In addition, in 1987 and 2012, two other rare metals ruthenium and iridium hydrogenation catalysts for the preparation of chiral α-substituted propionic acids were successively developed at first by Profs. Noyori25 and Zhou28 respectively (Fig. 1b).

In recent years, asymmetric hydrogenation catalyzed by earth-abundant metals (such as Mn, Fe, Co, Ni, and Cu) has made significant progress36,37,38,39,40,41,42,43,44,45,46,47,48,49. Among them, The latest reports by Profs. Chirik29 and Zhang30 on the cobalt-catalyzed asymmetric hydrogenation of α,β-unsaturated carboxylic acids (including α-substituted acrylic acids) in 2020 represent two excellent examples (Fig. 1b). Recently, nickel catalysts have received much attention when employed in asymmetric hydrogenation, particularly by Hamada50,51, Zhou52,53,54,55, Chirik56, Zhang57,58,59,60,61, our group62,63,64,65,66, and other research groups67,68,69,70,71,72. However, perhaps owing to the lower steric hindrance of disubstituted olefins, which complicates the control of stereoselectivity in the reaction, the majority of studies have concentrated on tri- and tetra-substituted olefins50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72. In recent years, we have discovered that the multiple attractive dispersion interactions (MADI) between catalyst and substrate have a great influence on the activity and selectivity of asymmetric catalytic reactions49,62,63,64,65,66. Thus, we envisioned whether this discovery could be applied to the asymmetric hydrogenation of such disubstituted substrates. Herein, we report an efficient enantioselective nickel-catalyzed hydrogenation of α-substituted acrylic acids to provide the corresponding chiral α-substituted propionic acid products with excellent results. The efficient hydrogenation (10,000 S/C) proceeds through an unusual process of protonolysis of the C-Ni bond. This involves intramolecular proton transfer from the carboxylic acid group of the substrate, to release the product and regenerate the catalyst, which is different from the reported dissociated methods using hydrogen or solvent hydrogen cations (Fig. 1b).

Results

Investigation of reaction conditions

Initially, 2-phenylacrylic acid (1a) was chosen as a model substrate for asymmetric hydrogenation using 1.0 mol% Ni(OAc)2·4H2O and P-chiral (R,R)-QuinoxP* with large electron-donating groups, which can generally improve the activities and enantioselectivities of metal catalysts62. The reaction was conducted under 30 bar H2 at 50 °C in TFE (2,2,2-trifluoroethanol) over 24 h. As a result, the reaction provided a moderate conversion and enantioselectivity (Table 1, entry 1, 70% conv., 76% ee). To our delight, another similar P-chiral ligand, (R,R)-BenzP*, showed excellent reactivity and enantioselectivity (entry 2, >99% conv., 96% ee). However, only 17% conversion was obtained using (R)-BINAP as the ligand (entry 3). Next, when several other commonly used chiral diphosphine ligands and other (single or mixed) solvents were tested, no improvement in this hydrogenation was observed (see Supplementary Table 1 for details). By lowering the temperature to 30 °C, 2a could be obtained with 98% ee, but the reaction did not proceed to completion (entry 4). When the reaction temperature is 30 °C and the H2 pressure is 50 bar, the reaction proceeded with 97% conversion and 97% ee (entry 5). In order to test the catalytic efficiency, we reduced the catalyst loading to 0.2 mol% (S/C = 500), and substrate 1a could still be completely converted to its corresponding product (entry 6). After screening nickel salts (see Supplementary Table 1 for details), the use of 0.20 mol% Ni(OAc)2·4H2O and (R,R)-BenzP* under 30 bar of H2 at 50 °C in TFE was selected as the optimal reaction conditions.

Table 1 Reaction optimizationa

Scope of asymmetric catalysis of α-substituted acrylic acids

Under the optimized reaction conditions, the substrate scope of the α-substituted acrylic acids 1 was explored (Fig. 2). All substrates provided the corresponding products with full conversions and excellent enantioselectivities (90-99.4% ees), with only a few substrates requiring a slight reduction in S/C (5 examples with 250 S/C and 2 examples with 100 S/C) perhaps due to their low solubility or slightly poor activity. When the substituents are located at the ortho-position of the aryl groups (1b-f), the substrates provided the corresponding products (2bf) with better enantioselectivities (97-99.4% ees) than 2a (96% ee). The aryl acrylic acids containing meta- (1g, 1h) and para-substituents (1in) also exhibited excellent stereoselectivities in this hydrogenation (92–96% ees). Then, the substrates bearing disubstituted aryl were also explored. The several typical substrates (1o-s) provided good catalytic results (92–99.2% ees). Next, a range of chain and cyclic alkyl substrates (1t-y) proceeded smoothly and provided the corresponding hydrogenation products with excellent enantioselectivities (90-95% ees). In addition, some heteroaromatic and trisubstituted substrates are not suitable for this catalytic system (see Supplementary Fig. 1 for details). The absolute configuration of product 2f was assigned to be R by X-ray crystallographic analysis (see Supplementary Fig. 8 for details).

Fig. 2: Substrate scopea.
figure 2

Reaction conditions unless otherwise noted: 1 (0.50 mmol), Ni(OAc)2·4H2O/(R,R)-BenzP* (S/C = 500), H2 (30 bar), TFE (2.0 mL), 50 oC, 24 h. bConditions: 1 (0.50 mmol), Ni(OAc)2·4H2O/(R,R)-BenzP* (S/C = 250). cConditions: 1 (0.20 mmol), Ni(OAc)2·4H2O /(R,R)-BenzP* (S/C = 100).

The study of catalyst efficiency and synthetic applications

To further evaluate the activity of the catalyst and applicability of this catalytic system, the catalyst loading was first tested. To our delight, the model substrate 1a, in the presence of a much lower catalyst loading (1/10000), was reacted completely on a gram scale to give 2a with 98% yield and 96% ee, albeit requiring a little longer to complete (Fig. 3a). To the best of our knowledge, this result represents the highest TON (turnover number) for the Ni- catalyzed asymmetric hydrogenation of olefins reported to date36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72. The product can also be further transformed into an anti-tumor candidate2 and potent inhibitors against an inflammatory phenotype3 (Fig. 3a) according to literature procedures. Dihydroartemisinic acid (R)-2z is the key intermediate for preparing Artemisinin, which is one of the most effective drugs for the treatment of malaria4. Thus, the asymmetric hydrogenation of artemisinic acid (1z) was conducted using this catalytic system. Fortunately, 1z was reduced completely to give dihydroartemisinic acid (R)-2z with 98% yield and 99.8:0.2 dr, even at a 0.020 mol% catalyst loading (S/C = 5000), indicating its potential for industrial application (Fig. 3b). As a non-steroid anti-inflammatory drug, (S)-Ibuprofen ((S)-2l) could be obtained via asymmetric hydrogenation of 1l using a catalyst loading of 0.10 mol% (1000 S/C) (Fig. 3c, 99% yield and 96% ee). (S)-Flurbiprofen ((S)-2aa) and (S)-Naproxen ((S)-2ab) could also be obtained with the same 95% ee at 50 and 100 S/C, respectively, due to poor solubility (Fig. 3d, e).

Fig. 3: The study of catalyst efficiency and practical applications.
figure 3

a A gram scale experiment of 1a and the further transformation of the corresponding product 2a. Conditions i: Ni(OAc)2·4H2O (0.050 mol%), (S,S)-BenzP* (0.010 mol%, S/C = 10000), H2 (30 bar), TFE (12 mL), 60 oC, 3 days. b Asymmetric hydrogenation of artemisinic acid (1z) and further synthesis of artemisinin. Conditions ii: Ni(OAc)2·4H2O (0.20 mol%), (R,R)-BenzP* (0.020 mol%, S/C = 5000), H2 (60 bar), TFE (12 mL), EtOAc (6.0 mL), 60 oC, 3 days. c Asymmetric hydrogenation to synthesize (S)-Ibuprofen. d Asymmetric hydrogenation to synthesize (S)-Flurbiprofen. e Asymmetric hydrogenation to synthesize (S)-Flurbiprofen.

The deuterium-labeling experiments

In order to explore a possible mechanism, deuterium-labelling experiments were conducted (Fig. 4). The asymmetric hydrogenation of 1f was conducted in TFE under 30 bar of D2. The product was obtained with >95% deuteration at the α-position of the carboxyl group and <5% deuteration at the methyl group (β-position of the carboxyl group). When CF3CH2OD was employed as the deuterated solvent, the product was obtained with 73% β-D and 7% α-D. These results suggest that the two added hydrogen atoms (α-H/β-H) of the products originate predominantly from H2 and the protic solvent, respectively. This is different from our previous studies concerning α-substituted vinylphosphonates, in which the two added hydrogen atoms of the product originate from H266.

Fig. 4: The deuterium-labeling experiments.
figure 4

The deuterium-labeling asymmetric hydrogenations of 1f was conducted using D2 or CF3CH2OD.

Mechanistic considerations

Combined with the results of the above deuterium-labelling experiments, we further investigated the reaction mechanism through DFT calculations (Fig. 5a, b). The heterolytic cleavage of hydrogen by the Ni complex generates a Ni-H complex, which then coordinates with the C=C bond of substrate 1a to form intermediate IM-1. Next, the migratory insertion of Ni(II)-H to the vinyl group occurs, giving intermediates IM-2S and IM-2R via transition states TS-1S (1.78 kcal/mol) and TS-1R (0.97 kcal/mol), respectively. This step is reversible due to the low activation energy. Afterwards, one molecule of TFE solvent coordinates with IM-2, forming IM-3S and IM-3R. After the protonolysis of the C-Ni bond of IM-3 via intramolecular proton transfer from the carboxylic acid group of the substrate via TS-2, the product 2a is generated and the Ni complex is released. This is the rate-determining and stereo-determining step, and the ΔΔG (2.08 kcal/mol) corresponds to the favored configuration R with 93% ee, matching the experimental data (Fig. 5b).

Fig. 5: Study on the catalytic mechanism.
figure 5

a Proposed catalytic cycle. b DFT calculation of Ni-catalyzed asymmetric hydrogenation of 1a.

Next, reaction order studies were carried out to further verify the rate-determining step (Fig. 6a–c, also see Supplementary Tables 27 and Supplementary Figs. 57 for details). The results indicate that the reaction is first order with respect to the active nickel species or substrate and zero order with respect to H2 pressure.

Fig. 6: The reaction order studies and kinetic equation derivation.
figure 6

a The reaction order of catalyst. b The reaction order of substrate 1a. c The reaction order of hydrogen pressure. d The process of kinetic equation derivation.

Kinetic equation derivation

Based on the above experimental and DFT computation results (Fig. 5), a series of kinetic equations could be derived as follows (Fig. 6d). According to the above experimental results of reaction order studies (Fig. 6a–c), a kinetic equation for this reaction can be written as eq.1. As the generation of Ni-H complex is rapid due to zero order reaction with regards to H2 pressure, this step should not be considered in the kinetic equation. Then, on account of the DFT result, there are two main elementary reactions in the hydrogenation (eq.2 and 3) in which the migratory insertion step (eq.2) is fast and reversible, while the intramolecular proton transfer step is slow and irreversible. These two elementary reactions perfectly satisfy the conditions required for the equilibrium hypothesis, so that eq.4 can be written. Subsequently, according to eq.3, the rate of generation of 2a relies on IM-2 and the solvent TFE, therefore eq.5 is obtained. Substituting eq.4 into eq.5 gives eq.6 after simplification, which is the same as eq.1. The same form of the kinetic equation derived from the DFT calculations and experimental results proves the rationality of our proposed mechanism.

The Control Experiments

Subsequently, in order to further verify the effect of the carboxylic acid group in the catalytic cycle (TS-2), control experiments were conducted by adding acid (AcOH), bases (Na2CO3 and Et3N) or using the corresponding ester substrate (1ac) accordingly (Fig. 7a–c). From Fig. 7a, it can be observed that almost no effect was seen by adding an extra 1.0 or 10.0 equivalents of acetic acid (51-52% conv. and 96% ee). This result suggests that the reaction is not a traditional proton-dissociation mechanism, in which increasing the acidity of the reaction in a protic solvent increases the reaction rate63,65. When 1.0 equivalent of base was added, the reaction activity was reduced significantly (0.50 equiv. Na2CO3 or 1.0 equiv. Et3N, 46% and 49% conv., respectively, Fig. 7b). As the amount of base increases (1.0 equiv. Na2CO3 or 2.0 equiv. Et3N, Fig. 7b), no reaction occurs. This result suggests that our reaction mechanism is also different from the previous reports, in which the substrate carboxylate anion coordinates with the metal center30,69. In addition, the hydrogenation of the esterified substrate 1ac did not occur under standard hydrogenation conditions (<5%, Fig. 7c). These results clearly showed that the proton of the carboxylic acid group of the substrate plays a crucial role in the hydrogenation and is in agreement with the calculated catalytic cycle.

Fig. 7: The control experiments of the Ni-catalyzed asymmetric hydrogenation of 1a and 1ac.
figure 7

a The control experiments by adding acid. b The control experiments by adding bases. c The control experiments by using substrate 1ac.

Hirshfeld partition (IGMH) analysis

In addition, we used an independent gradient model based on Hirshfeld partition (IGMH) analysis to provide the visualization of the secondary interactions between substrates and Ni catalysts species in TS-2R and TS-2S (see Supplementary Fig. 9 for details). Some C-H···H-C and C-H···O interactions are found in both transition states. However, these interactions are significantly weaker in TS-2S compared to TS-2R. Thus, this IGMH analysis suggests that these weak interactions may participate in stabilizing the transition state and enhancing enantioselectivity.

Discussion

In conclusion, an efficient earth-abundant metal nickel-catalyzed asymmetric hydrogenation of α-aryl and alkyl-substituted acrylic acids was developed. The corresponding chiral α-substituted propionic acids were obtained with excellent results (up to 99% yield, 99.4% ee, 10,000 S/C). Several acrylic acid drugs and drug intermediates were efficiently synthesized using this method. In particular, the key intermediate of Artemisinin, (R)-dihydroartemisinic acid, could be obtained with up to 99.8:0.2 dr and 5000 S/C. The mechanistic study suggested that the protonolysis of the C-Ni bond is the rate-determining step, which involves intramolecular proton transfer from the carboxylic acid group of the substrate.

Methods

General procedure for asymmetric hydrogenation of α-substituted acrylic acids

To a hydrogenation tube, Ni(OAc)2·4H2O (0.25 mg, 0.001 mmol), (R,R)-BenzP* (0.28 mg, 0.001 mmol) and the substrate 1 (S/C = 500) were added, and then the mixture was transferred to a nitrogen-filled glovebox. The degassed and anhydrous trifluoroethanol (TFE, 2.0 mL) was added. The reaction was performed with H2 (30 bar) at 50 °C for 24 h. After carefully releasing hydrogen gas, the pure product is obtained by column chromatography (DCM/MeOH). The product was reacted with K2CO3/Me2SO4 or DMAP/DCC/aniline to afford the corresponding methyl ester or amide, whose enantiomeric excess was determined by HPLC with a chiral column.