Cobalt Precursors for High-Throughput Discovery of Base Metal Asymmetric Alkene Hydrogenation Catalysts

See allHide authors and affiliations

Science  29 Nov 2013:
Vol. 342, Issue 6162, pp. 1076-1080
DOI: 10.1126/science.1243550

Lighter Hydrogenation Catalysts

Enzymes have evolved to use abundant metals such as iron, cobalt, and nickel for redox catalysis. However, synthetic catalysis has generally relied on the rarer, heavier relatives of these elements: ruthenium, rhodium, iridium, palladium, and platinum (see the Perspective by Bullock). Friedfeld et al. (p. 1076) used high-throughput screening to show that the right cobalt precursor can be activated for asymmetric hydrogenation catalysis by using the traditional ligands developed for the precious metals. Zuo et al. (p. 1080) focused on iron, demonstrating a highly effective asymmetric transfer hydrogenation catalyst that uses a ligand rationally designed after careful mechanistic study. Jagadeesh et al. (p. 1073) prepared supported iron catalysts that selectively reduce nitro substituents on aromatic rings to amines, thereby facilitating the preparation of a wide range of aniline derivatives.


Asymmetric hydrogenation of alkenes is one of the most widely used methods for the preparation of single enantiomer compounds, especially in the pharmaceutical and agrochemical industries. For more than four decades, precious metal complexes containing rhodium, iridium, and ruthenium have been predominantly used as catalysts. Here, we report rapid evaluation of libraries of chiral phosphine ligands with a set of simple cobalt precursors. From these studies, base metal precatalysts have been discovered for the hydrogenation of functionalized and unfunctionalized olefins with high enantiomeric excesses, demonstrating the potential utility of more earth-abundant metals in asymmetric hydrogenation.

Asymmetric hydrogenation of prochiral molecules catalyzed by chiral metal compounds has emerged as one of the most effective and widely used synthetic methods for the preparation of single-enantiomer compounds and has matured into a transformative technology for the pharmaceutical and agrochemical industries (13). This methodology has been applied to the commercial synthesis of the pesticide metolachlor (4) as well as routes to the Parkinson’s drug L-DOPA (1, 5) and blockbusters, including Pfizer’s pregablin (6) and Merck’s sitagliptin (79). The approval of nonracemic drug candidates by the U.S. Food and Drug Administration (FDA) has increased from 58% in 1992 to 75% in 2006 (10) and now constitutes a $15 billion industry (11). This history, coupled with ever-shortening periods for market exclusivity for first-in-class drugs (12), suggests that asymmetric hydrogenation will continue to be embraced for drug synthesis and manufacture (9, 10, 13).

Methods for catalyst generation, first reported in 1968 (14, 15), rely on addition of a chiral ligand to a precious metal precursor, which is typically a compound containing rhodium, iridium, or ruthenium. Reliable substitution chemistry coupled with available metal precursors (16) has enabled the synthesis of hundreds of precatalysts that can be rationally tuned to overcome challenges encountered with activity and selectivity. Automation of this process has enabled high-throughput experimentation and resulted in the discovery of optimized catalysts and conditions, culminating in the development of asymmetric hydrogenation on an industrial scale (12).

Despite these successes, the full potential of asymmetric hydrogenation has yet to be realized. Most precious metal catalysts are effective for only a small class of specifically functionalized olefins that engage in two-point binding to achieve high enantioselectivity, thus limiting the number of substrates that give high enantioselectivity (17, 18). Specialized iridium catalysts have been reported by Pfaltz, and later Andersson and Burgess, that overcome some of these limitations, hydrogenating certain unfunctionalized alkenes with high asymmetric induction (1921).

Catalysts that contain earth-abundant elements such as iron and cobalt are attractive because of potential cost and environmental advantages (22). The smaller atomic radii of the first-row transition metals and their distinct electronic structures offer opportunities for a new class of catalysts that may expand the utility of asymmetric hydrogenation. Examples of cobalt-catalyzed alkene hydrogenation have been known for some time by using borohydride salts or hydrogen as the terminal reductant, although their substrate scope is limited, and in some cases, the activities and selectivities are not synthetically viable (2327). More recently, bis(imino)pyridine–(28) and bis(phosphine)amine–ligated (29, 30) cobalt(II) alkyl complexes have been reported that exhibit improved turnover frequencies for the hydrogenation of simple, unactivated terminal and internal alkenes. Our laboratory has reported enantiopure, C1 symmetric variants of the former catalyst class that are effective for the hydrogenation of substituted styrenes with high activity and enantioselectivity (31). Modification of the cobalt precatalysts however, requires multistep synthesis and accompanying purification, presenting obstacles for automated high-throughput catalyst evaluation. Discovery of accessible cobalt precursors capable of efficient in situ catalyst formation with common chiral ligands would enable more rapid optimization. Here, we describe the discovery of a variety of cobalt precursors that, in combination with enantiopure bidentate phosphines identified via high-throughput screening, generate base metal precatalysts that are highly active and selective for the asymmetric hydrogenation of both functionalized and unfunctionalized alkenes.

The success of Co(II) complexes in catalytic hydrogenation prompted exploration of precursors with this oxidation state. Budzelaar and coworkers reported the synthesis of (py)2Co(CH2SiMe3)2 (py, pyridine) and displacement of the pyridine ligands by various tridentate N-donor ligands (32). These observations inspired examination of the scope of this reaction with chiral phosphines to generate single-enantiomer cobalt precatalysts. Given the commercial availability and success of enantiopure bidentate phosphines in asymmetric catalysis (2), 192 commercially available or readily synthesized ligands were assayed for cobalt-catalyzed asymmetric hydrogenation. The hydrogenation of methyl 2-acetamidoacrylate (MAC) was chosen as a representative substrate because of the ubiquity of the unsaturated α-amino acid motif in general, and MAC in particular, in the asymmetric hydrogenation literature (3). In a typical experiment, 10 mole percent (mol %) of the phosphine and cobalt precursor were stirred for 20 min followed by removal of the volatiles under vacuum and addition of a 0.041 M tetrahydrofuran (THF) solution of the alkene, followed by pressurization with 500 psi (34 atm) of H2. Relatively high metal-ligand loadings were used in the initial evaluations to maximize the probability for identifying successful combinations. The enantioselectivity of each reaction was assayed by means of chiral supercritical fluid chromatography (SFC) after 20 hours of reaction time. With (py)2Co(CH2SiMe3)2 as the cobalt source, only a few selected phosphines produced reasonable conversion and enantioselectivity. (R,R)-iPrDuPhos and (py)2Co(CH2SiMe3)2 furnished >99% conversion to product in 91.7% enantiomeric excess, with a preference for the (S) stereoisomer of the alkane.

To explore the possibility that pyridine might inhibit phosphine coordination and hence catalyst formation, we performed a second high-throughput screen using CoCl2 in combination with two equivalents of LiCH2SiMe3 and the 192 phosphines evaluated previously. We conducted the asymmetric hydrogenation of MAC using the conditions and analytical techniques described in the preceding paragraph. Selected phosphine-cobalt combinations that produced high conversion and enantioselectivity are highlighted in Fig. 1. A complete listing of the phosphines and their respective yields and enantioselectivities are reported in table S2. Unlike the experiment with (py)2Co(CH2SiMe3)2, this protocol revealed many phosphine-cobalt combinations that furnished both high conversion and enantioselectivity, indicating that pyridine is deleterious for catalyst formation in some cases. Bidentate phosphines with two-carbon atom linkers proved most effective because (R,R)-QuinoxP*, (R,R)-EtBPE, (R,R)-iPrDuPhos, (R,R)-BenzP*, and (S,S by various tridentate N-donor ligands#x2032;,R,R′)-TangPhos also yielded the alkane in over 90% conversion and 90% enantioselectivity.

Fig. 1 Chiral, bidentate phosphines that in combination with CoCl2/2LiCH2SiMe3 produce high activity and enantioselectivity for the hydrogenation of MAC.

A complete listing of all phosphines evaluated is reported in table S2. Percentages correspond to the conversion to product, whereas those in the parentheses represent the percent enantiomeric excess.

The success of CoCl2 activated with LiCH2SiMe3 and various chiral bidentate phosphines prompted exploration of other air-stable cobalt(II) precursors. Subsequent experiments were conducted with (R, R)-iPrDuPhos as a representative phosphine because of its commercial availability and widespread use in asymmetric catalysis. Each experiment was conducted with 10 mol % each of the cobalt source and the disphosphine in a 0.041 M THF solution of MAC under 500 psi (34 atm) of H2 at 22°C. A solution of the cobalt precursor and phosphine was mixed in methanol followed by solvent removal in vacuo and reconstitution in THF before addition of 20 mol % of activator. Results with various cobalt sources and activators are highlighted in Fig. 2, and a complete listing of results is reported in fig. S23. Various air-stable, commercially available cobalt(II) dihalides and dicarboxylates were examined along with a cobalt(III) precursor, Co(acac)3 (acac, acetylacetonate). A selection of commercially available alkyl and aryl Grigand and zinc reagents were examined as activators. The alkyl lithium reagent LiCH2SiMe3 was also included for comparison with previous screens.

Fig. 2 Cobalt-catalyzed asymmetric hydrogenation of MAC with air-stable, commercially available precursors with (R, R)-iPrDuPhos.

The numerical values are conversion to alkane, and the color code quantifies the enantioselectivity (ee, enantiomeric excess).

The data presented in Fig. 2 identified anhydrous cobalt(II) halides and carboxylates as generally effective base metal sources, suggesting favorable phosphine coordination kinetics. Accordingly, (R, R)-iPrDuPhosCoCl2 and (R, R)-iPrDuPhosCo(OBz)2 have been independently prepared and crystallographically characterized (fig. S8). Among the activators examined, LiCH2SiMe3 and ClMgCH2SiMe3 were broadly effective, which is likely a result of smooth alkylation chemistry of the in situ–generated phosphine cobalt(II) dihalide or bis(carboxylate) compound. Hydrated cobalt sources were also examined and generated active catalysts, alleviating the requirement for stringently anhydrous reagents. Vacuum was applied after addition of the phosphines and likely removed most if not all of the water liberated from the cobalt source. For both anhydrous and hydrated salts, the chlorides were the most broadly effective, generating highly active and enantioselective catalysts with each activator examined. The results of this evaluation demonstrate that active and often stereoselective base metal catalysts can be generated from a diverse range of air-stable cobalt sources and commercially available activators.

Experiments were also conducted to reduce catalyst loadings. The combination of 1 mol % (R,R)-iPrDuPhos with 1 mol % of CoCl2 activated with two equivalents of LiCH2SiMe3 was effective for the hydrogenation of MAC, reaching completion in 24 hours with a slightly reduced enantiomeric excess of 91.4%, favoring the (S)-enantiomer of the alkane. This configuration of the chiral phosphine ligand also yields the (S)-enantiomer of the alkane after hydrogenation of MAC with rhodium catalysts (33). Increasing the total concentration of the reaction solution to 0.41 M and performing the hydrogenation with 58.7 psi (4 atm) of H2 produced 98.7% conversion to (S)-alkane with 98.6% enantiomeric excess. Similarly, 5 mol % of Co(OBz)2 in combination with 5 mol % of (R,R)-iPrDuPhos and 10 mol % of LiCH2SiMe3 produced 98.5% conversion and 98.0% enantiomeric excess [(S)-alkane] over the course of 6 hours at 22°C with 58.7 psi (4 atm) of H2. Impressive results were also obtained with isolated QuinoxP* cobalt compounds. In the presence of 1 mol % of isolated [(R, R)-QuinoxP*]Co(CH2SiMe3)2, the hydrogenation of MAC produced >99% conversion (18 hours) to (R)-alkane in 99% enantiomeric excess. In a second experiment, two equivalents of pyridine were added relative to the cobalt precursor, and the hydrogenation of MAC under otherwise identical conditions proceeded to 99% conversion with 96.5% enantiomeric excess. Likewise, a similar experiment with isolated [(R,R)-iPrDuPhos]Co(CH2SiMe3)2 and two equivalents of added pyridine (based on Co) also produced 99% conversion and slight reduction of the enantioselectivity to 95.7%. In both cases, these results establish little inhibitory effect of added pyridine on cobalt catalysts for MAC hydrogenation.

In addition to MAC, the combination of (R,R)-(iPrDuPhos) with CoCl2 activated with two equivalents of LiCH2SiMe3 was evaluated for the asymmetric hydrogenation of other functionalized alkenes (Fig. 3). Seeking a base metal–catalyzed route to α-amino acids, the hydrogenation of methyl 2-acetamido-3-phenylacrylate was examined and proceeded to >99% conversion and 92.7% enantiomeric excess in 12 hours at 22°C with 500 psi (34 atm) of H2. Attempts to extend this catalytic protocol to hydrogenation of α-acetamidostyrene again produced excellent activity, reaching >99% conversion to alkane in 15 hours at 22°C. However, a modest enantiomeric excess of 40.8% was obtained, prompting a second iteration of catalyst evaluation. From these experiments, (S,S)-EtDuPhos in combination with CoCl2, activated with two equivalents of LiCH2SiMe3, was identified as an effective catalyst combination, reaching 99% conversion in 6 hours with 82.0% enantiomeric excess.

Fig. 3 Asymmetric hydrogenation of amino acid and enamide derivatives with DuPhos cobalt dialkyl precatalysts.

The versatility of the high-throughput screening approach and utility of the various cobalt precursors was evaluated in the hydrogenation of trans-methylstilbene, an alkene with little coordinating functionality. This class of substrate has proven one of the most challenging in asymmetric hydrogenation, and only a select class of specialized iridium catalysts are capable of achieving both high conversion and enantioselectivity (3437). Because of the inhibitory effect of pyridine observed in formation of cobalt catalysts for MAC hydrogenation, CoCl2/2LiCH2SiMe3 in combination with various chiral phosphines was initially examined. Few acceptable metal-ligand combinations were identified (table S6), and as a consequence, we returned to the initial precursor, (py)2Co(CH2SiMe3)2, which is a more soluble and well-defined compound. A set of experiments was conducted with a library containing 192 chiral bidentate phosphines in combination with 10 mol % (py)2Co(CH2SiMe3)2 with 500 psi (34 atm) of H2 at 22°C. A typical procedure involved premixing the phosphine with (py)2Co(CH2SiMe3)2 followed by removal of all volatile components by evacuation.

Presented in Fig. 4 are examples of phosphine-cobalt combinations that furnished the alkane in highest yields and enantioselectivities, a complete listing of results presented in table S6. Fewer phosphines than the MAC hydrogenations proved effective, highlighting the challenge associated with substrates lacking strongly coordinating functionality. Four-carbon-tethered, principally biaxial type phosphines were most effective, in contrast to MAC hydrogenations, for which two-carbon-linked ligands were preferred. The improved performance observed with the (py)2Co(CH2SiMe3)2 precursor is likely a result of more favorable ligand coordination kinetics with a soluble organometallic compound rather than with an anhydrous salt with an extended structure. Biphep derivative SL-A109-2 (38) was identified as an effective ligand for cobalt-catalyzed asymmetric hydrogenation, producing 83.1% conversion and 93.8% enantiomeric excess [(R)-alkane] after 20 hours at 22°C (Fig. 4). In a follow-up experiment, a reduced catalyst loading of 5 mol % resulted in an improved conversion and enantiomeric excess of 99 and 89%, respectively. Although higher catalyst loadings were used than for the known iridium catalysts, these results demonstrate that cobalt offers similar substrate versatility when paired with the appropriate phosphine ligand.

Fig. 4 Bidentate phosphines that in combination with (py)2Co(CH2SiMe3)2 produce highest activities and selectivities for the asymmetric hydrogenation of trans-methylstilbene.

A complete listing of the phosphines evaluated is presented in table S5. Percentages correspond to the conversion to product, whereas those in the parentheses represent the percent enantiomeric excess.

Additional studies were conducted to gain insight into the nature of the catalytically active cobalt species for the asymmetric hydrogenation of both MAC and trans-methyl stilbene. The success of (R,R)-iPrDuPhos prompted exploration of well-defined cobalt dialkyl complexes with this bis(phosphine). Combination of toluene solutions of (py)2Co(CH2SiMe3)2 and (R,R)-iPrDuPhos followed by filtration and recrystallization from pentane at –35°C furnished orange crystals of (R,R)-(iPrDuPhos)Co(CH2SiMe3)2 in 95% yield. Likewise, (R,R)-(iPrDuPhos)CoCl2 and (R,R)-(iPrDuPhos)Co(OBz)2 were isolated as red solids in 97 and 86% yield after addition of the free bis(phosphine) to a THF suspension of CoCl2 or Co(OBz)2, respectively. Magnetic measurements on all three compounds established low-spin Co(II) complexes, with S = 1/2 ground states. The solid-state structures were determined by means of x-ray diffraction and confirm the essentially planar geometries about the cobalt atom (fig. S8). The chloride complex is slightly more distorted than the alkyl derivative, with the two halide ligands displaced above and below the idealized cobalt-phosphine plane. The isolated enantiopure cobalt dialkyl complex was evaluated for the asymmetric hydrogenation of MAC and produced quantitative conversion to the alkane with 96.1% enantiomeric excess in 12 hours at 22°C. As with the in situ generated catalysts, the (S)-enantiomer of the product was favored.

Experiments were also conducted to explore the coordination chemistry of SL-A109-2 with the cobalt precursors. The solid-state structure of a related cobalt(II) dichloride compound, (SL-A101-1)CoCl2, has been determined (39), suggesting a likely 1:1 bis(phosphine):cobalt stoichiometry in the in situ catalytic experiments. To further probe the cobalt-phosphine stoichiometry, the hydrogenation of trans-methylstilbene was conducted with 5 mol % of the SL-A109-2/(py)2Co(CH2SiMe3)2 mixture in the presence of an additional equivalent (relative to cobalt) of SL-A109-2. No measureable effect in activity or enantioselectivity was observed in the catalytic hydrogenation. Addition of (py)2Co(CH2SiMe3)2 to a solution containing an equimolar mixture of both enantiomers of SL-A109 generated an active cobalt catalyst that furnished essentially no (<2%) enantioselectivity. In contrast, premixing one enantiomer of the phosphine, SL-A109-2, with (py)2Co(CH2SiMe3)2 followed by an addition of an equimolar quantity of the opposite bis(phosphine) antipode, SL-A109-1, resulted in only a modest erosion of the enantioselectivity to 76%, favoring the expected (R)-stereoisomer of the alkane. In a related experiment, performing the catalytic hydrogenation with 5 mol % of cobalt precursor and 10 mol % of SL-A109-2 maintained the enantiomeric excess at 87.6%. These results establish a 1:1 [SL-A109]:[Co] stoichiometry and demonstrate that once the bis(phosphine) is coordinated, it is not sufficiently labile on the time scale of the catalytic hydrogenation to enable exchange with free ligand in solution. Repeating the hydrogenation of trans-methylstilbene with 5 mol % of each SL-A109-2 and (py)2Co(CH2SiMe3)2 in the presence of two equivalents of added pyridine lowered the conversion to 17% and the enantioselectivity to 69.8%, demonstrating that unlike in the MAC hydrogenations, pyridine is deleterious for the performance of SL-A109–derived cobalt catalysts used for hydrogenation of largely unfunctionalized olefins (40). These observations motivate future studies toward understanding the coordination chemistry of these species so that catalysts with improved performance may be rationally synthesized.

Supplementary Materials

Materials and Methods

Figs. S1 to S27

Tables S1 to S6

References (4146)

References and Notes

  1. The last number in the name indicates the enantiomer of the ligand. SL-A109-1 corresponds to the (R) enantiomer, whereas SL-A109-2 is the (S) antipode.
  2. Similarly, performing the hydrogenation of trans-methylstilbene with 5 mol % each of SL-A109-2 and (py)2Co(CH2SiMe3)2 without removal of the volatiles, and hence in the presence of two equivalents of pyridine, lowered the conversion and enantioselectivity to 50 and 51%, respectively, indicating that incomplete removal of the volatile byproducts in catalyst generation could also be deleterious to overall performance.
  3. Similarly, performing the hydrogenation of trans-methylstilbene with 5 mol % each of SL-A109-2 and (py)2Co(CH2SiMe3)2 without removal of the volatiles, and hence in the presence of two equivalents of pyridine, lowered the conversion and enantioselectivity to 50 and 51%, respectively, indicating that incomplete removal of the volatile byproducts in catalyst generation could also be deleterious to overall performance.
  4. Acknowledgments: We thank the U.S. National Science Foundation for a Grant Opportunities for Academic Liaison with Industry (GOALI) grant (CHE-1265988) between Princeton and Merck. M.R.F. thanks the National Science Foundation for a Graduate Research Fellowship (DGE-1148900). We also thank Z. Turner, S. Semproni, and G. Margulieux (Princeton University) for assistance with x-ray crystallography and B. Chen, J. Cuff, L. Joyce, Z. Pirzada, W. Schafer, and H. Wang (Merck) for assistance with chiral assays. Metrical parameters for the solid-state structures are available free of charge from the Cambridge Crystallographic Data Centre under the reference numbers CCDC 958430, 958431, and 958432. P.J.C., M.R.F., and J.M.H. along with Princeton University have filed a U.S. patent application (13/838,835) for the compounds disclosed in this work.
View Abstract

Stay Connected to Science

Navigate This Article