Design of Chiral Ligands for Asymmetric Catalysis: from C2-Symmetric P,P- and N,N-Ligands to Sterically and Electronically Nonsymmetrical P,N-Ligands

PCA ETR PERSPECTIVE FEATURE SPECIAL

Design of chiral ligands for asymmetric catalysis: From C2-symmetric P,P- and N,N-ligands to sterically and electronically nonsymmetrical P,N-ligands

Andreas Pfaltz* and William J. Drury III Department of Chemistry, University of Basel, St. Johanns-Ring 19, CH-4056 Basel, Switzerland

Edited by Barry M. Trost, Stanford University, Stanford, CA, and approved February 26, 2004 (received for review November 4, 2003)

For a long time, C2-symmetric ligands have dominated in asymmetric catalysis. More recently, nonsymmetrical modular P,N-ligands have been introduced. These ligands have been applied successfully in various metal-catalyzed reactions and, in many cases, have outperformed P,P- or N,N-ligands.

ost asymmetric catalysts lar advantage in mechanistic studies be- that have been developed cause it facilitates analysis of the ligand– so far are metal complexes substrate interactions that may be M with chiral organic ligands. responsible for enantioselection. The chiral ligand modifies the reactivity The design principles that led Dang and selectivity of the metal center in and Kagan to this ligand had a marked such a way that one of two possible en- influence on the course of research in antiomeric products is formed preferen- asymmetric catalysis, and many diphos- tially. Based on this concept, many phine ligands that were introduced sub- metal complexes have been found that sequently were patterned after DIOP. catalyze various reactions with impres- Knowles (7), for example, prepared a sive enantioselectivity. Despite impres- dimeric analogue of one of his previ- sive progress in this field, the design of ously synthesized monophosphines, suitable chiral ligands for a particular which he termed DiPAMP (Fig. 2). application remains a formidable task. Based on this ligand, he developed an The complexity of most catalytic processes industrial catalytic asymmeric process, precludes a purely rational approach an Rh-catalyzed hydrogenation of a based on mechanistic and structural cri- dehydro-amino acid derivative, used as teria. Therefore, most new chiral cata- the key step in the production of 3,4- lysts are still found empirically, with dihydroxy-L-phenylalanine (L-Dopa). chance, intuition, and systematic screening Since then, the concept of C2 symmetry all playing important roles. Nevertheless, has led to many further highly efficient for certain reactions such as Rh-catalyzed diphosphines, such as BINAP (8) and hydrogenation (1, 2) or Pd-catalyzed Fig. 1. Privileged ligand structures. DuPhos (9), and it has been applied suc- allylic substitution (3, 4), the mechanism cessfully to other ligand classes with coor- is known, allowing at least a semira- C -symmetric ligand with two equivalent dinating N or O atoms (Fig. 1) (5, 10–13). tional approach to catalyst development. 2 We, too, were attracted by the advan- Moreover, useful general concepts have P atoms was to reduce the number of possible isomeric metal complexes, as tages and aesthetics of C2 symmetry been developed during the last three when we introduced the semicorrins decades that greatly facilitate the devel- well as the number of different substrate–catalyst arrangements and reac- (Fig. 3) as a new type of ligand. To our opment of new chiral ligands, even in delight, these ligands gave excellent results the absence of mechanistic information. tion pathways, when compared with a nonsymmetrical ligand. This conse- in the Cu-catalyzed cyclopropanation of Some of these concepts are described in quence of C symmetry can have a benefi- olefins and Co-catalyzed conjugate re- the following sections, mainly from the 2 ␣ ␤ cial effect on enantioselectivity because duction of , -unsaturated carboxylic perspective of our own research. ␲ the competing less-selective pathways acid derivatives (10). The planar - system and the two five-membered rings C2-Symmetric Ligands are possibly eliminated. Because fewer reaction intermediates must be taken confine the conformational flexibility, Of the thousands of chiral ligands pre- into account, C2 symmetry is of particu- which simplifies a prediction of the 3D pared so far, a relatively small number structure of semicorrin metal complexes. of structural classes stand out because The two substituents at the stereogenic of their broad applicability. These ‘‘priv- centers are positioned in close proximity ileged ligands,’’ as they may be called to the coordination site, and they shield (5), allow high levels of enantiocontrol the metal center from two opposite di- in many different metal-catalyzed reactions. A survey of their structures reveals that, at first sight, a surprisingly large This paper was submitted directly (Track II) to the PNAS number of them possess C2 symmetry ofﬁce. (Fig. 1). Abbreviation: PHOX, phosphinooxazoline. The C2-symmetric ligand DIOP (Fig. *To whom correspondence should be addressed. E-mail: 2) was introduced by Dang and Kagan Fig. 2. Kagan’s DIOP and Knowles’ DiPAMP [email protected]. in 1971 (6). The reason for choosing a ligands. © 2004 by The National Academy of Sciences of the USA

www.pnas.org͞cgi͞doi͞10.1073͞pnas.0307152101 PNAS ͉ April 20, 2004 ͉ vol. 101 ͉ no. 16 ͉ 5723–5726 Downloaded by guest on September 30, 2021 Fig. 3. Structure of a semicorrin metal complex.

rections (Fig. 3). Therefore, these sub- Fig. 6. Structure of PHOX ligands and x-ray data stituents are expected to have a strong of an allyl–Pd complex. direct effect on a reaction taking place in the coordination sphere. Variation of the semicorrin structure illustrated by the industrial synthesis of led to analogous bisoxazoline (BOX) the important herbicide metolachlor ligands; Fig. 1).† Ligands of this type (Fig. 5) (23). The key step, an Ir- were reported independently by research Fig. 4. Achiwa’s hydrogenation studies with de- catalyzed imine hydrogenation, was groups in 1990–1991 (14–21), and since symmetrized DIOP. improved dramatically by systematic then they have been established as one variation of the individual substituents of the most versatile ligand classes for at the P atoms. In contrast to applica- asymmetric catalysis (10, 11). Bisoxazo- ordinated substrate is primarily steric in tions in the pharmaceutical sector, ex- trans lines are attractive because various de- nature, whereas P exerts mainly an tremely high enantioselectivity was not rivatives can be readily prepared from electronic effect (electronic effects of a required here. The crucial issues in this simple amino alcohols as chiral precur- ligand are transmitted preferentially to case were the turnover number and rate. sors, allowing the ligand to be tailored the trans-coordination site). Similar ar- Under carefully optimized conditions to a specific catalytic process. It is unre- guments can be given for other interme- with the chiral ligand XYLIPHOS, more alistic to expect that one particular li- diates in the catalytic cycle. Because the than a million turnovers and extremely gand will exert perfect enantiocontrol in two phosphine groups influence the re- high rates could be achieved in very many reactions for many different sub- activity and selectivity of the metal cata- concentrated solution, making this com- strates. Therefore, it is crucial that the lyst in different manners, their structures mercial process highly attractive. synthesis is flexible and simple, to allow should be optimized individually, to ob- structural optimization of a ligand for a tain a perfect ligand. From C2-Symmetric Bisoxazolines to particular application. Achiwa and coworkers have illustrated Nonsymmetrical P,N-Ligands this by desymmetrizing the DIOP ligand An even more effective way to desym- (Fig. 4). Indeed, replacing one of the C2 Versus C1 Symmetry metrize a P,P- or N,N-ligand is to switch diphenylphosphine units by a more elec- Although the concept of C2 symmetry to mixed donor P,N-ligands, because of has been very successful, there is no tron-rich dicyclohexylphosphino group the distinctly different characteristics of resulted in a significant increase of both ␲ fundamental reason why C2-symmetric a ‘‘soft’’ P-ligand with -acceptor prop- ligands should necessarily be superior to catalyst activity and enantioselectivity. erties and a ‘‘hard’’ N-ligand acting pri- their nonsymmetrical counterparts. In Although this strategy of desymmetriz- marily as an ␴-donor.‡ This line of ing C -symmetric ligands appears to be fact, efficient nonsymmetrical ligands 2 thought led us and, independently, straightforward, it does not necessarily have been found that in some reactions Helmchen (24) and Williams (42) to a guarantee an improvement of the per- give even higher enantioselectivities new, highly versatile class of ligands, the formance of the catalyst. If catalytically than the best C -symmetric ligands. phosphinooxazoline (PHOX) ligands 1 2 active, isomeric metal complexes are Moreover, convincing arguments can be (Fig. 6). formed in which the coordinating atoms made for certain reactions as to why The reaction that we were investigat- (e.g., Pcis and Ptrans of ligand B; Fig. 4) nonsymmetrical ligands with two elec- ing at that time, an enantioselective Pd- have exchanged positions, all efforts to tronically and sterically divergent coordi- catalyzed allylic substitution, is shown in individually optimize the coordinating nating units should, in principle, permit Fig. 7 (3, 4). Starting from a racemic groups are futile. However, if this prob- more effective enantiocontrol than C - mixture of allylic acetates 2 and ent-2, 2 lem can be avoided (which is often diffi- symmetric ligands. containing two identical substituents at cult), the results may be spectacular, as Asymmetric hydrogenation with Rh the allylic termini, an allyl complex 3 is catalysts provides an instructive exam- formed. Because both allylic acetate 2 ple. As pointed out by Achiwa and co- and ent-2 are converted to the same inter- workers (22), the intermediates in the mediate, the stereochemical information is catalytic cycle are nonsymmetrical and, lost in this step. It is the subsequent consequently, the two phosphine groups step, a nucleophilic addition to the allyl interact with a metal-bound substrate in system, that determines which enantio- an electronically and sterically different mer is formed. Nucleophilic attack at C manner. In a substrate complex (Fig. 4), (1) leads to product 4, attack at C (3) to the interaction between Pcis and the co- the enantiomer ent-4. Consequently, the problem of inducing enantioselectivity is

†Other oxazoline-based ligands have been reported by Brunner and Obermann [Brunner, H. & Obermann U. ‡Application of chiral P,N-ligands was reported by Hayashi (1989) Chem. Ber. 122, 499–507] and Nishiyama et al. et al. [Hayashi, T., Tajika, M., Tamao, K. & Kumada, M. [Nishiyama, H., Sakaguchi, H., Nakamura, T., Horihata, M. Fig. 5. Industrial process for the production of (1976) J. Am. Chem. Soc. 98, 3718–3719]. However, these Kondo, M. & Itoh, K. (1989) Organometallics 8, 846–848]. (S)-metolachlor. ligands were chosen for different reasons.

5724 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0307152101 Pfaltz and Drury Downloaded by guest on September 30, 2021 Fig. 8. Modular construction of PHOX ligands 1.

isopropyl-allyl acetates, only moderate enantioselectivities were recorded for Fig. 7. Mechanism of enantioselective Pd- the ‘‘smaller’’ dimethyl- and diethyl- catalyzed allylic substitution. substituted analogues. In this respect, the PHOX ligands display opposite reactivity when compared with the diphos- equivalent to controlling the regioselect- phine ligands, developed by Trost, that ivity of nucleophilic attack. give excellent results with ‘‘small’’ sub- Initially, we tested bisoxazolines and strates but unsatisfactory enantiomeric related C2-symmetric ligands in this re- excess and yield with 1,3-diphenylallyl action. Although good results were ob- acetate (3, 4). However, because of the tained in certain cases, the scope of this modular nature of the PHOX ligands ligand class proved to be limited. There- (Fig. 8), many different derivatives Fig. 9. Ligand optimization for regioselective fore, we turned our attention to other could be readily synthesized, which and enantioselective allylic substitution. ligands such as the PHOX ligands 1 made it possible to optimize the ligand (Fig. 6) (24). From this ligand class, we structure for certain substrates that ini- hoped to gain an additional means of tially had given unsatisfactory results. In (Fig. 9) (24, 27). The analogous methyl- controlling the regioselectivity, based on this way, Helmchen (24) achieved high allyl ester gave unsatisfactory results. electronic effects. In contrast to allyl enantioselectivities with small substrates However, a further modification of complexes with C2-symmetric N,N- or such as 1,3-dimethylallyl or cylopentenyl P,N-ligands was reported recently, P,P-ligands, complexation by P,N- and cylohexenyl acetate. which resulted in high regioselectivity ligands should result in effective elec- We wondered whether the ligand and enantioselectivity for this substrate tronic discrimination of the allylic termini structure could also be adapted to the as well (28). because of the different trans-influence of problems of regiocontrol and enantio- Although the PHOX ligands were de- the two electronically dissimilar hetero control for monsubstituted allyl esters veloped originally for Pd-catalyzed al- atoms. (For other effective approaches (Fig. 9). In general, Pd catalysts induce lylic substitution, they could be applied to enantiocontrol in Pd-catalyzed allylic the formation of the linear, achiral substitution, see refs. 3 and 4.) Elec- product 8 with high preference over the tronic differentiation of this type had branched, chiral regioisomer 7. [Other been demonstrated by Faller et al. (25) metal catalysts (e.g., Mo, W, and Ir) (stoichiometric reaction of allyl–Mo show opposite regioselectivity in favor complexes with CO and (NO)ϩ as trans of the branched product. Examples of ligands) and by Åkermark et al. (26) chiral catalysts of this type are (NMR studies of allyl–Pd complexes). W–PHOX and Ir–PHOX complexes, as Crystal structure and NMR data con- well as Mo complexes with N,N-ligands firmed that complexation with a PHOX (3, 4).] By rendering the transition state ligand results in a strong electronic dif- more cationic in character, nucleophilic ferentiation of the allylic termini, as re- attack at the substituted allyl terminus flected by the different Pd–C distances should become more favorable. Conse- (Fig. 6) (3, 24). As we had hoped, Pd– quently, we introduced electronegative PHOX complexes were found to be substituents on the P atom to increase highly reactive and selective catalysts, the electrophilicity at the Pd center. providing up to 99% enantiomeric ex- Also, by introduction of sterically de- cess with a range of C-terminal and N- manding P-substituents, we wanted to terminal nucleophiles (Fig. 7). The se- shift the equilibrium between the inter- lectivities could be rationalized by a mediate allyl complexes toward isomer combination of steric and electronic 6. Because of the trans influence of the effects, the preference of nucleophilic P atom, the predominance of this iso- attack at the allylic C atom trans the mer should promote nucleophilic attack phosphino group being a key factor (24). at the substituted terminus. These con- siderations led us to ligands 9 and 10, The Importance of a Modular Ligand which in contrast to the original PHOX Structure ligands 1 induced formation of the Although PHOX ligands gave excellent branched products with high regioselec- Fig. 10. Ir-catalyzed asymmetric hydrogenation results in reactions of diphenyl- and di- tivity and excellent enantioselectivity of oleﬁns.

Pfaltz and Drury PNAS ͉ April 20, 2004 ͉ vol. 101 ͉ no. 16 ͉ 5725 Downloaded by guest on September 30, 2021 successfully to various other metal- Again, the modular nature of the ciency and good to excellent enantio- catalyzed processes, including Heck re- PHOX ligands made it possible to ex- selectivity. actions, Cu-catalyzed 1,4-additions, and tend the application range of these cata- Ru-catalyzed transfer hydrogenation of lysts. Among the many PHOX analogues Conclusion ketones (24). Another reaction class that that we tested, readily available phos- The concept of steric and electronic de- gave very promising results was Ir- phinites of type 12 and 13 proved to be symmetrization has led to a new class of catalyzed asymmetric hydrogenation of the most versatile ligands. Subsequently, chiral ligands, the PHOX ligands 1 and CAC and CAN bonds (29, 30). We we added phosphinoimidazolines, such related P,N-ligands. Because of the modular construction, these ligands thought that Ir–PHOX complexes might as type 14, to our collection of ligands could be adapted to many metal- behave like chiral analogues of the (32, 33), as well as pyridine- and catalyzed reactions and in many cases, Crabtree catalyst, an achiral (tricyclo- quinoline-derived P,N-ligands of type 15 they outperformed P,P- or N,N-ligands. hexylphosphine)(pyridine)Ir(I) complex and 16 (34), all of which gave very Chiral ligands based on other combina- that displays unusually high reactivity promising results. Recently, other re- tions of coordinating atoms, such as the toward trisubstituted and tetrasubsti- search groups have become interested in P,S; P,O; or N,S varieties, have also tuted olefins (31). Pleasingly, after opti- this class of catalysts and have reported been reported and, considering the mization of the catalyst structure and additional variants of Ir complexes with enormous diversity of possible ligand the reaction parameters, good to excel- P,N-ligands (35–40) and, as a further structures of this type, further work in lent enantiomeric-excess values and high modification, oxazoline ligands with a this area seems worthwhile. Although turnover numbers could be obtained in heterocyclic carbene unit instead of a the concept of C2 symmetry will remain the hydrogenation of imines and unfunc- phosphino group (41). Clearly, Ir com- an attractive design principle for new tionalized aryl-substituted olefins (Fig. plexes derived from P,N-ligands ligands, sterically and electronically non- 10). Until now, olefins of this kind could represent a new class of catalysts that symmetrical heterobidentate ligands are not be hydrogenated with high enantio- significantly expands the application likely to play an increasing role in the selectivity at such low catalyst loadings. range of asymmetric hydrogenation. Sev- development of asymmetric catalysis. In this respect, Ir-PHOX complexes eral types of functionalized and nonfunc- This work was supported by the Swiss Na- clearly distinguish themselves from Ru tionalized olefins, for which no suitable tional Science Foundation, the Federal Com- and Rh catalysts that require a polar catalysts were known previously, can mission of Technology and Innovation (KTI), coordinating group near the CAC bond. now be hydrogenated with high effi- and Solvias.

1. Brown, J. M. (1999) in Comprehensive Asymmetric 15. Mu¨ller, D., Umbricht, G., Weber, B. & Pfaltz, A. 30. Schnider, P., Koch, G., Pre´toˆt, R., Wang, G., Catalysis, eds. Jacobsen, E. N., Pfaltz, A. & (1991) Helv. Chim. Acta 74, 232–240. Bohnen, F. M., Kru¨ger, C. & Pfaltz, A. (1997) Yamamoto, H. (Springer, Berlin), Vol. 1, pp. 16. Evans, D. A., Woerpel, K. A., Hinman, M. M. & Chem. Eur. J. 3, 887–892. 121–182. Faul, M. M. (1991) J. Am. Chem. Soc. 113, 726–728. 31. Crabtree, R. H. (1979) Acc. Chem. Res. 12, 331–337. 2. Ohkuma, T., Kitamura, M. & Noyori, R. (2000) in 17. Corey, E. J., Imai, N. & Zhang, H.-Y. (1991) J. Am. 32. Menges, F. Neuburger, M. & Pfaltz, A. (2002) Org. Catalytic Asymmetric Synthesis, ed. Ojima, I. Chem. Soc. 113, 728–729. Lett. 4, 4713–4716. (Wiley–VCH, New York), 2nd Ed., pp. 1–110. 18. Helmchen, G., Krotz, A., Ganz K. T. & Hansen, D. 33. Busacca, C. A., Grossbach, D., So, R. C., O’Brien, 3. Pfaltz, A. & Lautens, M. (1999) in Comprehensive (1991) Synlett, 257–259. E. M. & Spinelli, E. M. (2003) Org. Lett. 5, Asymmetric Catalysis, eds. Jacobsen, E. N., Pfaltz, 19. Hall, J., Lehn, J.-M., DeCian, A. & Fischer, J. 595–598. A. & Yamamoto, H. (Springer, Berlin), Vol. 2, pp. (1991) Helv. Chim. Acta 74, 1–6. 34. Drury, W. J., III, Zimmermann, N., Keenan, M., 833–884. 20. Onishi, M. & Isagawa, K. (1991) Inorg. Chim. Acta Hayashi, M., Kaiser, S., Goddard, R. & Pfaltz, A. 4. Trost, B. M. & Lee, C. (2000) in Catalytic Asym- 179, 155–156. (2004) Angew. Chem. Int. Ed. 43, 70–74. metric Synthesis, ed. Ojima, I. (Wiley–VCH, New 21. Yang, R.-Y., Chen, Y.-H. & Dai, L.-X. (1991) Acta 35. Hou, D.-R., Reibenspies, T. J., Colacot, T. J. York), 2nd Ed., pp. 593–649. Chim. Sin. 49, 1038–1040. & Burgess, K. (2001) Chem. Eur. J. 7, 5. Yoon, P. Y. & Jacobsen, E. N. (2003) Science 299, 22. Inoguchi, K., Sakuraba, S. & Achiwa, K. (1992) 5391–5400. 1691–1693. Synlett, 169–178. 36. Cahill, J. P., Lightfoot, A. P., Goddard, R., Rust, 6. Dang, T. P. & Kagan, H. B. (1971) J. Chem. Soc. 23. Blaser, H.-U. (2002) Adv. Synth. Catal. 344, 17–31. J. & Guiry, P. J. (1998) Tetrahedron: Asymmetry 9, Chem. Commun., 481. 24. Helmchen, G. & Pfaltz, A. (2000) Acc. Chem. Res. 4307–4312. 7. Knowles, W. S. (2003) Adv. Synth. Catal. 345, 3–13. 33, 336–345. 37. Cozzi, P. G., Menges, F. & Kaiser, S. (2003) 8. Noyori, R. (2003) Adv. Synth. Catal. 345, 15–32. 9. Burk, M. J., Gross, M. F., Harper, G. P., Kalberg, 25. Faller, J. W., Chao, K. H. & Murray, H. H. (1984) Synlett, 833–836. C. S., Lee, J. R. & Martinez, J. P. (1996) Pure Appl. Organometallics 3, 1231–1240. 38. Xu, G. & Gilbertson, S. R. (2003) Tetrahedron Chem. 68, 37–44. 26. Åkermark, B., Krakenberger, B. & Hansson, S. Lett. 44, 953–955. 10. Pfaltz, A. (1993) Acc. Chem. Res. 26, 339–345. (1987) Organometallics 6, 620–628. 39. Tang, W., Wang, W. & Zhang, X. (2003) Angew. 11. Ghosh, A. K., Mathivanan, P. & Cappiello, J. 27. Pre´toˆt, R. & Pfaltz, A. (1998) Angew. Chem. Int. Chem. Int. Ed. 42, 943–946. (1998) Tetrahedron: Asymmetry 9, 1–45. Ed. 37, 323–325. 40. Bunlaksananusorn, T., Polborn, K. & Knochel, 12. Chen, Y., Yekta, S. & Yudin, A. K. (2003) Chem. 28. You, S.-L., Zhu, X.-Z., Luo, Y.-M., Hou, X.-L. & P. (2003) Angew. Chem. Int. Ed. 42, Rev. (Washington, D.C.) 103, 3155–3212. Dai, L.-X. (2001) J. Am. Chem. Soc. 123, 7471–7472. 3941–3943. 13. Seebach, D., Beck, A. K. & Heckel, A. (2001) 29. Pfaltz, A., Blankenstein, J., Hilgraf, R., Ho¨rmann, 41. Powell, M. T., Hou, D.-R., Perry, M. C., Cui, X. & Angew. Chem. Int. Ed. Engl. 40, 92–138. E. McIntyre, S., Menges, F., Scho¨nleber, M., Burgess, K. (2001) J. Am. Chem. Soc. 123, 8878– 14. Lowenthal, R. E., Abiko, A. & Masamune, S. Smidt, S. P., Wu¨stenberg, B. & Zimmermann, N. 8879. (1990) Tetrahedron Lett. 31, 6005–6008. (2003) Adv. Synth. Catal. 345, 33–43. 42. Williams, J. M. J (1996) Synlett, 705–710.

5726 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0307152101 Pfaltz and Drury Downloaded by guest on September 30, 2021