A New Principle in Combinatorial Asymmetric Transition‐Metal Catalysis: Mixtures of Chiral Monodentate P Ligands

Abstract

The right mix!: Surprisingly, the use of the mixture of two different chiral monodentate phosphorus ligands (e.g., a phosphonite/phosphite combination; see schematic representation; cod=1,5-cyclooctadiene) in Rh-catalyzed olefin hydrogenation leads to significantly higher enantioselectivities than in the case of the pure ligands themselves. Successful ligand optimization in the area of enantioselective transition-metal catalysis requires rational design, intuition, and experience, but also some degree of trial and error.1, 2 Combinatorial asymmetric catalysis was recently developed to speed up and/or to complement the process of optimization.3, 4 All of the examples thus far involve the generation and screening of libraries of chiral ligands whose modular construction allows structural diversity and therefore increases the chance of finding a “hit”. We describe a new concept in the area of combinatorial enantioselective transition-metal catalysis. The basic idea concerns the use of mixtures of chiral monodentate ligands. The method is relevant whenever in the transition state of the reaction at least two monodentate ligands (L) are coordinated to the metal (M) of the active catalyst MLx. For example, in the case of a mixture of two such ligands La and Lb, three different catalysts exist in equilibrium with one another, namely the two homocombinations MLaLa and MLbLb as well as the heterocombination MLaLb.5 Many examples for homocombinations are known in the literature,1a, 6 for example, the recently reported BINOL-based modular monophosphonites 1,7 monophosphites 2,8 and the monophosphoramidite 3,9 1 which often (but not always) lead to high enantioselectivities when used as ligands in Rh-catalyzed olefin hydrogenation (BINOL=2,2′-dihydroxy-1,1′-binaphthyl). In contrast, nothing has been reported about the use of heterocombinations MLaLb as catalysts. Since rapid ligand exchange is likely in most systems, the preparation of MLaLb in pure form in solution is not expected to be possible. However, the mixture of all three catalysts may well lead to enhanced enantioselectivity provided MLaLb is more active and more selective than either of the traditional catalysts MLaLa or MLbLb. Moreover, the relative amounts of the ligands La and Lb used may also influence the stereochemical outcome. Entry Ligands ee [%] (config.) Homocombinations 1 (R)-1 a/(R)-1 a 91.8 (S) 2 (R)-1 b/(R)-1 b 94.4 (S) 3 (R)-1 c/(R)-1 c 92.0 (S) 4 (R)-1 d/(R)-1 d 93.3 (S) 5 (R)-1 e/(R)-1 e 72.8 (S) 6[b] (R)-1 f/(R)-1 f 7.4 (S) 7 (S)-2 a/(S)-2 a 76.6 (R) 8 (S)-2 b/(S)-2 b 83.6 (R) 9 (R)-2 c/(R)-2 c 94.6 (S) 10 (S)-2 d/(S)-2 d 95.4 (R) 11[c] (S)-2 e/(S)-2 e 78.6 (R) 12[d] (S)-2 f/(S)-2 f 32.4 (R) 13 (S)-2 g/(S)-2 g 94.4 (R) 14 (S)-2 h/(S)-2 h 92.4 (R) Heterocombinations 15 (R)-1 a/(R)-1 b 92.6 (S) 16 (R)-1 a/(R)-1 c 97.9 (S) 17 (R)-1 a/(R)-1 d 97.8 (S) 18 (R)-1 c/(R)-1 d 94.1 (S) 19 (R)-1 d/(R)-1 e 75.8 (S) 20 (R)-1 d/(R)-1 f racemic 21 (R)-2 a/(R)-2 b 80.0 (S) 22 (R)-2 a/(R)-2 c 76.6 (S) 23 (R)-2 a/(R)-2 d 89.0 (S) 24 (R)-2 a/(R)-2 e 77.4 (S) 25 (R)-2 a/(R)-2 f 84.6 (S) 26 (R)-2 a/(R)-2 g 87.2 (S) 27 (R)-2 b/(R)-2 c 79.0 (S) 28 (R)-2 b/(R)-2 d 91.2 (S) 29 (R)-2 b/(R)-2 e 80.8 (S) 30 (R)-2 b/(R)-2 g 90.0 (S) 31 (R)-2 d/(R)-2 c 94.2 (S) 32 (R)-2 d/(R)-2 e 92.2 (S) 33 (R)-2 e/(R)-2 c 85.6 (S) 34 (R)-2 g/(R)-2 c 94.6 (S) 35 (R)-2 g/(R)-2 d 94.8 (S) 36 (R)-2 g/(R)-2 e 91.2 (S) 37 (R)-1 a/(R)-2 a 81.9 (S) 38 (R)-1 a/(R)-2 c 94.4 (S) 39 (R)-1 a/(R)-2 d 93.0 (S) 40 (R)-1 c/(R)-2 a 96.4 (S) 41 (R)-1 c/(R)-2 d 91.8 (S) 42 (R)-1 d/(R)-2 a 98.0 (S) 43 (R)-1 d/(R)-2 c 94.6 (S) 44 (R)-1 d/(R)-2 h 97.2 (S) 45 (R)-1 c/(R)-2 h 95.6 (S) It is apparent that certain ligand combinations La/Lb lead to notably high enantioselectivities. In the case of two different phosphonites (Table 1, entries 15–19) enantioselectivity in the test reaction 4→5 is nearly complete (ee=98 %) whenever one component bears a small substituent R at the phosphorus center (e.g., methyl as in 1 a) and the other is characterized by steric bulk (Table 1, entries 16 and 17). The results constitute significant improvements relative to the use of the corresponding homocombinations which lead to ee values of only 92–94 % (Table 1, entries 1, 3, 4). The chloro derivative 1 f fails to afford high enantioselectivity, irrespective of its use as a homocombination (Table 1, entry 6) or as a heterocombination (Table 1, entry 20). Acceptable ee values occur in the phosphite series (Table 1, entries 21–36), but improvements upon using heterocombinations are in most cases not significant. The greatest effect results when going from the homocombinations 2 a (76.6 % ee) and 2 f (32.4 % ee) to the heterocombination based on the same ligands 2 a/2 f (84.6 % ee) (Table 1, entry 25). In contrast, upon using proper combinations of phosphonites 1 and phosphites 2, pronounced enhancements in enantioselectivity are observed, especially if a bulky phosphonite (e.g., 1 c or 1 d) is combined with the sterically least demanding phosphite 2 a (Table 1, entries 40, 42). In the case of the hydrogenation of 4 b the combinatorial search was restricted to mixtures of phosphonites 1. Again, significant improvements in the ee values were accomplished ((R)-1 a/(R)-1 c: 96.7 % ee (S); (R)-1 a/(R)-1 d: 99.2 % ee (S); (R)-1 b/(R)-1 d: 94.6 % ee (S) relative to (R)-1 a/(R)-1 a: 89.9 % ee (S); (R)-1 b/(R)-1 b: 89.2 % ee (S); (R)-1 d/(R)-1 d: 69.1 % ee (S)). To study the possible influence of the relative amounts of the two ligands used, appropriate mixtures of 1 a and 1 d were employed in the Rh-catalyzed hydrogenation of 6 a (Rh:P ligand=1:2). The results summarized in Table 2 show that precise adjustment of the ratio is not necessary. However, when the relative amount of the methyl phosphonite 1 a dominates strongly, then the ee value drops significantly. Entry Ligands (R)-1 a/(R)-1 d ee [%] (config.) 1[b] 1:5 95.4 (S) 2 1:3 97.4 (S) 3 1:2 97.2 (S) 4 1:1 96.4 (S) 5 2:1 88.8 (S) 6 3:1 85.0 (S) 7 5:1 81.2 (S) Entry Ligands ee [%] (config.) homocombinations 1 (R)-1 a/(R)-1 a 90.2 (R) 2 (R)-1 b/(R)-1 b 71.4 (R) 3 (R)-1 c/(R)-1 c 21.9 (R) 4 (R)-1 d/(R)-1 d 57.3 (R) 5 (R)-1 e/(R)-1 e 28.8 (R) heterocombinations 6 (R)-1 a/(R)-1 b 82.4 (R) 7 (R)-1 a/(R)-1 c 88.6 (R) 8 (R)-1 a/(R)-1 d 96.4 (R) 9 (R)-1 b/(R)-1 d 92.2 (R) 10 (R)-1 c/(R)-1 d 69.1 (R) 11 (R)-1 c/(R)-1 e 50.0 (R) 12 (R)-1 d/(R)-1 e 57.4 (R) Our results illustrate a new principle in the area of combinatorial asymmetric transition-metal catalysis.11 Accordingly, it makes sense to consider the use of mixtures of chiral monodentate ligands in asymmetric transition metal catalysis, inspite of the fact that systems of this kind contain at least three different catalysts or precatalysts. For example, the NMR spectrum of the mixture of 1 a, 1 d, and [Rh(cod)2]BF4 (cod=1,5-cyclooctadiene) shows the presence of [Rh(1 a)2(cod)]BF4, [Rh(1 d)2(cod)]BF4 and [Rh(1 a)(1 d)(cod)]BF4 in a ratio of about 20:20:60.12 Initial kinetic studies indicate that this mixture is a more active catalyst system than the respective homocombinations [Rh(1 a)2(cod)]BF4 or [Rh(1 d)2(cod)]BF4. Additional studies are necessary to identify the source of increased enantioselectivity. The question whether the general concept is also valid in other types of asymmetric reactions, and whether mixtures of achiral ligands enhance catalyst performance in achiral transformations, also needs to be addressed. General procedure for the hydrogenation of 4 and 8 using mixtures of monodentate P ligands: A dry 50-mL Schlenk flask under an atmosphere of argon was charged with a mixture of a 1.7 mM solution of the first ligand (0.6 mL) and a 1.7 mM solution of the second ligand (0.6 mL) in dry dichloromethane. The solution was treated with a 2.0 mM solution of [Rh(cod)2]BF4 (0.5 mL) in dichloromethane and stirred for 5 min at room temperature. Then a 0.112 m solution of the substrate in dichloromethane (9 mL) was added. Vacuum was applied three times until the solvent began to evaporate gently, and hydrogen was introduced. Hydrogenation was carried out at 1.3 bar for the periods given. Following dilution, conversion was determined by gas chromatography (GC). To determine the ee values, about 1.5 mL of the reaction solution was passed through a small amount of silica gel prior to the GC or HPLC analysis. The hydrogenation experiments were carried out in a parallel manner using 20 or more flasks.