New bisphosphine ligands for asymmetric catalysis
The success of homogeneous asymmetric catalysis has been attributed to the structure and stereochemistry of the coordinated ligand(s). The most effective ligands are C2-symmetrical bisphosphines containing either a rigid chiral backbone linking two PPh2 units or a bisphosphine, DIPAMP containing two chiral phosphine units linked by an achiral backbone. The synthesis of P-chiral ligands of this type has been severely hindered by the lack of a general synthetic route allowing the incorporation of phosphorus chirality without the need for separation of diastereomeric precursors or resolution of intermediate enantiomers. The objective of this work was to develop a general synthetic route to homochiral bulky arylphosphines with substantial flexibility in the groups at phosphorus and extend the approach to new P-chiral bisphosphines. In one approach, diastereomerically pure (2R, 4S, 5R)-2,5-diphenyl-3,4-dimethyl-1,3,2-oxazaphospholidine was prepared directly from PhPCl2 using l-ephedrine as a chiral auxiliary. Stereospecific oxidation using ButOOH gave the corresponding P-oxide which was shown to have R-stereochemistry at phosphorus by single-crystal X-ray diffraction studies. The compound reacted regiospecifically with ortho-anisylmagnesium bromide to afford the product formed by P-O bond cleavage with >96% d.e. and with retention of configuration at phosphorus as demonstrated by single-crystal X-ray diffraction studies. The l-ephedrine residue was replaced by O-methyl under acid-catalysis with inversion of configuration and with >95% e.e., the reaction was monitored by 1H n.m.r. spectroscopy which gave t1/2 of ca. 30 min. Attempts to incorporate para-fluorophenol using similar conditions led to the isolation of the pyrophosphinate in low yield. The OMe residue in the methyl (ortho-anisyl)phenylphosphinate was readily displaced by aliphatic Grignard reagents giving the corresponding phosphine oxides with inversion of configuration and with >95% e.e. Displacement of methoxy using aryl magnesium bromides showed similar enantioselectivity but in lower chemical yield, however the corresponding arylmagnesium chlorides were more efficient. In a second approach, diastereomerically pure (2R, 4S, 5R)-2-chloro-3,4-dimethyl-5-phenyl-1,3,2-oxazaphospholidine was prepared from PCl3 and l-ephedrine. The compound underwent diastereoselective P-C1 cleavage with aryl Grignard and aryllithium reagents with net retention of configuration at phosphorus and with 90% d.e. Oxidation of the ortho-anisyl derivative afforded (2R, 4S, 5R)-2-(ortho-anisyl)-3,4-dimethyl-5-phenyl-1,3,2-oxazaphospholidine-2-oxide which was subsequently reacted with a range of bulky aryl Grignard reagents to afford the corresponding biarylphosphinamides with retention of configuration at phosphorus. Subsequent acid-catalysed methanolysis and displacement of the methoxy residue with PhMgCl afforded a range of bulky arylphosphine oxides with defined configuration at phosphorus with >95% e.e. as determined by 1H n.m.r. methods. (S)-ortho-anisyl (meta-anisyl)phenylphosphine oxide underwent regiospecific ortho-lithiation on the meta-anisyl ring which on quenching with D2O afforded the corresponding 2-deuteride in 80% yield. The 2-iodo analogue was also prepared although in low chemical purity and is a key precursor to new axially dissymmetric bisphosphines containing chiral phosphorus centres. Other approaches to P-chiral ferrocenyl ligands and biaryl ligands are also described and modifications for further development are implicated. An X-ray crystallographic study of six aryl-oxazaphospholidines is also presented and demonstrates the influence of the substituents at phosphorus in determining the conformation of the 1,3,2-oxazaphospholidine ring. A comparison with solution 1H n.m.r. data showed, in some cases, good correlation between the P-O-C-H dihedral angle and the corresponding solid state torsion angle.