Phospho-transfer catalysis: On the asymmetric hydrophosphonylation of aldehydes

Abstract

We report here a precise, in situ 31P{ 1H}-NMR method of assaying enantiopurity of α-hydroxyphosphonate esters, the products of the carbonyl hydrophosphonylation (Pudovik) reaction. This method is based upon a diazaphospholidine chiral derivatising agent (CDA) which satisfies all of the criteria for a precise assay; (i) derivatisation of α-hydroxyphosphonate esters is both rapid and clean, (ii) kinetic resolution is absent and (iii) 31P{ 1H} chemical shift dispersions are excellent (>5ppm). Calibration of this assay has been achieved by cross-referencing the 31P{ 1H}-NMR signals obtained for the CDA-derivatised ester of (MeO) 2P(O)CHPh(OH) to optical rotation measurements from scalemic material obtained upon lipase catalysed hydrolysis (F–AP 15, Rhizopus oryzae) of (MeO) 2P(O)CHPh(OAc). Analysis of NMR chemical shift and coupling parameters for a closely related series of derivatised α-hydroxyphosphonate esters support further configuration assignments on the basis of inference. We report also on the configurational stability of α-hydroxyphosphonate esters in the presence of acids, organonitrogen bases and metal salts. 2H-labelling and carbonyl crossover experiments reveal that low levels of epimerisation (<2%) at the alpha-carbon atom (C α) of α-hydroxyphosphonate esters is possible under certain conditions of catalysis and within certain limits. A design strategy for the construction of catalyst systems in the Pudovik reaction is outlined based upon a combination of Lewis acidic ( E) and Lewis basic ( N) sites. Four types of catalyst are outlined, members of two distinct Classes I and II according to the nature of the acid and base sites, along with our investigations of representative examples of each Class. A variety of Class I.1 systems based on β-amino alcohols (one hydrogen bonding E site and one organonitrogen N site), have been assayed in the model reaction between (MeO) 2P(O)H and PhCHO. Results suggest that catalysis of the Pudovik reaction is clean and efficient in certain cases but that catalytic activity is strongly dependent upon the nature of the basic ( N) nitrogen centre. Moreover, only low levels (<10%) of enantioselectivity are afforded by all amino alcohols assayed. Achiral variants of Class I.2 catalysts (multiple hydrogen bonding E and/or N sites) have been examined to model carbonyl and H-phosphonate binding; an amphoteric receptor based on a pyridine dicarboxamide scaffold has been synthesised and shown to bind benzaldehyde >50% more strongly ( K 11 0.53 mol −1 dm 3) than dimethyl-H-phosphonate ( K 11 0.34 mol −1 dm 3, 298 K) and to catalyse the hydrophosphonylation reaction between these two substrates with a second order rate constant comparable to that of triethylamine (both k 2 5.9×10 −2 mol −1 dm 3 h −1, 293 K). However, one of the major limitations of this model is that competitive product inhibition dominates after some 15 turnovers (75% completion). Model studies reveal that hydrophosphonylation catalysis via a nitrogen Lewis base is accelerated up to 10-fold upon the introduction of [Zn(OSO 2CF 3) 2] as co-catalyst. Consequently, Class II.1 systems employ metal salts [Zn(OSO 2CF 3) 2] as Lewis acidic E sites and chiral co-catalysts capable of binding to the metal and also acting as Lewis basic N sites. Such systems catalyse the addition of (MeO) 2P(O)H to PhCHO cleanly with modest turnover numbers (<10 turnovers; 50% completion) but with enhanced enantioselectivity over Class I catalysts (<40% e.e.). However, competitive product inhibition is still problematic. Class II.2 systems are related to Class II.1 but possess directly coordinated E and N sites with more basic N functions and consequently are far more active as catalysts than the other classes. This increased catalytic activity is exemplified by one of the simplest achiral members, diethylzinc which catalyses the 100% chemo- and regioselective addition of (MeO) 2P(O)H to PhCHO to afford (MeO) 2P(O)CHPh(OH) with an average turnover rate (over a 1 h reaction time at 298 K) of 115 h −1 compared to ca. 1 h −1 for NEt 3 under analogous conditions. Chiral variants are proposed.