K. B. Sharpless, S. S. Woodard
1983
Citations
0
Influential Citations
120
Citations
Quality indicators
Journal
Pure and Applied Chemistry
Abstract
The experimental rate equation for titanium—tartrate catalyzed asymmetric epoxidation by tert-butyl hydroperoxide is reported. The catalyst is a dimer, and a structure of C2 symmetry is proposed. The mechanism of the reaction is discussed with respect to kinetic resolution of racemic secondary allylic alcohols as well as the enantioselectivity of epoxidation of prochiral substrates. The alignment of a lone pair of the reactive alkyl peroxo-oxygen atom with the olefin orbital is postulated as an important interaction in the transition state. In 1980 we reported the discovery of the titanium-tartrate catalyzed asymmetric epoxidation of allylic alcohols by tert-butyl hydroperoxide (TBHP).1 This system performs well with a remarkable variety of allylic alcohol substrates, consistently providing the enantiofacial selection shown in Scheme I. Scheme I diethyl tartrate (unnatural) 11:0: (CH&3COOH, Ti(OiPr)4 ___ ('Li ('I !r" " 2'"2 ' 7Q9Q0/0 yield L-(+)-diethyl tartrate (natural) In addition, the catalyst is sensitive to pre—existing chirality in the substrate: the epoxidation of racemic secondary allylic alcohols proceeds rapidly with only one of the enantiomers, leaving the other, slower-reacting, enantiomer of the allylic alcohol behind.2 This reactivity pattern is demonstrated by the epoxidation of racemic 1823 )90 % e.e. 1824 K. B. SHARPLESS et al, (E)-cyclohexylpropenylcarbinol using L-(+)-diisopropyl tartrate as shown in Scheme II. Scheme II +DIPT LOH °yOH YH CyH CyH 98:2 "fast" 'siow 04 Lyiflr. threo -(+)-DIPT OH — S L o w OH 2O0C ny HCy 38:62 r.yiiiLQ threo In the time since the discovery of the reaction, much information has been obtained concerning the mechanism of the process, and several fascinating modifications to the "parent' catalyst system have been discovered.3 In this paper we present a summary of our Investigation of the kinetics of the asymmetric epoxidation and of the catalyst structure.4 We also propose a mechanistic model, which has led us to speculate on the details of the oxygen—transfer step. We begin by considering the nature of titanium alkoxide systems in general. They have several properties crucial to the success of the asymmetric epoxidation reaction: (1) exchange of monodentate alkoxide ligands is rapid in solution5; (2) titanium (IV) participates in four covalent bonds, which is exactly the number required for this reaction (two for the divalent chiral auxiliary (tartrate), and one each for 1BHP and the allylic alcohol), (3) titanium (IV) (d0) alkoxide systems display a range of coordination numbers and geometries in published structures in crystals and solutions,Sa,Sb,6 and so presumably their coordination chemistry is somewhat flexible; and (4) titanium (IV) alkoxides are weak Lewis acids, and thus serve to activate a coordinated alkyl peroxoligand toward nucleophilic attack by the olefin of a bound allylic alcohol.7'1° The third property, that of flexibility in coordination number and geometry, may be partly responsible for the catalyst's ability to accomodate substrates of such widely different steric demand. Of course other d° transition metal alkoxide systems also possess these properties, but they fail to give high enantiomeric excess when used with tartrate and TBHP in the standard fashion.8 It appears that Ti(IV) has a unique combination of properties that permits the formation of an effective catalyst structure with tartrate and allows the reactants to interact efficiently in compliance with what we belive are strict Mechanism of titanium—tartrate catalyzed asymmetric epoxidation 1825 molecular orbital requirements (vide infra). Mixing equimolar amounts of a titanium tetraalkoxide and a chiral tartrate diester releases two equivalents of alcohol into solution9 and forms a dominant species of stoichiometry [Ti(OR)2(tartrate)]x (see equation 1 in Scheme III). Addition of TBHP and allylic alcohol rapidly establishes the equilibria characterized by constants K1 and K2; these constants were found to be approximately 1 for TBHP and most allylic alcohols. When 1BHP and the allylic alcohol are juxtaposed in the coordination sphere of the same metal center1° (i.e. as in 1), the oxygen atom transfer occurs to give tert-butyl alcohol and the chiral epoxy alcohol bound as alkoxides. These product alkoxides are replaced by more allylic alcohol and TBHP and the catalytic cycle is completed as the loaded complex 1 is regenerated. Scheme III Ti (OR)4 + tortrate ester Ti(OR)2(tortrate) + 2ROH (I) TBHP Ti (OR) (tartrate) Ti (OR)(TBHP)(tartrote) 2 ROH K1 ROH ROH ol K' K 2 2 TBHP Ti(O R)(avi,c alcohol)(tortrate) , -__ Ti (TBH P)(atiic alcohol)(tartrofe) ROH 1 K' K1 K2 K1'K epoxidation Ti (OBd)(epoxy alcohol) (i artrate) In full accord with Scheme III is the experimental rate equation which was determined under pseudo-first-order conditions to be: k [TBHFJ [Ti(OR)2(tartrate [allylic alcohoU rate 2 [nhibitor alcohoji 1826 K. B. SHARPLESS el al. Note that the rate constant k is actually the product of the rate constants for the epoxidation step and the equilibrium constants K1 and K2. Thus the rate expression is consistent with the action of a system wherein reactants are assembled on the metal followed by a rate-determining product forming step. We are attempting to learn as much as possible about the structure of species formed when titanium tetraalkoxides and tartrate esters are mixed in solution. Unfortunately, our efforts to obtain crystals of these samples have not been successful as yet, so we are forced to rely on less direct methods. In any event, as the work of Halpern has exquisitely demonstrated,12 even the most complete structural characterization of the major component of a catalyst system may bear little relation to the structure of the actual catalyst. While ever mindful of this lesson, we believe for several reasons that the major species in our titanium—tartrate solution is actually the dominant catalyst for the reaction, as discussed below. It is important to understand the special characteristics of the 1:1 system in the reaction. With substrates that are relatively slow to epoxidize, the use of a Ti:tartrate ratio even slightly greater than 1:1 results in a marked loss in enantioselectivity, presumably because of the formation of species with less than 1 tartrate per Ti atom that catalyze epoxidation at a similar or faster rate and with a different selectivity than the 1:1 structure. Conversely, addition of more than one equivalent of tartrate to titanium causes the rate of the reaction to decrease by exactly the amount predicted by assuming that excess ligand forms a species of stoichiometry [Ti(tartrate)2]x that is catalytically inactive because of the inability of monodentate allylic alcohol and 1BHP to displace the divalent tartrate. Since titanium alkoxides are well known to exist in oligomeric forms in solution, the molecular weight of the 1:1 species was of primary concern. Also, we had noted that use of racemic (dl) tartrate as the ligand for titanium resulted in the formation of different ratios of diastereomeric products in the epoxidation of racemic secondary allylic alcohols than those produced when either dor !-tartrate was used alone. This suggests (but does not prove) that more than one tartrate is involved in the active complex. As measured by differential vapor phase osmometry (CH2C12) and by Rayleigh light scattering (cyclohexane),27 the complex is a dimer in solution. The electron impact mass spectrum is in full accord with the dimeric structure and gives no evidence for either monomer or a species larger than the dimer. That the dimer, and not some trace monomer or higher aggregate, is the actual dominant catalyst is strongly supported by the fact that the rate remains first—order in catalyst over a 10-fold range in concentration. The IR and 'H NMR of equimolar solutions of titanium tetraalkoxide and tartrate esters remain essentially invariant over at least a Mechanism of titanium—tartrate catalyzed asymmetric epoxidation 1827 20—fold change in concentration. (It is admittedly possible, but unlikely, that the catalyst is a very active, undetected species that responds to changes in concentration to give the observed first-order result. However, there seems to be no reason why the dimer should be far less reactive than other possible titanium—tartrate structures.) We believe that this dimeric catalyst has the ten-membered ring structure2in Fig. 1, analogous to the crystal structure of the complex of vanadium (IV) with tartaric acid, Na4[(V0)2(tart)2]4 (tart = C4H2064 ligand), found by Tapscott and coworkers.13 The NMR (both 'H and 13C) and IR spectra are consistent with such a structure, the latter showing both free (1738 cm)'4 and coordinated (1635 cm1) carbonyl stretching bands.