2.1 Method A

In a flame-dried Schlenk was placed ligand (0.05 mmol) under argon atmosphere. To this was added degassed toluene (1 ml) and Ti(Oi-Pr)₄ (3 mmol), and the mixture was stirred at 40 °C for 20 min. After being cooled to –78 °C, Et₂Zn (1.0 M, 3 ml, 3 mmol) was added to the solution. In a moment, the solution turned orange. To the resulting solution was added benzaldehyde (2.5 mmol) in toluene (0.2 ml) and the mixture was warmed to –24 °C, and was let to react at this temperature for 18 h. The reaction was quenched by adding 1 M HCl, and the product was extracted with diethyl ether. The solvent was removed by evaporation under reduced pressure. The residue was distilled under reduced pressure (95 °C/10 mm Hg), giving pure 1-phenyl-1-propanol 〚5〛. Enantiomeric excess was determined by HPLC analysis using Chiralcel OD-H column, and a 98:2 mixture of n-hexane and i-PrOH as eluent.

2.2 Method B

In a flame-dried Schlenk were placed ligand (0.05 mmol) and diethylzinc (1.2 equiv), which were combined with 3 ml of toluene under argon atmosphere. The Schlenk was cooled to –78 °C instead of –30 °C in 〚7〛. Ti(Oi-Pr)₄ (1.2 equiv) in hexane (1.5 ml) was added slowly by syringe. Upon addition of titanium tetraisopropoxide the solution became orange. The solution was stirred for 5 min and benzaldehyde was added (1 equiv). The resulting solution was checked for homogeneity, transferred to a –24 °C bath, and stirred for 18 h. The reaction was quenched with 1 M HCl, and the product was extracted with diethylether. The solvent was removed by evaporation under reduced pressure. The residue was distilled under reduced pressure (95 °C/10 mm Hg), giving pure 1-phenyl-1-propanol. Enantiomeric excess was determined by HPLC analysis (see above).

2.3 Discussion

Clearly, the asymmetric amplification is due to the catalyst preparation procedure. Does the inverse addition in the formation of the catalyst in method B give rise to a different catalytic species? We shall discuss our results in the light of the mechanistic assumption of the literature. Ohno et al. proposed for method A the mechanism described in Fig. 3. Walsh et al. 〚8〛 have shown that no reaction occurs by mixing 1 with Ti(Oi-Pr)₄, while Denmark et al. 〚14〛 established that addition of diethylzinc to sulfonamides such as 1 forms some strongly self-associated species 4, as indicated in Fig. 4. For these reasons, Walsh et al. suggested an alternative mechanism to the mechanism of Fig. 3 of Ohno et al. and proposed the initial formation of 4 (Fig. 4), with further transformation into the monomeric and catalytically active titanium species 3.

It remains unclear if it is the reversal of the order of mixing from 1, Ti(Oi-Pr)₄ and Et₂Zn (method A, Ohno procedure) to Et₂Zn, 1 and Ti(Oi-Pr)₄ (method B, Walsh procedure) which changes the structure of the actual catalytic species. Another possibility could be the different temperatures used in methods A and B to prepare the catalyst. Obviously, some difference must exist between the two methods, as reflected by the strong (+)-NLE occurring only for method A. What could be this difference?

In order to answer to this question we performed some kinetic studies on the both systems (Fig. 5). The reaction rate with method A was much slower if the ee of the auxiliary was small (compared to the enantiopure catalyst). In method B the reaction rate was not affected by ee_aux.

The slowdown of the rate in method A for decreasing ee_aux fits with the presence of a (+)-NLE. It is well documented that asymmetric amplification is associated with decreasing rates for low ee_aux, which is in agreement either with the ML_n kinetic model 〚15, 16〛 or with the reservoir effect 〚4〛. We recall that the reservoir has been defined by the formation of diastereomeric species external to the catalytic cycle, usually of racemic composition, while, in the kinetic model, a heterochiral catalyst less active than the homochiral catalyst is at the origin of the asymmetric amplification 〚17〛. We studied the influence of the temperature (and of the heating time) on the (+)-NLE in method A. As shown in Fig. 6, there is a clear beneficial effect by increasing temperature and time of the catalyst preparation. It could be related to the enhanced production of a ‘reservoir’ product of racemic composition, whose formation is slow at room temperature. The most simple hypothetical structure could be a dimeric titanium complex (5, Fig. 7), although higher molecular-weight polymers are also possible. With this hypothesis, there should be no difference between enantiopure catalysts prepared by methods A or B. As a test, we investigated the addition of diethylzinc on racemic hydratropaldehyde in the presence of catalyst A or B. The isolation procedure of products and the chromatographic analysis were carried out according to 〚18〛. We found an absence of kinetic resolution in the two experiments. Moreover, both reactions proceeded at identical reaction rates with enantiopure catalysts A and B.

It must be noticed that Ti(Oi-Pr)₄ was used in stoichiometric amounts by respect to aldehydes, e.g., in large excess by respect to the sulfonamide ligand 〚5–9〛. However, it has been observed that a large excess of titanium complex is unnecessary for achieving good ees and yields 〚6〛. We found, for example, that a decrease of the proportion of titanium in Ti(Oi-Pr)₄/1a/PhCHO from 1.1:0.01:1 (methods A and B) to 0.1:0.01:1 and then to 0.04:0.01:1 leads to 2 with 98% ee, 84% ee and 33% ee respectively, the yields remaining good. Stoichiometric amount of titanium catalyst is needed for achieving high ee, although sometimes some titanium complex is involved as catalyst and not as reagent. Seebach et al. 〚19〛 have shown that a spirotitanate derived from Taddol and Ti(Oi-Pr)₄ is a good enantioselective catalyst for the addition of diethylzinc on aldehydes. It gives a strong acceleration of the reaction with respect to Ti(Oi-Pr)₄ used in catalytic amounts. The absence of NLE is in agreement with a monomeric structure for the catalyst.

We are currently investigating the mechanistic details of this catalyst, which remains the favoured one for the enantioselective addition of organozincs on aldehydes 〚12〛, because of its robustness and the ease to get the two enantiomers of 1. It was recently shown that the meso analog of 1 combined to a tetra-alkoxide titanium complex derived from Ti(Oi-Pr)₄ and (S)-PhCH(Me)OH led to an enantioselective catalyst, an evidence for the importance of both the alkoxide and sulfonamide ligands around titanium 〚20, 21〛.