Statistically Evaluated Screening of Experimental Conditions

On the basis of these results new ranges of the parameters for the optimization work (and scope of operation) were set substrate/catalyst ratio (range 1700-2300), temperature (25-45 °C), pressure (4-10 bar), methanol concentration (800-1200 g/ mol 19). The stirring rate was not included for the subsequent statistical design. The vertex centroid plan (quadr. D-opt., 19 runs) recommended by Stavex was used. The goal of the optimization was to refine conditions in order to ensure a robust and technically feasible process, and the measure of performance was the yield of the desired optically active 20. 

A series of 19 experiments were run each for a reaction time of 3 hours, after which period they were worked up, in all cases the racemate crystallized out, complicating the analysis. The statistical analysis of the data gave a good model for the conversion (see Fig. 18) but was not satisfactory for the enantioselectivity, and consequently not for the yield. This is believed to be at least partially due to the inhomogeneity of the reaction mixtures leading to imprecision of the analytical measurements. 

The trends mentioned above could however be confirmed. The most significant factor determining the conversion is the temperature. High pressure combined with high temperature leads to a fast reaction but is detrimental to the enantioselectivity. The concentration of the reaction solution and the S/C ratio are less important within the ranges investigated. The optimal parameters proposed by the Stavex program for optimum yield with a reaction time of 3 hours were essentially those in the best experiment temperature 45 °C, pressure 4 bar, S/C 2300, MeOH 800 g/mol 19. An additional experiment run under these conditions yielded, after 98% conversion, 20 in 87.3% yield, 99.7% ee, content 98.6% with 

and highest unknown impurity 0.1 area-%. Of course the further hydrolysis step 20-17 was also optimized during the course of this study, leading to a straightforward and economical process. 

As the proposed optimal parameters were at the extremes of the ranges tested, it was foreseen in the next phase of process development, i.e., establishment of qualified ranges for cGMP production, to examine higher temperatures, lower pressures and higher S/C ratios. 

---