文摘
Classical experimental designs have been used as configurations of choice at GlaxoSmithKline within Chemical Development in establishing robust products and processes while achieving other goals such as reduction of cost, waste, process investigation times and most importantly improved quality. Occasionally, as in this paper, these first intent designs have been proved to be unworkable through irregularly shaped experimental regions induced through chemistry constraints. To achieve optimum efficiency, a fit for purpose computer-aided D-Optimal Design in considering the reduction of cyclic imides to amines has been investigated here, proving to be an invaluable additional tool in applying the appropriate design at the relevant time. In addition, partial least squares (PLS) analyses has been explored with a view of simultaneously modelling the responses whilst primarily identifying those factors having the largest impact across the responses of interest (quantity of reducing agent and solvent, temperature, and time). This combined with process analytical technology (PAT) through online IR monitoring has facilitated part of this cyclic process of investigations using scientific and business judgment.