A method for real time detection of reaction endpoints using a moving window t-test of in situ time course data

Abstract

In-line Fourier transform mid-infrared spectroscopy (FTIR) was used to monitor a reaction which had a highly inconsistent rate due to sensitivity to changes in process parameters and the quality of the starting materials. Accurate determination of the endpoint in real time was needed to prevent formation of an impurity by degradation of the product. A moving window t-test algorithm was used to predict and determine the endpoint by analysis of FTIR trends during the reaction. The method was selected because it determined the endpoint in the correct range for reactions with different process parameters and data collected with different FTIR instruments. During development, the FTIR and t-test algorithm were used to monitor reactions for feasibility studies, process optimization, use tests, robustness DoE, and scale up to pilot plant. The method for monitoring the reaction may be useful in a GMP manufacturing environment where sampling and analysis are time consuming.