Improved Process Understanding of Filtration Drying via a Rotary Evaporator-Based Process Analytical Technology (PAT) Interface

Abstract

The filtration drying of chemical solids is a critical unit operation in the isolation of active pharmaceutical ingredients and intermediates during the manufacturing process. Due to its central role in API isolation, it is important to understand the kinetics of filtration and drying processes to ensure minimal timeline impact and, importantly, the quality of the API produced. Process analytical technologies (PAT), such as Raman spectroscopy, near-infrared spectroscopy (NIRS), gas chromatography (GC), and mass spectrometry (MS), have been used to monitor filtration drying operations to provide real-time solvent content concentrations. However, each instrument has limitations, such as the need for at-scale calibration experiments with Raman and NIRS, and for GC and MS, the difficulty of correlating the headspace solvent concentration to the wet cake. Furthermore, current PAT methods to assess drying kinetics have large material demands, in even the smallest scale-down filter dryers, and lack representative measurements in a heterogeneous gas–solid sample matrix. To address some of these challenges, a drying workflow using a rotary evaporator (rotavap) with PAT interfaces has been developed in our lab. This unique rotavap PAT interface improves upon existing PAT drying methods because it reduces the material demand to as little as 1 g and, due to the rotational mechanism of the rotavap, ensures representative sampling by PAT of the wet cake during the drying curve collection. Three different drying case studies were carried out using the rotavap PAT interface and are illustrated herein. Initial slurry and final solid residual solvent concentrations were measured by GC or Karl Fischer (KF) methods and served not only as reference measurements for calibrating PAT sensors but were also used to rescale predicted solvent content by vibrational spectroscopy to capture accurate drying dynamics. Such PLS models were then applied to different experimental conditions across all three case studies, such as various drying temperatures and nitrogen sweep rates, to inform process understanding of the normal operating range of the filtration drying step. The alignment between the model predictions and offline reference measurements successfully demonstrates the value of the rotava p-based PAT interface for collecting filtration drying data in a material-sparing manner. This approach is expected to streamline the process development of API filtration drying steps by helping to determine the drying end point and also providing real-time drying data for kinetic modeling that could be used to predict drying outcomes at variable scales and across different filtration dryer designs.