Model-based analysis and optimization of organic solvent nanofiltration processes for pharmaceutical applications

As the pharmaceutical industry is reshaped by Quality by Design directives, the fundamental shift towards continuous manufacturing calls for novel solutions in downstream processes. Cross-flow organic solvent nanofiltration (OSN), an inherently continuous technology, is a low-energy and low-waste alternative for solute-solute separations. Several previous studies have demonstrated that process modeling can be successfully applied to predict the separation performance of OSN systems. Mathematical description has been used for applications such as purification, catalyst recovery and in situ solvent recycling. Building on these works we introduce a new, process- oriented framework to describe and optimize nanofiltration efficiency and performance in silico. Furthermore, as a case study, we present the first comprehensive model-based analysis and assessment of enantioselective OSN processes.

Lumped dynamic models were developed for various system configurations, including structurally diverse nanofiltration cascades and single-stage separations with side-stream recycling and in situ racemization. Steady-state models were used for sensitivity analysis and techno-economic assessment. A process-based transformation and analysis of the OSN Database (www.osndatabase.com) was performed to present a holistic approach to nanofiltration efficiency description.

We used model-based design space exploration to give an insight into the membrane characteristics necessary for efficient separations. Transforming the OSN Database directly into a rejection- selectivity dataset allowed us to present an efficient categorical comparison of membrane performance.

Model-based sensitivity analysis facilitated the detailed discussion and comparison of the purity – recovery trade-off for a variety of enantioselective nanofiltration configurations. These results led to the feasibility assessment of membrane-based stereochemical resolution processes. Comparative and quantitative process performance examples were examined for theoretical rejection scenarios and extended to cases from the literature of chiral membranes. Simulations

revealed how OSN cascades extend the inherent purity limitations of single-stage separations. A model-based prediction tool is available on www.osndatabase.com to aid researchers in connecting chiral materials development results with early-stage process performance assessments. Our research aimed at bridging the gap between material science and sustainable process development by providing a guideline for future endeavors in the field.