Data-driven future for nanofiltration: Escaping linearity

Abstract

Compared with traditional membrane separation methods such as distillation and chromatography, nanofiltration (NF) affords decreased waste generation and energy consumption. Despite the multiple advantages of NF and materials available for NF membranes, the industrial applicability of this process requires improvement. To address these challenges, we propose four important pillars for the future of membrane materials and process development. These four pillars are digitalization, structure–property analysis, miniaturization, and automation. We fill gaps in the development of NF membranes and processes by fostering the most promising contemporary technologies, e.g., the integration of process analytical technologies and the development of a parallel artificial nanofiltration permeability assay (PANPA) or large online databases. Moreover, we propose the extensive use of density functional theory-aided structure–property relationship methods to understand solute transport process at a molecular level. Realizing an inverse design would allow researchers and industrial scientists to develop custom membranes for specific applications using optimized properties.

Keywords

• Machine learning
• Big data
• Inverse design
• Data science
• Process analytical technologies