Screening of ILs by COSMO ‐ RS and machine learning for extraction of n‐propanol/n‐propyl ether:

Abstract

N‐propyl ether (DPE) is produced via acid‐catalyzed dehydration of n‐propanol (NPA), forming a minimum‐boiling azeotrope. Conventional separation processes face challenges in both efficiency and product purity. In this work, a multi‐scale approach was employed to develop an ionic liquids (ILs)‐based extractive separation process. Following a systematic framework from molecular structure and solvent screening to experimental validation, microscopic mechanism and process simulation, a green, efficient, and clean method for separating azeotrope using ILs was investigated. An initial screening of 378 ILs was conducted based on the COSMO‐RS model, and machine learning algorithms were introduced to achieve high‐throughput prediction of ILs viscosity. Selected ILs were evaluated for thermal stability and toxicity, confirmed by phase equilibrium experiments. Molecular simulation revealed extraction mechanisms at molecular‐electronic levels. Compared to conventional processes, the designed extraction process reduces TAC by 38.35% and cooling water consumption by 38.68%, offering a transferable strategy for the separation of alcohol‐ether azeotropes.