First principles prediction of NRTL binary interaction parameters for furfural derivatives

Biobased molecules constitute sustainable alternatives for fossil-based chemicals. However, to allow further valorization, the complex product mixtures resulting from biomass processing typically require downstream separation. Although many (semi)empirical thermodynamic models, like NRTL, are available for separation process design, their application for biobased molecules is hampered due to the lack of the necessary data to determine the binary interaction parameters used in these models. Therefore, in this work a case study comprising a set of furfural derivates has been used to showcase a methodology to efficiently determine these parameters. Based on first principles, COSMO-RS is used to generate a dataset of thermodynamic activities, which is subsequently used to regress the NRTL model against. Comparison of NRTL-HOC simulated vapor-liquid equilibria (VLE's) with experimental VLE's showed excellent agreement, confirming that COSMO-RS can be used to extend the use of (semi)-empirical models toward applications involving biobased molecules of which limited experimental data is available.