Computational prediction of cellulose solubilities in ionic liquids based on COSMO-RS

Abstract

A computational approach is presented for prediction of cellulose solubilities in ionic liquids (ILs) based on COSMO-RS (Conductor-like Screening Model for Real Solvents). Thermodynamically stable molecular structures were optimized from 2D structures of cellulose and ILs following specific force-field based search of conformation lowest in energy and quantum chemical optimizations of molecular geometry. The thermodynamic property of logarithmic activity coefficient (lnγ) and excess enthalpy (HE) were calculated by COSMO-RS based on the COSMO molecular surfaces of cellulose and ILs to qualitatively predict the ability of ILs for cellulose dissolution. To evaluate the method, four sets of ILs were used to calculate lnγ and HE based on four different cellulose models. The goodness-of-fit of linear regressions between the experimental cellulose solubilities and the calculated lnγ and HE shows that lnγ is more reliable than HE for prediction of the dissolving power of ILs to dissolve cellulose. However, HE is more suitable for prediction of the dissolution ability of halogen-based ILs. Moreover, all the cellulose models gave comparably good prediction results regarding of the dissolving power of ILs based on the calculated lnγ, but the cellobiose model was identified as the optimal model due to the relatively higher prediction ability (R2) across different IL datasets. The approach is time efficient and robust, which provides a novel method for large-scale screening of ILs for cellulose dissolution.