A QSPR model for estimating Henry's law constant of H2S in ionic liquids by ELM algorithm

Ionic liquids (ILs) as green solvents have been studied in the application of gas sweetening. However, it is a huge challenge to obtain all the experimental values because of the high costs and generated chemical wastes. This study pioneered a quantitative structure-property relationship (QSPR) model for estimating Henry's law constant (HLC) of H₂S in ILs. A dataset consisting of the HLC data of H₂S for 22 ILs within a wide range of temperature (298.15-363.15 K) were collected from published reports. The electrostatic potential surface area (S_EP) and molecular volume of these ILs were calculated and used as input descriptors together with temperature. The extreme learning machine (ELM) algorithm was employed for nonlinear modelling. Results showed that the determination coefficient (R²) of the training set, test set and total set were 0.9996, 0.9989,0.9994, respectively, while the average absolute relative deviation (AARD%) of them were 1.3383, 2,4820 and 1.5820, respectively. The statistical parameters for the measurement of the model exhibited the great reliability, stability, and predictive power of the ELM model. The Applicability Domain (AD) of the ELM model is also investigated. It proves that the majority of ILs in the training and test sets are located in the model's AD and verifies the reliability of the model. The proposed model is potentially applicable to guide the application of ILs for gas sweetening.