부산대학교 도서관

Deep eutectic solvents (DESs) with benign properties as green alternatives are being preferred for a multitude of energy conversion and environmental protection processes and designs. The tunability of the constituents makes the design of novel DESs for specific tasks possible, accompanied by the urgent requirement of accurate characterization and prediction of the properties. Modeling the structure–property relationships of DESs will offer feasible conduction to predict properties and design novel DESs. Particularly, the estimation of density, as a fundamental physical property, is essential for process operation and design. Herein, based on the hydrogen-bond interaction existing in DESs, a general bonding-group interaction contribution (BGIC) method was introduced to predict the densities of DESs by their structures. The optimized parameter values were determined by the training set (2553 data points, 70%) with the average absolute relative deviation (AARD) result of 1.49%. The test set (1094 data points, 30%) was used to evaluate the accuracy of the method with the AARD result of 1.56%. The predictive relative deviation (RD) can be controlled within ±20% except for one abnormal data point. The BGIC method was also extended to estimate densities of ternary DESs, and the resulting AARD was 2.29% for 174 data points. The results illustrate that the BGIC method has provided a valuable and reliable tool for predicting densities of DESs.