Machine learning models for solvent effects on electric double layer capacitance. (20th July 2019)

• Record Type: Journal Article
• Authors: Su, Haiping; Lian, Cheng; Liu, Jichuan; Liu, Honglai
• Is Part Of: Chemical engineering science. Volume 202(2019)
• Page Start: 186
• Page End: 193
• Publication Date: 2019-07-20
• Subjects: Solvent effects -- Electric double layer capacitance -- Machine learning -- Classical density functional theory; Chemical engineering -- Periodicals; Génie chimique -- Périodiques; Chemical engineering; Periodicals; Electronic journals; 660
• DOI: 10.1016/j.ces.2019.03.037
• Languages: English
• ISSNs: 0009-2509

Abstract

Graphical abstract: Highlights: The solvent effects on electrochemical capacitances were investigated by machine learning. The classical density functional theory gives a better understanding of the solvent effects. This method provides a benchmark for screening of new energy materials. Abstract: The role of solvent molecules in electrolytes for supercapacitors, representing a fertile ground for improving the capacitive performance of supercapacitors, is complicated and has not been well understood. Here, a combined method is applied to study the solvent effects on capacitive performance. To identify the relative importance of each solvent variable to the capacitance, five machine learning (ML) models were tested for a set of collected experimental data, including support vector regression (SVR), multilayer perceptions (MLP), M5 model tree (M5P), M5 rule (M5R) and linear regression (LR). The performances of these ML models are ranked as follows: M5P > M5R > MLP > SVR > LR. Moreover, the classical density functional theory (CDFT) is introduced to yield more microscopic insights into the conclusion derived from ML models. This method, by combining machine learning, experimental and molecular modeling, could potentially be useful for predicting and enhancing the performance of electric double layer capacitors (EDLCs).