A review of ionic liquids and deep eutectic solvents design for CO2 capture with machine learning

Ionic liquids (ILs) and deep eutectic solvents (DESs) are regarded as the next generation solvents for carbon capture which consist of cations and anions. Thousands of combinations of cations and anions can lead to varied properties of ILs/DESs, which makes it difficult to screen such ILs/DESs for CO2 in experiments. Computer-aided molecular design (CAMD) saves time and cost by reversing the search for the structure of ILs that are suitable for carbon capture. Compared with other thermodynamic models, machine learning (ML) models have the advan-tages of efficiency and accuracy in CAMD; hence, the number of studies on the application of ML models in the field of CAMD is growing each year. In this paper, a concise review of the application of ML to ILs/DESs-based CO2 capture technology is provided. The development process of ML models in (1) the prediction of the prop-erties of ILs/DESs using their structure; and (2) the prediction of the carbon capture effect using process pa-rameters is discussed. Perspectives on future research directions are proposed and key challenges are identified for screening suitable ILs/DESs using the capture effectiveness of a specific carbon capture process as an eval-uation criterion.

Automated summary

Ionic liquids (ILs) and deep eutectic solvents (DESs) are considered as potential solvents for carbon capture. The properties of ILs/DESs can vary due to the different combinations of cations and anions, making it challenging to screen suitable ILs/DESs for CO2 capture experimentally. Computer-aided molecular design (CAMD) using machine learning (ML) models provides an efficient and accurate approach to search for appropriate IL structures for carbon capture. This review paper discusses the application of ML models in predicting the properties of ILs/DESs and the carbon capture effectiveness, and proposes future research directions and challenges in screening suitable ILs/DESs.