(707j) Combined Experimental, Molecular Simulations and Machine Learning Approach to Optimize Aqueous Amine Solvents for CO2 Capture

Tertiary amines are attractive for industrial CO₂ capture because they tend to be more stable, more energy-efficient and show a higher CO₂ absorption capacity than other amines. One of the current challenges in the CO₂ capture industry is identifying tertiary amines with a higher CO₂ absorption rate, which is their main weakness. Understanding the relationship between amine structure and properties and CO₂ absorption rate is required for efficient amine screening. Despite numerous attempts, this goal has been elusive.

First, we have developed a computational approach based on molecular modeling to predict the CO₂ absorption rates and energies in aqueous tertiary amines. The predictions are validated using experimental data. The key to the new model's excellent accuracy is to include the solvation effects in the computation of the free energy barrier for the reaction of CO₂ absorption (https://doi.org/10.1021/acs.jcim.0c01386).

In a second step we have generated a diverse library of 100 tertiary amines which represent a large portion of chemotypes relevant for industrial amine-based chemical CO₂ absorption. Using the molecular simulations-based model, we have calculated the CO₂ absorption rates and energies for those 100 aqueous amine solvents. As a check, some points have been verified experimentally.

In a third step we have used the dataset of the CO₂ absorption rates and energies for the 100 aqueous amines to build quantitative structure-property relationship models. These models allowed us to perform a large-scale virtual screening of commercially available amines and to identify the amines with optimal CO₂ absorption rate and energy. For several amines the predictions have been verified experimentally. The approach and the results will be presented.