Computer-aided molecular design of imidazole-based absorbents for CO2 capture. (June 2016)

• Record Type: Journal Article
• Authors: Venkatraman, Vishwesh; Gupta, Mayuri; Foscato, Marco; Svendsen, Hallvard F.; Jensen, Vidar R.; Alsberg, Bjørn K.
• Is Part Of: International journal of greenhouse gas control. Volume 49(2016:Jun.)
• Page Start: 55
• Page End: 63
• Publication Date: 2016-06
• Subjects: QSPR -- De novo -- Applicability domain -- CO2 capture -- Imidazole; Greenhouse gases -- Environmental aspects -- Periodicals; Air -- Purification -- Technological innovations -- Periodicals; Gaz à effet de serre -- Périodiques; Gaz à effet de serre -- Réduction -- Périodiques; Air -- Purification -- Technological innovations; Greenhouse gases -- Environmental aspects; Periodicals; 363.73874605
• DOI: 10.1016/j.ijggc.2016.02.023
• Languages: English
• ISSNs: 1750-5836

Abstract

Abstract : Highlights: Evolutionary de novo design is used to search for better CO2 absorbent molecules. Structures are optimized with respect to high p K a values in the in silico evolution. p K a computation is approximated by machine learning models and verified by DFT. Machine learning models for viscosity, toxicity, density etc. filter the results. Structures with high p K a and favourable values for multiple properties are suggested. Abstract: Reactive absorption of CO2 by aqueous amine solutions is widely used in the oil, gas and petrochemical industries. However, the energy cost associated with CO2 capture is still a major hurdle to global implementation. Finding new absorbent systems with improved properties is therefore critical to the safety and efficiency of the carbon capture process. Owing to a wide range of tunable properties, imidazoles or imidazoles/amine mixtures have recently been identified as promising to address some of these challenges. In this work, evolutionary de novo design is used to propose new imidazole-based compounds suitable for carbon capture. At the core of this scheme is a genetic algorithm that optimizes the acid dissociation constant (p K a ), an important factor governing the performance of solvents. Calculation of the p K a using quantum chemical methods may produce accurate results but is often too computationally demanding to be suitable for a de novo design setup, where thousands of structures are evaluated over many iterations. To improve the efficacy of the evolutionary process while maintaining high accuracy, we apply a quantitative structure–property relationship (QSPR) model that relates molecular structure descriptors calculated at the semi-empirical level to experimentally determined p K a values. Several promising compounds with high p K a (>10) were identified by the evolutionary design approach and further validated using density functional theory calculations. QSPR models for equally relevant physical properties (such as density, viscosity, vapour pressure) and biodegradability were used as additional filters to ensure that the high p K a structures also have favourable values for these properties.