Researchers in the US and Germany have applied an experimentally validated machine learning method to rapidly design and develop advanced gas separation membranes. The work is said to represent the first time this method has been used to selectively design gas-filtering polymer membranes to reduce greenhouse gas emissions.
19 May 2020
Publishing their work in Science Advances, PI Sanat Kumar, Bykhovsky Professor of Chemical Engineering at Columbia Engineering, Columbia University, US, said; ‘Our work points to a new way of materials design and we expect it to revolutionise the field.’
The concept of using membrane technology to separate carbon dioxide, and other gases, for natural gas purification and carbon capture has been around for a while. But there are potentially hundreds of thousands of plastics that could be used, all varying in chemical structure. Manufacturing and testing this number of materials would be an expensive and time consuming process, and to date, only around 1000 have been evaluated as gas separation membranes.
Kumar, who has pioneered developments in improved polymer nano-composites, along with collaborators at the University of South Carolina, and the Max Planck Society in Germany, have created a machine learning algorithm that correlates the chemical structure of the 1000 tested polymers with their gas transport properties, to investigate what structure makes the best membrane. They then applied the algorithm to more than 10 000 known polymers to predict which would produce the best material for the separation process under investigation.
The algorithm identified some 100 polymers that had never been tested for gas transport, but were predicted to surpass the current membrane performance limits for carbon dioxide/methane separations.
To test the accuracy of the algorithm the research team synthesised two of the most promising polymer membranes predicted. It was found that the membranes ‘exceeded the upper levels for carbon dioxide/methane separation performance.’
The study’s co-author Connor Bilchack, post-doctoral fellow at the University of Pennsylvania said ‘Rather than experimentally test all the materials that exist for a particular application, you can instead test a smaller subset of materials which have the most promise. You then find the materials that combine the very best ingredients and that give you a shot at designing a better material…’ The research team added that the methodology has significant potential for commercial use into areas where membrane materials are used, such as water purification.
Science Advances DOI: 10.1126/sciadv.aaz4301