Computational Materials Discovery for Carbon Dioxide Capture Applications

Artificial intelligence (AI) is aiding the discovery of sustainable materials each step along the computational workflow. Machine learning (ML) supports the automated extraction of data from the literature, the creation of simulation data sets, the generative design of new materials, as well as their computational performance assessment. In this talk, I will present our team’s research in computational discovery of polymers [1, 2] and nano-porous materials [3-5] for carbon dioxide capture applications. I will discuss some of the challenges in the AI/ML design of complex materials and provide examples of how discovery outcomes can be computationally validated, prior to lab synthesis and characterization. In view of global sustainability challenges such as climate change, open-science strategies with publicly shared data and models are needed for accelerating computational materials discovery.



References



[1] Giro, R., Hsu, H., Kishimoto, A. et al. AI powered, automated discovery of polymer membranes for carbon capture. NPJ Comput Mater 9, 133 (2023). https://doi.org/10.1038/s41524-023-01088-3.



[2] Ferrari, B.S., Manica, M., Giro, R. et al. Predicting polymerization reactions via transfer learning using chemical language models. arXiv preprint (2023). https://arxiv.org/abs/2310.11423.



[3] Oliveira, F.L., Cleeton, C., Neumann Barros Ferreira, R. et al. CRAFTED: An exploratory database of simulated adsorption isotherms of metal-organic frameworks. Sci Data 10, 230 (2023). https://doi.org/10.1038/s41597-023-02116-z.



[4] Zheng, B., Lopes Oliveira, F., Neumann Barros Ferreira, R. et al. Quantum Informed Machine-Learning Potentials for Molecular Dynamics Simulations of CO₂’s Chemisorption and Diffusion in Mg-MOF-74. ACS Nano 17 (6), 5579-5587 (2023). https://doi.org/10.1021/acsnano.2c11102.



[5] Cipcigan, F., Booth, J., Neumann Barros Ferreira, R. et al. Discovery of Novel Reticular Materials for Carbon Dioxide Capture using GflowNets. arXiv preprint (2023). https://arxiv.org/abs/2310.07671.