Dr Yongliang (Harry) Yan at Newcastle University was awarded funding in the UKCCSRC’s Flexible Funding 2022 call to look at the “Applying Machine Learning in Screening Perovskite-based Oxygen Carriers for Chemical Looping Applications”.
Chemical looping is a versatile and emerging platform for cost-effective CO2 separation, and sustainable chemical and energy conversion. Oxygen carriers (OCs) constitute the cornerstone of chemical looping processes. The traditional process of developing the desired perovskite-based OCs is based on trial-and-error synthesis and characterisation, which is costly and time-consuming. To overcome this shortcoming, we developed a ML approach that can learn and build a model to make predictions for future OCs in chemical looping applications illustrated in Figure 1.
In this work, we applied artificial neural networks (ANNs) trained by the database of Materials Project to predict oxygen vacancy formation energies of perovskite oxides for chemical looping hydrogen production (CLHP). An analysis of ML model was conducted to identify the relationship between fundamental properties of the perovskite oxide and its oxygen vacancy formation energy. Based on their variance contribution, we found the heat of formation, volume and band gap are the strongest descriptors to predict the oxygen vacancy formation energies of perovskite oxides.
Then the predicted oxygen vacancy formation energies of perovskite oxides were used to evaluate the equilibrium conversions of gases for chemical looping reactions and identify suitable perovskites for CLHP. In this project, a shortlist of top five candidates has been selected by the ML models together with the human expertise, which could be synthesised and tested in the lab-scale reactor for CLHP with the further financial support.
Read more on Yongliang’s Flexible Funding 2022 project page.