A proxy for oxygen storage capacity from high-throughput screening and automated data analysis

Oxygen storage and release is a foundational part of many key pathways in heterogeneous catalysis, such as the Mars-van Krevelen mechanism. However, direct measurement of oxygen storage capacity (OSC) is time-consuming and difficult to parallelise. To accelerate the discovery of stable high OSC rare-earth doped ceria-zirconia oxygen storage catalysts, a high-throughput robotic-based co-precipitation synthesis route was coupled with sequentially automated powder X-ray diffraction (PXRD), Raman and thermogravimetric analysis (TGA) characterisation of the resulting materials libraries. Automated extraction of data enabled rapid trend identification and provided a data set for the development of an OSC prediction model, investigating the significance of each extracted quantity towards OSC. The optimal OSC prediction model produced incorporated variables from only fast-to-measure analytical techniques and gave predicted values of OSC that agreed with experimental observations across an independent validation set. Those measured quantities that feature in the model emerge as proxies for OSC performance. The ability to predict the OSC of the materials accelerates the discovery of high-capacity oxygen storage materials and motivates the development of similar high-throughput workflows to identify candidate catalysts for other heterogeneous transformations. A quantitative proxy model for the slow-to-measure oxygen storage capacity was developed using only fast-to-measure metrics taken from a workflow consisting of high-throughput synthesis, high-throughput screening techniques and automated analysis.