Explainable Artificial Intelligence Elucidates Synthesis-Structure-Property-Function Relationships in Nanostructured Catalysts

In the latest version of this repository (V3) there are 3 Excel files, 2 Jupyter notebook files, and 1 zipped folder with microscopic images which are submitted as a part of the research manuscript titled "Explainable Artificial Intelligence Elucidates Synthesis-Structure-Property-Function Relationships in Nanostructured Catalysts." The CRISP_dataset.xlsx file comprises data required to run the jupyternotebook titled "CRISP_final_model_.ipynb". The SPIRO_dataset.xlsx file comprises data required to run the jupyternotebook titled "SPIRO_final_model_.ipynb". The CRISP model is a classification model that predicts the formation of single atoms vs nanoparticles based on a set of synthetic parameters and bulk material properties as input features. The SPIRO model predicts the performance of single-atom catalysts for HER and OER based on atomic properties. Both models were developed as part of the study. The Charachterization_and_reaction_data.xlsx files include all the characterization data (including XRD, ICP, XANES, and EXAFS) and electrochemical performance measurement (including Cyclic voltammetry (CV), and Linear sweep voltammetry (LSV) for both OER and HER reactions investigated in this study. All the data are neatly curated in specific tabs within this Excel sheet. The zipped folder contains STEM images of the single-atom catalysts synthesized and tested in the study and presented in the manuscript.