This repository is tied to the manuscript "A Flexible Theory for Catalysis: Learning Alkaline Oxygen Reduction on Complex Solid Solutions within the Ag-Pd-Pt-Ru Composition Space" found at DOI:10.1002/anie.202307187 and serves to make the experimental and computational data publicly available.
To reproduce the data analysis and figures, it will be necessary to install the utils package. By running 00_get_dists_experiment.py and 00_get_dists_grid.py it is possible to obtain the adsorption energy distributions of *OH and *O, however the folder dist_libraries is also avaible for download at ERDA (4.11GB).
An overview of the repository:
- The features and regression folders hold the scripts construct_feats.py and train_GNN.py. These will create graph-features and train the GNN-model for adsorption energy inference.
- 00_get_dists_experiment.py and 00_get_dists_grid.py will simulate high-entropy alloy surrogate surfaces for both the experimentally samples compositions and for compositions uniformly distributed on a grid.
- 00_plot_volcano.py recreates figure 3b.
- 01_plot_lsvs.py recreates all plots of LSVs and adjusts the current densities of the AgPtRu materials library
- 02_shift_agptru.py fits the discrepancy between materials libraries as described in the supplementary information of the manuscript.
- 03_plot_libraries.py recreates both ternary and 3d plots.
- 04_fit_comp.py fits the activities to the as-deposited alloy compositions.
- 05_fit_volcano.py fits the activities using the theory-derived expression.
- 06_fit_histogram.py fits the activities using SKlearn regression models.
- 07_plot_predictions.py plots activity predictions in ternary plots.
- 08_flexible_comp.py recreates the ternary plots included in figure 4.