PCAT: Practical CAtalysis Toolkit

Python toolkit package for analyzing, pre-processing and post-processing with density functional theory, cluster expansion, graph neural network, Monte Carlo simulated annealing, genetic algorithm, and active learning workflows in the field of catalysis.

It supports advanced genetic algorithm operators and provides a comprehensive suite of analysis tools, including formation energy, free energy, 3D convex hull construction, scaling relations, activity volcano plots, Pourbaix diagrams, and projected density of states (PDOS). Some examples can be found in examples folder. Especally, active learning workflow can be found in workflow.

Installation

1. Clone the repository:

   git clone https://gitlab.com/changzhiai/pcat.git

2. Enter the installation path:

   cd pcat

3. Install PCAT package:

   python setup.py install

Publications using PCAT and how to cite

1. Density functional theory method with a kinetic model under a practical workflow was applied to screen doped Pd hydrides. The code can be found in the folder instances/instance1_dft and Jupyter-notebook and the details can be found in this paper:

   Metal-doped PdH(111) catalysts for CO₂ reduction. Changzhi Ai, Tejs Vegge and Heine Anton Hansen, ChemSusChem 2022, 15, e202200008.

2. Cluster expansion with Monte Carlo simulated annealing method was applied to study hydrogen impact on CO₂ reduction in PdH_x. The code can be found in the folder instances/instance2_ce_mcsa and the details can be found in this paper:

   Impact of hydrogen concentration for reduction on PdH_x: A combination study of cluster expansion and kinetics analysis. Changzhi Ai, Jin Hyun Chang, Alexander Sougaard Tygesen, Tejs Vegge and Heine Anton Hansen, Journal of Catalysis, 2023, 428, 115188.

3. Cluster expansion with Monte Carlo simulated annealing method was applied to high-throughput compositional screening of metal alloy hydride. The code can be found in the folder instances/instance3_ce_mcsa and the details can be found in this paper:

   High-throughput compositional screening of Pd_xTi_1-xH_y and Pd_xNb_1-xH_y hydrides on CO₂ reduction. Changzhi Ai, Jin Hyun Chang, Alexander Sougaard Tygesen, Tejs Vegge and Heine Anton Hansen, ChemSusChem, 2024, 17, e202301277.

4. Graph neural network with multitasking genetic algorithm method was applied to screen Pd_xTi_1-xH_y with adsorbates under Various CO₂ Reduction Reaction Conditions. The code can be found in the folder instances/instance4_ml_ga and the details can be found in this paper:

   Graph neural network-accelerated multitasking genetic algorithm for optimizing Pd_xTi_1-xH_y surfaces under various CO₂ reduction reaction conditions. Changzhi Ai, Shuang Han, Xin Yang, Tejs Vegge and Heine Anton Hansen, ACS Appl. Mater. Interfaces, 2024, 16, 12563–12572.

Contact

Changzhi Ai (changzhi@stanford.edu/changai@dtu.dk) was at the Section of Atomic Scale Materials Modelling, Department of Energy Conversion and Storage, Technical University of Denmark. He is currently a postdoctoral reasearcher, SUNCAT Center for Interface Science and Catalysis, Stanford University & SLAC National Accelerator Laboratory, USA.