ACAT: Alloy Catalysis Automated Toolkit

ACAT is a Python package for atomistic modelling of metal (alloy) (oxide) catalysts used in heterogeneous catalysis. The package is based on automatic identifications of adsorption sites and adsorbate coverages on surface slabs and nanoparticles. Synergizing with ASE, ACAT provides useful tools to build atomistic models and perform global optimization tasks for alloy surfaces and nanoparticles with and without adsorbates. The goal is to automate workflows for the high-throughput screening of alloy catalysts.

ACAT has been developed by Shuang Han at the Section of Atomic Scale Materials Modelling, Department of Energy Conversion and Storage, Technical University of Denmark (DTU) in Lyngby, Denmark.

To use ACAT, please read ACAT documentation (also notebook tutorial and examples). For all symmetry-inequivalent adsorption sites on the surfaces (and nanoparticles) supported in ACAT, please refer to the table of adsorption sites.

Developers:

Shuang Han (hanshuangshiren@gmail.com) - current maintainer

Dependencies

• python>=3.6
• networkx>=2.4
• ase
• asap3 (strongly recommended but not required, since asap3 does not support Windows)

Installation

Install via pip:

pip3 install acat

Alternatively, you can clone the repository:

git clone https://gitlab.com/asm-dtu/acat.git

then go to the installed path and install all dependencies:

pip3 install -r requirements.txt

Finally, install the main package:

python3 setup.py install

Acknowledgements

I would like to highly appreciate the support from BIKE project, where fundings are received from the European Union’s Horizon 2020 Research and Innovation programme under the Marie Skłodowska-Curie Action – International Training Network (MSCA-ITN), grant agreement 813748.

I also want to thank Dr. Steen Lysgaard for the useful scripts and Dr. Giovanni Barcaro, Dr. Alessandro Fortunelli for the useful discussions.

How to cite ACAT

If you find ACAT useful in your research, please cite this paper:

[1] Han, S., Lysgaard, S., Vegge, T. et al. Rapid mapping of alloy surface phase diagrams via Bayesian 
evolutionary multitasking. npj Comput. Mater. 9, 139 (2023). 

If you use ACAT's modules related to symmetric nanoalloy, please also cite this paper:

[2] S. Han, G. Barcaro, A. Fortunelli et al. Unfolding the structural stability of nanoalloys via 
symmetry-constrained genetic algorithm and neural network potential. npj Comput. Mater. 8, 121 (2022).

Notes

1. ACAT was originally developed for metal (alloy) surface slabs and nanoparticles. Therefore H, C, N, O, F, S and Cl atoms are treated as adsorbate molecules and metals are treated as catalyst by default. Now ACAT is generalized for any given surface structure through acat.settings.CustomSurface, which means (mixed) metal oxide surfaces are also allowed. However, note that the H, C, N, O, F, S and Cl atoms at the surface are still always treated as adsorabtes.

2. Some functions distinguishes between nanoparticle and surface slabs based on periodic boundary condition (PBC). Therefore, before using the code, it is recommended (but not required) to set all directions as non-periodic for nanoparticles and at least one direction periodic for surface slabs, and also add vacuum layers to all non-periodic directions. For periodic surface slabs the code will not work if the number of layers is less than 3 (which should be avoided anyways). Each layer always has the same number of atoms as the surface atoms. For stepped surface slabs one layer will have atoms at different z coordinates. However, note that there is no limitation to the size of the cell in the x and y directions. ACAT is able to identify adsorption sites for even a 1x1x3 cell with only one surface atom.

3. ACAT uses a regularized adsorbate string representation. In each adsorbate string, the first element must set to the bonded atom (i.e. the closest non-hydrogen atom to the surface). Hydrogen should always follow the element that it bonds to. For example, water should be written as 'OH2', hydrogen peroxide should be written as 'OHOH', ethanol should be written as 'CH3CH2OH', formyl should be written as 'CHO', hydroxymethylidyne should be written as 'COH', sulfur dioxide can be written either as 'SO2' (S down) or 'O2S' (S up). For fragment that belongs to each indivdual site, the regularized string representation is even more specific as introduced in the acat.utilities.generate_adsorbate_string function. If the string is not supported by the code, it will return the ase.build.molecule instead, which could result in a weird orientation. If the string is not supported by this code nor ASE, you can make your own molecules in the adsorbate_molecule function in acat.settings.

4. There is a bug that causes get_neighbor_site_list() to not return the correct neighbor site indices with ASE version <= 3.18. This is most likely due to shuffling of indices in some ASE functions, which is solved after the release of ASE 3.19.0.

5. If the adsorption site identification by SlabAdsorptionSites is unsatisfactory, it is most likely due to a lattice mismatch of the input surface with the reference surface used for code parameterization. Most of the time this can be resolved by setting optimize_surrogate_cell=True. If the result is still not good enough, you can always use the acat.settings.CustomSurface class to represent your surface.