CHGNet_phonopy

code to generate 2nd order interatomic force constants from phonopy using CHGNet

Usage

you can test the code by running the following:

python chgnet_phonopy_run.py \ --atoms_path='POSCAR' \ --relax=True \ --supercell-dims=222 \ --disp=0.02 \ --num_rand_disp=None \ --output_disp=True \ --pretrained_model=True \ --model_path=None \ --stability_criteria=-0.1 \ --output_ph_band=True
the code will generate these files:
1-**FORCE_CONSTANTS**: 2nd order IFCs in phonopy format
2-**stability**: it shows the words "stable" or "unstable" based on the "stability_criteria" argument
3- **band.conf**: file that can be used later by phonopy to plot the phonon dispersion
4- **orig_band.conf**: file that has the q-points or phonon wavevectors that were used to get the frequencies to output the **stability** file
5- **SPOSCAR_###**: supercell of the input structure where # represents the supercell dimension

**Note:** you have to put both files "*chgnet_phonopy_class.py*" and "*chgnet_phonopy_run.py*" in the same path since "*chgnet_phonopy_run.py*" inherits classes from "*chgnet_phonopy_class.py*". The code will output the above files in the same path where you put "*chgnet_phonopy_class.py*" and "*chgnet_phonopy_run.py*".

Args Explanation

--atoms_path: structure path ('./POSCAR' by default)
--relax: whether chgnet will relax the structure before the displacments or not (False by default)
--supercell-dims: supercell dimensions (222 by default)
--disp: atomic displacement amplitude in Angstroms (0.01 by default)
--num_rand_disp: # of random displacements. you might have to install alm to produce 2nd order IFCs (None by default)
--output_disp: whether to output the displacements in POSCAR format or not (True by default)
--pretrained_model: whether to use the pretrained chgnet model (True by default)
--model_path: new chgnet model path if the pretrained model is not used (None by default)
--stability_criteria: frequency stability threshold. If one frequency is less than that value, "unstable" is written on stability file (-0.1 by default)
--output_ph_band: output phonon dispersion plot in file phonopy_bands_dos.png(True by default)\

Required Packages

the code is tested on the following packages and versions: torch=2.0.1 pymatgen=2023.11.12 ase=3.23.0 chgnet=0.3.5 phonopy=2.20.0
The code can probably work with different versions of the above packages

Credit and Citation

• please cite this journal article below since the code was used to produce the article:
  Ojih, J.; Al-Fahdi, M.; Yao, Y.; Hu, J.; Hu, M. Graph Theory and Graph Neural Network Assisted High-Throughput Crystal Structure Prediction and Screening for Energy Conversion and Storage. Journal of Materials Chemistry A 2024, 12 (14), 8502–8515
• please consider reading my published work in Google Scholar using this link thank you :)
• also please let me know if more features are needed to be added and/or improved