NEBscape

NEBscape automates the sampling of minimum energy paths (MEPs) for surface reactions, combining global geometry optimization, atom-mapping, and NEB calculations into a single workflow — requiring minimal human input.

Installation from source

git clone https://gitlab.mpcdf.mpg.de/hjung/nebscape.git
pip install .

Required Inputs

This workflow prepares initial and final geometry with global optimization (e.g. minima hopping) and provides automatic atom-mapping and generates possible permutation of symmetric atoms. Only following three inputs are required.

1. List of reaction SMILES string in reaction_smiles.txt (e.g. O=[C][CH2][O]>>[C-]#[O+].[CH2]=O)
2. slab information in slab.xyz (Slab should use tag to indicate which atoms are fixed (tag=0) and which atoms are relaxed(tag=1))
3. Calculator (e.g. MLIP like MACE)

Setting

This workflow uses wfl and expyre for remote submission, so configure the following related settings

• config.json file has to be set for remote submission to HPC (Can be found in /example)
• According to config.json, the remote_info entries should be adapted to your environment.

Quick Start

A minimal, fully local example (no HPC/remote submission, ASE's built-in EMT calculator) is provided in example/0_EMT. It contains everything needed to run the workflow from input to interpolations for NEB :

Note: It uses autoadsorbate to place adsorbate and relaxes only without minima hopping. The EMT calculator can dissociate the adsorbate during relaxation, so Hookean constraints are imposed and the post-alignment relaxation is skipped. This example demonstrates how NEBscape prepares an initial reaction trajectory from the inputs.

Run it:

git clone https://gitlab.mpcdf.mpg.de/hjung/nebscape.git
cd nebscape
pip install .

cd example/0_EMT

# run the whole pipeline (No NEBs because EMT is not good with this system)
python setting.py --run isolated dimers prepare

Documentation

• Home - overview
• Installation - installation
• Setting up Local Environment - setting for wfl and expyre
• Input Preparation - input file preparation
  • Setting File - configuration settings
• Geometry Generation - generating initial geometries
  • Minima Hopping - running minima hopping
  • Autoadsorbate - adsorbate placement
• Interpolation Generation - generating initial minimum energy paths
  • Atom Mapping - mapping atoms between endpoints
  • Geometry Selection - selecting endpoint geometries
  • FS-to-IS Alignment - aligning final to initial state
  • Index Permutation - permuting atom indices
• Running NEB - running the NEB calculation

(We are currently working on improving the documentation / tutorial. Let us know if you need help or suggestions.)

References

If you use this code, please cite our paper:

Jung, H., Colombi Manzi, E., Goncalves, T.J. et al. From global optimization to transition state search: an automated workflow for surface reaction barriers. npj Comput Mater 12, 300 (2026). https://doi.org/10.1038/s41524-026-02298-1

@article{jung2026global,
  title={From global optimization to transition state search: an automated workflow for surface reaction barriers},
  author={Jung, Hyunwook and Colombi Manzi, Emanuel and Goncalves, Tiago J and Bukas, Vanessa J and De, Sandip and Margraf, Johannes T and Reuter, Karsten and Heenen, Hendrik H},
  journal={npj Computational Materials},
  volume={12},
  number={1},
  pages={300},
  year={2026},
  publisher={Nature Publishing Group UK London}
}

Contact

If you have any questions, please contact us at hjung@fhi.mpg.de Also for any bugs or issues, you can use gitlab issues.

License

The NEBscape code is published and distributed under the MIT License.