pith. machine review for the scientific record. sign in

arxiv: 2605.09394 · v1 · submitted 2026-05-10 · ⚛️ physics.chem-ph · cond-mat.mtrl-sci

Recognition: 2 theorem links

· Lean Theorem

Systematic Fine-Tuning of MACE Interatomic Potentials for Catalysis

Christopher Sutton, Derek Vigil-Fowler, G\'abor Cs\'anyi, Jacob Clary, Nima Karimitari, Ravishankar Sundararaman

Authors on Pith no claims yet

Pith reviewed 2026-05-12 02:26 UTC · model grok-4.3

classification ⚛️ physics.chem-ph cond-mat.mtrl-sci
keywords interatomic potentialscatalysisfine-tuningmachine learningreaction energiesbimetallic alloysmetal oxidestransferability
0
0 comments X

The pith

Fine-tuning large MACE models on targeted catalytic configurations outperforms from-scratch training and enables accurate screening of bimetallic alloys.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper compares nine machine-learned interatomic potentials for modeling catalytic reactions and shows that fine-tuning pre-trained models on specific data sets yields lower errors in reaction energies and barriers than training new models from scratch. Fine-tuning makes the potentials less sensitive to exact data sampling choices and allows them to transfer effectively to reactions outside the training distribution, such as oxygen evolution on metal oxides after training primarily on metals. A model fine-tuned on 49,860 configurations delivers the strongest results overall and supports screening of many bimetallic alloys with 0.15 eV mean absolute error for reaction energies, including adsorbates on high-index surfaces never seen during training.

Core claim

Fine-tuning of MACE foundation models on metallic catalysts with a mix of relaxation trajectories and 5-10% perturbed high-energy structures reduces errors more than twofold relative to from-scratch models. The resulting potentials achieve 0.30 eV MAE on out-of-distribution OER reactions on iridium oxide polymorphs and 0.19 eV barrier MAE for CO2 reduction on copper, while the largest fine-tuned model on 49,860 configurations provides the best cross-system performance and screens left-out bimetallic alloys at 0.15 eV MAE even for adsorbates on unseen Miller-index surfaces such as (532).

What carries the argument

Fine-tuning of MACE foundation models using relaxation trajectories plus 5-10% perturbed high-energy structures sampled from molecular dynamics or contour exploration, which improves transferability across metallic, oxide, and bimetallic catalytic systems.

If this is right

  • Large numbers of bimetallic alloy catalysts can be screened for reaction energies without collecting new training data for each composition.
  • Reaction modeling becomes feasible on high Miller-index surfaces without explicit inclusion of those surfaces in the training set.
  • Fine-tuned models support out-of-distribution reactions such as oxide oxygen evolution after training on metallic systems.
  • The overall data requirements for reaching sub-0.2 eV accuracy in catalysis modeling are reduced compared with from-scratch approaches.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Similar fine-tuning strategies could be tested on reactions outside catalysis, such as battery electrolyte decomposition or organic synthesis pathways.
  • Embedding these potentials into molecular dynamics runs may allow exploration of finite-temperature effects and rare events in catalytic mechanisms.
  • Extending the element set in the foundation model and fine-tuning data could produce broader-coverage potentials for multi-component catalyst design.

Load-bearing premise

The 141-reaction evaluation set together with the chosen training configurations sufficiently represent the chemical space of catalytic reactions to support claims of broad transferability.

What would settle it

A new collection of catalytic reactions or surface orientations where the fine-tuned model produces mean absolute errors for reaction energies or barriers that exceed 0.30 eV by a large margin.

read the original abstract

Once trained, machine-learned interatomic potentials (MLIPs) provide a fast and accurate way to study catalytic reaction pathways, but their performance strongly depends on the training set. Here, we compare nine MLIPs trained with different data sets and strategies, including from-scratch (FS) training and fine-tuning (FT) of large foundation models. The models are evaluated on reaction energies, $E_{r}$, and reaction energy barriers, $E_{a}$, for 141 reactions, including CO$_2$ reduction to C$_2$ and C$_3$ products, propane dehydrogenation, hydrogen intercalation on Pd, and out-of-distribution oxygen evolution reaction (OER) on metal oxides. FS models trained with 5%--10% perturbed high-energy configurations from molecular dynamics or contour exploration reduce the error by more than twofold compared with models trained only on relaxation trajectories. In contrast, FT MLIPs are less sensitive to sampling and transfer well to out-of-distribution reactions. An MLIP fine-tuned on metallic catalysts achieves a 0.30 eV MAE for OER on iridium oxide polymorphs, outperforming out-of-the-box MACE-MH-1 by 0.08 eV and the best FS model by 0.14 eV. A model fine-tuned to O and OH adsorption on metal oxides gives a 0.19 eV reaction-barrier MAE for out-of-distribution CO$_2$RR on Cu, comparable to an FS model trained on in-distribution C--C bond-breaking reactions. Finally, a large MLIP fine-tuned on 49,860 configurations gives the best overall performance across metallic and metal-oxide catalysts and was used to screen a large left-out set of bimetallic alloys, achieving a 0.15 eV MAE for $E_{r}$, even for adsorbates on unseen Miller-index surfaces such as (532). This work identifies the training configurations needed for accurate FS and FT MLIPs for catalytic reaction modeling.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 1 minor

Summary. The manuscript compares nine MACE-based MLIPs trained from scratch (FS) or via fine-tuning (FT) on varying datasets, including 49,860 configurations for the largest FT model. It reports that including 5-10% perturbed high-energy structures improves FS models more than twofold on reaction energies (Er) and barriers (Ea) for 141 reactions spanning CO2 reduction, propane dehydrogenation, Pd hydrogen intercalation, and OER on oxides. FT models show greater robustness to sampling choices and better transfer to out-of-distribution cases, with the largest FT model achieving the best overall performance and a 0.15 eV MAE on Er for a left-out bimetallic alloy screening set that includes adsorbates on unseen Miller-index surfaces such as (532).

Significance. The work supplies concrete, held-out MAE numbers across 141 reactions with direct FS-versus-FT and in-distribution versus OOD comparisons, which strengthens the practical guidance on training strategies. If the transferability results hold, the findings would help researchers select data-sampling protocols that improve accuracy for catalytic reaction modeling and large-scale screening without requiring exhaustive in-distribution data.

major comments (1)
  1. [Abstract] Abstract and training-configuration description: the headline transferability result (0.15 eV MAE on left-out bimetallics, including (532) surfaces) rests on the claim that the 141-reaction benchmark plus 5-10% perturbed high-energy structures from MD/contour exploration adequately span catalytic chemical space. No coverage statistics (e.g., bond-type or facet distributions) or ablation on the data-selection criteria are supplied, so it remains unclear whether the reported advantage over FS baselines and out-of-the-box MACE-MH-1 reflects genuine extrapolation or partial overlap between train and eval distributions.
minor comments (1)
  1. [Abstract] The abstract states that nine MLIPs were compared but does not list their exact training-set sizes or compositions in one place, which would improve readability of the FS-versus-FT contrast.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their constructive feedback. Below we respond point-by-point to the major comment, agreeing that additional coverage analysis would strengthen the transferability claims.

read point-by-point responses

  1. Referee: [Abstract] Abstract and training-configuration description: the headline transferability result (0.15 eV MAE on left-out bimetallics, including (532) surfaces) rests on the claim that the 141-reaction benchmark plus 5-10% perturbed high-energy structures from MD/contour exploration adequately span catalytic chemical space. No coverage statistics (e.g., bond-type or facet distributions) or ablation on the data-selection criteria are supplied, so it remains unclear whether the reported advantage over FS baselines and out-of-the-box MACE-MH-1 reflects genuine extrapolation or partial overlap between train and eval distributions.

    Authors: We agree that quantitative coverage statistics and ablations on data-selection criteria would better substantiate the transferability results. The 141-reaction benchmark spans CO2 reduction (including C-C bond formation), propane dehydrogenation, Pd hydrogen intercalation, and OER on oxides, with training sets drawn from relaxation trajectories augmented by 5-10% high-energy structures from MD and contour exploration. The OOD tests (metallic FT model on oxide OER; oxide FT model on Cu CO2RR) and the bimetallic screening set (including unseen (532) surfaces) are intended to probe extrapolation beyond the training distributions. However, the original manuscript does not supply explicit metrics such as bond-type or facet distributions. In the revised version we will add a dedicated subsection (and supplementary figure) reporting elemental compositions, bond-type frequencies (e.g., C-C, O-H, metal-adsorbate), and Miller-index coverage for both training and evaluation sets. We will also include a short discussion of the rationale for the 5-10% high-energy fraction, supported by the observed error reductions, and a limited ablation on sampling choices where space permits. These additions will clarify the degree of overlap versus genuine extrapolation while preserving the core findings. revision: yes

Circularity Check

0 steps flagged

No significant circularity; empirical results rest on held-out independent references

full rationale

The paper reports MAE values for reaction energies and barriers on explicitly held-out sets (141 reactions plus left-out bimetallic alloys and unseen Miller indices) using separate reference calculations. No equation or central claim reduces by construction to a fitted parameter from the same data, no self-definitional loop appears in the training/evaluation protocol, and no load-bearing uniqueness theorem or ansatz is imported via self-citation. The comparisons to FS models and the base MACE-MH-1 are external benchmarks, not internal redefinitions. All performance numbers are therefore falsifiable against independent DFT data.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The central claims rest on standard assumptions about MLIP accuracy and on the representativeness of the chosen training and test configurations; no new entities are postulated.

free parameters (1)
  • fraction of perturbed high-energy configurations
    The 5-10% fraction is selected to improve FS model accuracy on reaction barriers.
axioms (1)
  • domain assumption MLIPs trained on DFT data can approximate reaction energies and barriers with errors below 0.3 eV when the training distribution covers relevant configurations
    Invoked implicitly when claiming that fine-tuned models generalize to out-of-distribution OER and CO2RR.

pith-pipeline@v0.9.0 · 5695 in / 1357 out tokens · 55953 ms · 2026-05-12T02:26:23.141868+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

Reference graph

Works this paper leans on

61 extracted references · 61 canonical work pages · 1 internal anchor

  1. [1]

    Luna, P.D., Hahn, C., Higgins, D., Jaffer, S.A., Jaramillo, T.F., Sargent, E.H.: What would it take for renewably powered electrosynthesis to displace petrochem- ical processes? Science364(6438), 3506 (2019) https://doi.org/10.1126/science. aav3506

    work page doi:10.1126/science 2019

  2. [2]

    Nature Catalysis2(8), 648–658 (2019) https://doi.org/10.1038/s41929-019-0306-7

    Ross, M.B., De Luna, P., Li, Y., Dinh, C.-T., Kim, D., Yang, P., Sargent, E.H.: Designing materials for electrochemical carbon dioxide recycling. Nature Catalysis2(8), 648–658 (2019) https://doi.org/10.1038/s41929-019-0306-7

    work page doi:10.1038/s41929-019-0306-7 2019

  3. [3]

    Heterogeneous Catalysis and a Sustainable Future, pp. 1–5. John Wiley & Sons, Ltd (2014). Chap. 1. https://doi.org/10.1002/9781118892114.ch1 . https: //onlinelibrary.wiley.com/doi/abs/10.1002/9781118892114.ch1

    work page doi:10.1002/9781118892114.ch1 2014

  4. [4]

    and Jónsson, Hannes , year =

    Henkelman, G., Uberuaga, B.P., J´ onsson, H.: A climbing image nudged elastic band method for finding saddle points and minimum energy paths. The Journal of Chemical Physics113(22), 9901–9904 (2000) https://doi.org/10.1063/1.1329672 https://pubs.aip.org/aip/jcp/article- pdf/113/22/9901/19259681/9901 1 online.pdf

    work page doi:10.1063/1.1329672 2000

  5. [5]

    The Journal of Chemical Physics111(15), 7010–7022 (1999) https://doi.org/10.1063/1.480097

    Henkelman, G., J´ onsson, H.: A dimer method for finding saddle points on high dimensional potential surfaces using only first derivatives. The Journal of Chemical Physics111(15), 7010–7022 (1999) https://doi.org/10.1063/1.480097

    work page doi:10.1063/1.480097 1999

  6. [6]

    The Journal of Chemical Physics130(24), 244108 (2009) https://doi.org/10.1063/1.3156312

    Goodrow, A., Bell, A.T., Head-Gordon, M.: Transition state-finding strategies for use with the growing string method. The Journal of Chemical Physics130(24), 244108 (2009) https://doi.org/10.1063/1.3156312

    work page doi:10.1063/1.3156312 2009

  7. [7]

    The Journal of Physical Chemistry Letters10(15), 4401–4408 (2019) https://doi.org/10.1021/acs.jpclett.9b01428 https://doi.org/10.1021/acs.jpclett.9b01428

    Back, S., Yoon, J., Tian, N., Zhong, W., Tran, K., Ulissi, Z.W.: Convolu- tional neural network of atomic surface structures to predict binding energies for high-throughput screening of catalysts. The Journal of Physical Chemistry Letters10(15), 4401–4408 (2019) https://doi.org/10.1021/acs.jpclett.9b01428 https://doi.org/10.1021/acs.jpclett.9b01428. PMID:...

    work page doi:10.1021/acs.jpclett.9b01428 2019

  8. [8]

    Nature Catalysis1(9), 696–703 (2018) https://doi.org/10.1038/s41929-018-0142-1

    Tran, K., Ulissi, Z.W.: Active learning across intermetallics to guide discovery of electrocatalysts for CO2 reduction and H2 evolution. Nature Catalysis1(9), 696–703 (2018) https://doi.org/10.1038/s41929-018-0142-1

    work page doi:10.1038/s41929-018-0142-1 2018

  9. [9]

    Li, Z., Wang, S., Chin, W.S., Achenie, L.E., Xin, H.: High-throughput screening of bimetallic catalysts enabled by machine learning. J. Mater. Chem. A5, 24131– 24138 (2017) https://doi.org/10.1039/C7TA01812F

    work page doi:10.1039/c7ta01812f 2017

  10. [10]

    Current Opinion in Solid State and Materials Science35, 101214 (2025) https://doi.org/10.1016/j.cossms

    Jacobs, R., Morgan, D., Attarian, S., Meng, J., Shen, C., Wu, Z., Xie, C.Y., Yang, J.H., Artrith, N., Blaiszik, B., Ceder, G., Choudhary, K., Csanyi, G., Cubuk, E.D., Deng, B., Drautz, R., Fu, X., Godwin, J., Honavar, V., Isayev, O., Johansson, A., Kozinsky, B., Martiniani, S., Ong, S.P., Poltavsky, I., Schmidt, K., Takamoto, 42 S., Thompson, A.P., Wester...

    work page doi:10.1016/j.cossms 2025

  11. [11]

    Acero, Z

    Gilmer, J., Schoenholz, S.S., Riley, P.F., Vinyals, O., Dahl, G.E.: Neural Message Passing for Quantum Chemistry. arXiv (2017). https://doi.org/10.48550/ARXIV. 1704.01212 . https://arxiv.org/abs/1704.01212

    work page internal anchor Pith review doi:10.48550/arxiv 2017

  12. [12]

    Nature Communications 13(1), 2453 (2022)

    Batzner, S., Musaelian, A., Sun, L., Geiger, M., Mailoa, J.P., Kornbluth, M., Molinari, N., Smidt, T.E., Kozinsky, B.: E(3)-equivariant graph neural networks for data-efficient and accurate interatomic potentials. Nature Communications 13(1), 2453 (2022)

    work page 2022

  13. [13]

    Batatia,et al., The Design Space of E(3)-Equivariant Atom-Centered Interatomic Potentials

    Batatia, I., Batzner, S., Kov´ acs, D.P., Musaelian, A., Simm, G.N.C., Drautz, R., Ortner, C., Kozinsky, B., Csanyi, G.: The Design Space of E(3)-Equivariant Atom-Centered Interatomic Potentials (2022). http://arxiv.org/abs/2205.06643

    work page arXiv 2022

  14. [14]

    Lawrence and Ulissi, Zachary , year=

    Chanussot, L., Das, A., Goyal, S., Lavril, T., Shuaibi, M., Riviere, M., Tran, K., Heras-Domingo, J., Ho, C., Hu, W., Palizhati, A., Sriram, A., Wood, B., Yoon, J., Parikh, D., Zitnick, C.L., Ulissi, Z.: Open catalyst 2020 (oc20) dataset and community challenges. ACS Catalysis11(10), 6059–6072 (2021) https://doi.org/ 10.1021/acscatal.0c04525 https://doi.o...

    work page doi:10.1021/acscatal.0c04525 2020

  15. [15]

    Wood, Misko Dzamba, Xiang Fu, Meng Gao, Muhammed Shuaibi, Luis Barroso-Luque, Kareem Abdelmaqsoud, Vahe Gharakhanyan, John R

    Wood, B.M., Dzamba, M., Fu, X., Gao, M., Shuaibi, M., Barroso-Luque, L., Abdelmaqsoud, K., Gharakhanyan, V., Kitchin, J.R., Levine, D.S., Michel, K., Sriram, A., Cohen, T., Das, A., Rizvi, A., Sahoo, S.J., Ulissi, Z.W., Zitnick, C.L.: UMA: A Family of Universal Models for Atoms (2025). https://arxiv.org/abs/ 2506.23971

    work page arXiv 2025

  16. [16]

    ACS Catalysis13(5), 3066–3084 (2023) https://doi.org/10.1021/acscatal.2c05426 https://doi.org/10.1021/acscatal.2c05426

    Tran, R., Lan, J., Shuaibi, M., Wood, B.M., Goyal, S., Das, A., Heras- Domingo, J., Kolluru, A., Rizvi, A., Shoghi, N., Sriram, A., Therrien, F., Abed, J., Voznyy, O., Sargent, E.H., Ulissi, Z., Zitnick, C.L.: The open catalyst 2022 (oc22) dataset and challenges for oxide electrocatalysts. ACS Catalysis13(5), 3066–3084 (2023) https://doi.org/10.1021/acsca...

    work page doi:10.1021/acscatal.2c05426 2022

  17. [17]

    Levine, Zachary Ulissi, C

    Sahoo, S.J., Maraschin, M., Levine, D.S., Ulissi, Z., Zitnick, C.L., Varley, J.B., Gauthier, J.A., Govindarajan, N., Shuaibi, M.: The Open Catalyst 2025 (OC25) Dataset and Models for Solid-Liquid Interfaces (2025). https://arxiv.org/abs/ 2509.17862

    work page arXiv 2025

  18. [18]

    Levine, Muhammed Shuaibi, Evan Walter Clark Spotte-Smith, Michael G

    Levine, D.S., Shuaibi, M., Spotte-Smith, E.W.C., Taylor, M.G., Hasyim, M.R., Michel, K., Batatia, I., Cs´ anyi, G., Dzamba, M., Eastman, P., Frey, N.C., Fu, X., Gharakhanyan, V., Krishnapriyan, A.S., Rackers, J.A., Raja, S., Rizvi, A., Rosen, A.S., Ulissi, Z., Vargas, S., Zitnick, C.L., Blau, S.M., Wood, B.M.: The Open Molecules 2025 (OMol25) Dataset, Eva...

    work page arXiv 2025

  19. [19]

    Brabson, Xiaohan Yu, Sihoon Choi, Kareem Abdelmaqsoud, Elias Moubarak, Pim de Haan, Sindy Löwe, Johann Brehmer, John R

    Sriram, A., Brabson, L.M., Yu, X., Choi, S., Abdelmaqsoud, K., Moubarak, E., Haan, P., L¨ owe, S., Brehmer, J., Kitchin, J.R., Welling, M., Zitnick, C.L., Ulissi, Z., Medford, A.J., Sholl, D.S.: The Open DAC 2025 Dataset for Sorbent Discovery in Direct Air Capture (2025). https://arxiv.org/abs/2508.03162

    work page arXiv 2025

  20. [20]

    Barroso-Luque et al., Open Materials 2024 (OMat24) Inorganic Materials Dataset and Models

    Barroso-Luque, L., Shuaibi, M., Fu, X., Wood, B.M., Dzamba, M., Gao, M., Rizvi, A., Zitnick, C.L., Ulissi, Z.W.: Open Materials 2024 (OMat24) Inorganic Materials Dataset and Models (2024). https://arxiv.org/abs/2410.12771

    work page arXiv 2024

  21. [21]

    Scientific Data10(1), 11 (2023) https://doi.org/10.1038/ s41597-022-01882-6

    Eastman, P., Behara, P.K., Dotson, D.L., Galvelis, R., Herr, J.E., Horton, J.T., Mao, Y., Chodera, J.D., Pritchard, B.P., Wang, Y., De Fabritiis, G., Markland, T.E.: Spice, a dataset of drug-like molecules and peptides for training machine learning potentials. Scientific Data10(1), 11 (2023) https://doi.org/10.1038/ s41597-022-01882-6

    work page 2023

  22. [22]

    Scientific Data10(1), 145 (2023) https://doi.org/10.1038/ s41597-023-02043-z

    Zhao, Q., Vaddadi, S.M., Woulfe, M., Ogunfowora, L.A., Garimella, S.S., Isayev, O., Savoie, B.M.: Comprehensive exploration of graphically defined reaction spaces. Scientific Data10(1), 145 (2023) https://doi.org/10.1038/ s41597-023-02043-z

    work page 2023

  23. [23]

    Monkhorst, H.J., Pack, J.D.: Special points for brillouin-zone integrations. Phys. Rev. B13, 5188–5192 (1976) https://doi.org/10.1103/PhysRevB.13.5188

    work page doi:10.1103/physrevb.13.5188 1976

  24. [24]

    npj Computational Materials11(1), 352 (2025) https://doi.org/10.1038/ s41524-025-01834-9

    Kuner, M.C., Kaplan, A.D., Persson, K.A., Asta, M., Chrzan, D.C.: Mp- aloe: an r2scan dataset for universal machine learning interatomic poten- tials. npj Computational Materials11(1), 352 (2025) https://doi.org/10.1038/ s41524-025-01834-9

    work page 2025

  25. [25]

    Elena, Dávid P

    Batatia, I., Benner, P., Chiang, Y., Elena, A.M., Kov´ acs, D.P., Riebesell, J., Advincula, X.R., Asta, M., Avaylon, M., Baldwin, W.J., Berger, F., Bernstein, N., Bhowmik, A., Blau, S.M., C˘ arare, V., Darby, J.P., De, S., Pia, F.D., Deringer, V.L., Elijoˇ sius, R., El-Machachi, Z., Falcioni, F., Fako, E., Ferrari, A.C., Genreith- Schriever, A., George, J...

    work page arXiv 2024

  26. [26]

    The Journal of 44 Chemical Physics120(21), 9911–9917 (2004) https://doi.org/10.1063/1.1724816

    Goedecker, S.: Minima hopping: An efficient search method for the global mini- mum of the potential energy surface of complex molecular systems. The Journal of 44 Chemical Physics120(21), 9911–9917 (2004) https://doi.org/10.1063/1.1724816

    work page doi:10.1063/1.1724816 2004

  27. [27]

    npj Computational Materials9(1), 180 (2023)

    Schaaf, L.L., Fako, E., De, S., Sch¨ afer, A., Cs´ anyi, G.: Accurate energy barriers for catalytic reaction pathways: an automatic training protocol for machine learning force fields. npj Computational Materials9(1), 180 (2023)

    work page 2023

  28. [28]

    Cross learning between electronic structure theories for unifying molecular, surface, and inorganic crystal foundation force fields

    Batatia, I., Lin, C., Hart, J., Kasoar, E., Elena, A.M., Norwood, S.W., Wolf, T., Cs´ anyi, G.: Cross Learning between Electronic Structure Theories for Unifying Molecular, Surface, and Inorganic Crystal Foundation Force Fields (2025). https: //arxiv.org/abs/2510.25380

    work page arXiv 2025

  29. [29]

    QUANTUM ESPRESSO: a modular and open-source software project for quantum simulations of materials,

    Giannozzi, P., Baroni, S., Bonini, N., Calandra, M., Car, R., Cavazzoni, C., Ceresoli, D., Chiarotti, G.L., Cococcioni, M., Dabo, I., Dal Corso, A., Giron- coli, S., Fabris, S., Fratesi, G., Gebauer, R., Gerstmann, U., Gougoussis, C., Kokalj, A., Lazzeri, M., Martin-Samos, L., Marzari, N., Mauri, F., Mazzarello, R., Paolini, S., Pasquarello, A., Paulatto,...

    work page doi:10.1088/0953-8984/21/39/395502 2009

  30. [30]

    Computational Materials Science81, 446–452 (2014) https://doi.org/10.1016/j.commatsci.2013.08.053

    Garrity, K.F., Bennett, J.W., Rabe, K.M., Vanderbilt, D.: Pseudopotentials for high-throughput dft calculations. Computational Materials Science81, 446–452 (2014) https://doi.org/10.1016/j.commatsci.2013.08.053

    work page doi:10.1016/j.commatsci.2013.08.053 2014

  31. [31]

    Wellendorff, J., Lundgaard, K.T., Møgelhøj, A., Petzold, V., Landis, D.D., Nørskov, J.K., Bligaard, T., Jacobsen, K.W.: Density functionals for surface sci- ence: Exchange-correlation model development with bayesian error estimation. Phys. Rev. B85, 235149 (2012) https://doi.org/10.1103/PhysRevB.85.235149

    work page doi:10.1103/physrevb.85.235149 2012

  32. [32]

    The Journal of Chemical Physics144(16), 164109 (2016) https://doi.org/10.1063/1.4947024 https://pubs.aip.org/aip/jcp/article- pdf/doi/10.1063/1.4947024/13607146/164109 1 online.pdf

    Packwood, D., Kermode, J., Mones, L., Bernstein, N., Woolley, J., Gould, N., Ortner, C., Cs´ anyi, G.: A universal preconditioner for simulating con- densed phase materials. The Journal of Chemical Physics144(16), 164109 (2016) https://doi.org/10.1063/1.4947024 https://pubs.aip.org/aip/jcp/article- pdf/doi/10.1063/1.4947024/13607146/164109 1 online.pdf

    work page doi:10.1063/1.4947024 2016

  33. [33]

    The Journal of Chemical Physics140(21), 214106 (2014) https://doi.org/10.1063/1.4878664

    Smidstrup, S., Pedersen, A., Stokbro, K., J´ onsson, H.: Improved initial guess for minimum energy path calculations. The Journal of Chemical Physics140(21), 214106 (2014) https://doi.org/10.1063/1.4878664

    work page doi:10.1063/1.4878664 2014

  34. [34]

    Journal of Physics: Condensed Matter33(44), 445901 (2021) https: //doi.org/10.1088/1361-648X/ac1af0

    Waters, M.J., Rondinelli, J.M.: Energy contour exploration with potentiostatic kinematics. Journal of Physics: Condensed Matter33(44), 445901 (2021) https: //doi.org/10.1088/1361-648X/ac1af0

    work page doi:10.1088/1361-648x/ac1af0 2021

  35. [35]

    Advances in Neural Information Processing Systems35, 11423–11436 (2022) 45

    Batatia, I., Kovacs, D.P., Simm, G., Ortner, C., Cs´ anyi, G.: Mace: Higher order equivariant message passing neural networks for fast and accurate force fields. Advances in Neural Information Processing Systems35, 11423–11436 (2022) 45

    work page 2022

  36. [36]

    Nature Machine Intelligence7(1), 56–67 (2025) https://doi.org/10.1038/s42256-024-00956-x

    Batatia, I., Batzner, S., Kov´ acs, D.P., Musaelian, A., Simm, G.N.C., Drautz, R., Ortner, C., Kozinsky, B., Cs´ anyi, G.: The design space of E(3)-equivariant atom- centred interatomic potentials. Nature Machine Intelligence7(1), 56–67 (2025) https://doi.org/10.1038/s42256-024-00956-x

    work page doi:10.1038/s42256-024-00956-x 2025

  37. [37]

    Jour- nal of Physics: Condensed Matter29(27), 273002 (2017) https://doi.org/10.1088/ 1361-648X/aa680e

    Larsen, A.H., Mortensen, J.J., Blomqvist, J., Castelli, I.E., Christensen, R., Du lak, M., Friis, J., Groves, M.N., Hammer, B., Hargus, C., Hermes, E.D., Jen- nings, P.C., Jensen, P.B., Kermode, J., Kitchin, J.R., Kolsbjerg, E.L., Kubal, J., Kaasbjerg, K., Lysgaard, S., Maronsson, J.B., Maxson, T., Olsen, T., Pastewka, L., Peterson, A., Rostgaard, C., Sch...

    work page 2017

  38. [38]

    The Journal of Chemical Physics81(1), 511–519 (1984) https://doi.org/10.1063/1.447334 https://pubs.aip.org/aip/jcp/article- pdf/81/1/511/9722086/511 1 online.pdf

    Nos´ e, S.: A unified formulation of the constant temperature molecu- lar dynamics methods. The Journal of Chemical Physics81(1), 511–519 (1984) https://doi.org/10.1063/1.447334 https://pubs.aip.org/aip/jcp/article- pdf/81/1/511/9722086/511 1 online.pdf

    work page doi:10.1063/1.447334 1984

  39. [39]

    Hoover, W.G.: Canonical dynamics: Equilibrium phase-space distributions. Phys. Rev. A31, 1695–1697 (1985) https://doi.org/10.1103/PhysRevA.31.1695

    work page doi:10.1103/physreva.31.1695 1985

  40. [40]

    Scientific Data6(1) (2019) https://doi.org/10.1038/s41597-019-0080-z

    Mamun, O., Winther, K.T., Boes, J.R., Bligaard, T.: High-throughput calcula- tions of catalytic properties of bimetallic alloy surfaces. Scientific Data6(1) (2019) https://doi.org/10.1038/s41597-019-0080-z

    work page doi:10.1038/s41597-019-0080-z 2019

  41. [41]

    ACS Catalysis8(4), 3447–3453 (2018) https://doi.org/10.1021/acscatal.8b00201 https://doi.org/10.1021/acscatal.8b00201

    Schumann, J., Medford, A.J., Yoo, J.S., Zhao, Z.-J., Bothra, P., Cao, A., Studt, F., Abild-Pedersen, F., Nørskov, J.K.: Selectivity of synthe- sis gas conversion to c2+ oxygenates on fcc(111) transition-metal surfaces. ACS Catalysis8(4), 3447–3453 (2018) https://doi.org/10.1021/acscatal.8b00201 https://doi.org/10.1021/acscatal.8b00201

    work page doi:10.1021/acscatal.8b00201 2018

  42. [42]

    Proceedings of the National Academy of Sci- ences118(30), 2106527118 (2021) https://doi.org/10.1073/pnas.2106527118 https://www.pnas.org/doi/pdf/10.1073/pnas.2106527118

    Wang, T., Abild-Pedersen, F.: Achieving industrial ammonia synthesis rates at near-ambient conditions through modified scaling relations on a confined dual site. Proceedings of the National Academy of Sci- ences118(30), 2106527118 (2021) https://doi.org/10.1073/pnas.2106527118 https://www.pnas.org/doi/pdf/10.1073/pnas.2106527118

    work page doi:10.1073/pnas.2106527118 2021

  43. [43]

    ChemSusChem8(13), 2180–2186 (2015) https://doi.org/10.1002/cssc.201500322 https://chemistry- europe.onlinelibrary.wiley.com/doi/pdf/10.1002/cssc.201500322

    Montoya, J.H., Tsai, C., Vojvodic, A., Nørskov, J.K.: The chal- lenge of electrochemical ammonia synthesis: A new perspective on the role of nitrogen scaling relations. ChemSusChem8(13), 2180–2186 (2015) https://doi.org/10.1002/cssc.201500322 https://chemistry- europe.onlinelibrary.wiley.com/doi/pdf/10.1002/cssc.201500322

    work page doi:10.1002/cssc.201500322 2015

  44. [44]

    Journal of Catalysis374, 161–170 (2019) https://doi.org/10.1016/j.jcat

    Hansen, M.H., Nørskov, J.K., Bligaard, T.: First principles micro-kinetic model of catalytic non-oxidative dehydrogenation of ethane over close-packed metallic 46 facets. Journal of Catalysis374, 161–170 (2019) https://doi.org/10.1016/j.jcat. 2019.03.034

    work page doi:10.1016/j.jcat 2019

  45. [45]

    ACS Catalysis14(7), 5286–5296 (2024) https://doi.org/10

    Comer, B.M., Bothra, N., Lunger, J.R., Abild-Pedersen, F., Bajdich, M., Winther, K.T.: Prediction of o and oh adsorption on transition metal oxide surfaces from bulk descriptors. ACS Catalysis14(7), 5286–5296 (2024) https://doi.org/10. 1021/acscatal.4c00111 https://doi.org/10.1021/acscatal.4c00111

    work page doi:10.1021/acscatal.4c00111 2024

  46. [46]

    ACS Catalysis9(5), 4006–4014 (2019) https: //doi.org/10.1021/acscatal.9b00260 https://doi.org/10.1021/acscatal.9b00260

    Clark, E.L., Ringe, S., Tang, M., Walton, A., Hahn, C., Jaramillo, T.F., Chan, K., Bell, A.T.: Influence of atomic surface structure on the activity of ag for the elec- trochemical reduction of co2 to co. ACS Catalysis9(5), 4006–4014 (2019) https: //doi.org/10.1021/acscatal.9b00260 https://doi.org/10.1021/acscatal.9b00260

    work page doi:10.1021/acscatal.9b00260 2019

  47. [47]

    Chemistry of Materials33(15), 5872–5884 (2021) https://doi.org/10.1021/acs.chemmater

    Landers, A.T., Peng, H., Koshy, D.M., Lee, S.H., Feaster, J.T., Lin, J.C., Beeman, J.W., Higgins, D., Yano, J., Drisdell, W.S., Davis, R.C., Bajdich, M., Abild- Pedersen, F., Mehta, A., Jaramillo, T.F., Hahn, C.: Dynamics and hysteresis of hydrogen intercalation and deintercalation in palladium electrodes: A multi- modal in situ x-ray diffraction, coulome...

    work page doi:10.1021/acs.chemmater 2021

  48. [48]

    Li, J., Halldin Stenlid, J., Tang, M.T., Peng, H.-J., Abild-Pedersen, F.: Screening binary alloys for electrochemical co2 reduction towards multi-carbon products. J. Mater. Chem. A10, 16171–16181 (2022) https://doi.org/10.1039/D2TA02749F

    work page doi:10.1039/d2ta02749f 2022

  49. [49]

    Energy Environ

    Peng, H., Tang, M.T., Liu, X., Schlexer Lamoureux, P., Bajdich, M., Abild- Pedersen, F.: The role of atomic carbon in directing electrochemical co(2) reduction to multicarbon products. Energy Environ. Sci.14, 473–482 (2021) https://doi.org/10.1039/D0EE02826F

    work page doi:10.1039/d0ee02826f 2021

  50. [50]

    Nature Communications13(1), 1399 (2022) https://doi.org/10.1038/ s41467-022-29140-8

    Peng, H.-J., Tang, M.T., Halldin Stenlid, J., Liu, X., Abild-Pedersen, F.: Trends in oxygenate/hydrocarbon selectivity for electrochemical CO(2) reduction to C2 products. Nature Communications13(1), 1399 (2022) https://doi.org/10.1038/ s41467-022-29140-8

    work page 2022

  51. [51]

    npj Computational Materials8(1), 163 (2022) https://doi.org/10.1038/ s41524-022-00846-z

    Saini, S., Halldin Stenlid, J., Abild-Pedersen, F.: Electronic structure fac- tors and the importance of adsorbate effects in chemisorption on surface alloys. npj Computational Materials8(1), 163 (2022) https://doi.org/10.1038/ s41524-022-00846-z

    work page 2022

  52. [52]

    ACS Catalysis9(4), 3399–3412 (2019) https://doi.org/10.1021/acscatal.8b04848 https://doi.org/10.1021/acscatal.8b04848 47

    Snider, J.L., Streibel, V., Hubert, M.A., Choksi, T.S., Valle, E., Upham, D.C., Schumann, J., Duyar, M.S., Gallo, A., Abild-Pedersen, F., Jaramillo, T.F.: Revealing the synergy between oxide and alloy phases on the per- formance of bimetallic in–pd catalysts for co2 hydrogenation to methanol. ACS Catalysis9(4), 3399–3412 (2019) https://doi.org/10.1021/acs...

    work page doi:10.1021/acscatal.8b04848 2019

  53. [54]

    Applied Catalysis B: Environmental279, 119384 (2020) https://doi.org/10.1016/ j.apcatb.2020.119384

    Tang, M.T., Peng, H., Lamoureux, P.S., Bajdich, M., Abild-Pedersen, F.: From electricity to fuels: Descriptors for c1 selectivity in electrochemical co2 reduction. Applied Catalysis B: Environmental279, 119384 (2020) https://doi.org/10.1016/ j.apcatb.2020.119384

    work page arXiv 2020

  54. [55]

    ACS Applied Materials & Interfaces14(22), 25246–25256 (2022) https://doi.org/10.1021/acsami.2c00398 https://doi.org/10.1021/acsami.2c00398

    Tetteh, E.B., Gyan-Barimah, C., Lee, H.-Y., Kang, T.-H., Kang, S., Ringe, S., Yu, J.-S.: Strained pt(221) facet in a ptco@pt-rich catalyst boosts oxy- gen reduction and hydrogen evolution activity. ACS Applied Materials & Interfaces14(22), 25246–25256 (2022) https://doi.org/10.1021/acsami.2c00398 https://doi.org/10.1021/acsami.2c00398. PMID: 35609281

    work page doi:10.1021/acsami.2c00398 2022

  55. [56]

    ACS Catalysis11(10), 6290–6297 (2021) https://doi.org/ 10.1021/acscatal.0c05711 https://doi.org/10.1021/acscatal.0c05711

    Wang, T., Cui, X., Winther, K.T., Abild-Pedersen, F., Bligaard, T., Nørskov, J.K.: Theory-aided discovery of metallic catalysts for selective propane dehydro- genation to propylene. ACS Catalysis11(10), 6290–6297 (2021) https://doi.org/ 10.1021/acscatal.0c05711 https://doi.org/10.1021/acscatal.0c05711

    work page doi:10.1021/acscatal.0c05711 2021

  56. [57]

    Journal of the American Chemi- cal Society138(11), 3705–3714 (2016) https://doi.org/10.1021/jacs.5b12087 https://doi.org/10.1021/jacs.5b12087

    Yang, N., Medford, A.J., Liu, X., Studt, F., Bligaard, T., Bent, S.F., Nørskov, J.K.: Intrinsic selectivity and structure sensitivity of rhodium cat- alysts for c2+ oxygenate production. Journal of the American Chemi- cal Society138(11), 3705–3714 (2016) https://doi.org/10.1021/jacs.5b12087 https://doi.org/10.1021/jacs.5b12087. PMID: 26958997

    work page doi:10.1021/jacs.5b12087 2016

  57. [58]

    Performance and Analysis of the Alchemical Transfer Method for Binding-Free-Energy Predictions of Diverse Ligands

    Flores, R.A., Paolucci, C., Winther, K.T., Jain, A., Torres, J.A.G., Aykol, M., Montoya, J., Nørskov, J.K., Bajdich, M., Bligaard, T.: Active learning accelerated discovery of stable iridium oxide polymorphs for the oxygen evolution reaction. Chemistry of Materials32(13), 5854–5863 (2020) https://doi.org/10.1021/acs. chemmater.0c01894 https://doi.org/10.1...

    work page doi:10.1021/acs 2020

  58. [59]

    de Klerk, Eveline van der Maas, and Marnix Wagemaker

    Lee, K., Flores, R.A., Liu, Y., Wang, B.Y., Hikita, Y., Sinclair, R., Bajdich, M., Hwang, H.Y.: Epitaxial stabilization and oxygen evolution reaction activity of metastable columbite iridium oxide. ACS Applied Energy Materials4(4), 3074–3082 (2021) https://doi.org/10.1021/acsaem. 0c02788 https://doi.org/10.1021/acsaem.0c02788

    work page doi:10.1021/acsaem 2021

  59. [60]

    ACS Energy Letters7(7), 2228–2235 (2022) https://doi.org/10.1021/acsenergylett.2c00578 https://doi.org/10.1021/acsenergylett.2c00578

    Shi, X., Peng, H.-J., Hersbach, T.J.P., Jiang, Y., Zeng, Y., Baek, J., Winther, K.T., Sokaras, D., Zheng, X., Bajdich, M.: Efficient and stable acidic water oxi- dation enabled by low-concentration, high-valence iridium sites. ACS Energy Letters7(7), 2228–2235 (2022) https://doi.org/10.1021/acsenergylett.2c00578 https://doi.org/10.1021/acsenergylett.2c00578

    work page doi:10.1021/acsenergylett.2c00578 2022

  60. [61]

    ACS Applied Materials & Interfaces11(37), 34059–34066 (2019) https://doi.org/10.1021/acsami.9b13697 https://doi.org/10.1021/acsami.9b13697

    Strickler, A.L., Flores, R.A., King, L.A., Nørskov, J.K., Bajdich, M., Jaramillo, T.F.: Systematic investigation of iridium-based bimetallic thin film catalysts 48 for the oxygen evolution reaction in acidic media. ACS Applied Materials & Interfaces11(37), 34059–34066 (2019) https://doi.org/10.1021/acsami.9b13697 https://doi.org/10.1021/acsami.9b13697. PM...

    work page doi:10.1021/acsami.9b13697 2019

  61. [62]

    The Journal of Phys- ical Chemistry C125(48), 26437–26447 (2021) https://doi.org/10.1021/acs.jpcc

    Tang, M.T., Peng, H.-J., Stenlid, J.H., Abild-Pedersen, F.: Exploring trends on coupling mechanisms toward c3 product formation in co(2)r. The Journal of Phys- ical Chemistry C125(48), 26437–26447 (2021) https://doi.org/10.1021/acs.jpcc. 1c07553 https://doi.org/10.1021/acs.jpcc.1c07553 49

    work page doi:10.1021/acs.jpcc 2021