# How accurate are foundational machine learning interatomic potentials for heterogeneous catalysis?

arXiv: https://arxiv.org/abs/2512.16702 · HTML: https://mlsanity.com/paper/2512.16702/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 16 of its 74 references that are in the map.
- First posted on arXiv 18 Dec 2025.

## Builds on (most cited first, 16 of 16)

- [FiLM: Visual Reasoning with a General Conditioning Layer](https://mlsanity.com/paper/1709.07871/) (2017)
- [E(3)-Equivariant Graph Neural Networks for Data-Efficient and Accurate Interatomic Potentials](https://mlsanity.com/paper/2101.03164/) (2021)
- [MACE: Higher Order Equivariant Message Passing Neural Networks for Fast and Accurate Force Fields](https://mlsanity.com/paper/2206.07697/) (2022)
- [Equivariant message passing for the prediction of tensorial properties and molecular spectra](https://mlsanity.com/paper/2102.03150/) (2021)
- [GemNet: Universal Directional Graph Neural Networks for Molecules](https://mlsanity.com/paper/2106.08903/) (2021)
- [Learning Local Equivariant Representations for Large-Scale Atomistic Dynamics](https://mlsanity.com/paper/2204.05249/) (2022)
- [The Open Catalyst 2020 (OC20) Dataset and Community Challenges](https://mlsanity.com/paper/2010.09990/) (2020)
- [EquiformerV2: Improved Equivariant Transformer for Scaling to Higher-Degree Representations](https://mlsanity.com/paper/2306.12059/) (2023)
- [Reducing SO(3) Convolutions to SO(2) for Efficient Equivariant GNNs](https://mlsanity.com/paper/2302.03655/) (2023)
- [The Open Catalyst 2022 (OC22) Dataset and Challenges for Oxide Electrocatalysts](https://mlsanity.com/paper/2206.08917/) (2022)
- [Spherical Channels for Modeling Atomic Interactions](https://mlsanity.com/paper/2206.14331/) (2022)
- [High-performance training and inference for deep equivariant interatomic potentials](https://mlsanity.com/paper/2504.16068/) (2025)
- [AQCat25: Unlocking spin-aware, high-fidelity machine learning potentials for heterogeneous catalysis](https://mlsanity.com/paper/2510.22938/) (2025)
- [Bayesian optimization of atomic structures with prior probabilities from universal interatomic potentials](https://mlsanity.com/paper/2408.15590/) (2024)
- [Breaking scaling relations with inverse catalysts: a machine learning exploration of trends in CO2 hydrogenation energy barriers](https://mlsanity.com/paper/2504.16493/) (2025)
- [Transition States Energies from Machine Learning: An Application to Reverse Water-Gas Shift on Single-Atom Alloys](https://mlsanity.com/paper/2505.00574/) (2025)

## Built on by (most cited first, 1 of 1)

- [Reaction-network reasoning with frontier models for experimentally confirmed catalyst-selectivity hypotheses](https://mlsanity.com/paper/2607.08003/) (2026)

## Abstract (arXiv)

Foundational machine learning interatomic potentials (MLIPs) are being developed at a rapid pace, promising closer and closer approximation to ab initio accuracy. This unlocks the possibility to simulate much larger length and time scales. However, benchmarks for these MLIPs are usually limited to ordered, crystalline and bulk materials. Hence, reported performance does not necessarily accurately reflect MLIP performance in real applications such as heterogeneous catalysis. Here, we systematically analyze zero-shot performance of 80 different MLIPs, evaluating tasks typical for heterogeneous catalysis across a range of different data sets, including adsorption and reaction on surfaces of alloyed metals, oxides, and metal-oxide interfacial systems. We demonstrate that current-generation foundational MLIPs can already perform at high accuracy for applications such as predicting vacancy formation energies of perovskite oxides or zero-point energies of supported nanoclusters. However, limitations also exist. We find that many MLIPs catastrophically fail when applied to magnetic materials, and structure relaxation in the MLIP generally increases the energy prediction error compared to single-point evaluation of a previously optimized structure. Comparing low-cost task-specific models to foundational MLIPs, we highlight some core differences between these model approaches and show that -- if considering only accuracy -- these models can compete with the current generation of best-performing MLIPs. Furthermore, we show that no single MLIP universally performs best, requiring users to investigate MLIP suitability for their desired application.