A foundation model for atomistic materials chemistry

Atomistic simulations of matter, especially those that leverage first-principles (ab initio) electronic structure theory, provide a microscopic view of the world, underpinning much of our understanding of chemistry and materials science. Over the last decade or so, machine-learned force fields have transformed atomistic modeling by enabling simulations of ab initio quality over unprecedented time and length scales. However, early machine-learning (ML) force fields have largely been limited by (i) the substantial computational and human effort required to develop and validate potentials for each particular system of interest and (ii) a general lack of transferability from one chemical system to the next. Here, we show that it is possible to create a general-purpose atomistic ML model, trained on a public dataset of moderate size, that is capable of running stable molecular dynamics for a wide range of molecules and materials. We demonstrate the power of the MACE-MP-0 model-and its qualitative and at times quantitative accuracy-on a diverse set of problems in the physical sciences, including properties of solids, liquids, gases, chemical reactions, interfaces, and even the dynamics of a small protein. The model can be applied out of the box as a starting or "foundation" model for any atomistic system of interest and, when desired, can be fine-tuned on just a handful of application-specific data points to reach ab initio accuracy. Establishing that a stable force-field model can cover almost all materials changes atomistic modeling in a fundamental way: experienced users obtain reliable results much faster, and beginners face a lower barrier to entry. Foundation models thus represent a step toward democratizing the revolution in atomic-scale modeling that has been brought about by ML force fields.

Authors

Institutions

Publication Details

Journal
The Journal of Chemical Physics
Published
2025-11-13
DOI
https://doi.org/10.1063/5.0297006
Citations
93
Primary Topic
Machine Learning in Materials Science
Type
article
Field-Weighted Citation Impact
48.16
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
article

A foundation model for atomistic materials chemistry

Jules Tilly, Jolla Kullgren, Vlad Cărare, Christian Holm et al.
93 citations
The Journal of Chemical Physics
Machine Learning in Materials Science
48.16
article

A foundation model for atomistic materials chemistry

Jules Tilly, Jolla Kullgren, Vlad Cărare, Christian Holm, Aakash Ashok Naik, Venkat Kapil, Ioan-Bogdan Magdău, Rhys E. A. Goodall, Philipp Benner, Michele Ceriotti, Sanggyu Chong, Arghya Bhowmik, Edvin Fako, Karsten Reuter, Gábor Cśanyi, Christoph Schran, James R. Kermode, Samuel M. Blau, Hendrik H. Heenen, Samuel P. Niblett, Fabian Zills, Noam Bernstein, Kersti Hermansson, Mark Asta, Kristin Aslaug Persson, Zakariya El‐Machachi, Niamh O’Neill, Filippo Bigi, Ilyes Batatia, Petr Grigorev, William J. Baldwin, Annalena R. Genreith‐Schriever, Dávid Péter Kovács, Eric Sivonxay, Johannes T. Margraf, Cas van der Oord, Benjamin X. Shi, Fabio Falcioni, Christopher Sutton, Yuan Chiang, Matthew C. Kuner, Yangshuai Wang, Sam Walton Norwood, Shuang Han, Cheuk Hin Ho, Konstantin S. Jakob, Nima Karimitari, Volker L. Deringer, A. M. Elena, Angelos Michaelides, Christoph Ortner, Andrea C. Ferrari, J. Harry Moore, Aaron D. Kaplan, Flaviano Della Pia, James P. Darby, Thomas D. Swinburne, Sandip De, Fabian Berger, Martin Vondrák, Will Handley, Tejs Vegge, Janosh Riebesell, Janine George, Andrew Rosen, Matthew Avaylon, William C. Witt, Stephan Hofmann, Clare P. Grey, Lars L. Schaaf, Santiago Vargas, Tamás K. Stenczel, John L. A. Gardner, Guoda Liepuoniute, Namu Kroupa, Xavier R. Advincula, Rokas Elijošius, Domantas Kuryla, Viktor Svahn, Eszter Varga-Umbrich, Jad Jaafar, Hyunwook Jung, Chen Lin, Thomas Wolf, Panagiotis Kourtis, Louise A. M. Rosset, Mikołaj J. Gawkowski, Jonas Grandel
article en
93 citations

Abstract

Atomistic simulations of matter, especially those that leverage first-principles (ab initio) electronic structure theory, provide a microscopic view of the world, underpinning much of our understanding of chemistry and materials science. Over the last decade or so, machine-learned force fields have transformed atomistic modeling by enabling simulations of ab initio quality over unprecedented time and length scales. However, early machine-learning (ML) force fields have largely been limited by (i) the substantial computational and human effort required to develop and validate potentials for each particular system of interest and (ii) a general lack of transferability from one chemical system to the next. Here, we show that it is possible to create a general-purpose atomistic ML model, trained on a public dataset of moderate size, that is capable of running stable molecular dynamics for a wide range of molecules and materials. We demonstrate the power of the MACE-MP-0 model-and its qualitative and at times quantitative accuracy-on a diverse set of problems in the physical sciences, including properties of solids, liquids, gases, chemical reactions, interfaces, and even the dynamics of a small protein. The model can be applied out of the box as a starting or "foundation" model for any atomistic system of interest and, when desired, can be fine-tuned on just a handful of application-specific data points to reach ab initio accuracy. Establishing that a stable force-field model can cover almost all materials changes atomistic modeling in a fundamental way: experienced users obtain reliable results much faster, and beginners face a lower barrier to entry. Foundation models thus represent a step toward democratizing the revolution in atomic-scale modeling that has been brought about by ML force fields.

The Journal of Chemical PhysicsVol. 163(18)
University of Stuttgart (DE), Université Claude Bernard Lyon 1 (FR), Uppsala University (SE), United States Naval Research Laboratory (US), Centre National de la Recherche Scientifique (FR), Federal Institute For Materials Research and Testing (DE), University of British Columbia (CA), Lawrence Berkeley National Laboratory (US), University of South Carolina (US), Science and Technology Facilities Council (GB), Princeton University (US), Aix-Marseille Université (FR), University of Cambridge (GB), University of Michigan (US), Fritz Haber Institute of the Max Planck Society (DE), BASF (United States) (US), University of Warwick (GB), University of Oxford (GB), Universidad Evangélica del Paraguay (PY), Bridge University (SS), The Association of Early Pregnancy Units (GB), Bio Signal Group (United States) (US), Matériaux Ingénierie et Science (FR), Institute of Astronomy (RU), Insilixa (United States) (US), Centre d’Immunologie de Marseille-Luminy (FR), Science Oxford (GB), Centre Interdisciplinaire de Nanoscience de Marseille (FR), University College London (GB), Institut National des Sciences Appliquées de Lyon (FR), École Polytechnique Fédérale de Lausanne (CH), Charles R. Drew University of Medicine and Science (US), University of Bayreuth (DE), Friedrich Schiller University Jena (DE), Newcastle University (GB), Robert Bosch (Germany) (DE), Science and Engineering Research Council (SG), University of California, Berkeley (US), Technical University of Denmark (DK)
Openalex Percentile: Top 0%
Machine Learning in Materials Science
48.16
DOIOpen Access PDFOpenAlex
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.