arXiv subjects

Wojciech G. Stark

Publications and source records attributed to Wojciech G. Stark.

11 recordsLinked to original sources

arXiv2026

Role of anisotropic electronic friction in laser-driven hydrogen recombination on copper

Ultrafast light-driven chemical dynamics at surfaces are governed by energy transfer from excited electrons to vibrational degrees of freedom. When this nonadiabatic energy transfer is anisotropic, it can lead to dynamical steering effects that affect reaction probabilities or non-thermal final energy distributions in molecules. Here, enabled by a machine-learning-based simulation framework, we compare isotropic and anisotropic models of electronic friction during laser-driven hydrogen evolution on the (111) facet of copper. While anisotropic electronic friction determines the rate of energy transfer into the adsorbate and the fluence dependence of desorption probabilities, final translational, vibrational and rotational energy distributions are mainly governed by the potential energy landscape at the barrier and insensitive to nonadiabatic effects.

physics.chem-ph↗

arXiv2026

Six Open Questions in Machine-Learned Interatomic Potential Foundation Models

Machine-learned interatomic potentials (MLIPs) have had a profound impact on molecular modelling in recent years, promising to resolve the long-standing tension between the scale and accuracy of simulations. There has been a proliferation of new models and designs, and recently the paradigm of ``foundational'' MLIPs has become prevalent. Broadly speaking, foundation models are trained on large diverse datasets and promise to work well for new systems with minimal updates required. However, in such a new and fast moving field, there are many unanswered questions. In this article, we set out to articulate and explore what we see as the most important among these questions. We start by developing a working definition for foundational MLIPs and use this definition to frame the subsequent open questions. Despite the rapid progress in the field of MLIP models, we believe that these are fundamental questions which will continue to define cutting edge research in MLIPs in the years to come.

cond-mat.mtrl-sci↗

arXiv2025

Fine-tuning foundation models of materials interatomic potentials with frozen transfer learning

Machine-learned interatomic potentials are revolutionising atomistic materials simulations by providing accurate and scalable predictions within the scope covered by the training data. However, generation of an accurate and robust training data set remains a challenge, often requiring thousands of first-principles calculations to achieve high accuracy. Foundation models have started to emerge with the ambition to create universally applicable potentials across a wide range of materials. While foundation models can be robust and transferable, they do not yet achieve the accuracy required to predict reaction barriers, phase transitions, and material stability. This work demonstrates that foundation model potentials can reach chemical accuracy when fine-tuned using transfer learning with partially frozen weights and biases. For two challenging datasets on reactive chemistry at surfaces and stability and elastic properties of tertiary alloys, we show that frozen transfer learning with 10-20% of the data (hundreds of datapoints) achieves similar accuracies to models trained from scratch (on thousands of datapoints). Moreover, we show that an equally accurate, but significantly more efficient surrogate model can be built using the transfer learned potential as the ground truth. In combination, we present a simulation workflow for machine learning potentials that improves data efficiency and computational efficiency.

cond-mat.mtrl-sci↗

arXiv2025

Nonadiabatic reactive scattering of hydrogen on different surface facets of copper

Dissociative chemisorption is a key process in hydrogen-metal surface chemistry, where nonadiabatic effects due to low-lying electron-hole-pair excitations may affect reaction outcomes. Molecular dynamics with electronic friction simulations can capture weak nonadiabatic effects at metal surfaces, but require as input energy landscapes and electronic friction tensors. Here, we present full-dimensional machine learning surrogate models of the electronic friction tensor to study reactive hydrogen chemistry at the low-index surface facets Cu(100), Cu(110), Cu(111), and Cu(211). We combine these surrogate models with machine learning interatomic potentials to simulate quantum-state-resolved H$_2$ reactive scattering on pristine copper surfaces. The predicted sticking coefficient and survival probabilities are in excellent agreement with experiment. Comparison between adiabatic and nonadiabatic simulations reveals that the influence of electron-hole pair excitations on the scattering dynamics is weak and that the probability for dissociative adsorption is dominated by the shape of the underlying potential energy surface and the initial vibrational quantum state. Nonadiabatic effects only lead to subtle changes in rovibrationally inelastic state-to-state scattering probabilities. The differences between jellium-based isotropic local density friction and full ab-initio response-theory-based orbital-dependent friction are even more subtle. The presented machine learning models represent the most accurate publicly available, full-dimensional models for H$_2$ on copper to date.

physics.chem-ph↗

arXiv2025

Machine Learning and Data-Driven Methods in Computational Surface and Interface Science

Nanoscale design of surfaces and interfaces is essential for modern technologies like organic LEDs, batteries, fuel cells, superlubricating surfaces, and heterogeneous catalysis. However, these systems often exhibit complex surface reconstructions and polymorphism, with properties influenced by kinetic processes and dynamic behavior. A lack of accurate and scalable simulation tools has limited computational modeling of surfaces and interfaces. Recently, machine learning and data-driven methods have expanded the capabilities of theoretical modeling, enabling, for example, the routine use of machine-learned interatomic potentials to predict energies and forces across numerous structures. Despite these advances, significant challenges remain, including the scarcity of large, consistent datasets and the need for computational and data-efficient machine learning methods. Additionally, a major challenge lies in the lack of accurate reference data and electronic structure methods for interfaces. Density Functional Theory, while effective for bulk materials, is less reliable for surfaces, and too few accurate experimental studies on interface structure and stability exist. Here, we will sketch the current state of data-driven methods and machine learning in computational surface science and provide a perspective on how these methods will shape the field in the future.

cond-mat.mtrl-sci↗

arXiv2024

Room Temperature Hydrogen Atom Scattering Experiments Are Not a Sufficient Benchmark to Validate Electronic Friction Theory

In the dynamics of atoms and molecules at metal surfaces, electron-hole pair excitations can play a crucial role. In the case of hyperthermal hydrogen atom scattering, they lead to nonadiabatic energy loss and highly inelastic scattering. Molecular dynamics with electronic friction simulations where friction is computed under an isotropic homogeneous electron gas approximation have previously shown good agreement with measured kinetic energy loss distributions, suggesting that this level of theoretical description is sufficient to describe nonadiabatic effects of atomic scattering. In this work, we show that similar agreement with room temperature experiments can also be achieved with friction derived from density functional theory linear response calculations. The apparent agreement of the homogeneous electron gas approximation with experiment arises from a fortuitous cancellation of errors where friction is overestimated close to the surface and the spin transition is neglected. Only for scattering at low temperatures can both approximations be distinguished and differences rationalised in terms of the number of bounces of the atom on the surface. We identify the signatures of nonadiabatic energy loss of different levels of theory, which future low-temperature scattering experiments will be able to measure.

cond-mat.mtrl-sci↗

arXiv2024

Equivariant Representation of Configuration-Dependent Friction Tensors in Langevin Heatbaths

Dynamics of coarse-grained particle systems derived via the Mori-Zwanzig projection formalism commonly take the form of a (generalized) Langevin equation with configuration-dependent friction and diffusion tensors. In this article, we introduce a class of equivariant representations of tensor-valued functions based on the Atomic Cluster Expansion (ACE) framework that allows for efficient learning of such configuration-dependent friction and diffusion tensors from data. Besides satisfying the correct equivariance properties with respect to the Euclidean group E(3), the resulting heat bath models satisfy a fluctuation-dissipation relation. Moreover, our models can be extended to include additional symmetries, such as momentum conservation, to preserve the hydrodynamic properties of the particle system. We demonstrate the capabilities of the model by constructing a model of configuration-dependent tensorial electronic friction calculated from first principles that arises during reactive molecular dynamics at metal surfaces.

cond-mat.mtrl-sci↗

arXiv2024

Benchmarking of machine learning interatomic potentials for reactive hydrogen dynamics at metal surfaces

Simulations of chemical reaction probabilities in gas surface dynamics require the calculation of ensemble averages over many tens of thousands of reaction events to predict dynamical observables that can be compared to experiments. At the same time, the energy landscapes need to be accurately mapped, as small errors in barriers can lead to large deviations in reaction probabilities. This brings a particularly interesting challenge for machine learning interatomic potentials, which are becoming well-established tools to accelerate molecular dynamics simulations. We compare state-of-the-art machine learning interatomic potentials with a particular focus on their inference performance on CPUs and suitability for high throughput simulation of reactive chemistry at surfaces. The considered models include polarizable atom interaction neural networks (PaiNN), recursively embedded atom neural networks (REANN), the MACE equivariant graph neural network, and atomic cluster expansion potentials (ACE). The models are applied to a dataset on reactive molecular hydrogen scattering on low-index surface facets of copper. All models are assessed for their accuracy, time-to-solution, and ability to simulate reactive sticking probabilities as a function of the rovibrational initial state and kinetic incidence energy of the molecule. REANN and MACE models provide the best balance between accuracy and time-to-solution and can be considered the current state-of-the-art in gas-surface dynamics. PaiNN models require many features for the best accuracy, which causes significant losses in computational efficiency. ACE models provide the fastest time-to-solution, however, models trained on the existing dataset were not able to achieve sufficiently accurate predictions in all cases.

physics.chem-ph↗

arXiv2023

Machine learning interatomic potentials for reactive hydrogen dynamics at metal surfaces based on iterative refinement of reaction probabilities

Reactive chemistry of molecular hydrogen at surfaces, notably dissociative sticking and hydrogen evolution, plays a crucial role in energy storage and fuel cells. Theoretical studies can help to decipher underlying mechanisms and reaction design, but studying dynamics at surfaces is computationally challenging due to the complex electronic structure at interfaces and the high sensitivity of dynamics to reaction barriers. In addition, ab initio molecular dynamics, based on density functional theory, is too computationally demanding to accurately predict reactive sticking or desorption probabilities as it requires averaging over tens of thousands of initial conditions. High-dimensional machine learning-based interatomic potentials are starting to be more commonly used in gas-surface dynamics, yet, robust approaches to generate reliable training data and to assess how model uncertainty affects the prediction of dynamic observables are not well established. Here, we employ ensemble learning to adaptively generate training data while assessing model performance with full uncertainty quantification for reaction probabilities of hydrogen scattering on different copper facets. We use this approach to investigate the performance of two message-passing neural networks, SchNet and PaiNN. Ensemble-based uncertainty quantification and iterative refinement allow us to expose the shortcomings of the invariant pairwise-distance-based feature representation in the SchNet model for gas-surface dynamics.

physics.chem-ph↗

arXiv2022

NQCDynamics.jl: A Julia Package for Nonadiabatic Quantum Classical Molecular Dynamics in the Condensed Phase

Accurate and efficient methods to simulate nonadiabatic and quantum nuclear effects in high-dimensional and dissipative systems are crucial for the prediction of chemical dynamics in condensed phase. To facilitate effective development, code sharing and uptake of newly developed dynamics methods, it is important that software implementations can be easily accessed and built upon. Using the Julia programming language, we have developed the NQCDynamics.jl package which provides a framework for established and emerging methods for performing semiclassical and mixed quantum-classical dynamics in condensed phase. The code provides several interfaces to existing atomistic simulation frameworks, electronic structure codes, and machine learning representations. In addition to the existing methods, the package provides infrastructure for developing and deploying new dynamics methods which we hope will benefit reproducibility and code sharing in the field of condensed phase quantum dynamics. Herein, we present our code design choices and the specific Julia programming features from which they benefit. We further demonstrate the capabilities of the package on two examples of chemical dynamics in condensed phase: the population dynamics of the spin-boson model as described by a wide variety of semi-classical and mixed quantum-classical nonadiabatic methods and the reactive scattering of H2 on Ag(111) using the Molecular Dynamics with Electronic Friction method. Together, they exemplify the broad scope of the package to study effective model Hamiltonians and realistic atomistic systems.

physics.chem-ph↗

arXiv2021

Ab initio calculation of electron-phonon linewidths and molecular dynamics with electronic friction at metal surfaces with numeric atom-centered orbitals

Molecular motion at metallic surfaces is affected by nonadiabatic effects and electron-phonon coupling. The ensuing energy dissipation and dynamical steering effects are not captured by classical molecular dynamics simulations, but can be described with the molecular dynamics with electronic friction method and linear response calculations based on density functional theory. Herein, we present an implementation of electron-phonon response based on an all-electron numeric atomic orbital description in the electronic structure code FHI-aims. After providing details of the underlying approximations and numerical considerations, we present significant scalability and performance improvements of the new code compared to a previous implementation [Phys. Rev. B 94, 115432 (2016)]. We compare convergence behaviour and results of our simulations for exemplary systems such as CO on Cu(100), H$_2$ adsorption on Cu(111), and CO on Ru(0001) against existing plane wave implementations. We find that our all-electron calculations exhibit faster and more monotonic convergence behaviour than conventional plane-wave-based electron-phonon calculations. Our findings suggest that many electron-phonon linewidth calculations in literature may be underconverged. Finally, we showcase the capabilities of the new code by studying the contribution of interband and intraband excitations to vibrational linewidth broadening of aperiodic motion in previously unfeasibly large periodic surface models.

cond-mat.mtrl-sci↗