arXiv subjects

Fabian Berger

Publications and source records attributed to Fabian Berger.

5 recordsLinked to original sources

arXiv2026

Breaking Water at Graphene Defects

Water dissociation at solid surfaces underpins processes ranging from corrosion and catalysis to electrochemistry and photovoltaics. Defects often serve as reactive sites for dissociation, yet how solvation influences water dissociation at such sites remains poorly understood. Here, we use state-of-the-art machine-learned interatomic potentials to explore water dissociation at defective graphene-water interfaces. We show that solvation qualitatively changes the reaction mechanism at a graphene single vacancy (SV), opening pathways that are absent for an isolated water molecule. Whereas the gas-phase process proceeds via a single concerted channel, the solvated SV splits water through two competing pathways: a basic route forming SV-H and OH-(aq), and an acidic route forming SV-OH and H3O+(aq). These lower-barrier pathways produce distinct chemisorbed intermediates that enhance graphene-water adsorption. Accordingly, even a simple carbon vacancy gives rise to unexpectedly rich interfacial chemistry, coupling surface chemistry to interfacial charge and wettability, with implications for carbon functionalization and nanofluidic transport.

physics.chem-ph↗

arXiv2025

Mechanisms for the Formation of Active Sites in Single-Atom Alloys

Reactive dopant atoms embedded in inert host metal surfaces define the active sites in single-atom alloys (SAAs), yet SAA synthesis remains challenging. To address this, we elucidate how dopant adatoms deposited on Cu and Ag surfaces become incorporated into the metal and identify periodic trends from early to late transition metals (TMs) using density functional theory. Adatoms diffuse nearly freely across terraces, as diffusion barriers are small, whereas direct incorporation into terraces is unfavourable. In line with conventional wisdom, step edges and kink sites strongly facilitate dopant incorporation, confirming their critical role in alloy formation. Attachment of adatoms to steps and kinks from the lower terrace is favoured. Incorporation then proceeds either from this attached state or when adatoms approach a step edge from above, where reactions often proceed without barrier. Incorporation barriers are generally lower for early and central TMs, increase towards late TMs, and are slightly higher on Cu than on Ag surfaces. Repulsive interactions between Pd adatoms and dopants explain the experimental observation that a dopant-rich brim on the upper terrace of Cu surfaces inhibits incorporation from above. In contrast, attractive interactions, as found for Ru, anchor diffusing adatoms (even on terraces) and promote the formation of adatom islands, yet hinder incorporation next to the dopant and may impede the growth of embedded dopant clusters. By rationalising periodic trends and experimental observations, we show how specific surface sites and adatom--dopant interactions shape dopant incorporation, offering guidance on the surface environments most conducive to SAA synthesis for different dopant elements.

cond-mat.mtrl-sci↗

arXiv2025

How Accurate Are DFT Forces? Unexpectedly Large Uncertainties in Molecular Datasets

Training of general-purpose machine learning interatomic potentials (MLIPs) relies on large datasets with properties usually computed with density functional theory (DFT). A pre-requisite for accurate MLIPs is that the DFT data are well converged to minimize numerical errors. A possible symptom of errors in DFT force components is nonzero net force. Here, we consider net forces in datasets including SPICE, Transition1x, ANI-1x, ANI-1xbb, AIMNet2, QCML, and OMol25. Several of these datasets suffer from significant nonzero DFT net forces. We also quantify individual force component errors by comparison to recomputed forces using more reliable DFT settings at the same level of theory, and we find significant discrepancies in force components averaging from 1.7 meV/{\AA} in the SPICE dataset to 33.2 meV/{\AA} in the ANI-1x dataset. These findings underscore the importance of well converged DFT data as increasingly accurate MLIP architectures become available.

physics.chem-ph↗

arXiv2024

Cooperative CO$_2$ capture via oxalate formation on metal-decorated graphene

CO$_2$ capture using carbon-based materials, particularly graphene and graphene-like materials, is a promising strategy to deal with CO$_2$ emissions. However, significant gaps remain in our understanding of the molecular-level interaction between CO$_2$ molecules and graphene, particularly, in terms of chemical bonding and electron transfer. In this work, we employ random structure search and density functional theory to understand the adsorption of CO$_2$ molecules on Ca, Sr, Na, K, and Ti decorated graphene surfaces. Compared to the pristine material, we observe enhanced CO$_2$ adsorption on the decorated graphene surfaces. Particularly on group 2 metals and titanium decorated graphene, CO$_2$ can be strongly chemisorbed as a bent CO$_2$ anion or as an oxalate, depending on the number of CO$_2$ molecules. Electronic structure analysis reveals the adsorption mechanism to involve an ionic charge transfer from the metal adatom to the adsorbed CO$_2$. Overall, this study suggests that reducing CO$_2$ to oxalate on group 2 metals and titanium metal-decorated graphene surfaces is a potential strategy for CO$_2$ storage.

cond-mat.mtrl-sci↗

arXiv2023

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 ML force fields have largely been limited by: (i) the substantial computational and human effort of developing and validating 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 get reliable results much faster, and beginners face a lower barrier to entry. Foundation models thus represent a step towards democratising the revolution in atomic-scale modeling that has been brought about by ML force fields.

physics.chem-ph↗