arXiv subjects

Laurent Karim Béland

Publications and source records attributed to Laurent Karim Béland.

18 recordsLinked to original sources

arXiv2026

A Gaussian process coarse-grained potential for Na-montmorillonite

Hydraulic transport in compacted bentonite is diffusion-controlled and governed by the hydration and microstructure of sodium montmorillonite (Na-MMT). Experiments cannot resolve how platelet interactions govern pore structure, transport and stiffness, while existing coarse-grained models smooth hydration oscillations or require manual corrections. We develop a tabulated potential combining Morse interactions between platelet centre and edge sites with a Gaussian process regression correction trained on all-atom potentials of mean force. It captures the hydration-induced complexity of the potential-of-mean-force profiles, including the three-water (3-W) hydration minimum and transfers across geometries, layer-charge variants and unseen configurations. Applied to monodisperse and polydisperse Na-MMT assemblies at dry densities of 0.8-1.3 g cm^-3, the model captures the 3-W to 1-W transition, loss of non-interlayer porosity and evolution of pore structure, random-walk tortuosity, diffusion and stiffness. Predicted diffusion agrees with compacted Na-bentonite measurements.

cond-mat.mtrl-sci↗

arXiv2026

Machine-Learning Potentials for sodium-potassium chloride mixtures: Predicting thermophysical properties and phase behavior of multicomponent salts

Predicting the properties of multicomponent molten salts using density functional theory (DFT) remains challenging because the spatial and temporal scales required to evaluate transport properties and phase behavior are computationally prohibitive. In this work, we develop a moment tensor potential trained using a a DFT dataset of NaCl, KCl, NaCl-KCl mixtures, and the NaK alloy, enabling large-scale molecular dynamics simulations across wide ranges of temperatures and compositions. We systematically evaluate the effect of D3 dispersion corrections and apply the resulting potential to predict liquid densities, diffusion coefficients, radial distribution functions, heat capacities, thermal conductivities, and the NaCl-KCl phase diagram. The model successfully reproduces many temperature- and composition-dependent trends. However, systematic deviations in several absolute properties persist, highlighting the importance of experimental validation and calibration. These findings support a hybrid modeling framework in which first-principles-informed machine-learning potentials provide transferable predictive capability and mechanistic insight, while experimental data incorporated during model development or subsequent engineering assessments is necessary to improve quantitative accuracy.

cond-mat.mtrl-sci↗

arXiv2025

Accelerating Moment Tensor Potentials through Post-Training Pruning

Moment Tensor Potentials (MTPs) are machine-learning interatomic potentials whose basis functions are typically selected using a level-based scheme that is data-agnostic. We introduce a post-training, cost-aware pruning strategy that removes expensive basis functions with minimal loss of accuracy. Applied to nickel and silicon-oxygen systems, it yields models up to seven times faster than standard MTPs. The method requires no new data and remains fully compatible with current MTP implementations.

cond-mat.mtrl-sci↗

arXiv2025

Benchmarking 34 OpenKIM Nickel Potentials with an Emphasis on Surfaces and Extended Defects

We present an automated benchmarking suite for face-centered-cubic (FCC) nickel that evaluates 47 quantitative metrics spanning both standard tests (equation of state, elastic constants, surface energies and phonons) and application-specific scenarios such as defect formation and migration, grain boundaries, step edges, close-range interactions, and vacancy cluster energetics. Using this framework, we assess 34 interatomic potentials from the OpenKIM repository, including pairwise, embedded-atom, modified-embedded-atom, angular-dependent, and spectral neighbor analysis potentials (SNAP). Results are compared against ab initio benchmarks compiled from the literature. Most potentials accurately reproduce lattice parameters, elastic constants, and surface energies, whereas predictive accuracy degrades for migration barriers and short-range compression. Principal-component analysis identifies correlated property groups and a partially orthogonal component associated with migration and short-range physics, revealing Pareto trade-offs between accuracy domains. SNAP models occupy the lowest-error frontier, although several embedded-atom potentials remain competitive across many metrics. The framework provides a reproducible baseline for potential selection, highlights systematic limitations across formalisms, and supports benchmarking-in-the-loop strategies for developing next-generation machine-learning potentials for Ni and Ni-based alloys.

cond-mat.mtrl-sci↗

arXiv2025

A Kokkos-Accelerated Moment Tensor Potential Implementation for LAMMPS

We present a Kokkos-accelerated implementation of the Moment Tensor Potential (MTP) for LAMMPS, designed to improve both computational performance and portability across CPUs and GPUs. This package introduces an optimized CPU variant--achieving up to 2x speedups over existing implementations--and two new GPU variants: a thread-parallel version for large-scale simulations and a block-parallel version optimized for smaller systems. It supports three core functionalities: standard inference, configuration-mode active learning, and neighborhood-mode active learning. Benchmarks and case studies demonstrate efficient scaling to million-atom systems, substantially extending accessible length and time scales while preserving the MTP's near-quantum accuracy and native support for uncertainty quantification.

cond-mat.mtrl-sci↗

arXiv2025

Small-Cell-Based Fast Active Learning of Machine Learning Interatomic Potentials

Machine learning interatomic potentials (MLIPs) are often trained with on-the-fly active learning, where sampled configurations from atomistic simulations are added to the training set. However, this approach is limited by the high computational cost of ab initio calculations for large systems. Recent works have shown that MLIPs trained on small cells (1-8 atoms) rival the accuracy of large-cell models (100s of atoms) at far lower computational cost. Herein, we refer to these as small-cell and large-cell training, respectively. In this work, we iterate on earlier small-cell training approaches and characterize our resultant small-cell protocol. Potassium and sodium-potassium systems were studied: the former, a simpler system benchmarked in detail; the latter, a more complex binary system for further validation. Our small-cell training approach achieves up to two orders of magnitude of cost savings compared to large-cell (54-atom) training, with some training runs requiring fewer than 120 core-hours. Static and thermodynamic properties predicted using the MLIPs were evaluated, with small-cell training in both systems yielding strong ab initio agreement. Small cells appear to encode the necessary information to model complex large-scale phenomena--solid-liquid interfaces, critical exponents, diverse concentrations--even when the training cells themselves are too small to accommodate these phenomena. Based on these tests, we provide analysis and recommendations.

cond-mat.mtrl-sci↗

arXiv2025

Amorphous silicon structures generated using a moment tensor potential and the activation relaxation technique nouveau

Preparing realistic atom-scale models of amorphous silicon (a-Si) is a decades-old condensed matter physics challenge. Herein, we combine the Activation Relaxation Technique nouveau (ARTn) to a Moment Tensor Potential (MTP) to generate seven a-Si models containing between 216 and 4096 atoms. A thorough analysis of their short-range and medium-range structural properties is performed, alongside assessments of excess energy and mechanical properties. The seven ARTn-MTP models are compared with available experimental data and other high quality a-Si models present in the literature. The seven ARTn-MTP a-Si models are in excellent agreement with available experimental data. Notably, several of our models, including the 216-atom, 512-atom, and 1000-atom a-Si models, exhibit low coordination defects without any traces of crystalline grains. Historically overlooked in previous research, our study underlines the need to assess the validity of the continuous random-network hypothesis for the description of perfect amorphous model by characterizing local crystalline environment and to explore the crystallisation process of a-Si through modelling.

cond-mat.dis-nn↗

arXiv2024

Investigating the effect of Cu$^{2+}$ sorption in montmorillonite using density functional theory and molecular dynamics simulations

Montmorillonite (MMT) is the main mineral component of bentonite, which is currently proposed as a sealing material in deep geological repositories (DGRs) for used nuclear fuel. In the Canadian program, which will utilize copper-cladded used fuel containers, safety analysis considers the effect of copper corrosion, during which Cu$^{2+}$ ions could potentially be adsorbed by the surrounding MMT. In such a scenario, ion exchange between Na$^+$ and Cu$^{2+}$ is expected. In this study, a multiscale approach that combines electronic density functional theory (DFT) and force-field-based molecular dynamics (MD) simulations was employed to study the effect of introducing Cu$^{2+}$ ions to MMT. An extension to the ClayFF force field is parametrized and validated using DFT to model how Cu$^{2+}$ interacts with clay systems. MD simulations were performed to calculate the interaction free energies between MMT platelets containing Cu$^{2+}$ ions (Cu-MMT) and compared them to inter-platelet interaction energies in Na-MMT and Ca-MMT. Our calculations suggest Cu-MMT develops swelling pressures between those of Ca-MMT and Na-MMT. Furthermore, our MD simulations suggest that Cu$^{2+}$ has MMT interlayer mobility that is significantly slower than that of Ca$^{2+}$.

cond-mat.mtrl-sci↗

arXiv2024

A unified moment tensor potential for silicon, oxygen, and silica

Si and its oxides have been extensively explored in theoretical research due to their technological and industrial importance. Simultaneously describing interatomic interactions within both Si and SiO$_2$ without the use of \textit{ab initio} methods is considered challenging, given the charge transfers involved. Herein, this challenge is overcome by developing a unified machine learning interatomic potentials describing the Si/ SiO$_2$/ O system, based on the moment tensor potential (MTP) framework. This MTP is trained using a comprehensive database generated using density functional theory simulations, encompassing a wide range of crystal structures, point defects, extended defects, and disordered structure. Extensive testing of the MTP is performed, indicating it can describe static and dynamic features of very diverse Si, O, and SiO$_2$ atomic structures with a degree of fidelity approaching that of DFT

cond-mat.mtrl-sci↗

arXiv2023

A set of moment tensor potentials for zirconium with increasing complexity

Machine learning force fields (MLFFs) are an increasingly popular choice for atomistic simulations due to their high fidelity and improvable nature. Here, we propose a hybrid small-cell approach that combines attributes of both offline and active learning to systematically expand a quantum mechanical (QM) database while constructing MLFFs with increasing model complexity. Our MLFFs employ the moment tensor potential formalism. During this process, we quantitatively assessed structural properties, elastic properties, dimer potential energies, melting temperatures, phase stability, point defect formation energies, point defect migration energies, free surface energies, and generalized stacking fault (GSF) energies of Zr as predicted by our MLFFs. Unsurprisingly, model complexity has a positive correlation with prediction accuracy. We also find that the MLFFs wee able to predict the properties of out-of-sample configurations without directly including these specific configurations in the training dataset. Additionally, we generated 100 MLFFs of high complexity (1513 parameters each) that reached different local optima during training. Their predictions cluster around the benchmark DFT values, but subtle physical features such as the location of local minima on the GSFE surface are washed out by statistical noise.

physics.comp-ph↗

arXiv2021

Interstitialcy-based reordering kinetics of Ni$_3$Al precipitates in irradiated Ni-based super alloys

Neutron irradiation tends to promote disorder in ordered alloys through the action of the thermal spikes that it generates, while simultaneously introducing point defects and defect clusters. As they migrate, these point defects will promote reordering of the alloys, acting against irradiation-induced disordering. In this study, classical molecular dynamics and a highly parallel accelerated sampling method are used to study the reordering kinetics of Ni$_3$Al under the diffusion of self-interstitial atoms (SIA). By monitoring the order parameter and potential energy from atomistic simulations, we show that the SIA acts as a reordering agent in Ni$_3$Al. A mean-field rate theory model of the interstitialcy-based reordering kinetics is introduced, which reproduces simulation data and predicts reordering at temperatures as low as 500 K.

cond-mat.mtrl-sci↗

arXiv2021

A molecular dynamics study of path-dependent grain boundary properties in nanocrystals prepared using different methods

Grain boundaries are thermodynamically unstable. Hence, their properties should be path-dependent: grain boundaries in nanocrystals prepared using different methods might exhibit different properties. Using molecular dynamics simulations, we investigated grain boundaries of nanocrystals formed by quenching,solidification with pre-induced nucleation sites, and Voronoi tessellation. Some properties were found to be path depend: the quenched model has lower boundary energy per atom, smaller boundary excess free volume per atom, and slower grain growth than the Voronoi model. We surmise that these differences are attributed to the abundant annealing twins in the quenched model. On the other hand, other properties are path independent, such as Young's modulus, Poisson's ratio, and the ratio between grain boundary energy and boundary excess free volume. The results of this study further the understanding of the structural-property relationship of nanocrystals and provide guidance to future simulation-based studies of nanocrystalline materials.

cond-mat.mtrl-sci↗

arXiv2014

Understanding long-time vacancy aggregation in iron: a kinetic activation-relaxation technique study

Vacancy diffusion and clustering processes in body-centered-cubic (bcc) Fe are studied using the kinetic activation-relaxation technique (k-ART), an off-lattice kinetic Monte Carlo method with on-the-fly catalog building capabilities. For monovacancies and divacancies, k-ART recovers previously published results while clustering in a 50-vacancy simulation box agrees with experimental estimates. Applying k-ART to the study of clustering pathways for systems containing from one to six vacancies, we find a rich set of diffusion mechanisms. In particular, we show that the path followed to reach a hexavacancy cluster influences greatly the associated mean-square displacement. Aggregation in a 50-vacancy box also shows a notable dispersion in relaxation time associated with effective barriers varying from 0.84 to 1.1 eV depending on the exact pathway selected. We isolate the effects of long-range elastic interactions between defects by comparing to simulations where those effects are deliberately suppressed. This allows us to demonstrate that in bcc Fe, suppressing long-range interactions mainly influences kinetics in the first 0.3 ms, slowing down quick energy release cascades seen more frequently in full simulations, whereas long-term behavior and final state are not significantly affected.

cond-mat.mtrl-sci↗

arXiv2014

Slow relaxation of cascade-induced defects in Fe

On-the-fly kinetic Monte Carlo (KMC) simulations are performed to investigate slow relaxation of non-equilibrium systems. Point defects induced by 25 keV cascades in $α$-Fe are shown to lead to a characteristic time-evolution, described by the \emph{replenish and relax} mechanism. Then, we produce an atomistically-based assessment of models proposed to explain the slow structural relaxation by focusing on the aggregation of 50 vacancies and 25 self-interstital atoms (SIA) in 10-lattice-parameter $α$-Fe boxes, two processes that are closely related to cascade annealing and exhibit similar time signature. Four atomistic effects explain the timescales involved in the evolution: defect concentration heterogeneities, concentration-enhanced mobility, cluster-size dependent bond energies and defect-induced pressure. These findings suggest that the two main classes of models to explain slow structural relaxation, the Eyring model and the Gibbs model, both play a role to limit the rate of relaxation of these simple point-defect systems.

cond-mat.mtrl-sci↗

arXiv2014

Probing potential energy surface exploration strategies for complex systems

The efficiency of minimum-energy configuration searching algorithms is closely linked to the energy landscape structure of complex systems. Here we characterize this structure by following the time evolution of two systems, vacancy aggregation in Fe and energy relaxation in ion-bombarded c-Si, using the kinetic Activation-Relaxation Technique (k-ART), an off-lattice kinetic Monte Carlo (KMC) method, and the well-known Bell-Evans-Polanyi (BEP) principle. We also compare the efficiency of two methods for handling non-diffusive flickering states -- an exact solution and a Tabu-like approach that blocks already visited states. Comparing these various simulations allow us to confirm that the BEP principle does not hold for complex system since forward and reverse energy barriers are completely uncorrelated. This means that following the lowest available energy barrier, even after removing the flickering states, leads to rapid trapping: relaxing complex systems requires crossing high-energy barriers in order to access new energy basins, in agreement with the recently proposed replenish-and-relax model [Béland et al., PRL 111, 105502 (2013)] This can be done by forcing the system through these barriers with Tabu-like methods. Interestingly, we find that following the fundamental kinetics of a system, though standard KMC approach, is at least as efficient as these brute-force methods while providing the correct kinetics information.

cond-mat.mtrl-sci↗

arXiv2014

Strain effect and intermixing at the Si surface: A hybrid quantum and molecular mechanics study

We investigate Ge mixing at the Si(001) surface and characterize the $2\times N$ Si(001) reconstruction by means of hybrid quantum and molecular mechanics calculations (QM/MM). Avoiding fake elastic dampening, this scheme allows to correctly take into account long range deformation induced by reconstruted and defective surfaces. We focus in particular on the dimer vacancy line (DVL) and its interaction with Ge adatoms. We first show that calculated formation energies for these defects are highly dependent on the choice of chemical potential and that the latter must be chosen carefully. Characterizing the effect of the DVL on the deformation field, we also find that the DVL favors Ge segregation in the fourth layer close to the DVL. Using the activation-relaxation technique (ART nouveau) and QM/MM, we show that a complex diffusion path permits the substitution of the Ge atom in the fourth layer, with barriers compatible with mixing observed at intermediate temperature.

cond-mat.mtrl-sci↗

arXiv2013

Replenish and relax: explaining logarithmic annealing in disordered materials

Fatigue and aging of materials are, in large part, determined by the evolution of the atomic-scale structure in response to strains and perturbations. This coupling between microscopic structure and long time scales remains one of the main challenges in materials study. Focusing on a model system, ion-damaged crystalline silicon, we combine nanocalorimetric experiments with an off-lattice kinetic Monte Carlo simulation to identify the atomistic mechanisms responsible for the structural relaxation over long time scales. We relate the logarithmic relaxation, observed in a number of systems, with heat-release measurements. The microscopic mechanism associated with logarithmic relaxation can be described as a two-step replenish and relax process. As the system relaxes, it reaches deeper energy states with logarithmically growing barriers that need to be unlocked to replenish the heat-releasing events leading to lower energy configurations.

cond-mat.mtrl-sci↗

arXiv2011

Kinetic Activation Relaxation Technique

We present a detailed description of the kinetic Activation-Relaxation Technique (k-ART), an off-lattice, self-learning kinetic Monte Carlo algorithm with on-the-fly event search. Combining a topological classification for local environments and event generation with ART nouveau, an efficient unbiased sampling method for finding transition states, k-ART can be applied to complex materials with atoms in off-lattice positions or with elastic deformations that cannot be handled with standard KMC approaches. In addition to presenting the various elements of the algorithm, we demonstrate the general character of k-ART by applying the algorithm to three challenging systems: self-defect annihilation in c-Si (crystalline silicon), self-interstitial diffusion in Fe and structural relaxation in a-Si (amorphous silicon).

physics.comp-ph↗