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.

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.

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 2× 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.

In this study, we have developed a molecular dynamics (MD) framework to simulate the combined effects of the irradiation-induced defects production, their accumulation, and the generation of helium in metals and alloys. This framework is applied to alloy 800H, a candidate structural material for various advanced reactor designs, to evaluate the combined effects of irradiation-induced damage and helium generation in the microstructure. Our results show a strong impact of He on the accumulation of Frenkel pairs (FP), and on the nucleation and growth of nanocavities. In particular, the higher the generation rate of He, the larger the population of accumulated FPs and the higher the density and size of nanocavities at a given irradiation dose. Inversely, the importance of irradiation damage on the nucleation and growth of He bubbles is highlighted, indicating the strong correlation between both phenomena. This MD framework offers a clearer path towards a more detailed atomistic simulation of the evolution of the microstructure under irradiation but also can consider the generation of fission products and transmutations.

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 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 lower computational cost. Herein, we refer to these as small-cell and large-cell training, respectively. In 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 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 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.

Alloy 800H is currently used as structural material in light-water cooled nuclear reactors, and it is also considered as candidate materials for various advanced reactor designs. Under operation, the exposure of alloy 800H to neutron irradiation results in the formation of helium (He), mainly from Ni by the transmutation reaction (n, alpha). In this work, we present a molecular dynamics (MD) simulation study of the behavior and effects of He on the microstructure and mechanical properties of alloy 800H. Our results show that the population of clusters made of 5 to 9 He atoms is nearly constant throughout the simulation time (10 ns), while the population of larger clusters increases as the simulation time increases. The growth of clusters is controlled by either the dissociation and diffusion of smaller clusters towards nearby larger clusters or the merging of larger clusters initially located nearby to each other. A significant accumulation of He is observed at the grain boundaries (GB), while a depletion zone is found at the neighboring regions. As a result, the density of He cluster is significantly higher at the GBs as compared to the intra-granular regions. The nucleation and growth of He clusters also results in the formation of Frenkel pairs (FP), whose associated self-interstitial atoms (SIA) agglomerate into interstitial clusters in the alloy 800H matrix. As a consequence, dislocation segments, mostly of the Shockley type, are generated in the microstructure, and often located next to He clusters. The combination of the aforementioned defect structures and the high density of He clusters at the GBs results in a substantial degradation of the mechanical properties of alloy 800H single crystal and bicrystals.

Using a small-cell active learning approach, we generate a moment tensor potential (MTP) trained on only 609 configurations, jointly describing solid/liquid Na, gaseous Cl, and crystalline/molten NaCl. This MTP implicitly captures the effect of atomic charge variations on energies and forces based on local atomic configurations. Extensive testing of this potential points to a high-fidelity description of the structural and transport properties of Na and NaCl. Furthermore, this potential was used to calculate the standard reduction potential and solubility limit of Na in molten NaCl. These computed properties are in good agreement with available experimental data and ab initio calculations. Our proposed approach can be utilized to predict the electrochemical and physical properties of molten salts with arbitrary compositions and solutes, as well as the molten salt corrosion of metals.

Zirconium (Zr) alloys are widely used as structural materials in the core of nuclear reactors. However, the absorption of hydrogen by these alloys during the reactor operation may lead to the precipitation of hydrides. Such hydrides have considerable effects on the microstructure and mechanical properties of the alloys, thus significantly impacting their overall performance. In this work, we investigate the behavior of hydrogen in the presence of symmetric tilt grain boundaries in Zr using molecular dynamics simulations. To systematically explore the conditions for the formation of hydrides, we consider hydrogen concentration levels of up to 50 at.%. We find that hydrogen concentrations of 30 at.% or lower are below the threshold for the precipitation of hydrides, while fcc coordinated hydrides precipitate at 50 at.%, which is very close to the solubility limit of hydrogen in Zr at 800 K. We find that the grain boundaries (GBs) play a pivotal role in the behavior of hydrogen, as well as the formation of hydrides in Zr systems. In most cases, large amounts of hydrogen are found to accumulate at GBs. However, both homogeneous and heterogeneous nucleation are observed during the formation of hydrides at the grain interiors and GBs, respectively. Nonetheless, in some cases the GB inhibit the formation of hydrides.

Molten salts could play an important role in energy storage, in the form of liquid batteries, and heat storage for solar and nuclear power. However, their widespread application is hindered by a limited understanding of the mechanisms by which they corrode metallic containers. This knowledge gap necessitates atomic -scale studies on salt -metal interactions. Molecular dynamics simulations are well suited for such research but require interatomic potential capable of accurately modeling both ionic and neutral states of salt and metal elements. Herein, we developed a moment tensor potential (MTP) with this capability, employing a small -cell training approach. The proposed MTP is compact: It is described by 449 parameters fitted on 609 configurations; 30% of these are oneor two -atom configurations. Extensive testing of our MTP points to a high-fidelity description of the structural and transport properties of solid/liquid Na, gaseous Cl, and crystalline/molten NaCl. Furthermore, we applied this MTP to calculate the standard reduction potential and solubility limit of Na in molten NaCl, achieving results that closely align with experimental and ab initio simulation data. This approach offers a robust framework for exploring the electrochemical and physical properties of molten salts across various compositions and solutes.

Fe-Ni-Cr-based alloys, including alloy 800H, are used as structural materials in nuclear reactors such as the PWRs and BWRs, but also are considered as candidate materials for advanced reactors. In this work, we report a molecular dynamics (MD) study of the effect of irradiation and alloy composition on the microstructure and mechanical behavior of alloy 800H. Our results of primary irradiation damage suggest a larger spontaneous recombination distance for Cr compared to other alloy components, thereby suggesting that Cr may increase the tolerance of the alloy 800H to primary defect production. The progressive accumulation of irradiation events through cascade overlap results in the growth of defects and their agglomeration into clusters, but also induces the formation of dislocations. Irradiation-induced mechanical degradation was also investigated for the single crystal, and based on our observations two degradation mechanisms were proposed: the nucleation and glide of dislocations, promoted by the dynamics of interstitial atoms and vacancies defects under applied strain field, and the rapid activation of slip systems by the combined effect of the foregoing and generated dislocations, vacancies and interstitial atoms. We also found that increasing Ni or Cr substantially limits the irradiation-induced mechanical degradation. Finally, our MD findings highlight the strengthening of alloy 800H by increasing Ni content, and the degradation of the Young modulus induced by the increase of Cr content.

Uranium-zirconium (U-Zr) metallic fuel is considered as fuel candidate for sodium-cooled fast reactors due to its compatibility with sodium, intrinsic passive safety characteristics and higher burnup capability compared to the oxide systems. In the present study, we performed molecular dynamics (MD) and static simulations to investigate the stability, cohesive and structural properties of U-Zr grain boundaries, with 10 and 20 at.% of Zr content. The effect of the Zr composition, temperature and atomic configurations were investigated, and the fracture mechanism highlighted. The cohesive energies of special low energy GBs indicate that the intergranular fracture mechanism plays a significant role in the brittle regime. Moreover, we investigated the role of the atomic configurations in determining the GB properties of the alloys. Indeed, for a given misorientation angle, we found that different atomic configurations do not only yield different energies, but also different strain distribution, atomic coordination and different net expansions (free volume at the boundary). In general, we observed that Zr atoms tend to segregate in low energy configurations. We also observed a systematic increase in GB and surface energies with the Zr concentration, in which the surface energies are systematically higher than the corresponding GB energies. The reported findings highlight important effects of the atomic configurations on properties of investigated GB structures. Therefore, it clearly indicates that the atomic configurations of GBs must be correctly determined for the accurate modeling of the alloys properties.

Zirconium alloys are widely used as structural materials in the core of nuclear reactors. In such conditions, the continuous exposure to neutron irradiation of these alloys can have considerable effects on their microstructure and mechanical properties, which significantly impact their overall performance. In this work, we investigate the cumulative effects of irradiation damage in α-Zr using molecular dynamics (MD) simulations. To this end, displacement cascades are approximated using the computationally efficient stochastic core–shell model. Application of this approach results in structural damage which is representative of those generated using the more computationally demanding simulations of explicit collision cascades. Therefore, the examination of the damage at the atomic-scale, and the effects of long-term irradiation exposure on the material properties of zirconium can be assessed. We first evaluate the influence of radiation damage on the mechanical response of single crystal α-Zr. Subsequently, the effects in α-Zr systems including symmetric tilt grain boundaries are investigated.

Direct betavoltaic energy conversion is a specialized energy harvesting technology, converting beta radiation from a radiation source directly into electricity using a semiconductor. Of the most common beta emitting isotopes, tritium betavoltaics hold promise owing to the high specific activity of solid tritium compounds, low shielding requirements and relatively high availability. Betavoltaic devices offer great promise to produce continuous quantities of nanowatt to microwatt power over the course of several years, particularly for low-power sensor, medical, and space applications where sunlight is too sparse for solar cell use or where battery replacement is challenging. Canadian Nuclear Laboratories (CNL) is currently developing betavoltaic devices based on tritium. CNL has unique facilities to produce, fabricate and test betavoltaic devices. Computational techniques have been used to address challenges of longevity and electron-hole pair generation in semiconductor materials, in particular (In,Ga)P. Long-term studies on test wafers in different tritium-containing environments and the effects on power output and longevity will also be discussed.

Zirconium alloys are commonly employed in nuclear power applications. Under typical operating conditions, hydrogen ingress can lead to the formation of brittle Zr hydrides in the alloy. To study this behavior, transmission electron microscopy (TEM) is routinely used to image hydrides in Zr-alloys. However, the analysis of these TEM micrographs is a complex time-consuming task. Here, we employed a mask region-based convolutional neural network (Mask R-CNN) to automate an essential part of the analysis process: the identification and annotation of hydrides. In addition, although training a neural network usually requires large training datasets (in the order of thousands of images), the proposed framework was developed using a limited training dataset with the recourse of transfer learning. This work shows that the Mask R-CNN is capable of correctly and quickly labeling thermo-mechanically cycled hydrides in TEM images of pressure tube material.

Steels are used extensively in many industries due to their excellent balance of mechanical strength, manufacturability, costs, and acceptable corrosion resistance. Steels are being considered for structural components that are exposed to particularly challenging environments in advanced nuclear reactors. In this context, force-field molecular dynamics (MD) simulations were performed on austenitic steel surrogates with Fe40Cr25Ni35 and Fe50Cr20Ni30 compositions to compute the lattice parameter, elastic constants, bulk modulus, Poisson ratio, the dislocation velocity as a function of shear stress, and the phonon drag coefficient for use in discrete dislocation dynamics (DDD) simulations. The accurate computation of these parameters is essential to obtain correct results by the evolution of the dislocation network of the materials using DDD simulations. The dislocation velocity was extracted from the MD calculations for both steel compositions at 30 different values of stress. Because no significant migration of dislocations was observed at stresses below 200 MPa, the dislocation velocity and mobility were calculated at stresses of 200 MPa and higher. Subsequently, the dislocation density and strain–stress relationship were then computed using the DDD approach. The elastic and plastic deformations in the Fe40Cr25Ni35 system were found to be considerably larger than those of the Fe50Cr20Ni30 system. Our study illustrates the ability of atomistic and dislocation dynamics simulations to elucidate qualitative descriptions of the elasticity and plasticity in steel materials, and thus can assist experimental efforts to evaluate the impact of deformation in austenitic steels.

Indium gallium phosphide (GaxIn1−xP) semiconductors are attractive for betavoltaic batteries because of their excellent structural, thermophysical, and electronic properties. In the present work, we investigate the structural strength and electronic properties of zinc-blende GaxIn1−xP systems, for 0≤x≤1, using first-principles calculations. The most stable structures for different compositions are determined by a systematic evaluation of the atomic configurations. The dynamical stability of the structures is verified by examination of the phonon spectra. To assess the tolerance to radiation damage, we evaluated the minimum threshold displacement energy (Ed) for each atom type in the system. The Ed is estimated from simulations of recoil atom events using ab-initio molecular dynamics. We find a substantial effect of the composition on the electronic properties. In particular, the bandgap is relatively low for 0<x<0.5. The Ed was found to mainly depend on the crystallographic direction for In and P, whereas a considerable effect of the composition is observed for Ga. Overall, GaInP is found to be the more resistant to the radiation-induced degradation and the bandgap is also less affected by structural damage. Therefore, our results indicate that Ga0.25In0.75P is overall an excellent candidate for tritium-based betavoltaic batteries.

Understanding irradiation effects on the structure and properties of materials is crucial for the development of advanced nuclear technologies. Molecular dynamics (MD) simulations are suitable for the study of irradiation-induced damage in materials with atomistic resolution. However, MD simulation of structural damage through collision cascades is computationally expensive, and therefore it is impractical for the study of cumulative irradiation-induced effects in materials due to long-term exposure to irradiation. To overcome this limitation, we propose a stochastic core–shell (SCS) approximation of collision cascades, which is capable of reproducing the defect structures produced by primary radiation damage in MD simulations. The SCS retains the characteristic damage for a given primary knock-on atom energy and system temperature, but without the explicit simulation of the preceding collision cascade. Thus, the method provides an efficient approach to investigate irradiation-induced damage in materials using MD simulations. We demonstrate the application of the SCS method to the investigation of irradiation-induced microstructural effects in α-zirconium.

Indium gallium phosphide (GaxIn1−xP) semiconductors are attractive for betavoltaic batteries because of their excellent structural, thermophysical, and electronic properties. In the present work, we investigate the structural strength and electronic properties of zinc-blende GaxIn1−xP systems, for 0≤x≤1, using first-principles calculations. The most stable structures for different compositions are determined by a systematic evaluation of the atomic configurations. The dynamical stability of the structures is verified by examination of the phonon spectra. To assess the tolerance to radiation damage, we evaluated the minimum threshold displacement energy (Ed) for each atom type in the system. The Ed is estimated from simulations of recoil atom events using ab-initio molecular dynamics. We find a substantial effect of the composition on the electronic properties. In particular, the bandgap is relatively low for 0

The behavior of oxygen in uranium oxycarbide (UCO) has been investigated using density functional theory. To assess the role of carbon on the stability of oxygen point defects, we first determined the formation energies of oxygen vacancy and interstitial for different carbon configurations. Subsequently, we evaluated the barriers for migration of various oxygen defects in the UCO matrix. Our results show that carbon atoms create strong attraction centers for oxygen atoms, as a consequence the spontaneous formation of oxygen Frenkel pairs in the fluorite structure is promoted. Moreover, the strong affinity of oxygen atoms for sites in the first coordination shell of a carbon center leads to the formation of CO, and thus reducing the mobility of oxygen atoms in the UCO fuel. Upon formation, the CO remains fixed at the carbon site due to significant hybridization with uranium atoms, therefore behaving as trap sites for oxygen atoms. Our findings indicate that the formation of CO molecules increases the retention of oxygen atoms inside the UCO matrix. Consequently, carbon atoms in the UCO matrix contribute to mitigating the deleterious effects that oxygen release from the fuel kernel has on the structural integrity of the silicon carbide layer in tristructural isotropic (TRISO) particles.

We have investigated the structure, energetics and mechanical deformation of symmetric tilt grain boundaries (STGB) in hexagonal close-packed Zr (α-Zr). Molecular dynamics (MD) simulations are performed to determine the equilibrium structures of STGBs with [0001], [11¯00] and [112¯0] tilt axes with respect to the misorientation angle. We characterized the atomic structure and energy of STGBs, determined within the coincidence site lattice model, for the tilt axes considered. Subsequently, we performed simulations of uniaxial tensile loading to evaluate their mechanical response under mechanical deformation. Our results provide a comprehensive characterization of STGBs in α-Zr in a wide variety of conditions, and therefore substantially contributing to the understanding of the mechanical properties of α-Zr-based materials.