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 the 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, the model complexity has a positive correlation with prediction accuracy. We also find that the MLFFs were 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 GSF energy surface are washed out by statistical noise.

Accelerated kinetic Monte Carlo: A case study; vacancy and dumbbell interstitial diffusion traps in concentrated solid solution alloys.

Vacancy and self-interstitial atomic diffusion coefficients in concentrated solid solution alloys can have a non-monotonic concentration dependence. Here, the kinetics of monovacancies and ⟨100⟩ dumbbell interstitials in Ni-Fe alloys are assessed using lattice kinetic Monte Carlo (kMC). The non-monotonicity is associated with superbasins, which impels using accelerated kMC methods. Detailed implementation prescriptions for first passage time analysis kMC (FPTA-kMC), mean rate method kMC (MRM-kMC), and accelerated superbasin kMC (AS-kMC) are given. The accelerated methods are benchmarked in the context of diffusion coefficient calculations. The benchmarks indicate that MRM-kMC underestimates diffusion coefficients, while AS-kMC overestimates them. In this application, MRM-kMC and AS-kMC are computationally more efficient than the more accurate FPTA-kMC. Our calculations indicate that composition dependence of migration energies is at the origin of the vacancy's non-monotonic behavior. In contrast, the difference between formation energies of Ni-Ni, Ni-Fe, and Fe-Fe dumbbell interstitials is at the origin of their non-monotonic diffusion behavior. Additionally, the migration barrier crossover composition-based on the situation where Ni or Fe atom jumps have lower energy barrier than the other one-is introduced. KMC simulations indicate that the interplay between composition dependent crossover of migration energy and geometrical site percolation explains the non-monotonic concentration-dependence of atomic diffusion coefficients.

Automated Detection of Helium Bubbles in Irradiated X-750.

Imaging nanoscale features using transmission electron microscopy is key to predicting and assessing the mechanical behavior of structural materials in nuclear reactors. Analyzing these micrographs is often a tedious and labour intensive manual process. It is a prime candidate for automation. Here, a region-based convolutional neural network is adapted to detect helium bubbles in micrographs of neutron-irradiated Inconel X-750 reactor spacer springs. We demonstrate that this neural network produces analyses of similar accuracy and reproducibility to that produced by humans. Further, we show this method as being four orders of magnitude faster than manual analysis allowing for generation of significant quantities of data. The proposed method can be used with micrographs of different Fresnel contrasts and magnification levels.

Effect of He on the Order-Disorder Transition in Ni_{3}Al under Irradiation.

The order-disorder transition in Ni-Al alloys under irradiation represents an interplay between various reordering processes and disordering due to thermal spikes generated by incident high energy particles. Typically, ordering is enabled by diffusion of thermally generated vacancies, and can only take place at temperatures where they are mobile and in sufficiently high concentration. Here, in situ transmission electron micrographs reveal that the presence of He-usually considered to be a deleterious immiscible atom in this material-promotes reordering in Ni_{3}Al at temperatures where vacancies are not effective ordering agents. A rate-theory model is presented, that quantitatively explains this behavior, based on parameters extracted from atomistic simulations. These calculations show that the V_{2}He complex is an effective agent through its high stability and mobility. It is surmised that immiscible atoms may stabilize reordering agents in other materials undergoing driven processes, and preserve ordered phases at temperature where the driven processes would otherwise lead to disorder.

Atomistic and mesoscale simulation of sodium and potassium adsorption in cement paste.

An atomistic and mesoscopic assessment of the effect of alkali uptake in cement paste is performed. Semi-grand canonical Monte Carlo simulations indicate that Na and K not only adsorb at the pore surface of calcium silicate hydrates (C-S-H) but also adsorb in the C-S-H hydrated interlayer up to concentrations of the order of 0.05 and 0.1 mol/kg, respectively. Sorption of alkali is favored as the Ca/Si ratio of C-S-H is reduced. Long timescale simulations using the Activation Relaxation Technique indicate that characteristic diffusion times of Na and K in the C-S-H interlayer are of the order of a few hours. At the level of individual grains, Na and K adsorption leads to a reduction of roughly 5% of the elastic moduli and to volume expansion of about 0.25%. Simulations using the so-called primitive model indicate that adsorption of alkali ions at the pore surface can reduce the binding between C-S-H grains by up to 6%. Using a mesoscopic model of cement paste, the combination of individual grain swelling and changes in inter-granular cohesion was estimated to lead to overall expansive pressures of up to 4 MPa-and typically of less than 1 MPa-for typical alkali concentrations observed at the proximity of gel veins caused by the alkali-silica reaction.

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, yet these algorithms often include a number of steps of which the effect is not always clear. Decoupling these steps and their impacts can allow us to better understand both their role and the nature of complex energy landscape. Here, we consider a family of minimum-energy algorithms based, directly or indirectly, on the well-known Bell-Evans-Polanyi (BEP) principle. Comparing trajectories generated with BEP-based algorithms to kinetically correct off-lattice kinetic Monte Carlo schemes allow us to confirm that the BEP principle does not hold for complex systems since forward and reverse energy barriers are completely uncorrelated. As would be expected, following the lowest available energy barrier leads to rapid trapping. This is why BEP-based methods require also a direct handling of visited basins or barriers. Comparing the efficiency of these methods with a thermodynamical handling of low-energy barriers, we show that most of the efficiency of the BEP-like methods lie first and foremost in the basin management rather than in the BEP-like step.

Replenish and relax: explaining logarithmic annealing in ion-implanted c-Si.

We study ion-damaged crystalline silicon by combining 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 disordered systems, with heat-release measurements. The microscopic mechanism associated with this 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.

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 (KMC) 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).

Optimized energy landscape exploration using the ab initio based activation-relaxation technique.

Unbiased open-ended methods for finding transition states are powerful tools to understand diffusion and relaxation mechanisms associated with defect diffusion, growth processes, and catalysis. They have been little used, however, in conjunction with ab initio packages as these algorithms demanded large computational effort to generate even a single event. Here, we revisit the activation-relaxation technique (ART nouveau) and introduce a two-step convergence to the saddle point, combining the previously used Lanczós algorithm with the direct inversion in interactive subspace scheme. This combination makes it possible to generate events (from an initial minimum through a saddle point up to a final minimum) in a systematic fashion with a net 300-700 force evaluations per successful event. ART nouveau is coupled with BigDFT, a Kohn-Sham density functional theory (DFT) electronic structure code using a wavelet basis set with excellent efficiency on parallel computation, and applied to study the potential energy surface of C(20) clusters, vacancy diffusion in bulk silicon, and reconstruction of the 4H-SiC surface.