Search

Understanding and accurately predicting hydrogen diffusion in materials is challenging due to the complex interactions between hydrogen defects and the crystal lattice. These interactions span large length and time scales, making them difficult to address with standard ab initio techniques. This work addresses this challenge by employing accelerated machine learning (ML) molecular dynamics simulations through active learning. We conduct a comparative study of different ML-based interatomic potential schemes, including VASP, MACE, and CHGNet, utilizing various training strategies such as on-the-fly learning, pre-trained universal models, and fine-tuning.By considering different temperatures and concentration regimes, we obtain hydrogen diffusion coefficients and activation energy values which align remarkably well with experimental results, underlining the efficacy and accuracy of ML-assisted methodologies in the context of diffusive dynamics. Particularly, our procedure significantly reduces the computational effort associated with traditional transition state calculations or ad-hoc designed interatomic potentials. The results highlight the limitations of pre-trained universal solutions for defective materials and how they can be improved by fine-tuning. Specifically, fine-tuning the models on a database produced during on-the-fly training of VASP ML force-field allows the retrieving of DFT-level accuracy at a fraction of the computational cost.

Large language models manifest the ability of few-shot adaptation to a sequence of provided examples. This behavior, known as in-context learning, allows for performing nontrivial machine learning tasks during inference only. In this work, we address the question: can we leverage in-context learning to predict out-of-distribution materials properties? However, this would not be possible for structure property prediction tasks unless an effective method is found to pass atomic-level geometric features to the transformer model. To address this problem, we employ a compound model in which GPT-2 acts on the output of geometry-aware graph neural networks to adapt in-context information. To demonstrate our model's capabilities, we partition the QM9 dataset into sequences of molecules that share a common substructure and use them for in-context learning. This approach significantly improves the performance of the model on out-of-distribution examples, surpassing the one of general graph neural network models., Comment: 12 pages, 4 figures

In this work we propose simple, effective and computationally efficient transfer learning approaches for structure-property relation predictions in the context of materials, with highly informative input from different modalities. As materials properties stand from their electronic structure, representations are extracted directly from datasets of electronic charge density profile images using Neural Networks. We demonstrate transferability of the existing pre-trained Convolutional Neural Networks and Large Language Models knowledge to physics domain data, exploring a wide set of compositions for the regression of energetics- or structure- related properties, and the role of semantic crystallographic information in the context of multimodal approaches. We test the applicability of the CLIP multimodal model, and employ as well a training protocol for building a more interpretable and versatile stacked custom solution from different pre-trained modalities. The study offers a promising avenue for enhancing the effectiveness of descriptor identification in physical systems, shedding light on the power of multimodal transfer learning for materials property prediction. Instead of using the well-established GNN-based approaches, we explore the transfer learning of image- and text-based architectures, which can impact decision making for new low-cost AI methods in the field of Materials and Chemoinformatics.

Crystal dislocation dynamics, especially at high temperatures, represents a subject where experimental phenomenological input is commonly required, and parameter-free predictions, starting from quantum methods, have been beyond reach. This is especially true for phenomena like stacking faults and dislocation cross-slip, which are computationally intractable with methods like density functional theory, as $\sim 10^5-10^6$ atoms are required to reliably simulate such systems. Hence, this work extends quantum-mechanical accuracy to mesoscopic molecular dynamics simulations and opens unprecedented possibilities in material design for extreme mechanical conditions with direct atomistic insight at the deformation mesoscale. To accomplish this, we construct a Bayesian machine-learned force field (MLFF) from ab initio quantum training data, enabling direct observations of high-temperature and high-stress dislocation dynamics in single-crystalline Cu with atomistic resolution. In doing so, a generalizable training protocol is developed for construction of MLFFs for dislocation kinetics, with wide-ranging applicability to other single element systems and alloys. The resulting FLARE MLFF provides excellent predictions of static bulk elastic properties, stacking fault widths and energies, dynamic evolutions and mobilities of edge and screw dislocations, as well as cross-slip energy barriers for screw dislocations, all of which are compared to available experimental measurements. This work marks the first reliable quantitative determination of dislocation mobilities and cross-slip barriers, demonstrating a substantial advancement over previous empirical and machine learned force field models., Comment: 15 pages, 7 figures, 3 tables

Material characterization in nano-mechanical tests requires precise interatomic potentials for the computation of atomic energies and forces with near-quantum accuracy. For such purposes, we develop a robust neural-network interatomic potential (NNIP), and we provide a test for the example of molecular dynamics (MD) nanoindentation, and the case of body-centered cubic crystalline molybdenum (Mo). We employ a similarity measurement protocol, using standard local environment descriptors, to select ab initio configurations for the training dataset that capture the behavior of the indented sample. We find that it is critical to include generalized stacking fault (GSF) configurations, featuring a dumbbell interstitial on the surface, to capture dislocation cores, and also high-temperature configurations with frozen atom layers for the indenter tip contact. We develop a NNIP with distinct dislocation nucleation mechanisms, realistic generalized stacking fault energy (GSFE) curves, and an informative energy landscape for the atoms on the sample surface during nanoindentation. We compare our NNIP results with nanoindentation simulations, performed with three existing potentials -- an embedded atom method (EAM) potential, a gaussian approximation potential (GAP), and a tabulated GAP (tabGAP) potential -- that predict different dislocation nucleation mechanisms, and display the absence of essential information on the shear stress at the sample surface in the elastic region. We believe that these features render specialized NNIPs essential for simulations of nanoindentation and nano-mechanics with near-quantum accuracy.

Materials design has traditionally evolved through trial-error approaches, mainly due to the non-local relationship between microstructures and properties such as strength and toughness. We propose 'alloy informatics' as a machine learning based prototype predictive approach for alloys and compounds, using electron charge density profiles derived from first-principle calculations. We demonstrate this framework in the case of hydrogen interstitials in face-centered cubic crystals, showing that their differential electron charge density profiles capture crystal properties and defect-crystal interaction properties. Radial Distribution Functions (RDFs) of defect-induced differential charge density perturbations highlight the resulting screening effect, and, together with hydrogen Bader charges, strongly correlate to a large set of atomic properties of the metal species forming the bulk crystal. We observe the spontaneous emergence of classes of charge responses while coarse-graining over crystal compositions. Nudge-Elastic-Band calculations show that RDFs and charge features also connect to hydrogen migration energy barriers between interstitial sites. Unsupervised machine-learning on RDFs supports classification, unveiling compositional and configurational non-localities in the similarities of the perturbed densities. Electron charge density perturbations may be considered as bias-free descriptors for a large variety of defects.

Materials discovery, especially for applications that require extreme operating conditions, requires extensive testing that naturally limits the ability to inquire the wealth of possible compositions. Machine Learning (ML) has nowadays a well established role in facilitating this effort in systematic ways. The increasing amount of available accurate DFT data represents a solid basis upon which new ML models can be trained and tested. While conventional models rely on static descriptors, generally suitable for a limited class of systems, the flexibility of Graph Neural Networks (GNNs) allows for direct learning representations on graphs, such as the ones formed by crystals. We utilize crystal graph neural networks (CGNN) to predict crystal properties with DFT level accuracy, through graphs with encoding of the atomic (node/vertex), bond (edge), and global state attributes. In this work, we aim at testing the ability of the CGNN MegNet framework in predicting a number of properties of systems previously unseen from the model, obtained by adding a substitutional defect in bulk crystals that are included in the training set. We perform DFT validation to assess the accuracy in the prediction of formation energies and structural features (such as elastic moduli). Using CGNNs, one may identify promising paths in alloy discovery., Comment: 34 pages, 14 figures, Supplementary Material