arXiv subjects

Brian H. Lee

Publications and source records attributed to Brian H. Lee.

8 recordsLinked to original sources

arXiv2026

Multi-Fidelity Predictive Model for Shock Response of Energetic Materials Using Conditional U-Net

Mapping microstructure to properties is central to materials science. Perhaps most famously, the Hall-Petch relationship relates average grain size to strength. More challenging has been deriving relationships for properties that depend on subtle microstructural features and not average properties. One such example is the initiation of energetic materials under dynamical loading, dominated by energy localization on microstructural features such as pores, cracks, and interfaces. We propose a conditional convolutional neural network to predict the shock-induced temperature field as a function of shock strength, for a wide range of microstructures, and obtained via two different simulation methods. The proposed model, denoted MISTnet2, significantly extends prior work that was limited to a single shock strength, model, and type of microstructure. MISTnet2 can contribute to bridging atomistics with coarse-grain simulations and enable first principles predictions of detonation initiation and safety of this class of materials.

cond-mat.mtrl-sci↗

arXiv2025

Understanding and predicting trends in adsorption energetics on monolayer transition metal dichalcogenides

Two-dimensional (2D) transition metal dichalcogenides (TMDs) have recently been shown to demonstrate non-volatile resistive switching (NVRS), offering significant advantages such as high-density integration and low energy consumption due to their atomic-scale thinness. In this study, we focus on the adsorption and desorption of metal adatoms, which can modulate the electrical resistivity by several orders of magnitude. We develop material-based relationships of the adsorption energy with electronic and atomic structure descriptors by examining the effects of various transition-metal adsorbates on the surface of TMDs. Our results reveal that adsorption energies of transition metals exhibit consistent trends across different TMDs (MoS$_2$, MoSe$_2$, WS$_2$, WSe$_2$) and can be explained using simple descriptors of the atomic and electronic structure. We propose several models to describe this adsorption process, providing a deeper understanding of a crucial step in the resistive switching mechanism based on formation and dissolution of point defects. Finally, we connect our computed adsorption energies to the switching energy. These findings will help guide rational materials selection for the development of NVRS devices using 2D TMDs.

cond-mat.mtrl-sci↗

arXiv2024

Thermodynamic Fidelity of Generative Models for Ising System

Machine learning has become a central technique for modeling in science and engineering, either complementing or as surrogates to physics-based models. Significant efforts have recently been devoted to models capable of predicting field quantities but the limitations of current state-of-the-art models in describing complex physics are not well understood. We characterize the ability of generative diffusion models and generative adversarial networks (GAN) to describe the Ising model. We find diffusion models trained using equilibrium configurations obtained using Metropolis Monte Carlo for a range of temperatures around the critical temperature can capture average thermodynamic variables across the phase transformation and extrapolate to higher and lower temperatures. The model also captures the overall trends of physical properties associated with fluctuations (specific heat and susceptibility) except at the non-ergodic low temperatures and non-trivial scale-free correlations at the critical temperature, albeit with some difference in the critical exponent compared to Monte Carlo simulations. GANs perform more poorly on thermodynamic properties and are susceptible to mode-collapse without careful training. This investigation highlights the potential and limitations of generative models in capturing the complex phenomena associated with certain physical systems.

cond-mat.stat-mech↗

arXiv2024

Accelerating active learning materials discovery with FAIR data and workflows: a case study for alloy melting temperatures

Active learning (AL) is a powerful sequential optimization approach that has shown great promise in the discovery of new materials. However, a major challenge remains the acquisition of the initial data and the development of workflows to generate new data at each iteration. In this study, we demonstrate a significant speedup in an optimization task by reusing a published simulation workflow available for online simulations and its associated data repository, where the results of each workflow run are automatically stored. Both the workflow and its data follow FAIR (findable, accessible, interoperable, and reusable) principles using nanoHUB's infrastructure. The workflow employs molecular dynamics to calculate the melting temperature of multi-principal component alloys. We leveraged all prior data not only to develop an accurate machine learning model to start the sequential optimization but also to optimize the simulation parameters and accelerate convergence. Prior work showed that finding the alloy composition with the highest melting temperature required testing 15 alloy compositions, and establishing the melting temperature for each composition took, on average, 4 simulations. By developing a workflow that utilizes the FAIR data in the nanoHUB database, we reduced the number of simulations per composition to one and found the alloy with the lowest melting temperature testing only three compositions. This second optimization, therefore, shows a speedup of 10x as compared to models that do not access the FAIR databases.

cond-mat.mtrl-sci↗

arXiv2024

Multi-Task Multi-Fidelity Learning of Properties for Energetic Materials

Data science and artificial intelligence are playing an increasingly important role in the physical sciences. Unfortunately, in the field of energetic materials data scarcity limits the accuracy and even applicability of ML tools. To address data limitations, we compiled multi-modal data: both experimental and computational results for several properties. We find that multi-task neural networks can learn from multi-modal data and outperform single-task models trained for specific properties. As expected, the improvement is more significant for data-scarce properties. These models are trained using descriptors built from simple molecular information and can be readily applied for large-scale materials screening to explore multiple properties simultaneously. This approach is widely applicable to fields outside energetic materials.

cs.LG↗

arXiv2024

Graph neural network coarse-grain force field for the molecular crystal RDX

Condense phase molecular systems organize in wide range of distinct molecular configurations, including amorphous melt and glass as well as crystals often exhibiting polymorphism, that originate from their intricate intra- and intermolecular forces. While accurate coarse-grain (CG) models for these materials are critical to understand phenomena beyond the reach of all-atom simulations, current models cannot capture the diversity of molecular structures. We introduce a generally applicable approach to develop CG force fields for molecular crystals combining graph neural networks (GNN) and data from an all-atom simulations and apply it to the high-energy density material RDX. We address the challenge of expanding the training data with relevant configurations via an iterative procedure that performs CG molecular dynamics of processes of interest and reconstructs the atomistic configurations using a pre-trained neural network decoder. The multi-site CG model uses a GNN architecture constructed to satisfy translational invariance and rotational covariance for forces. The resulting model captures both crystalline and amorphous states for a wide range of temperatures and densities.

cond-mat.mes-hall↗

arXiv2023

Lennard Jones Token: a blockchain solution to scientific data curation

Data science and artificial intelligence have become an indispensable part of scientific research. While such methods rely on high-quality and large quantities of machine-readable scientific data, the current scientific data infrastructure faces significant challenges that limit effective data curation and sharing. These challenges include insufficient return on investment for researchers to share quality data, logistical difficulties in maintaining long-term data repositories, and the absence of standardized methods for evaluating the relative importance of various datasets. To address these issues, this paper presents the Lennard Jones Token, a blockchain-based proof-of-concept solution implemented on the Ethereum network. The token system incentivizes users to submit optimized structures of Lennard Jones particles by offering token rewards, while also charging for access to these valuable structures. Utilizing smart contracts, the system automates the evaluation of submitted data, ensuring that only structures with energies lower than those in the existing database for a given cluster size are rewarded. The paper explores the details of the Lennard Jones Token as a proof of concept and proposes future blockchain-based tokens aimed at enhancing the curation and sharing of scientific data.

cs.NI↗

arXiv2023

Mapping microstructure to shock-induced temperature fields using deep learning

The response of materials to dynamical, or shock, loading is important to planetary science, aerospace engineering, and energetic materials. Thermal-activated processes, including chemical reactions and phase transitions, are significantly accelerated by the localization of the energy deposited into hotspots. These results from the interaction of a supersonic wave with the materials microstructure and are governed by complex, coupled processes, including the collapse of porosity, interfacial friction, and localized plastic deformation. These mechanisms are not fully understood and today we lack predictive models to, for example, predict the shock to detonation transition from chemistry and microstructure alone. We demonstrate that deep learning techniques can be used to predict the resulting shock-induced temperature fields in complex composite materials obtained from large-scale molecular dynamics simulations with the initial microstructure as the only input. The accuracy of the Microstructure-Informed Shock-induced Temperature net (MISTnet) model is higher than the current state of the art at a fraction of the computation cost.

cond-mat.mtrl-sci↗