Publications

Understanding Li-ion transport through the solid electrolyte interphase (SEI) is essential for improving the stability of lithium-metal batteries, as nonuniform ionic transport through the inorganic SEI can promote spatially localized Li deposition and dendrite formation. In this work, we develop a deep-learning-assisted framework to model Li-ion transport across the inorganic SEI by combining density functional theory (nudged elastic band) calculations with graph neural network learning. A systematic diffusion dataset was first generated for eight major inorganic SEI components, namely, LiF, LiCl, LiBr, LiI, Li₂O, Li₂S, Li₃N, and Li₂CO₃, covering both bulk (grain) diffusion and grain-boundary diffusion over representative low-energy surfaces and interfaces. This dataset includes homogeneous and heterogeneous interfaces, providing a unified design space for Li-ion migration in SEI environments. A path-aware graph variational autoencoder (GVAE) was then used to learn latent representations of NEB trajectories, and the learned embeddings were incorporated one into a predictive GNN model for minimum-energy-path and migration-barrier estimation. The combined GVAE-GNN framework achieved strong predictive performance for both grain and grain-boundary diffusion, with test-set R² values of 0.93 and 0.94, respectively. Feature-importance and latent-space analyses further show that migration behavior is governed not only by composition, but also by local structural factors such as exposed surface, saddle-point character, and grain-boundary energetics. The results reveal clear transport trends across SEI chemistries, with halide-rich systems generally associated with narrower low-barrier distributions, while Li₃N, Li₂CO₃, and structurally mismatched heterogeneous interfaces exhibit broader and higher-barrier landscapes. This study give a physics-informed machine-learning framework for mapping and predicting Li-ion migration in complex SEI structures and provides insight into how SEI chemistry and microstructure jointly control interfacial ion transport.

The accumulation of micro- and nanoplastics (MPs/NPs) in croplands threatens food quality and human health. This study investigates the effectiveness of pristine biochar (BC) and iron-doped biochar (Fe-BC), produced from pine sawdust via one-step pyrolysis, in reducing NPs uptake by lettuce (Lactuca sativa L. var. adela). Fe-BC exhibited greater porosity, higher surface area, and a slightly positive surface charge compared to BC. Using Pd-doped NPs, we confirmed that NPs can penetrate roots and translocate to leaves, accumulating more in older tissues. Soil application of 3 wt % BC or Fe-BC significantly lowered leaf NPs concentrations, with Fe-BC showing a greater reduction, by approximately 60% (from 0.90 to 0.36 mg/kg). While both BC and Fe-BC demonstrated a capacity for alleviating NPs-induced metabolic disturbances and partially restoring soil enzyme activities, Fe-BC presents a more promising amendment for mitigating NPs contamination and protecting crop quality.

Refractory multi-principal element alloys (RMPEAs) have garnered considerable interest for their exceptional high-temperature strength and promising applications in demanding environments. However, the differing and complex compositions among many RMPEAs pose significant challenges to understanding the deformation mechanisms that govern their plastic deformation. In this work, we address these challenges by conducting atomistic simulations to determine the local slip resistances (LSRs) for edge and screw dislocations on the {110}, {112}, and {123} slip planes in 12 body-centered cubic (BCC), equal-molar RMPEAs. To elucidate the relationship between LSR and the underlying physical properties of the alloys, we employ machine learning methods, enabling a systematic analysis of how compositional variations affect the dislocation behavior. Building on these insights, we develop an analytical model based on thermally activated theory to predict the macroscopic yield stress of RMPEAs. Our results demonstrate that increasing the fraction of hexagonal close-packed (HCP) elements above 50% in the alloy composition notably reduces both the unstable stacking fault energy (USFE) and ideal shear strength (ISS), further lowering the LSRs of screw dislocations on all three slip planes. Furthermore, higher elastic anisotropy (quantified by the Zener ratio) diminishes USFE, ISS, and LSR, while severe lattice distortion (LD) reduces the screw-to-edge LSR ratio but enhances the -to- and -to- LSR ratios. By integrating an autoencoder with a random forest model, we analyze the correlation between material properties and LSR, revealing that elastic constants and LD coefficients play the most critical roles in determining LSR. Our thermally activated, dislocation-based model, which integrates the contributions of both edge and screw dislocations on all three slip planes, provides reliable predictions of tensile yield stress for BCC RMPEAs. This framework offers a robust foundation for alloy design and optimization in high-performance applications.

Microplastic pollution fosters the development of distinct microbial biofilm communities, termed the plastisphere, that vary across environmental contexts. Here, we used 16S rRNA gene sequencing combined with machine learning (ML) approaches to explore plastisphere microbial diversity and the interactions between potential plastic-degrading bacteria (PDBs) and non-plastic-degrading bacteria (NDBs) across ocean, surface water, and wastewater habitats. Our findings reveal that wastewater plastispheres harbor the most diverse and compositionally even microbial communities, likely driven by complex nutrient loads, pollutant inputs, and high microbial seeding potential. Genus-level analysis of potential PDBs indicated habitat-specific taxa, including Pseudomonas, Acinetobacter, and Aquabacterium in wastewater, Flavobacterium and Alteromonas in ocean, and Psychrobacter and Novosphingobium in surface waters. Network analyses using Pearson’s correlation and Random Forest modeling uncovered consistent co-occurrence patterns between potential PDBs and diverse NDB taxa such as Clostridium_sensu_stricto_5, Lachnospiraceae_UCG-001, and Cloacibacterium, suggesting potential facilitative interactions, including redox modulation, nutrient exchange, and biofilm support. ML tools proved effective in identifying key taxa and potential ecological interactions, but their application remains limited by taxonomic resolution, lack of functional validation, and insufficient integration of environmental metadata. These findings underscore the ecological complexity of plastisphere communities and the need for community-level approaches in plastic biodegradation research.

Refractory multi-principal element alloys (RMPEAs) represent an advanced subclass of MPEAs, distinguished by their complex multi-element compositions and significant inclusion of refractory elements. This study investigates the mechanical behavior of RMPEAs under compressive and tensile loading using molecular dynamics (MD) and hybrid MD/Monte Carlo (MD/MC) simulations. Mechanical responses are analyzed across a broad temperature range (300 K to 1673 K) for 16 alloy configurations, comprising one quinary, five quaternaries, and ten ternaries in post-MD/MC equilibrated states. A particular emphasis is placed on the roles of chemical short-range order and lattice distortion in influencing anisotropic mechanical properties. Compressive yield stresses are consistently higher than tensile yield stresses across all temperatures and compositions, demonstrating plastic anisotropy. Machine learning (ML) models reveal that temperature, lattice distortion, and Warren–Cowley parameter (WCP) values for pairs like Ta–Ta and Nb–Ta, and concentrations of Mo, W, and Nb are the most significant factors influencing yield stresses. Additionally, the ML models identify atomic size mismatch as the dominant contributor to lattice distortion, with WCP values playing a secondary role. The study also uncovers that while the lattice distortion promotes plastic anisotropy, the chemical short-range order decreases it, highlighting the critical interplay between atomic structure and mechanical behavior in RMPEAs.

This paper presents the Machine Learned Diffusion Coefficient Estimator, a comprehensive machine learning framework designed to predict diffusion coefficients in impure metallic (IM) and multi-component alloy (MCA) media. The framework incorporates five machine learning models, each tailored to specific diffusion modes:(1) impurity and (2) self-diffusion in IM media, and (3) self,(4) impurity, and (5) chemical diffusion in MCA media. These models use statistical aggregations of atomic descriptors for both the diffusing elements and the diffusion media, along with the temperature of the diffusion process, as features. Models are trained using the random forest and deep neural network algorithms, with performance evaluated through the coefficient of determination (R 2), mean squared error (MSE), and uncertainty estimates. The models within this framework achieve an impressive R 2 score above 0.90 with MSE ...

Climate change and the global energy crisis have led to an increasing need for greenhouse gas remediation and clean energy sources. The electrochemical CO₂ reduction reaction (CO₂RR) is a promising solution for both issues as it harvests waste CO₂ and chemically reduces it to more useful forms. However, the high overpotential required for the reaction makes it electrochemically unfavorable. Here, we fabricate a novel electrode composed of TiO₂ nanoparticles grown in situ on MXene charge acceptor 2D sheets with excellent CO₂RR characteristics. A straightforward solvothermal method was used to grow the nanoparticles on the Ti₃C₂T_x MXene flakes. The electrochemical performance of the TiO₂/MXene electrodes was analyzed. The Faradaic efficiencies of the TiO₂/MXene electrodes were determined, with a value of 99.41% at −1.9 V (vs. Ag/AgCl). Density functional theory mechanistic analysis was used to reveal the most likely mechanism resulting in the production of one CO molecule along with a carbonate anion through ∗CO, ∗O, and activated CO₂²⁻ intermediates. Bader charge analysis corroborated this pathway, showing that CO₂ gains a greater negative charge when TiO₂/MXene serves as a catalyst compared to MXene or TiO₂ alone. These results show that TiO₂/MXene nanocomposite electrodes may be very useful in the conversion of CO₂ while still being efficient in both time and cost.

The solid electrolyte interphase (SEI) is a passive layer, typically a few hundred angstroms thick, that forms on the electrode surface in the first few battery cycles when the electrode is in contact with the electrolyte in lithium-metal batteries. Composed of a combination of lithium salts and organic compounds, the SEI plays a critical role in battery performance, serving as a channel for Li-ion shuttling. Its structure typically comprises an inorganic component-rich sublayer near the electrode and an outer organic component-rich sublayer. Understanding heat transport through the SEI is crucial for improving battery pack safety, particularly since the Li-ion diffusion coefficient exhibits an exponential temperature dependence. This study employs first-principles calculations to investigate phonon-mediated temperature-dependent lattice thermal conductivity across the inorganic components of the SEI, including, LiF, Li₂O, Li₂S, Li₂CO₃, and LiOH. This study is also extended to the dependence of the grain size on thermal conductivity, considering the mosaic-structured nature of the SEI.

β-Ga₂O₃ is an important ultra-wide bandgap oxide semiconductor that has recently gained significant attention in radiation detection. This study examined the impact of helium-ion irradiation on the anisotropic thermal conductivity of iron-doped β-Ga₂O₃. A laser-based spatial domain thermoreflectance technique was used to measure the thermal conductivity map, which was validated against simulation results derived from density functional theory-based phonon transport simulations. Our experimental results revealed that the irradiation damage led to 40 % reduction in the thermal conductivity along [001] direction and 25 % along [100] directions, which significantly reduced the anisotropy of thermal conductivity. Phonon transport simulations indicated that the thermal conductivity decreases when helium atom is at interstitial sites or vacancy sites, and such reduction is particularly evident when helium atom occupies vacancy sites. This work underscores the role of irradiation-induced microstructural changes in the thermal transport properties of β-Ga₂O₃, which is crucial for its applications in sensor devices for extreme environments.

Metallic nanowires are widely employed as small-scale structural materials due to their characteristically small volume and high strength compared with their bulk counterparts. Nowadays, the mechanical properties of nanowires in pure metals are well understood with the help of experiments and simulations. However, the deformation of nanowires in metallic alloys remains elusive. In recent years, a new class of alloys called refractory multi-principal element alloys (RMPEAs) emerged. RMPEAs are alloys that form solid solution phases and consist of three or more principal elements, most of which are refractory metals. In this paper, we perform atomistic simulations to investigate the uniaxial deformation of nanowires in 16 body-centered cubic RMPEAs. For each RMPEA, three nanowires consisting of atoms randomly distributed in three different ways are used. The main finding is that dislocation slips on {110} planes and twinning on {112} planes, respectively, control the compressive and tensile plastic deformation of the nanowires. To provide references, we also study the deformation of nanowires in natural and A-atom potential-based artificial pure metals. Results show that the deformation of RMPEA nanowires cannot be predicted by simply extrapolating from those of pure metal nanowires, highlighting the significance of directly simulating RMPEAs using multiple random atomic structures. It is also found that RMPEAs possess a reduced tension-compression asymmetry compared with pure metals, regardless of the underlying plastic deformation mechanism.