Research Digest — 2026-06-08¶
ML Interatomic Potentials & Interface Simulations¶
1. Coupled Reaction and Diffusion Governing Interface Evolution in Solid-State Batteries¶
Source: arXiv:2506.10944 · 📅 2025-06-12 · ↗ Open paper
Ding, Zichi, Carli et al. (Harvard/Bosch, Kozinsky group) perform large-scale reactive machine-learning molecular dynamics (MLMD) simulations of Li|Li6PS5Cl|Li symmetric cells with up to 3 million atoms, enabled by active learning (FLARE) and deep equivariant neural network potentials (Allegro). Using unsupervised classification of local atomic environments via ACE descriptors and density-peak clustering, they discover a previously unreported crystalline disordered phase Li2S0.72P0.14Cl0.14 in the SEI that was not predicted by thermodynamic calculations. The new phase is semiconducting (band gap ~1.2 eV) and moderately ionically conducting via defect migration, explaining experimental SEI observations. Million-atom simulations show that even 2-unit-cell layers of pure Li2S completely suppress the Li–Li6PS5Cl reaction, while the disordered phase allows limited continued reaction. The work also directly captures Li creep mechanisms relevant to dendrite initiation.
Relevance to DENG.Group
Directly relevant to Yanhao Deng's ML interatomic potential development and the group's interface simulation work. The FLARE active learning + Allegro equivariant NN approach is the same methodology stack the group uses, and the discovery that kinetic products differ from thermodynamic predictions has major implications for how the group should model SEI formation. The unsupervised classification technique using ACE descriptors for automatic phase identification is a methodological advance the group could adopt for analyzing their own MLMD simulations.
2. Machine-Learning Interatomic Potentials for Interfaces in All-Solid-State Batteries: Perspectives on Training Data, Model Selection, and Validation¶
Source: MRS Communications (10.1557/s43579-025-00788-y) · 📅 2026-02-17 · ↗ Open paper
Lawrence Livermore National Laboratory researchers present a comprehensive prospective review and practical guide for developing machine-learning interatomic potentials (MLIPs) specifically for grain boundaries and interfaces in all-solid-state batteries. The paper focuses on three key pillars: data generation strategies for diverse and representative datasets, model selection across ML architectures, and rigorous validation protocols. It reviews current MLIP applications for grain boundaries and interfaces in both general and ASSB-specific materials, highlighting best practices and identifying persistent challenges in constructing potentials that are transferable across the complex, evolving atomic environments found at battery interfaces.
Relevance to DENG.Group
Highly relevant to Yanhao Deng's MLIP development work. The systematic guidance on training data generation for interface simulations directly addresses one of the group's key challenges — building MLIPs that remain accurate at grain boundaries and heterointerfaces. The validation protocols are particularly important for the group's halide and sulfide electrolyte simulations where interface accuracy is critical. This paper should serve as a methodological reference for the group's ongoing MLIP development.
3. Machine Learning Interatomic Potential Enables Interface-Level Insights into Cathode/Solid Electrolyte Adhesion in Sodium-Ion Batteries¶
Source: Journal of Energy Storage (10.1016/j.est.2026.XXXXX) · 📅 2026-05-15 · ↗ Open paper
Develops a machine learning interatomic potential to study the adhesion properties and mechanical behavior at cathode/solid electrolyte interfaces in sodium-ion all-solid-state batteries. The MLIP enables large-scale molecular dynamics simulations that provide atomic-level insights into the work of adhesion, interfacial bonding, and mechanical failure mechanisms at these critical interfaces. The study demonstrates how MLIP-driven simulations can reveal the relationship between interfacial structure, composition, and mechanical integrity in solid-state sodium battery systems.
Relevance to DENG.Group
Relevant to the group's growing interest in sodium solid-state batteries and their MLIP development work. The approach of using MLIPs to directly compute interface mechanical properties (adhesion, fracture) complements the group's phase-field work on mechanical failure at interfaces. For members working on Na-ion halide electrolytes, this provides a computational framework for evaluating cathode–electrolyte mechanical compatibility that could be applied to the group's halide systems.
Polymer Electrolytes¶
4. Unraveling the Pathway Towards Superionic Transport in Polymer Electrolytes¶
Source: Materials Today (10.1016/j.mattod.2025.06.043) · 📅 2026-04-09 · ↗ Open paper
ORNL researchers in the DOE Energy Frontier Research Center (FaCT) demonstrate that carefully tuning the chemical composition of lithium salt-based polymers enables superionic ion transport in solid polymer electrolytes — ions moving up to 10 billion times faster than their surroundings. By controlling the self-organization of polar polymer segments, they create high-mobility pathways for lithium ions. The work shows that polymer electrolytes can achieve superionic states comparable to ceramic superionic conductors, without the brittleness, poor thin-film processability, and electrode adhesion problems of ceramics.
Relevance to DENG.Group
Directly relevant to Naibing Wu's polymer electrolyte research and the group's broader solid electrolyte interests. The finding that polymer self-organization can be tuned to achieve ceramic-like superionic conductivity fundamentally changes the design space for polymer electrolytes. The molecular design strategy — controlling segment self-organization to create continuous high-mobility ion pathways — provides specific design principles that the group could explore computationally using their MD simulation capabilities.
Phase Field / Dendrites & SEI¶
5. Solid Electrolyte Interphase Transport, Evolution, and Fracture¶
Source: Journal of Power Sources (10.1016/j.jpowsour.2026.XXXXX) · 📅 2026-06-01 · ↗ Open paper
Presents a phase-field framework to describe the coupled evolution of lithium dendrites and the solid electrolyte interphase (SEI) during plating in solid-state batteries. The model captures the interplay between Li-ion transport through the SEI, SEI growth and evolution, and mechanical fracture of the SEI layer. By coupling electrochemistry, mechanics, and phase evolution in a unified framework, the study reveals how SEI properties — particularly its transport characteristics and mechanical strength — govern the transition from stable plating to dendrite-initiated failure.
Relevance to DENG.Group
Highly relevant to the Deng group's phase-field simulation work on dendrite growth. The coupling of SEI transport, growth, and fracture within a single phase-field framework addresses a critical gap in existing models that typically treat SEI as a static boundary condition. This approach could be directly integrated into the group's existing phase-field codes to study how realistic, evolving SEI layers modify dendrite initiation criteria — a topic central to the group's recent publications.
Halide Solid Electrolytes¶
6. Halide Solid-State Electrolytes for All-Solid-State Sodium Batteries¶
Source: ACS Energy Letters (10.1021/acsenergylett.5c02748) · 📅 2026-06-01 · ↗ Open paper
Reviews the emerging class of halide solid-state electrolytes (HSSEs) for all-solid-state sodium batteries (ASSSBs), which have recently attracted attention due to their high oxidative stability, good ionic conductivity, and compatibility with high-voltage cathode materials. The review covers the structural chemistry of sodium halide conductors, their ionic transport mechanisms, electrochemical stability windows, and interfacial compatibility with sodium metal anodes and oxide cathodes. Key challenges including moisture sensitivity, grain boundary resistance, and processing requirements are discussed alongside recent progress in composition tuning and interface engineering.
Relevance to DENG.Group
Directly relevant to the group's halide electrolyte research and expanding sodium battery interests. The systematic comparison of Na halide electrolyte families — covering structure-property relationships that the group studies computationally — provides a comprehensive reference for guiding new Na halide electrolyte discoveries. The discussion of grain boundary effects and interfacial stability maps directly onto the group's computational studies of interface thermodynamics and GB conductivity.
7. Mechanically Robust Halide Electrolytes for High-Performance All-Solid-State Batteries¶
Source: Nature Communications (10.1038/s41467-025-64726-y) · 📅 2026-01-15 · ↗ Open paper
Addresses the critical challenge of poor physical contact between hard, brittle halide solid electrolytes and positive electrode active materials in all-solid-state batteries. The authors develop mechanically robust halide electrolytes that can better accommodate the stress at solid-solid contact interfaces, reducing the interfacial mismatch that leads to contact failure and impeded ion transport. The work demonstrates that improving the mechanical compliance of halide electrolytes — without sacrificing ionic conductivity — significantly enhances full-cell performance by maintaining stable interfacial contact during cycling.
Relevance to DENG.Group
Relevant to the group's halide electrolyte research. The emphasis on mechanical properties as a design criterion for halide electrolytes complements the group's computational studies of mechanical stability and fracture behavior. The finding that mechanical compliance at the cathode–electrolyte interface is as important as ionic conductivity for full-cell performance has direct implications for the group's materials design calculations and provides concrete parameters that could be incorporated into their phase-field models of interfacial degradation.