Research Digest — 2026-10-13¶
AI Methods for Solid Electrolyte Discovery¶
1. Automatic Generation of Inorganic Solid-State Electrolytes via Unified Multi-Modal Network¶
Source: Small (Wiley) · 📅 2026-10-08 · ↗ Open paper
Introduces USMNet, a unified structure-aware multimodal neural network that combines crystal-structure and compositional representations to predict ion transport properties of inorganic solid electrolytes, addressing the sparsity and structure-dependence of experimental ionic conductivity data. The model is coupled to an automated screening/generation workflow that proposes new solid electrolyte candidates.
Relevance to DENG.Group
Relevant to the whole group's screening efforts. Naibing Wu could test whether the multimodal structure+composition approach extends to polymer and composite electrolytes, where conductivity data is even sparser than inorganics; it also complements Yanhao Deng's ML-potential-based transport pipeline and the halide-electrolyte discovery interests of Yan Li and Mengke Li.
Interfaces, Interphases & Degradation¶
2. Distinguishing the mixed conduction interphase: a machine learning molecular dynamics study on the Na3PS4/Na battery interface¶
Source: Machine Learning: Science and Technology (IOP) · 📅 2026-09-14 · ↗ Open paper
Machine-learning molecular dynamics study of the Na3PS4/Na metal interface that resolves and distinguishes the mixed-conduction interphase forming at the contact — separating ionic from electronic transport contributions in the interfacial layer that controls Na plating/stripping stability in sulfide-electrolyte cells.
Relevance to DENG.Group
Methodologically the closest analogue to Yan Li and Mengke Li's halide-electrolyte degradation and transport simulations: the same MLMD workflow for identifying interphase chemistry and mixed conduction at metal|sulfide interfaces transfers directly to Li|halide systems. Also a useful reference for Umang Agarwal's heterogeneous electrolyte/electrode interface project, and for benchmarking Li vs Na interface chemistry.
3. Mechanical atomic-scale defects enable stable Ni-rich single-crystal cathodes¶
Source: Joule (Cell Press) · 📅 2026-10-01 · ↗ Open paper
Shows that Ni-rich single-crystal cathode particles can cycle stably under mechanical confinement when atomic-scale mechanical defects are engineered into the particles, and proposes mechanical defects as a new design axis for next-generation cathode materials. Combines atomic-scale characterization with chemo-mechanical analysis of how defects relieve stress during cycling.
Relevance to DENG.Group
Chemo-mechanical coupling is a shared theme with Shoutong Jin's phase-field mechanics/dendrite work and Cheng Peng's defect and grain-boundary perspective: this paper argues defects can be beneficial (stabilizing) rather than purely deleterious — an interesting counterpoint to grain-boundary engineering strategies in single-crystal vs polycrystalline particles, and relevant when modeling stress generation at cathode|electrolyte interfaces.