Research Digest — 2026-06-03¶
Halide Solid Electrolytes¶
1. Aluminum chloride-based catholytes for stable high-voltage solid-state sodium batteries¶
Source: Journal of Materials Chemistry A (10.1039/D5TA08632A) · 📅 2026-05-27 · ↗ Open paper
Investigates NaAlCl4-based solid-state catholytes for sodium all-solid-state batteries, studying high-voltage interactions with layered oxide cathodes (NaNi0.5Mn0.5O2). The authors find that bulk fluorination of NaAlCl4 improves ionic conductivity to 0.1 mS/cm but does not enhance oxidative stability, whereas a NaF surface coating on the cathode effectively mitigates interfacial degradation and enables stable high-voltage cycling.
Relevance to DENG.Group
Relevant to the group's broader interest in halide electrolytes and interface stability. The finding that surface coating strategies outperform bulk fluorination for high-voltage stability provides practical guidance for Na-ion solid-state cell design. The XPS analysis of halide-cathode interfacial decomposition parallels approaches the group could apply to Li-halide systems.
ML Interatomic Potentials¶
2. Comparing fine-tuning strategies of MACE machine learning force field for modeling Li-ion diffusion in LiF for batteries¶
Source: arXiv:2510.05020 (updated April 2026) · 📅 2026-04-09 · ↗ Open paper
Benchmarks MACE foundational model (MACE-MPA-0) against a well-trained DeePMD potential for predicting interstitial Li diffusivity in LiF, a key SEI component. The pre-trained MACE model achieves activation energy predictions (0.22 eV) close to the DeePMD reference (0.24 eV), while fine-tuning with only 300 data points further improves accuracy to 0.20 eV. This demonstrates that foundational MLIPs can match task-specific models trained on 40,000+ data points.
Relevance to DENG.Group
Highly relevant to Yanhao Deng's ML interatomic potential research. The finding that MACE foundational models need only ~300 fine-tuning data points to match DeePMD performance has direct implications for the group's workflow — potentially reducing DFT training data requirements by 100x. The LiF test case is also directly relevant to the group's SEI modeling work.
3. Domain oriented universal machine learning potential enables fast exploration of chemical space of battery electrolytes¶
Source: Nature Communications (s41467-025-67982-x) · 📅 2026-05-27 · ↗ Open paper
Develops a universal ML potential for liquid battery electrolytes trained via iterative learning on randomly composed datasets spanning a broad chemical space. The model accurately predicts transport properties (ionic conductivity, viscosity) and solvation structures across diverse electrolyte compositions. A novel coordination dynamics analysis framework quantifies solvation strength through coordination lifetime, providing a direct measure of ion-solvent interaction strength.
Relevance to DENG.Group
Relevant to Yanhao Deng's ML potential development and the group's electrolyte modeling work. The universal potential approach and iterative training strategy could be adapted for solid electrolyte systems. The coordination lifetime metric for quantifying solvation strength is a useful analytical tool that could be extended to characterize Li⁺ environments in solid polymer and composite electrolytes, relevant to Naibing Wu's work.
Interfaces & Electrode Stability¶
4. 2026 Roadmap on Next-Generation Solid Electrolytes for Battery Technologies¶
Source: Energy Research & Social Science (10.1088/2752-5724/ae5120) · 📅 2026-05-30 · ↗ Open paper
A comprehensive 2026 roadmap covering the current state of the art in sulfide- and halide-based solid electrolytes for Li and Na systems, examining post-lithium chemistries, advanced characterization techniques, and manufacturing scale-up challenges. The review provides a forward-looking perspective on the key scientific and engineering bottlenecks for solid-state battery commercialization.
Relevance to DENG.Group
Essential reference for the entire Deng group. As a roadmap, it provides the big-picture context for where the field is heading and where the group's research fits within the broader landscape. Useful for grant proposals, group meeting discussions, and strategic planning of research directions.