Research Digest — 2026-09-07¶
Interfaces & Anode Engineering¶
1. Void suppressive lithium anodes for all-solid-state batteries¶
Source: Nature Materials · 📅 2026-08-26 · ↗ Open paper
Investigates the kinetics and thermodynamics of void formation at the Li/solid-electrolyte interface, showing they are controlled by interfacial lithium concentration and diffusivity. A grain-refining strategy — introducing a Mg-1 wt% La inoculant into molten Li — increases diffusivity and pushes the critical current density to 2.2 mA cm-2 at 1.1 mAh cm-2.
Relevance to DENG.Group
Directly relevant to Shoutong Jin's phase-field work on void/dendrite evolution under current load, and to Umang Agarwal's electrolyte-electrode interface mechanics: it provides experimental validation that interface Li chemical potential and diffusivity (not just stack pressure) govern void nucleation. Cheng Peng may also note the grain-refinement angle, an analog of GB-engineering strategies in solid electrolytes.
Electrode Degradation & Spectroscopy Theory¶
2. Decoding Oxygen K-edge Fingerprints of NCM-811 Degradation via Ab Initio Many-Body Theory and High-Throughput Screening of Crystal Proxies¶
Source: arXiv · 📅 2026-09-02 · ↗ Open paper
Combines oxygen K-edge X-ray absorption spectroscopy with ab initio many-body perturbation theory and high-throughput DFT screening to identify spectroscopic fingerprints of layered, spinel, and rock-salt degradation products of NCM-811 cathodes. Highlights discrepancies between idealized bulk references and measured spectra of real degraded samples, building a route to decode degradation products from operando-like measurements.
Relevance to DENG.Group
A methods template for linking atomistic simulations to spectroscopy of battery degradation - pertinent to Yan Li and Mengke Li's halide electrolyte degradation studies, where similar XAS/XPS fingerprinting of decomposition products appears. Yanhao Deng can also mine the DFT-proxy screening workflow for training-data generation strategies.
SEI & Interphase Modeling¶
3. Predictive Simulation of Interphases on Li Metal Surface¶
Source: arXiv · 📅 2026-08-10 · ↗ Open paper
Attempts a universal simulation method to predict interphasial chemistries formed on Li metal from arbitrary electrolyte formulations, combining a transferable polarizable force field with a universal machine-learning force field. Across chemically diverse electrolytes the method reproduces experimental trends such as fluorinated solvents promoting LiF-rich interphases, aiming to replace Edisonian electrolyte trial-and-error.
Relevance to DENG.Group
Squarely on-theme for Umang Agarwal's electrolyte/electrode interphase work and Naibing Wu's electrolyte simulation projects: a concrete demonstration of coupling classical and universal ML force fields for reactive interphase formation. Also a useful benchmark of universal-potential reliability for the reactive chemistries the group simulates.
4. Solvation Restructuring Accelerates Early SEI Nucleation in Lithium Metal Batteries¶
Source: arXiv · 📅 2026-08-14 · ↗ Open paper
Uses AIMD-trained Deep Potential molecular dynamics to connect electrolyte solvation structure to early SEI nucleation in LiTFSI/DMC electrolytes. Increasing salt concentration shifts the liquid from solvent-separated ion pairs toward contact ion pairs and aggregates, with 3.5 M showing the strongest anion coordination; this anion-rich environment accelerates interfacial Li-F/Li-O bond formation and yields a denser, LiF/Li2O-rich nascent interphase.
Relevance to DENG.Group
Methodologically instructive for Naibing Wu (Deep Potential training on AIMD for electrolyte chemistry) and for Yan Li and Mengke Li's work on how local coordination chemistry controls interfacial decomposition in halide systems - the salt-concentration-to-interphase-composition mapping is a transferable design intuition.
Tools & Workflows¶
5. Bayesian Optimization for Self-Driving Materials Laboratories: From Algorithms to Physics-Informed Workflows¶
Source: arXiv · 📅 2026-08-26 · ↗ Open paper
A review of Bayesian optimization as the decision engine for self-driving laboratories, written around the practical messiness of real campaigns: failed or missing experiments, noise and drift, mixed variables, constraints, multi-objective trade-offs, variable cost and fidelity, and batch/asynchronous operation. Covers Gaussian-process surrogate choices and how to inject prior physics knowledge into the loop.
Relevance to DENG.Group
Useful roadmap for the group's growing simulation-driven screening pipeline: the same BO-with-constraints machinery applies when Yanhao Deng's ML-potential workflows or active-learning loops must allocate expensive DFT/MD evaluations. Tim Pook may find the workflow-engineering perspective relevant for cluster automation.
6. Closed-Loop Solid-State Synthesis Planning for Materials Discovery With Large Language Models¶
Source: Advanced Materials · 📅 2026-08-10 · ↗ Open paper
Extracts key synthesis information (target compounds, precursors, processing parameters) from 4,407 open-access solid-state synthesis papers into a structured recipe dataset, then builds an LLM platform that proposes and iteratively refines synthesis plans in a closed loop. Demonstrates dramatic reduction in trial-and-error for new inorganic materials synthesis.
Relevance to DENG.Group
Complements the group's AI-for-materials thread: as Jerry's group expands from simulation toward guiding synthesis collaborators, this shows how literature-mined structured datasets plus LLM agents can bridge computational predictions to actual sample preparation - e.g., for validating computationally screened halide or sulfide electrolyte candidates.