arXiv subjects

Shuaishuai Yuan

Publications and source records attributed to Shuaishuai Yuan.

2 recordsLinked to original sources

arXiv2026

Region-Resolved Interface Descriptors in III-nitride-nanocluster photocatalysts

Understanding how nanocluster cocatalysts modify the electronic structure of III-nitride surfaces is central to the rational design of efficient photocatalytic interfaces. Here, we introduce a region-resolved first-principles framework for analyzing GaN-, InGaN-, and ScGaN-based slabs decorated with six-atom elemental nanoclusters. Using a region-resolved projected local density of states (PLDOS) analysis, we separate electronic response of nanocluster-covered regions from that of uncovered semiconductor surface regions, revealing that semiconductor--nanocluster interfaces behave as laterally heterogeneous systems. In this picture, nanocluster-covered regions are associated with local band bending, charge injection, and nanocluster-induced gap states, whereas uncovered nitride regions retain surface-localized states and in-plane dipoles that may influence adsorbate interactions and water-activation pathways. These distinct regional responses motivate the definition of region-resolved interface descriptors, which are further integrated into an Interface Score: a physics-informed comparative metric that combines injection, activation, and interfacial-coupling contributions to rank heterogeneous model interfaces on a common descriptor basis. Machine-learning models trained on global and region-resolved descriptors are used primarily as interpretability tools to assess their correlation with single-H adsorption energetics. The results show that spatially averaged descriptors capture broad adsorption-energy trends, while region-resolved descriptors expose complementary local information related to interfacial electrostatics and charge separation. Together, this work establishes a descriptor framework for moving beyond spatially averaged electronic-structure analysis and for bridging first-principles calculations with experimental interpretation of heterogeneous photocatalytic interfaces.

cond-mat.mtrl-sci↗

arXiv2026

LatticeMind: A Validation-Gated Agentic Framework for Density Functional Theory Workflows

First-principles calculations are rarely single solver calls: they are multistage workflows in which structures, settings, and saved densities must pass correctly between dependent stages. Language models can write a valid single input, but reliability collapses once stages are chained: a dropped dependency, wrong file handoff, or hallucinated keyword can invalidate results even when the solver runs, and such errors surface only after costly execution. We introduce LatticeMind, a validation-gated agentic framework that converts natural-language requests into executable, checkable density functional theory (DFT) projects. A workflow contract specifies required stages, files, and dependencies; deterministic validators inspect the project before any solver launch; failed checks trigger bounded repair; and reports remain traceable to project files. On a 44-case RESCU benchmark spanning total-energy, density-of-states, band-structure, magnetic, phonon workflows, three non-flagship LLM backends reproduced every reference value; 119 of 132 runs passed on the first validated plan and the rest after one repair cycle. Ablations identify the workflow contract and pre-execution validation as the principal reliability mechanisms; a conventional trial-and-repair baseline reaches 71% at twice the token cost per success. The same architecture reproduces every reference value on a 19-case VASP benchmark, demonstrating transfer across solvers. Beyond these task families, LatticeMind generated and validated the 23 inputs of a multistage study of Rh-induced electronic responses at a Rh$_6$/In$_x$Ga$_{1-x}$N photocatalyst interface. These results establish workflow contracts, deterministic validation, bounded repair, and traceable records as a practical foundation for reliable AI-assisted scientific computing, where preventable errors are caught before execution and the solver is never the first validator.

cond-mat.mtrl-sci↗