Ilmu Komputer & AI editorial
Open AccessOA2026
PVmatAgent: A Large Language Model (LLM) Agent for Perovskite Photovoltaic Material Design and Analysis
An autonomous LLM-based computational agent integrating 11 domain-specific tools for perovskite photovoltaic material design, with hallucination truncation and retrieval-augmented generation.
Hang-Yuan Deng; Yu-Yan Wu; You-Le Wang; Lei Zhang· Materials Genome Engineering Advances· 2026· DOI 10.1002/mgea.70102
The core problem
The rapid advancement of large language models (LLMs) has opened new opportunities for materials informatics. However, LLMs fall short in photovoltaic (PV) material design due to their lack of domain grounding, unreliable outputs, and inability to perform integrated computational tasks. To address these issues, this work presents PVmatAgent, an autonomous LLM-based computational agent designed specifically for PV materials design and analysis. The system aims to provide a practical paradigm for LLM-based autonomous agents in AI-driven optoelectronic materials discovery.
Innovation
The system was validated on four representative scenarios. Results demonstrate that PVmatAgent effectively executes computational tasks and provides corrective recommendations grounded in literature evidence for photovoltaic material design. The integration of multiple tools allows for comprehensive analysis, from structure retrieval to performance evaluation, while the hallucination truncation mechanism ensures numerical reliability.
The rapid advancement of large language models (LLMs) has opened new opportunities for materials informatics. However, LLMs fall short in photovoltaic (PV) material design due to their lack of domain grounding, unreliable outputs, and inability to perform integrated computational tasks. To address these issues, this work presents PVmatAgent, an autonomous LLM-based computational agent designed specifically for PV materials design and analysis. The system aims to provide a practical paradigm for LLM-based autonomous agents in AI-driven optoelectronic materials discovery.
PVmatAgent integrates 11 domain-specific tools organized into 9 functional modules. These include:
Why it matters
PVmatAgent addresses key limitations of LLMs in materials science by combining domain-specific tools with a RAG knowledge base and a hallucination truncation mechanism. This approach enables autonomous, reliable, and integrated computational workflows for perovskite PV material design. The successful validation on four scenarios suggests that such LLM-based agents can accelerate AI-driven optoelectronic materials discovery. Future work may expand the toolset and improve generalization to other material classes.
Who should read this
CS practitioners and researchers
Opening member content…