LLM Agents for Knowledge Discovery in Atomic Layer Processing

Andreas Werbrouck, Marshall B. Lindsay, Matthew Maschmann, Matthias J. Young

Keywords: Agentic Systems, Knowledge Discovery, Thin Film Deposition, Atomic Layer Processing

TL;DR: We gave an LLM agent without access to a simulated chemical reactor with plenty of processes to discover. It's only task was to explore and synthesize what it observed.

Abstract: Large Language Models (LLMs) have garnered significant attention for several years now. Recently, their use as independently reasoning agents has been proposed. In this work, we test the potential of such agents for knowledge discovery in materials science. We repurpose LangGraph's tool functionality to supply agents with a black box function to interrogate. In contrast to process optimization or performing specific, user-defined tasks, knowledge discovery then consists of freely exploring the system, posing and verifying statements about the behavior of this black box, without explicit objective besides the generation and verification of generalizable statements. We provide proof of concept for this approach through a children's parlor game, demonstrating the role of trial-and-error and persistence in knowledge discovery, and a strong path-dependence of results. We then apply the same strategy to show that LLM agents can explore, discover, and exploit diverse chemical interactions in an advanced Atomic Layer Processing reactor simulation using intentionally limited probe capabilities and without explicit instructions.

Submission Track: Findings, Tools & Open Challenges

Submission Category: Automated Synthesis + Automated Material Characterization

AI4Mat RLSF: Yes

Submission Number: 84