OSDA Agent: Leveraging Large Language Models for De Novo Design of Organic Structure Directing Agents

Zeolites are crystalline porous materials that have been widely utilized in petrochemical industries as well as sustainable chemistry areas. Synthesis of zeolites often requires small molecules termed Organic Structure Directing Agents (OS-DAs), which are critical in forming the porous structure. Molecule generation models can aid the design of OSDAs, but they are limited by single functionality and lack of interactivity. Meanwhile, large language models (LLMs) such as GPT-4, as general-purpose artificial intelligence systems, excel in instruction comprehension, logical reasoning, and interactive communication. However, LLMs lack in-depth chemistry knowledge and first-principle computation capabilities, resulting in uncontrollable outcomes even after fine-tuning. In this paper, we propose OSDA Agent, an interactive OSDA design framework that leverages LLMs as the brain, coupled with computational chemistry tools. The OSDA Agent consists of three main components: the Actor, responsible for generating potential OSDA structures; the Evaluator, which assesses and scores the generated OSDAs using computational chemistry tools; and the Self-reflector, which produces reflective summaries based on the Evaluator's feedback to refine the Actor's subsequent outputs. Experiments on representative zeolite frameworks show the generationevaluation-reflection-refinement workflow can perform de novo design of OSDAs with superior generation quality than the pure LLM model, generating candidates consistent with experimentally validated OSDAs and optimizing known OSDAs. INTRODUCTION Zeolites, a class of microporous silicate-based materials, have been widely used as highly efficient catalysts and adsorbents, in petrochemical industries and sustainable chemistry processes (Davis, 2002) . The unique properties of zeolites stem from their porous frameworks and the synthesis of zeolites often requires the use of small molecules named Organic Structure Directing Agents (OS-DAs) (Moliner et al., 2013) . OSDAs act as templates influencing the size, shape, and connectivity of the pores of zeolites. The design of effective OSDAs is crucial for tailoring the properties of the target zeolite, but this process has traditionally been guided by empirical knowledge and laborintensive trial-and-error methods. Recently, artificial intelligence has been extensively applied across various scientific disciplines, including chemistry (