FURI | Fall 2024, Summer 2024

Prediction of New Methane-Capture Materials Using Large Language Models

This project explores the application of large language models (LLMs) — such as gpt-3.5-turbo-1106 via OpenAI — to predict the chemical composition of new materials to be used to capture methane from the atmosphere. The LLM model is trained on and capable of retrieving various chemical and material property data, and gives outputs predicting the properties and structures of new materials based on known methane-capture formulas. The model is capable of considering chemical property specifications fed to it as natural language text by users, allowing its outputs to be applicable to a wide range of material-development scenarios and constraints.