AIonopedia: an LLM agent orchestrating multimodal learning for ionic liquid discovery
Yuqi Yin, Yibo Fu, Siyuan Wang, Peng Sun, Hongyu Wang, Xiaohui Wang, Lei Zheng, Zhiyong Li, Zhirong Liu, Jianji Wang, Zhaoxi Sun
2025-11-17
Summary
This paper introduces AIonopedia, a new artificial intelligence system designed to help scientists discover new ionic liquids, which are special salts with unique properties.
What's the problem?
Finding new ionic liquids with specific desired characteristics is really difficult because it's hard to predict their properties. There isn't a lot of data available, existing prediction methods aren't very accurate, and the process of designing and testing these liquids is often disorganized and inefficient.
What's the solution?
The researchers created AIonopedia, which uses a powerful type of AI called a Large Language Model. This AI was trained on a large collection of information about ionic liquids and can predict their properties with high accuracy. It also uses a smart search strategy to suggest new molecular designs. Importantly, they actually tested some of the AI’s suggestions in a lab to see if they worked in the real world.
Why it matters?
This work is important because it offers a way to speed up the discovery of new ionic liquids. These liquids have many potential applications, like in batteries, chemical reactions, and carbon capture, so a faster discovery process could lead to breakthroughs in these areas. The fact that the AI’s predictions were confirmed in the lab shows it’s a reliable tool for real-world research.
Abstract
The discovery of novel Ionic Liquids (ILs) is hindered by critical challenges in property prediction, including limited data, poor model accuracy, and fragmented workflows. Leveraging the power of Large Language Models (LLMs), we introduce AIonopedia, to the best of our knowledge, the first LLM agent for IL discovery. Powered by an LLM-augmented multimodal domain foundation model for ILs, AIonopedia enables accurate property predictions and incorporates a hierarchical search architecture for molecular screening and design. Trained and evaluated on a newly curated and comprehensive IL dataset, our model delivers superior performance. Complementing these results, evaluations on literature-reported systems indicate that the agent can perform effective IL modification. Moving beyond offline tests, the practical efficacy was further confirmed through real-world wet-lab validation, in which the agent demonstrated exceptional generalization capabilities on challenging out-of-distribution tasks, underscoring its ability to accelerate real-world IL discovery.