University of Toronto Engineering researchers have launched an innovative, open-access tool that aims to streamline this process, called MOF-ChemUnity. This tool is designed to make it easier to discover and apply metal-organic frameworks (MOFs). This new campus can greatly enhance scientific research in a variety of disciplines. It dives deep into drug delivery, catalysis, and carbon capture, along with other dazzling uses of MOFs. Due to their amazing possibilities, MOFs have even been nominated for the 2025 Nobel Prize in Chemistry.
MOFs are extraordinary materials with incredible surface areas. Some have surface areas that can even achieve an astonishing 7,000 square meters per gram! Consequently, just one gram of MOF could in theory spread across an entire football field. The tunability and diversity of MOFs has resulted in research momentum across more than 25 application areas. As a result, the pace of research studies zeroing in on these structures has markedly increased.
MOF-ChemUnity: A Comprehensive Resource
MOF-ChemUnity was created by UMD assistant professor Mohamad Moosavi with the help of his graduate students. This powerful new resource provides a unique, systematic, and comprehensive approach to simplifying and synthesizing knowledge about MOFs. At its core, it uses a knowledge graph that catalogs about 0.5 million pieces of data and relationships for more than 15,000 unique MOFs. As an example, this new graph provides an unprecedented resource for directly linking common chemical names found in scientific literature to their individual crystal structures.
Professor Moosavi emphasized the value of this tool by stating, “A knowledge graph connects pieces of information like a web, linking things, like a MOF, its metal node, synthesis protocol, and adsorption property through their relationships—’made from’, ‘synthesized’, ‘used for.’”
MOF-ChemUnity also addresses a key barrier to using AI in scientific disciplines. This is accomplished with a unique multi-agent large language model (LLM) workflow that aggressively minimizes hallucination. Moosavi explained, “This approach reduces hallucination, which is one of the major obstacles in applying large language models to scientific domains.”
Enhancing AI-Driven Research
By integrating AI throughout the MOF-ChemUnity framework, we allow researchers to do more than just store data. It is better enabling us to understand how materials, properties, and applications are all interrelated with one another. “This lets AI not just store data but understand and reason about how materials, properties and applications are connected—exactly what MOF-ChemUnity enables,” Moosavi stated.
MOF-ChemUnity is releasing their dataset and code to the public on GitHub. We are witnessing a fundamental change in the way scientific knowledge is organized and made accessible. This open-access model has been incredibly advantageous for researchers. It prepares the ground for AI systems that can synthesize huge swathes of knowledge in multiple disciplines. “Human researchers are limited by the number of papers they can read, but MOF-ChemUnity takes a first step toward enabling AI systems that can process data across fields,” Moosavi noted.
A New Era for Scientific Discovery
MOF-ChemUnity also lays important groundwork for accelerating future research across materials science and AI-enabled fields. It acts as a common, public starting point that both human researchers and AI systems can build on to speed up discovery. “It establishes a new paradigm for literature-informed discovery, and we envision it as the beginning of generalized knowledge systems that can accelerate research across many fields,” Moosavi remarked.
Our vision of MOF-ChemUnity benefits goes much deeper than just for specific research projects. By breaking down silos in scientific research, it encourages collaboration and cross-disciplinary approaches that could lead to innovative solutions for global challenges. Moosavi further emphasized this point by stating, “This work will help break down silos in scientific research.”