A novel, open-access tool developed by University of Toronto Engineering researchers is poised to revolutionize how scientists navigate the rapidly expanding world of metal-organic frameworks (MOFs). These versatile materials, prized for their ultra-high surface area and tunable chemistry, hold immense potential in fields ranging from drug delivery and catalysis to carbon capture. However, keeping pace with the accelerating volume of MOF research has become a major challenge – one that this new tool, named MOF-ChemUnity, directly addresses.
The Promise and Complexity of MOFs
Metal-organic frameworks stand out due to their exceptional surface area – some variants boast up to 7,000 m²/g, meaning a single gram contains enough internal surface to cover a football field. This unique structure enables a wide array of applications. MOFs can act as molecular sieves, selectively capturing carbon dioxide from gas mixtures, or as sensors, detecting trace amounts of molecules. Their catalytic properties can speed up industrial reactions, while their porous structure allows controlled drug release.
The importance of MOFs was recently underscored by the 2025 Nobel Prize in Chemistry. Yet, the sheer growth of research – spanning over 25 application domains – has created a knowledge bottleneck, hindering both human researchers and the AI tools designed to assist them.
Introducing MOF-ChemUnity: A Structured Knowledge Graph
Professor Mohamad Moosavi’s team at the Department of Chemical Engineering & Applied Chemistry, and the Vector Institute, developed MOF-ChemUnity to overcome this challenge. The tool functions as a structured, scalable knowledge graph that systematically extracts and links information from MOF research papers, crystal structure repositories, and computational materials databases.
At its core, MOF-ChemUnity employs a multi-agent large language model (LLM) workflow to connect chemical names in scientific literature to their corresponding crystal structures. This allows synthesis procedures, material properties, and potential applications to be represented in a consistent, machine-readable format.
“Scientific discovery begins with reading and synthesizing the literature, but this remains one of the most difficult steps to automate,” explains Moosavi. “MOF-ChemUnity creates a unified foundation that both researchers and AI systems can build on.”
Reducing AI “Hallucinations” Through Grounded Literature
The team demonstrated the impact of their system by integrating the knowledge graph with LLMs to create a literature-informed AI assistant for MOFs. Unlike standard AI systems prone to generating plausible but incorrect statements (“hallucinations”), this assistant draws on verified experimental and computational records.
In blind evaluations conducted by MOF experts, the assistant’s responses were consistently rated as more accurate, interpretable, and trustworthy than those produced by baseline LLMs like GPT-4o. By grounding AI responses in curated and linked literature, the system significantly reduces the risk of unreliable scientific reasoning.
“This approach reduces hallucination, which is one of the major obstacles in applying large language models to scientific domains,” says Moosavi.
Open-Access and Future Implications
The U of T team – led by Moosavi, alongside key contributors Thomas Pruyn and Amro Aswad – has made the dataset and code openly available on GitHub, fostering continued progress in materials science and AI-driven research.
Moosavi envisions MOF-ChemUnity as a stepping stone toward a broader shift in how scientific knowledge is organized and accessed. The tool breaks down silos in research, enabling AI systems to process data across fields, exceeding the limitations of human researchers.
“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. It establishes a new paradigm for literature-informed discovery.”
This work not only accelerates materials discovery but also lays the groundwork for generalized knowledge systems that can revolutionize research across multiple scientific disciplines
