After 36 years, the Nobel Prize is finally awarded for MOF structures: As AI deciphers chemistry, metal-organic frameworks are stepping into the era of generative research.
On October 8, 2025, Susumu Kitagawa, Richard Robson, and Omar Yaghi, who have made contributions to the research in the field of metal-organic frameworks, were awarded the Nobel Prize in Chemistry. Over the past three decades, the field of metal-organic frameworks has evolved from structural design to industrialization, laying the foundation for computable chemistry. Today, artificial intelligence is reshaping MOF research with generative models and diffusion algorithms, opening a new era of chemical design.
On October 8, 2025, the Nobel Prize in Chemistry was unveiled in Sweden. The Royal Swedish Academy of Sciences decided to award the 2025 Nobel Prize in Chemistry to Professor Susumu Kitagawa from Kyoto University in Japan, Professor Richard Robson from the University of Melbourne, and Professor Omar Yaghi from the University of California, Berkeley, in recognition of their research contributions in the field of "metal-organic frameworks" (MOF, Metal–Organic Frameworks). This research field, which has been verified by the market for over 30 years, has now become an annual footnote in world science.
"The new molecular structures created by Susumu Kitagawa, Richard Robson, and Omar Yaghi contain large cavities that allow molecules to pass through. Therefore, they can extract moisture from desert air, remove pollutants from water, or capture carbon dioxide and store hydrogen," said Heiner Linke, the chairman of the Nobel Committee for Chemistry, in the official Nobel Prize news. Metal-organic frameworks have great potential and may bring previously unforeseen opportunities for customizing materials with new functions.
Beyond the Nobel Prize, the significance of MOF has long transcended the field of materials science itself, creating an era for humanity to re-understand the material world. When humanity discovered the "programmable" three-dimensional space at the molecular scale, chemistry has gradually shifted from "discovery" to "design," transitioning to the logic of data, algorithms, and AI.
More cutting-edge AI papers: https://hyper.ai/papers
From coordination polymers to MOF: A 36-year chemical puzzle
In fact, from the 1980s to finally winning the Nobel Prize in 2025, the research on MOF has witnessed a three-stage evolution from the exploration of chemical concepts to systematic design.
In 1989, Richard Robson first proposed the structural concept of three-dimensional coordination polymers, attempting to utilize the inherent properties of atoms in a new way. He used coordination bonds to connect metal nodes with organic ligands into a periodic network structure, combining positively charged copper ions with a four-armed molecule to construct a molecule with a chemical group at the end of each arm that can attract copper ions. "When they combine, they condense into an orderly and extremely empty crystal, just like a diamond full of countless voids."
This research result was published in the "Journal of the American Chemical Society" (JACS) under the title "Infinite Polymeric Frameworks Consisting of Three Dimensionally Linked Rod-like Segments." Among them, Ben F. Hoskins, one of the early co-authors of Robson, participated in publishing this groundbreaking paper, providing first-hand experimental evidence for the budding concept of MOF.
The paper jointly published by Richard Robson and Ben F. Hoskins
In the following 15 years, the research teams of Omar Yaghi and Susumu Kitagawa respectively published multiple papers in journals such as Nature and Science. They continuously achieved revolutionary breakthroughs in structural construction and functional regulation, establishing the new porous material system of MOF, and the research field gradually took shape and entered the stage of system expansion.
During this stage, the academic community began to conduct grafting and other treatments on the basic structure of MOF. Susumu Kitagawa proved that gases can flow in and out of this molecular structure and proposed the concepts of "flexible frameworks" and "breathing MOFs." This transformed MOF from a rigid pore material to a dynamic structural material, laying the foundation for the transformation of MOF from a rigid material to an intelligent responsive material.
In 1999, Omar Yaghi, known as the "father of MOF," created a very stable MOF (MOF-5) and proved that it can be modified through rational design to endow it with brand-new properties. As a representative metal-organic framework material, MOF-5 has been widely used in early research such as hydrogen storage and gas adsorption. His concept of "reticular chemistry" has promoted chemical synthesis into an era of predictable structures. Based on this, Mohamed Eddaoudi published papers such as "Metal-Organic Frameworks from Design Strategies to Applications" during the same period, advancing the experimental and synthetic chemistry of early high-specific-surface-area MOFs such as MOF-5.
During this period, Michael O’Keeffe also collaborated with Omar Yaghi to publish a research paper titled "Towards a Taxonomy of MOF Structures," systematically describing the underlying structures of crystals and MOF from a topological perspective.
The paper results of Mohamed Eddaoudi
As MOF research has boomed and become the core research object in fields such as gas storage, drug delivery, catalysis, and sensing, MOF materials have entered the stage of industrialization and have shown application potential in fields such as gas storage, carbon capture, and biomedicine. Many mainstream commercial MOF structures with high stability have begun to be industrialized. For example, the Zr-based UiO series developed jointly by the research team of Susumu Kitagawa and Uppsala University in Sweden has become a commercially available MOF with high thermal stability.
Since 2022, Atomis Co., Ltd., where Susumu Kitagawa serves as a scientific advisor, has joined hands with Yachiyo Engineering Co., Ltd. to develop a new energy gas distribution system called the "intelligent gas network" based on MOF technology. They hope to use the molecular structure of MOF to adsorb and release difficult-to-control methane gases such as biogas and natural gas at the nanoscale at room temperature without relying on pipeline infrastructure, achieving lightweight gas transportation.
The conceptual flowchart of the gas transportation project of Yachiyo Co., Ltd.
In the past 10 years, MOF has been widely used in industrial carbon capture. For example, the research group of George Shimizu from the University of Calgary developed a MOF material called CALF-20. In the paper published in Science by this research group, it was mentioned that CALF-20 can maintain its performance under conditions of moisture, oxidation, and exhaust gas, different from the defect that many previous MOFs are easily deactivated in a humid environment. Therefore, CALF-20 is used by Svante, a Canadian company, to capture carbon dioxide to remove greenhouse gases from the exhaust gas of cement production. In addition, the electronics industry has also begun to use MOF materials to absorb some toxic gases during semiconductor production.
The paper results related to the research on CALF-20
Regarding the special properties of MOF materials, Heiner Linke, the chairman of the Nobel Committee and a nanophysicist from Lund University in Sweden, humorously said, "This material is almost like Hermione's handbag in Harry Potter." The official Nobel Prize also stated in its official report that after the breakthrough discoveries of the three laureates, the tens of thousands of different MOFs constructed by chemists may help solve some of the major challenges facing humanity.
Currently, according to a survey in Science in July 2025, MOF has become the subject of over 100,000 academic papers globally.
When MOF is understood by algorithms: The resonance between chemistry and AI
Amid the wave of the integration of artificial intelligence and various fields, many researchers have begun to explore more possibilities of "AI + MOF." Previously, the research team of You Zhipeng from Nanchang University published a paper titled "Artificial Intelligence in Metal–Organic Frameworks from 2013 to 2024: A Bibliometric Analysis." They used bibliometric methods and knowledge graph visualization software to analyze the scientific research papers on MOFs AI in the Web of Science database from 2013 to the middle of 2024. Judging from the fitting curve, researchers' interest in the field of artificial intelligence in MOFs is increasing. Since 2016, the research on "AI + MOF" has even witnessed explosive growth, with the number of papers continuing to rise, indicating a promising future for this interdisciplinary direction.
The model fitting curve of the annual distribution and growth trend of "MOF + AI" articles from 2013 to 2024
The structural characteristics of MOF and the digitalization of chemistry
Currently, the structural characteristics of MOF are continuously promoting the digitalization of chemistry: The structure of "tunable metal nodes + organic ligands + topological grids" in MOF makes it a discretized chemical space that can be enumerated and parameterized, and it is a research object that material AI can understand.
The reason why MOF has become an "ideal object for material AI" stems from its natural modular and parameterizable characteristics. According to the description in the paper "From Data to Discovery: Recent Trends of Machine Learning in Metal–Organic Frameworks" published by the Korea Advanced Institute of Science and Technology in JACS, a MOF basically consists of three types of separable components, namely metal nodes (Metal Nodes), organic ligands (Linkers), and spatial topology (Nets). These three correspond to three types of enumerable discrete variables respectively:
* Different metal clusters/coordination numbers;
* Synthesizable organic ligand chemical skeletons;
* Topologically selectable connection methods.
The paper of the Korea Advanced Institute of Science and Technology
The combination of these dimensions makes the "scalability" of the MOF space increase exponentially. At the same time, since the candidate structures have clear chemical semantics, as candidate materials, MOF provides "programmable work steps" for the digital participation of machine learning. For example, metal clusters can be used as nodes, ligands as edges or hyperedges, and topological information can be encoded as network topological indicators. Therefore, graph neural networks (GNN) can directly learn properties such as adsorption energy and thermal stability from the structure; while scalar descriptions such as unit cell parameters, pore size distribution, surface area, and pore volume can be used as supervision labels or objective functions for multi-objective optimization.
In short, it is undeniable that compared with the situation where chemical structures cannot be enumerated, the modular grammar of MOF "discretizes" the chemical space into machine-readable rules, providing an ideal creation paradigm for material AI and opening up a new path for the digitalization of chemistry.
Two-way symbiosis: AI reshapes MOF research
The research teams from Tianjin University of Technology, the Chinese Academy of Sciences, and the Agency for Science, Technology and Research in Singapore mentioned in their jointly published paper "AI-driven Advances in Metal–organic Frameworks: from Data to Design and Applications" that "the latest advances in artificial intelligence and machine learning have brought transformative capabilities to the field of MOF. Key databases, deep learning architectures, generative models, and hybrid artificial intelligence simulation frameworks have reshaped the design and screening of high-performance MOF, enabling accurate property prediction, automated structure generation, and large-scale synthesis planning." In fact, in the past 5 years, many research teams have indeed achieved phased results in sub-fields such as property prediction and automated generation of MOF structures based on AI.
The paper of the joint team from Tianjin University of Technology, the Chinese Academy of Sciences, and the Agency for Science, Technology and Research in Singapore
The conceptual diagram of the research progress on the design, synthesis, performance prediction, and application of metal-organic framework materials driven by artificial intelligence
In 2024, the research teams from the Korea Advanced Institute of Science and Technology and the Pohang University of Science and Technology developed the first deep generative model MOFFlow specifically designed for MOF structure prediction to predict the generation framework of MOF structures. As a continuous regularization process, this framework utilizes the modular characteristics of MOF and adopts the method of flow matching. It regards metal nodes and organic ligands as rigid bodies and makes predictions in the SE space to reduce the structural complexity. Its architecture is as follows:
* Split the MOF into components such as metal nodes and organic ligands and define a consistent local coordinate system for them;
* Input them into the MOFFlow model, set a simple prior distribution as the starting point, define the transformation target from the prior to the target structure distribution, and use the flow matching framework to learn and arrange the random components reasonably;
* Embed the atomic information within the components as features based on the encoder, use graph neural networks to model the geometric and topological relationships between the components, and predict the rotation, translation, and lattice parameters of the components;
* Conduct sampling and structure reconstruction;
* MOFFlow outputs data, matches and compares the generated structure with