A generative artificial intelligence framework based on a molecular diffusion model for the design of metal-organic frameworks for carbon capture

Hyun Park,Xiaoli Yan,Ruijie Zhu,Eliu A. Huerta,Santanu Chaudhuri,Donny Cooper,Ian Foster,Emad Tajkhorshid
DOI: https://doi.org/10.1038/s42004-023-01090-2
IF: 7.211
2024-02-15
Communications Chemistry
Abstract:Metal-organic frameworks (MOFs) exhibit great promise for CO 2 capture. However, finding the best performing materials poses computational and experimental grand challenges in view of the vast chemical space of potential building blocks. Here, we introduce GHP-MOFassemble, a generative artificial intelligence (AI), high performance framework for the rational and accelerated design of MOFs with high CO 2 adsorption capacity and synthesizable linkers. GHP-MOFassemble generates novel linkers, assembled with one of three pre-selected metal nodes (Cu paddlewheel, Zn paddlewheel, Zn tetramer) into MOFs in a primitive cubic topology. GHP-MOFassemble screens and validates AI-generated MOFs for uniqueness, synthesizability, structural validity, uses molecular dynamics simulations to study their stability and chemical consistency, and crystal graph neural networks and Grand Canonical Monte Carlo simulations to quantify their CO 2 adsorption capacities. We present the top six AI-generated MOFs with CO 2 capacities greater than 2m mol g −1 , i.e., higher than 96.9% of structures in the hypothetical MOF dataset.
chemistry, multidisciplinary
What problem does this paper attempt to address?
The problem that this paper attempts to solve is: **How to accelerate the design of metal - organic frameworks (MOFs) materials with high carbon dioxide adsorption capacity and synthesizable linkers, in order to address the computational and experimental challenges faced by traditional experimental methods in searching for the best MOF materials in the vast chemical space.** Specifically, the paper introduces GHP - MOFassemble, which is a generative artificial intelligence framework based on the molecular diffusion model, aiming to rationally and rapidly design MOFs with high CO₂ adsorption capacity and synthesizable linkers. The main objectives of the paper include: 1. **Generate new MOF linkers**: Generate new and unique MOF linkers through the diffusion model DiffLinker, and assemble them with pre - selected metal nodes (such as Cu paddlewheel, Zn paddlewheel, Zn tetramer) into MOFs with a primitive cubic topology. 2. **Screen and validate the generated MOFs**: Screen and validate the AI - generated MOFs to ensure their uniqueness, synthesizability and structural validity. Use molecular dynamics simulations to study their stability and chemical consistency, and quantify their CO₂ adsorption capacity through crystal graph neural networks and Grand Canonical Monte Carlo (GCMC) simulations. 3. **Discover high - performance MOFs**: Identify the top six AI - generated MOFs with a CO₂ adsorption capacity greater than 2 mmol g⁻¹, and the performance of these MOFs is better than 96.9% of the structures in the hypothetical MOF dataset. Through this method, researchers hope to more efficiently find high - performance MOFs materials suitable for carbon capture applications in the vast chemical space.