MOF Synthesis Prediction Enabled by Automatic Data Mining and Machine Learning

When and Where

Nov 28, 2022
5:00pm - 5:15pm

Hynes, Level 2, Room 206

Presenter

Yi Luo

Saientan Bag

Orysia Zaremba

Adrian Cierpka

Jacopo Andreo

Stefan Wuttke

Pascal Friederich

Manuel Tsotsalas

Co-Author(s)

Yi Luo¹,Saientan Bag²,Orysia Zaremba³,Adrian Cierpka⁴,Jacopo Andreo³,Stefan Wuttke³,Pascal Friederich⁴,Manuel Tsotsalas¹

Institute of Functional Interfaces, Karlsruhe Institute of Technology¹,Institute of Nanotechnology, Karlsruhe Institute of Technology²,Basque Center for Materials, Applications & Nanostructures³,Institute of Theoretical Informatics, Karlsruhe Institute of Technology⁴

Abstract

Yi Luo¹,Saientan Bag²,Orysia Zaremba³,Adrian Cierpka⁴,Jacopo Andreo³,Stefan Wuttke³,Pascal Friederich⁴,Manuel Tsotsalas¹

Metal–organic frameworks (MOFs) are porous materials formed via the connection between metal-centered nodes with organic linkers.1 With the utilization of different metal nodes and linkers, the topology, pore size, and functional groups of MOFs can be adjusted for high specific surface area, numerous active sites, and mass transfer channels.2 This flexibility in the MOF design allows great potential for MOF-based and MOF-derived material in the application in energy storage and conversion.3 Computer-assisted methods have been applied in the acceleration for the design of MOF-based materials.4–6 However, the potential of using machine learning (ML) methods to suggest parameters in MOF synthesis experiments is not well explored. Here, we show an approach of data mining and machine learning (ML) method for rationalization and acceleration of the MOF discovery process by directly predicting the synthesis condition of a new MOF based on its crystal structure.7 There are three steps in our approach: 1) Creation of a MOF synthesis database (SynMOF) via automatic extraction of synthesis parameters from scientific literature; 2) Training of multiple ML models on the SynMOF database; 3) Prediction of synthesis conditions for new MOF structures by the ML models. These early-stage ML models, exhibit a good prediction performance, surpassing human expert predictions, which was shown through a synthesis survey from 11 MOF experts worldwide. References (1) Furukawa, H.; Cordova, K. E.; O’Keeffe, M.; Yaghi, O. M. The Chemistry and Applications of Metal-Organic Frameworks. Science 2013, 341 (6149). (2) Du, R.; Wu, Y.; Yang, Y.; Zhai, T.; Zhou, T.; Shang, Q.; Zhu, L.; Shang, C.; Guo, Z. Porosity Engineering of MOF-Based Materials for Electrochemical Energy Storage. Advanced Energy Materials 2021, 11 (20), 2100154. (3) Adegoke, K. A.; Maxakato, N. W. Porous Metal-Organic Framework (MOF)-Based and MOF-Derived Electrocatalytic Materials for Energy Conversion. Materials Today Energy 2021, 21, 100816. (4) Moosavi, S. M.; Chidambaram, A.; Talirz, L.; Haranczyk, M.; Stylianou, K. C.; Smit, B. Capturing Chemical Intuition in Synthesis of Metal-Organic Frameworks. Nat Commun 2019, 10 (1), 539. (5) Luo, Y.; Ahmad, M.; Schug, A.; Tsotsalas, M. Rising Up: Hierarchical Metal–Organic Frameworks in Experiments and Simulations. Advanced Materials 2019, 31 (26), 1901744. (6) Ahmad, M.; Luo, Y.; Wöll, C.; Tsotsalas, M.; Schug, A. Design of Metal-Organic Framework Templated Materials Using High-Throughput Computational Screening. Molecules 2020, 25 (21), 4875. (7) Luo, Y.; Bag, S.; Zaremba, O.; Cierpka, A.; Andreo, J.; Wuttke, S.; Friederich, P.; Tsotsalas, M. MOF Synthesis Prediction Enabled by Automatic Data Mining and Machine Learning**. Angewandte Chemie International Edition 2022, 61 (19), e202200242.