Materials Research: Artificial Intelligence Accelerates Syntheses

To automatically predict the synthesis of MOFs (pictured is a so-called SURMOF structure with an embedded molecule), the researchers have created a web tool. Copright: KIT / Andrea Fabry

KIT Researchers from the Helmholtz Research Field Information Demonstrate the Application of Machine Learning in the Development of Metal-Organic Framework Compounds. (Source: Karlsruhe Institute of Technology – Press Releases)

Energy and environmental protection, medicine, information and communication: these and many other areas depend on innovative materials. Data-based synthesis strategies can significantly accelerate the development of novel materials and improve their properties. Researchers at the Karlsruhe Institute of Technology (KIT) from Program 3: “Materials System Engineering” of the Research Field Information have used artificial intelligence to identify synthesis strategies for previously unknown metal-organic frameworks (MOFs). These highly porous crystalline materials can be tailored for a wide variety of applications such as material  separation, gas storage, catalysis, and sensor technology.

World’s first MOF synthesis database

In the journal Angewandte Chemie, the researchers now report how machine learning (ML) can be used to streamline MOF development. “In this process, the synthesis conditions of a MOF are predicted directly based on the crystal structure,” explains Manuel Tsotsalas from KIT’s Institute of Functional Interfaces and member of Topic 3: “Adaptive and Bioinstructive Materials Systems” in Program 3 of the Research Field Information. The study was conducted together with KIT’s Institute for Theoretical Computer Science. The data-driven prediction is possible thanks to the world’s first MOF synthesis database. To create it, the required parameters were extracted from the literature using natural language processing algorithms. The trained and optimized ML algorithms based on the database clearly outperformed the prediction performance of human experts even in the initial phase.

The original press release can be found at:

Materialforschung: Künstliche Intelligenz beschleunigt Synthesen (only in german)

The original publication can be found at:

Yi Luo, Saientan Bag, Orysia Zaremba, Adrian Cierpka, Jacopo Andreo, Stefan Wuttke, Pascal Friederich, Manuel Tsotsalas: MOF Synthesis Prediction Enabled by Automatic Data Mining and Machine Learning. Angew. Chem. Int. Ed., 2022. DOI: https://doi.org/10.1002/anie.202200242

Localization in the Helmholtz Research Field Information:

Helmholtz Research Field Information, Program 3: Materials Systems Engineering, Topic 3: Adaptive and Bioinstructive Materials Systems

Contact:

Dr. Manuel Tsotsalas
Group leader SURCOFs und SURGELs at the Institute of Functional Interfaces (IFG)
Karlsruher Institute for Technology (KIT)
Phone: +49-721-608-2-8107
E-Mail: manuel.tsotsalas@kit.edu