News from Nanotechjapan


Experimental Verification for the Concept of Organic Molecule Design by AI ---Opening a Road to Design Functional Molecules for Organic Electronics---

 On August 24, 2018, Riken. National Institute of Materials Science (NIMS), the University of Tokyo, and Japan Science and Technology Agency (JST) announced that a collaborative research group from Riken, NIMS, and the University succeeded in designing synthesizable organic molecules with desired properties by Artificial Intelligence (AI). Details were published in ACS Central Science with Masato Sumita at Riken as lead author*.

 Computer assist is attracting attention as a promising tool in designing desired organic molecules. Most of the previous methods, however, require chemical rules comprehensively installed in the computer system, which is almost impossible to fulfil completely. Artificial intelligence is able to make the computer to learn required chemical rules by "deep learning". The research group has coupled a machine-learning-based molecule generator with density functional theory (DFT) calculations for AI-assisted molecular design.

 The AI system was trained with 13,000molecules to learn chemical rules, and a 10 day run on the 12-core (Intel Xeon E5-2689v3 CPU) server discovered 86 potential photo-functional molecules around target lowest excitation levels at the wavelength 200, 300, 400, 500, and 600nm. Among them, six molecules were synthesized, and five of them were experimentally confirmed to have the desired absorption in the ultraviolet-visible (UV-vis) spectroscopy. For instance, a molecule composed of a benzene ring with  -OH and -N-N=O branches showed UV absorption peaks at 350nm and 400nm as expected from DFT calculations.

*Masato Sumita, Xiufeng Yang, Shinsuke Ishihara, Ryo Tamura, and Koji Tsuda, "Hunting for organic molecules with artificial intelligence: Molecules optimized for desired excitation energies", ACS Central Science, Article ASAP, DOI: 10.1021/acscentsci.8b00213; Publication Date (Web): August 20, 2018