Abstract
One of the challenges in materials informatics is establishing a way to describe molecular functions governed by multi-electronic states. Here, we show a study of advanced spectral optimization including beyond 20 electronic states toward ordinally uncolored organic electrochromic material design, where the color (magenta) is one of the three primary colors that had never been achieved due to the advanced spectroscopic requirements in a redox process. Using qualitatively accurate and computationally cheap semiempirical molecular orbital descriptors, desired structures were efficiently narrowed from 1.2 million triphenylamine derivatives and realized by subsequent organic syntheses with spectroelectrochemical experiments. The universality of the protocol would allow the simultaneous optimization of multiple functions in organic electronics materials.
Supplementary materials
Title
Machine Learning Details and Experimental Details
Description
1. Machine Learning Details
1.1. Descriptors
1.2. Labels
1.3. Model
1.4. Molecular Library
1.5. Learning Curve
2. Experimental Details
2.1. Nuclear Magnetic Resonance (NMR)
2.2. Mass Spectra (MS)
2.3. Fourier Transform Infrared Spectra (FT-IR)
2.4. UV-Vis Absorption Spectra
2.5. Electrochemical Spectroscopy
2.6. X-ray Crystallography
2.7. Synthesis and Characterization
Supporting References
Actions