Machine learning assisted designing of phenothiazine and phenoxazine-based dyes. Database generation and property prediction

The development of efficient dyes for organic electronic devices, requires the careful design of molecular structures with optimal electronic and photophysical properties. In this study, we employ machine learning (ML) to assist in the design and optimization of phenothiazine and phenoxazine-based dyes, which are promising candidates for these applications due to their tunable electronic properties. 5k phenothiazine and phenoxazine-based dyes are designed and ML models are used to predict their absorption maxima values. 30 promising candidates with red-shifted are selected. All the selected dyes have phenothiazine group. The results highlight the potential of ML-assisted design to accelerate the discovery of high-performance dyes for use in next-generation optoelectronic devices. Our findings provide a roadmap for future efforts in the design of organic dyes, with potential applications.