Designing New Crystallizable Organic Semiconductors for Optoelectronic Applications: Insights From a Computational Study and Machine Learning

Organic semiconductors have revolutionized electronics, but their amorphous nature hinders performance and stability. Crystallinity overcomes these limitations; however, the design of materials that combine high crystallinity, optimal thermal properties, and ease of synthesis remains a significant challenge. A machine learning (ML)‐assisted approach combined with the density functional theory (DFT) study was employed to generate a vast chemical space of crystallizable organic semiconductors. By breaking retrosynthetically interesting chemical species (BRICS), over 1700 new organic semiconductors with promising synthetic accessibility (SA) were designed. ML algorithms, specifically extra trees (ET) and random forest, were used to predict the melting temperatures ( T m ) of these semiconductors, yielding good R ‐squared ( R 2 ) values of 0.94–0.96. Dimensionality reduction analysis reveals that the t‐distributed stochastic neighbor embedding (t‐SNE) components 1 and 2 of these semiconductors ranged within the same value (−5 to 5), indicating a high degree of similarity. Furthermore, analysis of SA showed that new organic semiconductors with SMILES lengths between 15 and 40 are more likely to be easily synthesized. Based on these findings, 20 new candidate semiconductors were identified for practical synthesis and analysis. DFT calculations were employed to study the optoelectronic properties of chromophores. This study demonstrates the power of ML‐assisted design in generating crystallizable organic semiconductors with enhanced SA. The findings are expected to contribute significantly to the development of high‐performance organic electronics.