Scholarlink

<p>In this study, various machine learning (<i>ML</i>) algorithms have been applied to predict the melting temperature (<i>T</i><sub><i>m</i></sub>) of organic semiconductors. The study reveals that <i>fr_allylic_oxide</i> and <i>VSA-Estate10</i> exhibit a high Pearson correlation with <i>T</i><sub><i>m</i></sub>. among the trained <i>ML</i> models, Random Forest (<i>RF</i>) and Gradient Boosting (<i>GB</i>) achieve a good R-Squared (<i>R</i><sup><i>2</i></sup>) values of 0.89 and 0.78 with <i>RMSE</i> values of 6.37 and 14.61, respectively. Further analysis using <i>SHAP</i> values reveals that <i>VSA_Estate2</i> and <i>MolLogP</i> are the most impactful features on model performance. Additionally, it is found that the synthetic accessibility (<i>SA</i>) of the semiconductors is up to 0.25 to highlight the need for balance between crystallinity and noncrystallinity. The t-Distributed Stochastic Neighbor Embedding (<i>t-SNE</i>) map of the dataset showed equal components to indicate a complex interplay between molecular features. The current study demonstrates the potential of <i>ML</i> in predicting <i>T</i><sub><i>m</i></sub> values and provides insights into the relationships between molecular structure and thermal properties of organic semiconductors. These findings have significant implications for the design and synthesis of organic electronic materials with tailored thermal properties.</p> Graphical abstract <p>This study employs machine learning to predict the glass transition temperature (Tm) of 971 organic chromophores. Key features, including fr_allylic_oxide and VSA-Estate10, are identified as significant predictors. Our findings reveal a delicate balance between crystallinity and noncrystallinity, providing insights for designing organic semiconductors with tailored thermal properties.</p> <p></p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A machine learning study to predict long range order/disoroder of organic semiconductors: a study on maintaining delicate crystallinity balance

  • 期刊论文

  • 发表日期：

    2025年11月3日

查看全文
  • Hussein A. K. Kyhoiesh,
  • Hassan E. Abd Elsalam,
  • Azal S. Waheeb,
  • Khalid J. Al-Adilee,
  • Munthir A. Abdulhussain,
  • Dakhil N. Taha,
  • Ahmed A. Al-Kubaisi,
  • Islam H. El Azab

摘要

In this study, various machine learning (ML) algorithms have been applied to predict the melting temperature (Tm) of organic semiconductors. The study reveals that fr_allylic_oxide and VSA-Estate10 exhibit a high Pearson correlation with Tm. among the trained ML models, Random Forest (RF) and Gradient Boosting (GB) achieve a good R-Squared (R2) values of 0.89 and 0.78 with RMSE values of 6.37 and 14.61, respectively. Further analysis using SHAP values reveals that VSA_Estate2 and MolLogP are the most impactful features on model performance. Additionally, it is found that the synthetic accessibility (SA) of the semiconductors is up to 0.25 to highlight the need for balance between crystallinity and noncrystallinity. The t-Distributed Stochastic Neighbor Embedding (t-SNE) map of the dataset showed equal components to indicate a complex interplay between molecular features. The current study demonstrates the potential of ML in predicting Tm values and provides insights into the relationships between molecular structure and thermal properties of organic semiconductors. These findings have significant implications for the design and synthesis of organic electronic materials with tailored thermal properties.

Graphical abstract

This study employs machine learning to predict the glass transition temperature (Tm) of 971 organic chromophores. Key features, including fr_allylic_oxide and VSA-Estate10, are identified as significant predictors. Our findings reveal a delicate balance between crystallinity and noncrystallinity, providing insights for designing organic semiconductors with tailored thermal properties.