Deep Learning Models for Colloidal Nanocrystal Synthesis

作者：Kai Gu, Yingping Liang, Jiaming Su, Peihan Sun, Peng Jia, Naihua Miao, Zhimei Sun, Ying Fu, Haizheng Zhong, Jun Zhang · 发表于：ACS Nano · 年份：2025 · DOI：10.1021/acsnano.5c09134 · 被引用次数：6 · 研究领域：Machine Learning in Materials Science、Quantum Dots Synthesis And Properties、Nanocluster Synthesis and Applications

Colloidal synthesis of nanocrystals usually includes complex chemical reactions and multistep synthesis processes. Despite the great success in the past 30 years, it remains challenging to clarify the correlations between the synthetic parameters of the chemical reaction and the physical properties of nanocrystals. Here, we developed a deep learning-based nanocrystal synthesis model that correlates synthetic parameters with the final size and shape of target nanocrystals, using a data set of 3508 recipes covering 348 distinct nanocrystal compositions. The size and shape labels were obtained from transmission electron microscope images using a segmentation model trained with a semi-supervised algorithm on a data set comprising around 1.2 million nanocrystals. By applying the reaction intermediate-based data augmentation method and elaborated descriptors, the synthesis model was able to predict the nanocrystal's size with a mean absolute error of 1.39 nm, while reaching an 89% average accuracy for shape classification. The synthesis model shows knowledge transfer capabilities across different nanocrystals with the input of new recipes. With that, the influence of chemicals on the final size of nanocrystals was further evaluated, revealing the descending order of importance of the nanocrystal composition, precursor or ligand, and solvent. Overall, the deep learning-based nanocrystal synthesis model offers a powerful tool to expedite the development of high-quality nanocrystals.