A chemical autonomous robotic platform for end-to-end synthesis of nanoparticles

Abstract Traditional nanomaterial development faces inefficiency and unstable results due to labor-intensive trial-and-error methods. To overcome these challenges, we developed a data-driven automated platform integrating artificial intelligence (AI) decision modules with automated experiments. Spec...

Full description

Saved in:
Bibliographic Details
Main Authors: Fan Gao, Hongqiang Li, Zhilong Chen, Yunai Yi, Shihao Nie, Zihao Cheng, Zeming Liu, Yuanfang Guo, Shumin Liu, Qizhen Qin, Zhengjian Li, Lisong Zhang, Han Hu, Cunjin Li, Liang Yang, Yunhong Wang, Guangxu Chen
Format: Article
Language:English
Published: Nature Portfolio 2025-08-01
Series:Nature Communications
Online Access:https://doi.org/10.1038/s41467-025-62994-2
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Abstract Traditional nanomaterial development faces inefficiency and unstable results due to labor-intensive trial-and-error methods. To overcome these challenges, we developed a data-driven automated platform integrating artificial intelligence (AI) decision modules with automated experiments. Specifically, the platform employs a Generative Pre-trained Transformer (GPT) model to retrieve methods/parameters and implements an A* algorithm centered closed-loop optimization process. It achieves optimized diverse nanomaterials (Au, Ag, Cu2O, PdCu) with controlled types, morphologies, and sizes, demonstrating efficiency and repeatability. Using the A* algorithm, we comprehensively optimized synthesis parameters for multi-target Au nanorods (Au NRs) with longitudinal surface plasmon resonance (LSPR) peak under 600-900 nm across 735 experiments, and for Au nanospheres (Au NSs)/Ag nanocubes (Ag NCs) in 50 experiments. Reproducibility tests showed deviations in characteristic LSPR peak and full width at half maxima (FWHM) of Au NRs under identical parameters were ≤1.1 nm and ≤ 2.9 nm, respectively. Researchers only need initial script editing and parameter input, significantly reducing human resource requirements. Comparative analysis confirms the A* algorithm outperforms Optuna and Olympus in search efficiency, requiring significantly fewer iterations.
ISSN:2041-1723