Wang Chengshi, Kim Yeon-Ju, Vriza Aikaterini, Batra Rohit, Baskaran Arun, Shan Naisong, Li Nan, Darancet Pierre, Ward Logan, Liu Yuzi, Chan Maria K Y, Sankaranarayanan Subramanian K R S, Fry H Christopher, Miller C Suzanne, Chan Henry, Xu Jie
Nanoscience and Technology Division, Argonne National Laboratory, Lemont, IL, USA.
Department of Metallurgical and Materials Engineering, Indian Institute of Technology Madras, Chennai, India.
Nat Commun. 2025 Feb 17;16(1):1498. doi: 10.1038/s41467-024-55655-3.
The manipulation of electronic polymers' solid-state properties through processing is crucial in electronics and energy research. Yet, efficiently processing electronic polymer solutions into thin films with specific properties remains a formidable challenge. We introduce Polybot, an artificial intelligence (AI) driven automated material laboratory designed to autonomously explore processing pathways for achieving high-conductivity, low-defect electronic polymers films. Leveraging importance-guided Bayesian optimization, Polybot efficiently navigates a complex 7-dimensional processing space. In particular, the automated workflow and algorithms effectively explore the search space, mitigate biases, employ statistical methods to ensure data repeatability, and concurrently optimize multiple objectives with precision. The experimental campaign yields scale-up fabrication recipes, producing transparent conductive thin films with averaged conductivity exceeding 4500 S/cm. Feature importance analysis and morphological characterizations reveal key design factors. This work signifies a significant step towards transforming the manufacturing of electronic polymers, highlighting the potential of AI-driven automation in material science.
通过加工来操控电子聚合物的固态特性在电子学和能源研究中至关重要。然而，将电子聚合物溶液高效加工成具有特定性能的薄膜仍然是一项艰巨的挑战。我们推出了Polybot，这是一个由人工智能驱动的自动化材料实验室，旨在自主探索实现高导电性、低缺陷电子聚合物薄膜的加工途径。利用重要性引导的贝叶斯优化，Polybot能在复杂的七维加工空间中高效导航。特别是，自动化工作流程和算法能有效探索搜索空间、减轻偏差、采用统计方法确保数据可重复性，并精确地同时优化多个目标。实验活动得出了放大制造配方，生产出平均电导率超过4500 S/cm的透明导电薄膜。特征重要性分析和形态表征揭示了关键设计因素。这项工作标志着电子聚合物制造转型迈出了重要一步，凸显了人工智能驱动的自动化在材料科学中的潜力。
