Omidvar Mojan, Zhang Hangfeng, Ihalage Achintha Avin, Saunders Theo Graves, Giddens Henry, Forrester Michael, Haq Sajad, Hao Yang
School of Electronic Engineering and Computer Science, Queen Mary University of London, London, UK.
QinetiQ, Cody Technology Park, Farnborough, Hampshire, UK.
Nat Commun. 2024 Aug 2;15(1):6554. doi: 10.1038/s41467-024-50884-y.
Accelerating perovskite solid solution discovery and sustainable synthesis is crucial for addressing challenges in wireless communication and biosensors. However, the vast array of chemical compositions and their dependence on factors such as crystal structure, and sintering temperature require time-consuming manual processes. To overcome these constraints, we introduce an automated materials discovery approach encompassing machine learning (ML) assisted material screening, robotic synthesis, and high-throughput characterization. Our proposed platform for rapid sintering and dielectric analysis streamlines the characterization of perovskites and the discovery of disordered materials. The setup has been successfully validated, demonstrating processing materials within minutes, in stark contrast to conventional procedures that can take hours or days. Following setup validation with established samples, we showcase synthesizing single-phase solid solutions within the barium family, such as (BaSr)CeO, identified through ML-guided chemistry.
加速钙钛矿固溶体的发现和可持续合成对于应对无线通信和生物传感器领域的挑战至关重要。然而，大量的化学成分及其对晶体结构和烧结温度等因素的依赖性需要耗时的人工过程。为了克服这些限制，我们引入了一种自动化材料发现方法，该方法包括机器学习（ML）辅助材料筛选、机器人合成和高通量表征。我们提出的快速烧结和介电分析平台简化了钙钛矿的表征和无序材料的发现。该装置已成功验证，与可能需要数小时或数天的传统程序形成鲜明对比，能够在几分钟内处理材料。在用既定样品进行装置验证之后，我们展示了在钡族中合成单相固溶体，例如通过ML引导化学鉴定的（BaSr）CeO。
