BRITE Pivot: Machine Learning Enabled Rapid and Robust Three-Dimensional Nanomanufacturing

三维（3D）打印是近年来最重要的制造技术发展之一，其应用范围从原型设计和产品可视化到构建功能材料和器件。纳米级3D打印已被用于生产各种复杂的三维纳米结构和纳米器件，具有前所未有的性能和功能。然而，目前的3D纳米打印方法，如逐点激光打印，速度慢，质量多变，可重复性差，是它们进入商业规模制造的障碍。机器学习（ML）和人工智能（AI）非常适合改善和确保打印结构的质量。这项促进工程变革和公平进步的研究理念（BRITE）支点奖开发了人工智能引导、机器学习支持的3D纳米打印方法，以提高打印结构和设备的速度、规模、打印质量和稳健性。这些用于3D纳米制造的人工智能和机器学习工具实现了传感器和可穿戴设备等新应用，使多个经济部门受益，并有助于美国在先进纳米制造方面的竞争力和全球领导地位。该项目通过从代表性不足的群体中招募学生，并让他们参与人工智能、机器学习和3D纳米打印技术的研究和教育，从而在先进制造业中培养多元化的劳动力，从而促进多样性、公平性和包容性。PI在基于激光的纳米制造和器件制造方面拥有丰富的经验和专业知识。最近，首席研究员（PI）小组的努力导致了一种快速、连续、逐层的3D纳米打印技术的发展。该项目允许PI通过深入研究和应用这些方法来开发快速和强大的3D纳米打印技术，从而获得ML和AI方法方面的新专业知识。这包括对各种ML和AI方法的研究和评估，以及快速3D纳米打印的适当方法的选择和实施。该研究涉及开发将基于物理的模型与先进的ML算法（如自适应学习和迁移学习）相结合的方法。基于飞秒激光的3D纳米打印工艺需要了解控制激光束特性的基本原理，例如光束大小和扫描速率，其在光学系统中的传播以及激光在光聚合物中的诱导反应。基于物理的模型和激光光聚合实验为过程控制和优化的ML和AI分析工具的开发提供了关键数据和指导。此外，该项目研究了使用相对较少的实验数据开发ML和AI工具的方法。利用各种打印结构的结果，开发了ML和AL算法来指导高精度打印任意自定义结构的过程。该项目推进了人工智能引导、机器学习支持的飞秒激光光聚合技术，为快速、稳健的3D纳米制造提供了平台。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

Three-dimensional (3D) printing is one of the most important manufacturing technology developments in recent years for applications ranging from prototyping and product visualization to building functional materials and devices. Nanoscale 3D printing has been used to produce a wide range of complex 3D nanostructures and nanodevices with unprecedented properties and functionalities. However, the slow speed, variable quality and poor reproducibility of current 3D nanoprinting methods, such as point-by-point laser printing, are barriers for their adoption to commercial-scale manufacturing. Machine learning (ML) and artificial intelligence (AI) are ideally suited for improving and assuring the quality of the printed structures. This Boosting Research Ideas for Transformative and Equitable Advances in Engineering (BRITE) Pivot award develops AI-guided, ML-enabled 3D nanoprinting methods to improve the speed, scale, print quality and robustness of the printed structures and devices. These AI and ML tools for 3D nanomanufacturing enable new applications, such as sensors and wearables, that benefit several sectors of the economy and contribute to US competitiveness and global leadership in advanced nanomanufacturing. This project contributes to diversity, equity and inclusion by recruiting students from underrepresented groups and engaging them in research and education in AI, ML and 3D nanoprinting technologies, thus developing a diverse workforce in advanced manufacturing. The PI has extensive experience and expertise in laser-based nanomanufacturing and device manufacture. Recent efforts by the principal investigator (PI)’s group have resulted in the development of a rapid, continuous, layer-by-layer 3D nanoprinting technology. This project allows the PI to acquire new expertise in ML and AI methods through in-depth investigations and applications of these methods to develop rapid and robust 3D nanoprinting technologies. This includes studies and evaluations of various ML and AI methods and selection and implementation of appropriate methods for rapid 3D nanoprinting. The research involves developing methods that combine physics-based models with advanced ML algorithms such as adaptive learning and transfer learning. The femtosecond laser-based 3D nanoprinting process involves understanding the fundamentals of controlling the laser beam characteristics, such as beam size and scan rate, its propagation in the optical system and the laser-induced reactions in the photopolymers. The physics-based models and laser photo-polymerization experiments provide key data and guidance for the development of ML and AI analytic tools for process control and optimization. Furthermore, the project studies approaches to develop ML and AI tools using a relatively small amount of experimental data. Using results of various printed structures, ML and AL algorithms are developed to guide the process of printing arbitrary, user-defined structures with high precision. This project advances knowledge in AI-guided, ML-enabled femtosecond laser-based photo-polymerization as a platform for rapid and robust 3D nanomanufacturing.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.