ERI: A Machine Learning Framework for Preventing Cracking in Semiconductor Materials

半导体材料的性能和质量对广泛应用的先进技术至关重要。生产这些材料的一个重大挑战是冷却过程。在生产阶段，半导体材料在冷却时容易开裂。这些裂缝会导致最终产品失效，降低可靠性，提高制造成本。该工程研究启动（ERI）奖支持旨在防止半导体冷却过程中形成裂纹的基础研究。该项目的目标是开发一种将机器学习技术与力学基本原理相结合的新方法来预测裂纹的形成。这项研究将促进高质量半导体材料的生产。该项目也将为STEM教育领域做出重大贡献。将建立一个广泛访问的虚拟机械测试实验室，它将使用交互式虚拟工具来教育学生有关测试材料。在这项研究中，还将做出特别努力，让那些在STEM领域历史上代表性不足的学生参与进来。该项目的目标是开发一种基于力学的机器学习框架，以预测和量化半导体材料中的界面开裂，特别是在冷却过程中碳化硅/氮化铝（SiC/AlN）界面。认识到界面缺陷和残余应力是开裂的关键因素，研究目的是利用先进的机器学习和模拟技术来识别开裂机制并主动预防它。机器学习模型将使用断裂行为的原子模拟进行训练，为半导体材料的设计提供创新的见解。这项研究的潜在贡献是很多的，不仅旨在减轻半导体界面的损伤，从而彻底改变它们的设计和生产，而且还开发了一个具有不确定性量化的集成机器学习框架，这将在预测其他材料的行为和性质方面具有更广泛的适用性。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。

The performance and quality of semiconductor materials are critical to advanced technologies for a wide range of applications. A significant challenge in the production of these materials is the cooling process. During the production phase, semiconductor materials are prone to cracking as they cool. These cracks can lead to failures in the final products, decreased reliability, and higher manufacturing costs. This Engineering Research Initiation (ERI) award supports fundamental research aiming to prevent the formation of cracks during the semiconductor cooling process. The objective of this project is to develop a novel method that integrates machine learning techniques with fundamental principles of mechanics to predict crack formation. This research will enhance production of high-quality semiconductor materials. This project will also make significant contributions to the field of STEM education. A widely accessible Virtual Mechanical Testing Lab will be established, which will use interactive virtual tools to educate students about testing materials. Special efforts will also be made to engage students who have historically been underrepresented in STEM fields in this research.The goal of this project is to develop a mechanics-informed machine learning framework to predict and quantify interfacial cracking in semiconductor materials, specifically at silicon carbide/aluminum nitride (SiC/AlN) interfaces during the cooling process. Recognizing that interfacial defects and residual stresses are critical factors in cracking, the research aim is to use advanced machine learning and simulation techniques to identify the mechanisms of cracking and proactively prevent it. The machine learning model will be trained using atomistic simulations of cracking behaviors, providing innovative insights into the design of semiconductor materials. The potential contributions of this research are numerous, aiming not only to mitigate damage in semiconductor interfaces, thereby revolutionizing their design and production, but also to develop an integrated machine learning framework with uncertainty quantification, which will have broader applicability in predicting behaviors and properties of other materials.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.