Collaborative Research: Data-Driven Metrology and Inspection Technology for Semiconductor Wafer-Level Manufacturing

合作研究：用于半导体晶圆级制造的数据驱动计量和检测技术

• 批准号: 2125826
• 负责人: Martin Jun
• 金额: $ 30万
• 依托单位: Purdue University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2021
• 资助国家: 美国
• 起止时间: 2021-11-01 至 2024-10-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2125826&HistoricalAwards=false
• 关键词: Collaborative; Research; Data; Driven; Metrology

This grant supports research advancing wafer-level semiconductor manufacturing and inspection technology, establishing the data and technical architecture needed to ensure sustainable solutions and scaling digital innovation across the wafer metrology and inspection processes. This research will generate new knowledge and principles used in the wafer/thin-film inspection, metrology, design and manufacturing needed in the electronics industry. Modeling methodologies are created for the inspection capability of various defect types at wafer scale. Semiconductor metrology and inspection tools are presently stand-alone machines operated independently and there is an increasing need for creating an automated and integrated metrology and inspection across semiconductor manufacturing processes. This project can accelerate the semiconductor industry’s digital transformation through hardware and software integration, connectivity, intelligence, visualization, and flexible automation. An integrated and intelligent framework for semiconductor wafer/thin-film metrology and inspection technologies is developed to monitor, diagnose and control the quality of wafer-level defects, by using super-resolution 3D imaging process, as well as thin-film material properties. This grant supports the semiconductor manufacturing workforce development, providing research and education opportunities for undergraduate and graduate students including underrepresented groups to gain knowledge and hands-on experience in semiconductor technology. The semiconductor process automation and digitalization based on strobo-spectroscopy and dexel-based deep learning algorithms provide for a wafer/thin-film inspection and metrology capability to detect the wafer-level or packaging-level anomalies. A strobo-spectroscopy capability combined with a spectral imaging technology allows for the synchronized spectroscopic analysis and high-speed imaging capturing of both the spectral response and spatial images as the probe scans the wafer surface. The combined spectral response and camera images are converted to 3D data representations to train dexel-based deep learning algorithms and predict wafer grade, defect type, and defect locations. The dexel-based approach to 3D wafer topography data through 3D correlation Neural Network (CNN) and Recurrent Neural Network (RNN) architectures is established to improve computational speed and prediction accuracy. By combining strobo-spectroscopy and deep learning algorithms, this research will fill a critical knowledge gap in automated inspection technology and in the fundamental identification of the wafer and thin-film abnormalities and variation in material properties.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.

这笔赠款支持研究推进晶圆级半导体制造和检测技术，建立确保可持续解决方案所需的数据和技术架构，并在晶圆计量和检测过程中扩展数字创新。这项研究将产生电子行业所需的晶圆/薄膜检测、计量、设计和制造中使用的新知识和原理。建模方法是为晶圆级各种缺陷类型的检测能力而创建的。半导体计量和检测工具目前是独立运行的独立机器，并且越来越需要在半导体制造过程中创建自动化和集成的计量和检测。该项目可以通过软硬件集成、连接性、智能化、可视化和灵活自动化来加速半导体行业的数字化转型。开发了半导体晶圆/薄膜计量和检测技术的集成智能框架，通过使用超分辨率3D成像工艺以及薄膜材料特性来监测、诊断和控制晶圆级缺陷的质量。这笔赠款支持半导体制造劳动力的发展，为本科生和研究生（包括代表性不足的群体）提供研究和教育机会，以获得半导体技术的知识和实践经验。 The semiconductor process automation and digitalization based on strobo-spectroscopy and dexel-based deep learning algorithms provide for a wafer/thin-film inspection and metrology capability to detect the wafer-level or packaging-level anomalies.频闪光谱功能与光谱成像技术相结合，可以在探针扫描晶圆表面时同步光谱分析和高速成像捕获光谱响应和空间图像。组合的光谱响应和相机图像被转换为​​ 3D 数据表示，以训练基于 dexel 的深度学习算法并预测晶圆等级、缺陷类型和缺陷位置。通过 3D 相关神经网络 (CNN) 和递归神经网络 (RNN) 架构建立基于 dexel 的 3D 晶圆形貌数据方法，以提高计算速度和预测精度。通过结合频闪光谱和深度学习算法，这项研究将填补自动检测技术以及晶圆和薄膜异常和材料特性变化的基本识别方面的关键知识空白。该奖项反映了 NSF 的法定使命，并通过使用基金会的智力优点和更广泛的影响审查标准进行评估，被认为值得支持。