时域有限差分法
波长
光学
紫外线
公制(单位)
计算
极端紫外线
材料科学
光电子学
堆栈(抽象数据类型)
计算机科学
物理
激光器
算法
工程类
运营管理
程序设计语言
作者
Bryan M. Barnes,Hui Zhou,Mark-Alexander Henn,Martin Y. Sohn,Richard T. Silver
出处
期刊:Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series
日期:2017-06-26
被引量:2
摘要
The sizes of non-negligible defects in the patterning of a semiconductor device continue to decrease as the dimensions for
these devices are reduced. These “killer defects” disrupt the performance of the device and must be adequately controlled
during manufacturing, and new solutions are required to improve optics-based defect inspection. To this end, our group
has reported [Barnes et al., Proc. SPIE 1014516 (2017)] our initial five-wavelength simulation study, evaluating the
extensibility of defect inspection by reducing the inspection wavelength from a deep-ultraviolet wavelength to wavelengths
in the vacuum ultraviolet and the extreme ultraviolet. In that study, a 47 nm wavelength yielded enhancements in the
signal to noise (SNR) by a factor of five compared to longer wavelengths and in the differential intensities by as much as
three orders-of-magnitude compared to 13 nm. This paper briefly reviews these recent findings and investigates the
possible sources for these disparities between results at 13 nm and 47 nm wavelengths. Our in-house finite-difference
time-domain code (FDTD) is tested in both two and three dimensions to determine how computational conditions
contributed to the results. A modified geometry and materials stack is presented that offers a second viewpoint of defect
detectability as functions of wavelength, polarization, and defect type. Reapplication of the initial SNR-based defect
metric again yields no detection of a defect at λ = 13 nm, but additional image preprocessing now enables the computation
of the SNR for λ = 13 nm simulated images and has led to a revised defect metric that allows comparisons at all five
wavelengths.
科研通智能强力驱动
Strongly Powered by AbleSci AI