基于多波长迭代算法的极紫外光刻掩模检测关键技术研究

发布时间：2018-05-20 19:13

本文选题：缺陷检测 + 相位恢复　；参考：《南京理工大学》2017年硕士论文

【摘要】：极紫外(EUV)(λ=13.5nm)光刻技术是世界范围内最被看好的实现22nm最小刻线的下一代光刻技术。其光刻掩膜版在制作过程中参入纳米量级的灰尘污染、杂质颗粒,引入的鼓包、凹坑出现在基底表面或埋藏在多层膜内部或掩模表面,都会造成光刻样品出现严重的缺陷。光学非干涉相位恢复检测法完全基于光强信息测量,系统光路结构简单,降低仪器分辨率要求而不丢失相关信息,是实现极紫外掩膜缺陷检测的重要途径。本文在现有非干涉相位恢复理论的基础上,针对缺陷相位信息的恢复,展开基于迭代算法缺陷检测方法研究。在传统多波长迭代算法的基础上,引入角谱传输理论和梯度加速函数,提出一种快速收敛的相位恢复迭代算法——多波长梯度加速迭代算法,并采用matlab编写了算法程序。基于所提多波长梯度加速迭代算法进行仿真模拟,分别对相位分布在[-0.84,0]的高斯型相位面以及[-π,0]的"lena"相位面进行恢复,表征复原精度的相对均方根值均达到10-3数量级,验证了该方法的可行性。并与传统多波长迭代恢复算法进行对比验证,该加速算法相较于传统算法在收敛速度上提高了 2倍以上且具有更高的复原精度。最后对算法中波长数目、波段范围、传输距离、恢复相位区间参数对算法性能影响进行讨论。搭建了多波长梯度加速迭代算法实验装置,采用带通滤光片与白光光源形成七种不同波长入射光。对入射光输入面波前相位面信息进行恢复,并将输出面光强信息的算法计算值与测量值进行比较得到振幅误差E=0.078%,说明该算法的正确性;分别对分辨率板一定宽度的三条对线、字母e装片两种透射性元件的相位面进行恢复计算宽度、高度信息,得到的实验结果符合预期。
[Abstract]:Ultraviolet (位 ~ (13. 5) nm) photolithography is the most promising next generation lithography technology to realize the minimum line of 22nm in the world. In the process of making the lithography mask plate, dust pollution of nanometer magnitude, impurity particles, incoming bulge, pits appearing on the substrate surface or burying inside the multilayer film or mask surface will cause serious defects in the lithography sample. The optical non-interference phase recovery detection method is based on the measurement of light intensity information. The optical system is simple in structure and reduces the resolution requirement of the instrument without losing relevant information. It is an important way to realize the detection of the defect in the extreme ultraviolet mask. In this paper, based on the existing non-interference phase recovery theory and the restoration of defect phase information, the research of defect detection method based on iterative algorithm is carried out. Based on the traditional multi-wavelength iterative algorithm, the angle spectrum transmission theory and gradient acceleration function are introduced, and a fast convergent phase recovery iterative algorithm named multi-wavelength gradient accelerated iterative algorithm is proposed. The algorithm is programmed by matlab. Based on the proposed iterative algorithm of multi-wavelength gradient acceleration, the Gao Si type phase surface with phase distribution in [-0.84 0] and the "lena" phase surface with [- 蟺 0] are respectively restored. The relative root mean square (RMS) values of the restoration accuracy are in the order of 10 ~ (-3). The feasibility of the method is verified. Compared with the traditional multi-wavelength iterative recovery algorithm, the proposed algorithm improves the convergence speed by more than 2 times and has a higher recovery accuracy than that of the traditional algorithm. Finally, the effects of wavelength number, band range, transmission distance and recovery phase range on the performance of the algorithm are discussed. A multi-wavelength gradient accelerated iterative algorithm was set up, and seven different wavelengths of incident light were formed by bandpass filter and white light source. The phase surface information of the input surface wave front of the incident light is restored, and the calculated value of the light intensity information of the output surface is compared with the measured value, and the amplitude error E _ (0.078) is obtained, which shows the correctness of the algorithm. The phase surface of the two transmissible elements is restored to calculate the width and height, and the experimental results are in line with the expectation.
【学位授予单位】：南京理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN23

【相似文献】