改进遗传算法在CMP终点检测的应用

提出了一种基于反射法的化学机械抛光(CMP)在线终点判定方法。首先通过改进的单纯形混合遗传算法(GA),对SiO₂薄膜的膜厚与折射率进行了拟合,再以200～1200nm波段内反射率平均值作为特征值,得到了不同膜厚条件下的反射率平均值参照表。以实时反射率与预设膜厚反射率的差方和、特征值大小及其变化趋势这2个条件作为CMP终点判定条件。实验结果表明:混合算法拟合膜厚平均相对误差小于0.21%,折射率平均相对误差小于1.07%;特征值计算速度快,满足在线动态抛光的时间要求。

Application of Improved Genetic Algorithm in CMP Endpoint Detection

In order to improve the accuracy and efficiency of chemical mechanical polishing (CMP) endpoint detection, a method based on reflection method for CMP online endpoint determination was proposed. Firstly, the film thickness and refractive index of SiO₂ film were fitted by the improved simplex hybrid genetic algorithm (GA). Secondly, the average reflectance in the range of 200～1200nm was used as the eigenvalue, and a reference table of reflectance average values under different film thickness conditions was obtained. The difference between the real-time reflectance and the preset film thickness reflectance, and the eigenvalue with its changing trend were used as the CMP endpoint detection conditions. The experimental results demonstrate that the average relative error of the hybrid algorithm is less than 0.21%, the average relative error of the refractive index is less than 1.07%, and the eigenvalue calculation speed is fast, which meets the time requirement of online dynamic polishing.