Abstract In the polysilicon chemical vapor deposition reactor, the operating parameters are complex to affect the polysilicon’s output. Therefore, it is very important to address the coupling problem of multiple parameters and solve the optimization in a computationally efficient manner. Here, we adopted Response Surface Methodology (RSM) to analyze the complex coupling effects of different operating parameters on silicon deposition rate (R) and further achieve effective optimization of the silicon CVD system. Based on finite numerical experiments, an accurate RSM regression model is obtained and applied to predict the R with different operating parameters, including temperature (T), pressure (P), inlet velocity (V), and inlet mole fraction of H2 (M). The analysis of variance is conducted to describe the rationality of regression model and examine the statistical significance of each factor. Consequently, the optimum combination of operating parameters for the silicon CVD reactor is: T = 1400 K, P = 3.82 atm, V = 3.41 m/s, M = 0.91. The validation tests and optimum solution show that the results are in good agreement with those from CFD model and the deviations of the predicted values are less than 4.19%. This work provides a theoretical guidance to operate the polysilicon CVD process.

中文翻译：

---

响应面法优化多晶硅化学气相沉积反应器的操作参数

摘要 在多晶硅化学气相沉积反应器中，操作参数复杂，影响多晶硅产量。因此，解决多个参数的耦合问题并以计算效率高的方式解决优化问题非常重要。在这里，我们采用响应面方法 (RSM) 来分析不同操作参数对硅沉积速率 (R) 的复杂耦合效应，进一步实现硅 CVD 系统的有效优化。基于有限的数值实验，获得了准确的 RSM 回归模型，并应用于预测不同运行参数下的 R，包括温度 (T)、压力 (P)、入口速度 (V) 和入口 H2 摩尔分数 (M) . 方差分析是为了描述回归模型的合理性，检验各因素的统计显着性。因此，硅 CVD 反应器的最佳操作参数组合为：T = 1400 K，P = 3.82 atm，V = 3.41 m/s，M = 0.91。验证试验和最优解表明，结果与CFD模型的结果吻合良好，预测值偏差小于4.19%。这项工作为操作多晶硅 CVD 工艺提供了理论指导。验证试验和最优解表明，结果与CFD模型的结果吻合良好，预测值偏差小于4.19%。这项工作为操作多晶硅 CVD 工艺提供了理论指导。验证试验和最优解表明结果与CFD模型的结果吻合良好，预测值偏差小于4.19%。这项工作为操作多晶硅 CVD 工艺提供了理论指导。