当前位置: X-MOL 学术J. Cryst. Growth › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Optimization of the melt/crystal interface shape and oxygen concentration during the Czochralski silicon crystal growth process using an artificial neural network and a genetic algorithm
Journal of Crystal Growth ( IF 1.7 ) Pub Date : 2020-10-01 , DOI: 10.1016/j.jcrysgro.2020.125828
Xiaofang Qi , Wencheng Ma , Yifan Dang , Wenjia Su , Lijun Liu

Abstract The melt/crystal interface shape and oxygen concentration during the Czochralski silicon crystal growth process significantly influence the crystal quality. In this paper, an optimization system with the combination of an artificial neural network and a genetic algorithm is proposed to optimize the growth parameters during the growth process. Flattening the melt/crystal interface and reducing the oxygen concentration along the interface were chosen as the optimization targets. Two important growth parameters, the crystal rotation rate and crucible rotation rate, were chosen as optimization variables. First, a global heat and mass transfer model was developed to simulate the crystal growth process and then tested with experimental data. The verified heat and mass transfer model was then used to train an artificial neural network with the aim of rapidly assessing the complex nonlinear dependence of the interface shape and oxygen concentration on the growth parameters. The trained neural network combined with a genetic algorithm was then used to obtain the optimal growth parameters. Both deflection of the melt/crystal interface and the oxygen concentration along the interface decreased after optimization. Finally, the optimal growth parameters were checked in the heat and mass transfer model to evaluate the performance of the optimization system. The proposed method will also be useful for optimization of other crystal growth processes.

中文翻译:

使用人工神经网络和遗传算法优化直拉硅晶体生长过程中熔体/晶体界面形状和氧浓度

摘要 直拉硅晶体生长过程中熔体/晶体界面形状和氧浓度显着影响晶体质量。本文提出了一种人工神经网络和遗传算法相结合的优化系统来优化生长过程中的生长参数。选择平整熔体/晶体界面并降低界面上的氧浓度作为优化目标。两个重要的生长参数,晶体旋转速率和坩埚旋转速率,被选为优化变量。首先,开发了全局传热和传质模型来模拟晶体生长过程,然后用实验数据进行测试。然后使用经过验证的传热和传质模型来训练人工神经网络,目的是快速评估界面形状和氧浓度对生长参数的复杂非线性依赖性。训练好的神经网络与遗传算法相结合,然后用于获得最佳生长参数。优化后熔体/晶体界面的偏转和沿界面的氧浓度均降低。最后,在传热传质模型中检查最佳生长参数,以评估优化系统的性能。所提出的方法也可用于优化其他晶体生长过程。训练好的神经网络结合遗传算法用于获得最佳生长参数。优化后熔体/晶体界面的偏转和沿界面的氧浓度均降低。最后,在传热传质模型中检查最佳生长参数,以评估优化系统的性能。所提出的方法也可用于优化其他晶体生长过程。训练好的神经网络与遗传算法相结合,然后用于获得最佳生长参数。优化后熔体/晶体界面的偏转和沿界面的氧浓度均降低。最后,在传热传质模型中检查最佳生长参数,以评估优化系统的性能。所提出的方法也可用于优化其他晶体生长过程。
更新日期:2020-10-01
down
wechat
bug