基于神经网络的AZO薄膜制备工艺参数的正交优化设计

Orthogonal optimization design of process parameter for AZO thin film preparation based on neural network

采用磁控溅射法制备AZO薄膜,研究和讨论了溅射功率、溅射时间和溅射气压3个工艺参数对AZO薄膜光学和电学性能的影响。采用正交优化设计,对3个工艺参数进行优化,测量了透射率和电阻率,以此作为薄膜光电性能的评价指标,通过极差值分析确定了制备薄膜的最佳工艺参数。影响薄膜透射率的最主要因素为溅射气压;影响电阻率的最主要因素为溅射时间。获得制备高透射率低电阻率的AZO薄膜的最佳工艺组合方案为溅射功率为400W、溅射时间为1000s、溅射气压为1.0Pa。将反馈型(BP)神经网络应用于磁控溅射AZO薄膜光学性能(可见光区的平均透射率)和电学性能(电阻率)的研究。输入样品数据对神经网络进行训练,建立AZO薄膜光电性能随溅射参数变化的预测模型。

AZO films were prepared by magnetron sputtering,the effects of the three technological parameters of sputtering power,sputtering time and sputtering pressure on the optical and electrical properties of AZO films were studied.Three technological parameters were optimized by orthogonal test design.The transmittance and resistivity were measured,and used as evaluation index to determine the optimum parameters of the film by extreme value analysis table.The main factor influencing the transmission was the substrate pressure.The main factors affected the resistivity was the sputtering time.For the preparation of AZO thin films with high transmittance and low resistivity,the optimum combination scheme was that sputtering power was 400W,sputtering time of 1000s,the substrate pressure was 1.0Pa. A feedback(BP)neural? network was applied to study the optical properties(average transmittance in visible region) and electrical properties(resistivity)of magnetron sputtering AZO thin films.