基于SnO_(2)∶Sb薄膜沉积工艺参数优化的支持向量回归分析  

Support Vector Regression Analysis Based on the Process ParameterOptimization of SnO_(2)∶Sb Thin Film Deposition

作　　者：陈远豪 肖黎 梁昌兴 罗月婷 龚恒翔[1,2] CHEN Yuanhao;XIAO Li;LIANG Changxing;LUO Yueting;GONG Hengxiang(Institute of Photovoltaic New Energy Application Technology and Equipment,Chongqing University of Technology,Chongqing 400054,China;Physics Experiment Center,Chongqing University of Technology,Chongqing 400054,China;Chongqing Key Laboratory of Green Energy Material Technology and Systems,Chongqing University of Technology,Chongqing 400054,China)

机构地区：[1]重庆理工大学光伏新能源应用技术与设备研究所,重庆400054 [2]重庆理工大学物理实验中心,重庆400054 [3]绿色能源材料技术与系统重庆市重点实验室,重庆400054

出　　处：《材料导报》2023年第11期33-38,共6页Materials Reports

基　　金：重庆市教育委员会科学技术研究基金(KJQN202201137);重庆市自然科学基金(博士后基金)(cstc2021jcyj-bshX0219);重庆理工大学研究生教育高质量发展项目(gzljg2023315);重庆理工大学研究生创新项目(clgycx20203141)。

摘　　要：开发低成本高质量透明导电氧化物薄膜材料的生长技术对现代光电产业发展十分重要。面对多维的薄膜生长工艺参数空间,在寻求最优薄膜生长参数过程中如何有效降低时间、材料成本是研究人员迫切关注的问题。基于雾化辅助CVD法在石英衬底上制备SnO_(2)∶Sb薄膜,利用实验设计方法,获得不同工艺参数下制备的SnO_(2)∶Sb薄膜实验数据集。应用基于贝叶斯优化的支持向量回归方法,建立SnO_(2)∶Sb薄膜透明导电性能的支持向量回归预测模型,结合预测模型对整个工艺参数空间进行探索。利用有限27组工艺参数组合可在四维参数空间中找到高质量SnO_(2)∶Sb薄膜的有效制备参数。在最优工艺参数下制备薄膜的可见光透过率可达86.61%,方块电阻为21.1Ω·□^(-1),膜厚约380 nm。为基于雾化辅助CVD法制备薄膜材料的最优制备工艺探索提供一条有效途径,可有效节约开发成本。Developing the low-cost transparent conductive oxide film material with high quality is of vital important for the development of modern optoelectronic devices.Upon facing the multi-dimensional thin film growth parameter space,how to reduce the time and material cost effectively in the process of seeking the optimal thin film is an urgent concern for researchers.Based on this,a process parameter optimization approach is proposed focusing on the process of growing SnO_(2)∶Sb thin films on quartz substrates using the mist-CVD method.The mist-CVD system is adopted because its environmentally friendly process to prepare metal oxide of high quality with simple equipment configuration,and shows potential for industrial development.Firstly,A experimental design methods is adopted to design experiments for preparing the transparent conductive thin films considering process parameters of doping concentration,deposition temperature,HCl content and deposition time.Furthermore,the value of figures of merits is introduced to evaluate the transparent conducting property of SnO_(2)∶Sb film.Moreover,to establish the effective prediction model of the transparent conductivity of SnO_(2)∶Sb film under different process parameters,the support vector regression method based on Bayesian optimization is used.The two-dimensional cloud image distribution results show the effective prediction results of the influence of preparation parameters on properties of SnO_(2)∶Sb films,which including 4-dimensional parameter space with a limited combination of 27 sets of experimental results.Finally,the prominent process parameters preparing SnO_(2)∶Sb films with high figures of merits is predicted from the two-dimensional cloud image.The SnO_(2)∶Sb films with high transparency and conductivity can be obtained based on the mist-CVD method adopting the optimized process parameter.The average thickness of the SnO_(2)∶Sb films is about 380 nm,the average visible light transmittance of the sample can reach 86.61%,and the sheet resistance

关 键 词：支持向量回归 SnO_(2) ∶Sb薄膜 工艺参数优化 雾化辅助CVD 

分 类 号：TB303[一般工业技术—材料科学与工程]