基于ISOMAP-DE-SVM的Cz单晶硅等径阶段掉苞预测

摘要/Abstract

摘要： 针对目测法无法及时发现直拉单晶硅在等径生长阶段发生的掉苞问题,本文提出一种基于ISOMAP-DE-SVM的掉苞预测模型,可以在掉苞现象发生之前发出警告。首先剔除方差较小的参数,采用斯皮尔曼相关系数法剔除冗余参数,采用最大互信息法检验剩余参数的非线性相关性;然后将关键参数的均值和标准差作为等度量映射和多维放缩的输入,得到两份样本数据;最后将这两份样本数据分别输入到经过差分算法、遗传算法优化的支持向量机预测模型,得到4份预测结果。预测结果表明:基于ISOMAP-DE-SVM的预测模型具有收敛速度快、准确度高的特点,平均预测准确率可以达到96%;同时,所使用的方法揭示了单晶硅等径阶段的数据具有非线性特点。通过实际应用验证表明模型具有一定的工程实用价值。

关键词: 直拉法, 单晶硅, 等径生长, 支持向量机, 等度量映射, 掉苞, 预测

Abstract: Due to the failure of visual inspection to detect the phenomenon of broken edge during the equal-diameter growth process of single crystal silicon, a method based on the ISOMAP-DE-SVM was proposed to give early-warning before the phenomenon of broken edge. Firstly, the parameters with small variance were excluded, redundant parameters were rejected by the spearman correlation coefficient, the nonlinear correlation of remaining parameters was tested by maximal information cofficient. Then, the mean and standard deviation of the key parameters were input into isometric mapping and multiple dimensional scaling, and two samples were obtained. Finally, two samples were input into the support vector machine prediction model optimized by difference algorithm and genetic algorithm respectively, and four results were obtained. The prediction results show that the prediction model based on ISOMAP-DE-SVM can effectively predict the phenomenon of broken edge of single crystal silicon, the prediction model has the characteristics of fast convergence speed and high accuracy, and the average prediction rate can reach 96%. Meanwhile, the method reveals that the data in the equal-diameter growth process of single crystal silicon has nonlinear characteristics. In the practical application verification, it is shown that the model has certain engineering practical value.

Key words: Czochralski method, single crystal silicon, equal-diameter growth, support vector machine, isometric mapping, broken edge, prediction

中图分类号:

O78
TN304.1

侯少华, 张宏帅, 姜宝柱, 朱宾宾, 田增国. 基于ISOMAP-DE-SVM的Cz单晶硅等径阶段掉苞预测[J]. 人工晶体学报, 2023, 52(1): 25-33.

HOU Shaohua, ZHANG Hongshuai, JIANG Baozhu, ZHU Binbin, TIAN Zengguo. Broken Edge Prediction in the Equal-Diameter Growth Process of Cz Single Crystal Silicon Based on ISOMAP-DE-SVM[J]. JOURNAL OF SYNTHETIC CRYSTALS, 2023, 52(1): 25-33.

参考文献

[1] 董建明,张波,刘进,等.直拉法硅单晶生长中断棱与掉苞问题的探讨[J].材料导报,2013,27(S1):157-159.
DONG J M, ZHANG B, LIU J, et al. Investigation on the growth problems in Cz-Si crystal[J]. Materials Review, 2013, 27(S1): 157-159(in Chinese).
[2] 苏文佳,李九龙,杨伟,等.直拉法单晶硅中位错影响因素研究进展[J].硅酸盐学报,2021,49(4):723-735.
SU W J, LI J L, YANG W, et al. Research progress on influencing factors of dislocation in Czochralski silicon[J]. Journal of the Chinese Ceramic Society, 2021, 49(4): 723-735(in Chinese).
[3] 裴志军,纪秀峰,刘峰.4英寸＜111＞硅单晶制备中的“断棱”与“掉苞”问题[J].半导体杂志,1998(3):20-23.
PEI Z J, JI X F, LIU F. Growth problems analysis of Ф100 mm ＜111＞ silicon single crystal[J]. Semiconductor Magazine, 1998(3): 20-23(in Chinese).
[4] SATUNKIN G A. Mathematical modelling and control system design of Czochralski and liquid encapsulated Czochralski processes: the basic low order mathematical model[J]. Journal of Crystal Growth, 1995, 154(1/2): 172-188.
[5] ARMAOU A, CHRISTOFIDES P D. Crystal temperature control in the Czochralski crystal growth process[J]. AIChE Journal, 2001, 47(1): 79-106.
[6] LIU D, LIANG J L. A Bayesian approach to diameter estimation in the diameter control system of silicon single crystal growth[J]. IEEE Transactions on Instrumentation and Measurement, 2011, 60(4): 1307-1315.
[7] 黄伟超.多场作用下Cz法晶体生长过程建模与数值模拟[D].西安:西安理工大学,2018.
HUANG W C. Modeling and numerical simulation of Czochralski crystal growth under multi-physical fields coupling[D]. Xian: Xian University of Technology, 2018(in Chinese).
[8] KATO S, KIM S, KANO M, et al. Gray-box modeling of 300 mm diameter Czochralski single-crystal Si production process[J]. Journal of Crystal Growth, 2021, 553: 125929.
[9] REN J C, LIU D, WAN Y. Modeling and application of Czochralski silicon single crystal growth process using hybrid model of data-driven and mechanism-based methodologies[J]. Journal of Process Control, 2021, 104: 74-85.
[10] 肖立志.机器学习数据驱动与机理模型融合及可解释性问题[J].石油物探,2022,61(2):205-212.
XIAO L Z. The fusion of data-driven machine learning with mechanism models and interpretability issues[J]. Geophysical Prospecting for Petroleum, 2022, 61(2): 205-212(in Chinese).
[11] 杜佳晨.基于数据挖掘的单晶硅等径生长过程“掉苞”预测方法研究[D].杭州:浙江大学,2019.
DU J C. Research on “broken edge” prediction method for equal-diameter growth process of single crystal silicon based on data mining[D]. Hangzhou: Zhejiang University, 2019(in Chinese).
[12] ZHANG J, LIU H, CAO J W, et al. A deep learning based dislocation detection method for cylindrical crystal growth process[J]. Applied Sciences, 2020, 10(21): 7799.
[13] 赵华东,翟晓彤,田增国,等.基于MIC的Cz单晶硅放肩阶段关键特征参数辨识[J].人工晶体学报,2020,49(4):607-612.
ZHAO H D, ZHAI X T, TIAN Z G, et al. Identification of key characteristic parameters of Cz-Si monocrystal during shoulder growth process based on MIC[J]. Journal of Synthetic Crystals, 2020, 49(4): 607-612(in Chinese).
[14] 李欣鸽.CZ硅单晶等径生长阶段关键参数优化研究[D].郑州:郑州大学,2020.
LI X G. Study on the optimization of key parameters of CZ single crystal in equal diameter growth stage[D]. Zhengzhou: Zhengzhou University, 2020(in Chinese).
[15] 黄亮,彭清,谢长君,等.基于差分进化优化的支持向量机燃料电池故障诊断[J].电源技术,2021,45(10):1316-1319.
HUANG L, PENG Q, XIE C J, et al. Fuel cell fault diagnosis for support vector machines optimized based on differential evolution algorithm[J]. Chinese Journal of Power Sources, 2021, 45(10): 1316-1319(in Chinese).
[16] 刘鑫,韩强,周永帅,等.基于GA优化SVM参数的白酒分类识别方法应用研究[J].包装与食品机械,2022,40(2):64-68.
LIU X, HAN Q, ZHOU Y S, et al. Research on application of the classification and identification method of Baijiu based on GA optimized SVM parameters[J]. Packaging and Food Machinery, 2022, 40(2): 64-68(in Chinese).
[17] 吴贵军,范鹏生,陈浩辰,等.基于深度学习的数据分类预测及应用[J].无线互联科技,2022,19(8):126-127.
WU G J, FAN P S, CHEN H C, et al. Data classification prediction and application based on deep learning[J]. Wireless Internet Technology, 2022, 19(8): 126-127(in Chinese).
[18] 高旭旭.基于深度学习的分类预测算法研究及实现[D].北京:北京邮电大学,2019.
GAO X X. Research and application of classification prediction algorithms based on deep learning[D]. Beijing: Beijing University of Posts and Telecommunications, 2019(in Chinese).
[19] XIAO C W, YE J Q, ESTEVES R M, et al. Using Spearmans correlation coefficients for exploratory data analysis on big dataset[J]. Concurrency and Computation: Practice and Experience, 2016, 28(14): 3866-3878.
[20] RESHEF D N, RESHEF Y A, FINUCANE H K, et al. Detecting novel associations in large data sets[J]. Science, 2011, 334(6062): 1518-1524.
[21] 周志华.机器学习[M].北京:清华大学出版社,2016.
ZHOU Z H. Machine Learning[M].Beijing: Tsinghua University Press, 2006(in Chinese).
[22] PRICE K V, STRON R M, LAMPINEN J. Differential evolution: a practical approach to global optimization[M]. Springer Science & Business Media, 2005.
[23] HANCER E, XUE B, ZHANG M J. Differential evolution for filter feature selection based on information theory and feature ranking[J]. Knowledge-Based Systems, 2018, 140: 103-119.
[24] ABUROMMAN A A, IBNE REAZ M B. A novel weighted support vector machines multiclass classifier based on differential evolution for intrusion detection systems[J]. Information Sciences, 2017, 414: 225-246.
[25] 牛岩,魏雨露,刘思琪.数据标准化方法对SVM数据预测结果的影响研究[J].土地开发工程研究,2020,5(4):18-23.
NIU Y, WEI Y L, LIU S Q. Research on the impact of data standardization methods on the data prediction results of SVM[J]. Land Development and Engineering Research, 2020, 5(4): 18-23(in Chinese).
[26] ANOWAR F, SADAOUI S, SELIM B. Conceptual and empirical comparison of dimensionality reduction algorithms (PCA, KPCA, LDA, MDS, SVD, LLE, ISOMAP, LE, ICA, t-SNE)[J]. Computer Science Review, 2021, 40: 100378.