機器學習應用於氫化非晶矽（a-Si：H）鈍化膜之即時電漿化學氣相沉積製程監控;Machine learning of In-situ plasma monitoring in PECVD deposited amorphous (a-Si:H) passivation film

NCU Institutional Repository > 工學院 > 光機電工程研究所 > 博碩士論文 > Item 987654321/76967

jsp.display-item.identifier=請使用永久網址來引用或連結此文件: http://ir.lib.ncu.edu.tw/handle/987654321/76967

题名:	機器學習應用於氫化非晶矽（a-Si：H）鈍化膜之即時電漿化學氣相沉積製程監控;Machine learning of In-situ plasma monitoring in PECVD deposited amorphous (a-Si:H) passivation film
作者:	黃泓睿;Huang, Hung-Jui
贡献者:	光機電工程研究所
关键词:	電漿輔助化學氣相沉積;氫化非晶矽鈍化膜;主成分分析;機器學習;史密斯圖
日期:	2018-07-16
上传时间:	2018-08-31 11:54:53 (UTC+8)
出版者:	國立中央大學
摘要:	研究利用自製電漿輔助化學氣相沉積(PECVD)引入矽甲烷(SiH4)、氫氣(H2)、氬氣(Ar)製備超薄氫化非晶矽本質鈍化層(< 10 nm)於矽基板上，利用光放射光譜端點檢測透過電漿診斷即時提供電漿變化趨勢大幅減少製程測試實驗，而縮短製程開發成本。根據研究氫稀釋比（R= H2 / SiH4）對沉積a-Si:H，研究過程中，利用光放射光譜監測電漿全普光譜量測，其中SiH強度代表機器學習中的主成分分析法（PCA）結合工業4.0趨勢，使得製程更加穩定。即時電漿時序光譜發現電漿解離瞬間（暫態）之SiH強度與RF的輸入電壓及反射電壓間的關係，同時由Z-Scan看出腔體阻抗質和相位，並搭配光放射光譜儀與四極柱質譜儀做即時性的電漿診斷分析，找出電漿組態對於阻抗匹配間的關係，藉由此結論，研究出改善薄膜品質的主要關鍵，進而更有效率的縮短製程時間。其中量測薄膜性質如：FTIR(氫含量)、lifetime(載子生命週期)等，並在案例研究中得到驗證。研究結果顯示，當主成分分析法（PCA）結果當Health Value超過0.2，載子生命週期可超過600us或是更高。再由史密斯圓的匹配過程，當暫態的時間越短約莫1-2秒，也可獲得更好的載子生命週期更高達750us，上述研究及分析結果可獲得適當之製程參數和阻抗匹配條件於本PECVD系統中。 ;The use of self-made plasma-assisted chemical vapor deposition (PECVD) to introduce indium methane (SiH4), hydrogen (H2), and argon (Ar) to produce an ultra-thin hydrogenated amorphous germanium passivation layer (< 10 nm) on a germanium substrate, The use of optical emission spectroscopic end-point detection to provide immediate plasma changes through plasma diagnostics drastically reduces process testing experiments and reduces process development costs. According to the study of hydrogen dilution ratio (R=H2/SiH4), a-Si:H was deposited. During the study, the plasma spectroscopic measurement was performed by light emission spectroscopy, and the SiH* intensity represented the principal component analysis in machine learning ( PCA) combined with the trend of industry 4.0, making the process more stable. The instantaneous plasma time-series spectrum reveals the relationship between the SiH* intensity of the plasma dissociation instant (transient) and the RF input voltage and reflected voltage. At the same time, the Z-Scan sees the cavity impedance and phase, and is matched with the light emission spectrometer. The quadrupole mass spectrometer performs an immediate plasma diagnostic analysis to find out the relationship between the plasma configuration and the impedance matching. From this conclusion, the main key to improve the film quality is studied, and the process time is shortened more efficiently. Among them, film properties such as FTIR (hydrogen content) and lifetime (carrier life cycle) were measured and verified in case studies. The results of the study show that when the PCA results are greater than 0.2, the carrier lifetime can exceed 600 us or more. By Smith′s matching process, when the transient time is shorter about 1-2 seconds, a better carrier lifetime can be obtained up to 750us. The above research and analysis results can obtain appropriate process parameters and impedance matching conditions in this PECVD system.
显示于类别:	[光機電工程研究所 ] 博碩士論文

文件中的档案:

档案	描述	大小	格式	浏览次数
index.html		0Kb	HTML	264	检视/开启

在NCUIR中所有的数据项都受到原著作权保护.

社群 sharing

数据加载中.....