雙極脈衝直流反應式濺鍍氮化鋁薄膜與光學發射光譜輔助機器學習結晶性質分類器建立;Optical emission spectroscopy based machine learning classifiers of AlN crystal properties deposited by pulsed direct current reactive magnetron sputtering

NCUIR > Engineering College > Graduate Institute of Mechanical Engineering > Electronic Thesis & Dissertation > Item 987654321/81630

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/81630

Title:	雙極脈衝直流反應式濺鍍氮化鋁薄膜與光學發射光譜輔助機器學習結晶性質分類器建立;Optical emission spectroscopy based machine learning classifiers of AlN crystal properties deposited by pulsed direct current reactive magnetron sputtering
Authors:	張學而;Chang, Hsueh-Er
Contributors:	機械工程學系
Keywords:	反應式濺鍍;氮化鋁;光學發射光譜;機器學習
Date:	2019-07-19
Issue Date:	2019-09-03 16:34:03 (UTC+8)
Publisher:	國立中央大學
Abstract:	本研究中使用雙極脈衝直流反應式濺鍍沉積氮化鋁薄膜於矽基板上，並建立一個機器學習薄膜品質分類器來進行薄膜品質預測。實驗中利用調整濺鍍製程參數氬氣和氮氣的混氣比，進行不同結晶性質的氮化鋁薄膜製備及薄膜品質量測，完成的樣本將提供給機器學習分類器訓練使用。製程中利用光放射光譜儀(Optical Emission Spectroscopy, OES)測量電漿放射出的鋁光譜強度進行分析，在各種混氣比條件下，得到瞬態反應濺鍍模式(N2:Ar = 7:30)及介電反應濺鍍模式(N2:Ar = 45:15)區間，在這二種區間進行薄膜樣本的濺鍍。FTIR光譜顯示出濺鍍出的薄膜具有Al-N鍵的振動模式引起的吸收帶，峰寬隨脈衝頻率上升(75 kHz~250 kHz)而變窄(166~192 cm-1)，可視為氮化鋁薄膜結構的改善。最後通過XRD量測證實氮化鋁薄膜的結晶相，以此歸類出結晶與非晶的氮化鋁薄膜。本研究最終目標為建立出氮化鋁薄膜的結晶性質分類器。通過對製程中光放射光譜數據進行主成分分析(PCA)，得到不同製程條件之樣品數據的第一~第四特徵向量(PC1~PC4)，經結晶性質比對，訓練數據採用分辨率較高的PC2、PC3。將主成分分析完成的數據集利用支援向量機(SVM)進行訓練分類，並使用不同核函數：高斯核函數(Gaussian kernel)、徑向基核函數(Radial basis function kernel, RBF)及多項式核函數(Polynomial kernel)來進行分類邊界計算，訓練出不同核函數的分類器模型。所訓練之分類器模型藉由輸入其他驗證樣本的光放射光譜數據，並比對驗證樣本實際量測的結晶性質結果，可驗證不同核函數分類器對於預測氮化鋁薄膜結晶性質的準確性，結果顯示，高斯核函數演算法能夠建立較高準確率(90%)的氮化鋁結晶性質分類器。 ;In this study, we report pulsed direct current reactive magnetron sputtering deposited aluminum nitride (AlN) films with the employment of in-situ diagnostics tool of optical emission spectroscopy (OES) for plasma chemistry monitoring. Effect of flow ratio of nitrogen/argon on structural evolutions of the deposited AlN films were systematically investigated by various tools of Fourier-transform infrared spectroscopy (FTIR) and X-ray diffraction spectroscopy (XRD). In addition, principal component analysis (PCA) was performed to establish the correlations between the plasma chemistry (OES spectra) and crystal structure of deposited AlN thin films. The second and third principal component (PC2 and PC3) of the sample data was combined with the support vector machine (SVM) algorithm method to group the crystallization and amorphous OES spectral data. The support vector machine (SVM) method can classificated by using three different kernel function methods, include the Gaussian kernel, the polynomial kernel and the radial basis function kernel. The results show that the Gaussian kernel function can establish a Classifiers of AlN crystal properties with high accuracy (90%).
Appears in Collections:	[Graduate Institute of Mechanical Engineering] Electronic Thesis & Dissertation

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	127	View/Open

社群 sharing

Loading...