利用拉曼光譜分析與高斯混合模型特徵化多壁奈米碳管之研究

Study on Multi-walled Carbon Nanotubes Characterization Using Raman Spectral Analysis and Gassian Mixture Modeling

周仁祥(Jen-Hsiang Chou)

指導教授 ： 王順源

國立臺北科技大學/電資學院/電機工程系博士班/博士(2013年)

摘要

微波電漿化學氣相沉積法可以生長多樣式的鑽石薄膜與奈米碳管。然而，在微波電漿化學氣相沉積系統中，通常需要依賴經驗調校出滿足預期的電漿型態。因此，本研究裝設反射功率感測器與CCD攝影機觀測電漿影像，並設定氣體流量、微波功率、制動距離、沉積時間、真空腔體壓力和溫度等製程參數生成多壁奈米碳管，藉由調控沿著x-y方向之電場-磁場調整器(E-H tuner)，控制不同條件的電漿型態。為了在微波電漿化學氣相沉積系統中，能生成更佳品質的多壁奈米碳管，本研究採用多壁奈米碳管的長寬比值和拉曼光譜中之ID/IG 比值作為討論品質性能指標。本論文更提出以經驗模態分解法結合高斯混合模型，用以分析多壁奈米碳管之拉曼光譜，其光譜訊號可由一趨勢線與多個高斯函數的線性組合而成，並特徵化出多壁奈米碳管之D式樣與G式樣。本方法與高斯/勞倫浮點式擬合法比較，其實驗結果顯示本方法對於鑑別D式樣與G式樣具有較佳的效能。
另外，在微波電漿化學氣相沉積系統中，電漿模型化與調控扮演相當重要的角色，其中極為關鍵的因素是控制電漿型態與位置之微波反射功率的調控機制。然而，模型化與調控微波反射功率是具有高複雜且仍存在許多未知資訊的挑戰。在本論文中，我們提出以二維高斯混合模型建構相對應電磁場之微波反射率分布之模型，其實驗結果顯示微波反射功率可以簡化成數個二維高斯函數，且提供了可即時調控與預測電漿型態參數的參考，使本裝置具有高重現製程多壁奈米碳管薄膜的功能。

關鍵字

多壁奈米碳管 ； 拉曼光譜分析 ； 高斯混合模型

並列摘要

Microwave plasma chemical vapor deposition (MPCVD) is used to grow various types of diamond films and carbon nanotubes at various temperatures. However, only actual experience is sufficiently reliable to obtain satisfactory plasma using MPCVD. Therefore, this study uses a reflected power sensor as a reference parameter and a charge-coupled device to observe the plasma image. Manufacturing parameters—such as gas flow rate, input microwave power, working distance, deposition time, chamber pressure, and substrate temperature—were all fixed to grow multi-walled carbon nanotubes (MWCNTs). Independent variables were controlled by adjustments of E-H tuner positions along the x- and y- axes, which directly affect plasma conditions. MWCNT quality indices, MWCNT aspect ratios, and the ID/IG intensity ratio of the MWCNT Raman spectra are considered to grow better-quality MWCNTs using the self-assembled MPCVD system. This study uses empirical mode decomposition and Gaussian mixture modeling (GMM) to analyze functionalized MWCNT Raman spectra. Raman spectra data can be represented by a linear combination of a trend and several Gaussian functions that can be used to characterize MWCNT D and G band patterns. The performance of the method is compared using the floating Gaussian-Lorentz fitting method performance. The results show that the proposed method performs good decomposition for identifying D and G band patterns in MWCNTs.
Plasma modeling and control issues are important for MPCVD systems. The tunable reflected microwave power of the MPCVD system is crucial for controlling plasma shape and position. However, modeling the tunable reflected power of microwave plasma is highly complex and is poorly understood. This study uses a 2D GMM to model the microwave power distribution corresponding to the adjustable electromagnetic field. The estimated modeling results show that microwave power data can be simplified as a linear combination of certain Gaussian functions that provide a predictable and controlled basis for real-time tuning of manufacturing parameters and plasma sharpening. The experiment results show that each E-H tuner position fabricates highly reproducible MWCNT films after GMM.

並列關鍵字

Multi-walled Carbon Nanotubes ； Raman Spectral Analysis ； Gaussian Mixture Modeling

參考文獻

[1] S. Iijima, “Helical microtubules of graphitic carbon,” Nature, vol. 354, no. 6348, pp. 56–58, 1991.

[2] S. Reich, C. Thomsen, and J. Maulzsch, Carbon Nanotubes: Basic Concepts and Physical Properties, Wiley-VCH, Weinheim, Germany, 2004.

[3] O. Lourie and H. D. Wagner, “Evaluation of Young' s modulus of carbon nanotubes by micro-Raman spectroscopy,” J . Mater. Res. vol.13, pp. 2418-2422, 1998.

[4] B. Wu, J. Zhang, Z. Wei, S. Cai, and Z. Liu, “Chemical alignment of oxidatively shortened single-walled carbon nanotubeson silver surface,” J. Phys. Chem. B, vol. 22, pp. 5075-5078, 2001.

[5] D. P. Long, J. L Lazorcik, and R. Shashidhar, “Magnetically directed self-assembly of carbon nanotube devices,” Adv. Mater. vol. 16, no. 9-10, pp. 814-819, 2004.