本研究建構了一套具有自動對焦功能之拉曼量測系統，解決了計畫及實驗室對材料分析儀器的需求。系統由硬體以及軟體所構成，硬體使用光譜儀取代了光譜分析元件，並搭配電控移動平台控制樣品位置；軟體使用Python建立使用者介面，通過擷取光譜資訊及控制移動平台，實現即時調整光譜儀量測參數、波長與拉曼位移轉換及自動對焦功能，並同步顯示當下的光譜資訊。 首先以商用拉曼光譜儀驗證本研究使用的4H-SiC主要拉曼波峰位於776 cm-1，訊噪比為26，並以此做為參考數值。在自製拉曼量測系統方面，量測到的4H-SiC主要拉曼波峰位於778 cm-1，與商用拉曼光譜儀的拉曼位移偏差為2 cm-1，小於系統的解析度6.5 cm-1，證實了拉曼量測系統的準確性。除此之外，成功辨識Si以及雷射加工4H-SiC的拉曼光譜，證實本研究系統能夠量測三種不同材料的拉曼光譜，驗證了拉曼量測系統的可靠性。接著，比較手動對焦與自動對焦在粗糙表面的性能，結果顯示，手動對焦與自動對焦的標準差分別為5.1及2.0 m，說明了自動對焦方法在粗糙表面下具有更好的穩定性，驗證了自動對焦的導入能夠解決手動對焦粗糙樣品不穩定的問題。在自動對焦可靠性驗證中，自動對焦找到的最佳對焦位置能夠量測到最強的拉曼訊號，在5 mm偏移下，拉曼訊號強度僅減少25%，證實了自動對焦系統的可靠性。最後，探討系統的誤差來源，包含：拉曼量測系統造成的誤差、手動與自動對焦的誤差、表面粗糙度的誤差，說明拉曼量測系統的誤差不影響材料成分辨識結果。;This study designed and constructed a Raman measurement system with an automatic focusing function， addressing the need for material analysis instruments in research projects and laboratories. The system consists of hardware and software. The hardware utilizes a spectrometer in place of a spectral analysis component and is paired with an electrically controlled moving platform to adjust the sample position. The software builds a user interface using Python. By capturing spectral data and controlling the moving platform， it achieves real-time adjustment of spectrometer measurement parameters， wavelength-to-Raman shift conversion， and autofocus functions， while simultaneously displaying the current spectral data. First， a commercial Raman spectrometer was used to verify that the primary Raman peak of 4H-SiC is located at 776 cm⁻¹， with a signal-to-noise ratio of 26， using these values as reference data. Regarding the homemade Raman measurement system， the primary 4H-SiC Raman peak was measured at 778 cm⁻¹， with a Raman shift deviation of 2 cm⁻¹ compared to the commercial Raman spectrometer. Since this deviation is less than the system’s resolution of 6.5 cm⁻¹， the accuracy of the Raman measurement system was confirmed. In addition， the successfully identified Raman spectra of Si and laser-processed 4H-SiC confirm that the system can measure the Raman spectra of three different materials， verifying the reliability of the Raman measurement system. Next， the performance of manual focusing and autofocusing on rough surfaces was compared. The results showed that the standard deviations for manual focusing and autofocusing were 5.1 and 2.0 µm， respectively， indicating that the autofocusing method offers better stability on rough surfaces and verifying that the introduction of autofocusing can resolve the instability issues associated with manual focusing on rough samples. In the verification of autofocus reliability， the optimal focus position determined by autofocus yielded the strongest Raman signal. Even with a 5 mm offset， the Raman signal intensity decreased by only 25%， confirming the reliability of the autofocus system. Finally， the study discusses sources of system error， including errors caused by the Raman measurement system， errors associated with manual and autofocus， and errors related to surface roughness， thereby explaining the causes of measurement errors.
