以熱裂解化學沉積法成長碳化矽薄膜之研究

Abstract

近年來，碳化矽材料在高功率半導體應用領域被視為最具潛力之材料之一。最主要是因為其優異之材料特性，如寬能隙(>2.3eV), 高崩潰電場(>3MV/cm)以及高熱傳導係數(4.9 W K-1 cm-1)等。因為以上之材料特性，以碳化矽所製作之電子元件，非常適合應用於高溫環境以及高功率需求。另外，碳化矽之電子飽和速度可達2-2.7x107 cm2/V-s，是一般矽晶材料的兩倍，在微波元件應用上，可得到較高之通道電流，因此倍受矚目。 本研究主要以熱裂解化學沉積法於c軸藍寶石基板上成長碳化矽薄膜，並探討其成長參數對薄膜特性之影響。所成長之樣品經一系列X光繞射儀(XRD)、光學穿透譜(UV-vis)、原子力顯微鏡(AFM) 、X射線光電子能譜儀(XPS) 、光激發光(PL)以及掃描式電子顯微鏡(SEM)分析，並歸納出控制成長總氣壓、以及丙烷流量對於碳化矽薄膜特性之影響。在調變總氣壓之實驗中，確定低總壓成長條件下會導致矽晶的沉積，而在高總壓條件下明顯可抑制矽晶的析出。當總氣壓達1400mTorr時可得到最佳品質之碳化矽薄膜。另一方面，在調變丙烷流量實驗中發現，隨著丙烷流量遞增，碳化矽薄膜之結晶品質則隨之下降。而當丙烷流量為3sccm時可成長出最佳之碳化矽薄膜。 此外，為達成高品質的碳化矽成長，藉由在藍寶石基板上加入氮化鋁緩衝層，以及控制氫清潔時間與碳化時間，並歸納出對於碳化矽薄膜特性之影響。在單純增加氮化鋁緩衝層實驗中得到較佳品質之碳化矽薄膜。同時，經由XRD和PL的檢驗，證實在有無緩衝層氮化鋁的藍寶石基板上，均為雙相成長(twin domain)結構之3C-SiC。 在氫蝕刻之調變實驗中，在有緩衝層氮化鋁的藍寶石基板，確定氫蝕刻會導致較高品質碳化矽的成長。相反的，氫蝕刻在藍寶石基板上，未有明顯改善碳化矽薄膜之結晶品質的效果。另一方面，於調變碳化時間之實驗中發現，當無碳化階段時，於有氮化鋁緩衝層的藍寶石基板上，則可成長出較佳品質之碳化矽薄膜。 最後，為驗證碳化矽作為緩衝層對於氮化鎵的成長效應，我們在未經優化成長條件下比較，氮化鎵薄膜成長在有碳化矽緩衝層的藍寶石基板上，較直接成長在藍寶石基板，在應力方面有明顯改善。另一方面，若於矽基板成長氮化鎵，在同時成長且未優化之狀況下，有使用碳化矽緩衝層所成長之氮化鎵其磊晶品質大幅提高，這些結果顯示碳化矽對於氮化鎵成長而言是一優異之緩衝層。由於目前氮化鎵在矽基板上成長面臨高品質無裂痕的挑戰，以碳化矽做為緩衝層有可能是未來一個有潛力方案。

Recently, SiC material is regarded as one of the most promising materials for high temperature, high power, radiation-resistant electronics applications due to its wide bandgap (>2.3eV) and high breakdown electric field (>3MV/cm) and high thermal conductivity (4.9W/cm-K) characteristics. For the microwave applications, the high channel current can be obtained due to the high saturated electron drift velocity of SiC which is 2-2.7×10^7 cm2/V-s (2 times that of silicon). The investigation on the growth of SiC by thermal CVD is the focus of the research, we focus on the studies of growth parameters such as chamber pressure, the flow rate of C3H8, hydrogen etching and carbonization time on the film quality. To achieve the goal, the structure and characteristics of SiC films were studied by x-ray diffraction (XRD), atomic force microscopy (AFM), ultraviolet-visible spectroscopy (UV-vis), X-ray photoemission spectroscopy (XPS) and scanning electron microscopy (SEM). According to the results in this study, a comprehensive understanding on the growth of SiC is achieved. For the study of chamber pressure on SiC growth, it was found that the diffraction intensity of Si(111) was suppressed and the intensity of SiC(111) was increased as chamber pressure increased. When the chamber pressure reached 1400 mtorr (sample E), a maximum diffraction intensity of SiC(111) with the lowest FWHM of rocking curve and smallest surface roughness was obtained. On the other hand, for the study of C3H8 flow rate on SiC growth, the quality of SiC was enhanced as the flow rate C3H8 was reduced. When SiC was grown under the C3H8 flow rate of 3 sccm, the sample had the best quality with XRC FWHM of 104 arcmin for SiC(111). To achieve high quality SiC epitaxy, the experimented results shows that the crystalline quality of 3C-SiC film can be improved by introducing the AlN buffer layer on sapphire. Moreover, the twin domains growth of 3C-SiC both on sapphire and AlN/sapphire were confirmed by XRD phi-scan and PL. For the effect of hydrogen etching on the growth of SiC, the results show that the SiC grown without hydrogen pre-treatment on sapphire has better quality than SiC with 5-10min hydrogen etching on sapphire. In contrast, the pre-treatment of hydrogen etching is required to improve the crystalline quality of SiC grown on AlN/sapphire. Besides, for the growth of SiC film on AlN/sapphire, carbonization is not necessary due to the higher electron affinity of Si than C. Finally, for GaN grown with SiC buffer on sapphire, the FWHM value of (102) GaN is smaller, suggesting less screw dislocations exist in the GaN film. It implies that SiC buffer has obvious effect on the stress release of GaN grown on sapphire. In addition, for the growth of GaN with SiC buffer on silicon, under non-optimum condition, the GaN grown with SiC buffer layer has better crystalline quality compared with GaN directly grown on Si. The crystal quality of GaN using SiC buffer in this study is compatible with those published data of GaN on Si by AlN buffer. The results demonstrate that SiC should be a promising buffer layer for high quality GaN growth on Si and should be further studied in the future.