以氫化物氣相磊晶系統使用氮化鎵奈米柱模板製作獨立式氮化鎵基板應用於發光二極體之研究

Abstract

由於缺乏晶格匹配的基板，三族氮化物元件只能採用異質磊晶的方式，利用晶格常數不同的材料來當作基板，因此有缺陷密度過高，以及因為熱膨脹係數不同造成晶體彎曲甚至破裂的問題。對於低電流密度的元件（例如LED），這些問題的影響不大，但是對於高電流密度元件（例如藍光LD）、高功率元件或UV元件而言，影響就相當可觀，若有單晶三族氮化物基板來進行同質磊晶，不但可以降低晶格常數不匹配所造成的缺陷密度、熱膨脹係數不同所造成的彎曲；更可大幅簡化元件的製程及提昇元件的可靠度。未來在光電元件產業的發展上，單晶三族氮化物基板必然具有很大的影響。 本論文的研究方向在於使用氫化物氣相磊晶法成長氮化鎵厚膜，其中最大的問題在於如何克服因熱膨脹係數不同所產生的熱應力所導致的晶片破裂，我們運用了氮化鎵奈米柱為磊晶成長模板，以奈米側向磊晶成長技術克服了應力的問題，也有效的降低氮化鎵厚膜內的缺陷密度及殘存應力。經由氫化物氣相磊晶成長完之氮化鎵厚膜後，在降溫過程中，受氮化鎵厚膜與藍寶石基板不同的熱膨脹係數會產生應力，因為奈米柱較脆弱會被此應力折斷，而使得氮化鎵厚膜從原生基板上分開，而得到獨立氮化鎵基板。我們分別使用了掃描式電子顯微鏡(SEM)，陰極螢光(CL)，光致螢光(PL)，拉曼光譜(Raman)，高解析X光等工具來量測氮化鎵基板的特性。其差排密度大約為107cm-2左右， X光搖擺曲線的半高寬大約是196秒。從陰極螢光、光致螢光及拉曼光譜分析，獨立式的氮化鎵基板內的應力幾乎完全釋放。由儀器驗證氮化鎵厚膜從基板上分離後，晶格品質獲得明顯的改善，應力也獲得釋放，整個厚膜也處於近乎無應力累積的狀態。 隨後我們針對LED分別成長於獨立式的氮化鎵基板與藍寶石基板上做一比較，量測成長於不同基板上之LED的光電特性。成長於獨立式的氮化鎵基板之氮化銦鎵/氮化鎵多重量子井發光二極體，由於是同質性材料磊晶成長，所以缺陷密度明顯的減少很多，因而提升了發光二極體的發光效率。相較於相同結構成長於藍寶石基板，成長於獨立式的氮化鎵基板之發光二極體有較低的順向啟動電壓，較小的理想值常數，且有較高的出光功率。對於成長於獨立式的氮化鎵基板之發光二極體其外部量子效率也多有提升，尤其是在高注入電流下，光電特性的下墜效應有明顯的減少。

Due to a lack of lattice-matched substrates, III-nitride devices can only use heteroepitaxy and materials with different lattice constants as the substrate, resulting in a high defect density. Furthermore, differences in thermal expansion coefficients induce bowing or even crack of the crystal. These problems do not affect devices with low current density (such as LEDs); however, they significantly influence devices with high current density (such as blue LD) and high power or the UV devices. Using the single-crystal III-nitride substrate for homoepitaxy can not only reduce the high defect density due to mismatched lattice constants and crystal bowing caused by the different thermal expansion coefficient, but can also simplify the device fabricating procedures and improve the reliability of the devices. The use of single-crystal III-nitride substrates is certain to impact the optoelectronic devices development in the future. This thesis focuses primarily on the growth of GaN thick films using hydride vapor phase epitaxy (HVPE). However, overcoming the crystal crack caused by the thermal strain of the different thermal expansion coefficients remains the substantial problem to be solved. The researchers adopted GaN nanorods as the growth template to overcome strain problems and reduce the defect density and remaining strain inside GaN thick films through sidewall lateral epitaxial overgrowth (SLEO). After the GaN thick films grow on HVPE, the different thermal expansion coefficients between the sapphire substrate and GaN thick films generates strain when the temperature cooling down of the HVPE system. Due to the nanorods very weak, GaN nanorods are broken by the strain. GaN thick films are thus separated from the sapphire substrates, becoming freestanding GaN substrates. This study employed SEM, CL, PL, Raman, and HRXRD to measure the properties of GaN substrates. The threading dislocation of GaN substrates is approximately 107 cm-2 with a 196 aresec FWHM of the X-ray rocking curve. The Raman, PL, and CL spectrum analysis indicated that the strain inside freestanding GaN substrates was almost released. The test results demonstrated that after GaN thick films were separated from the substrate, the quality of the lattice improved significantly; the strain was released and the entire film became unstrained. The researchers compared LEDs grown on FS-GaN and sapphire substrates to test the optoelectronic characteristics of LEDs on different substrates. For InGaN/GaN multiple quantum well light-emitting diodes (LEDs) grown on the FS-GaN substrates, the defect densities in the homoepitaxially grown LEDs were substantially reduced, leading to improved light emission efficiency. Compared with LEDs grown on sapphire, this study obtained a lower forward voltage, smaller diode ideality factor, and higher light output power in the same structure grown on FS-GaN. The external quantum efficiency (EQE) of LEDs grown on FS-GaN was improved, particularly at high injection currents, significantly reducing the efficiency droop phenomenon at high current density.