次世代新穎氮化銦鎵/氮化鎵半極性奈米金字塔發光二極體之磊晶成長與元件特性之研究

Abstract

本論文根據compliance growth的原理，將圓柱狀的氮化鎵奈米柱改良成奈米金字塔與奈米鉛筆，理論上能承受更高的應力，試圖解決單晶成長時高度晶格不匹配的問題，著重在新穎的長波長發光二極體，以及高品質高銦含量氮化銦鎵薄膜的開發。 論文第一部分，我們成功發展了波長524nm，585nm以及640nm的綠光，黃光以及紅光光原。綠光奈米發光二極體外部量子效率在電流密度30A/cm2下為2.4%,此特性已領先全球目前奈米發光二極體的發表。當元件效率到達頂點，其效率直到170A/cm2 的高電流密度操作下幾乎沒有效率低垂的問題，即efficiency droop是0%。這是由於B係數(輻射複合效率)比一般平面式的c-plane發光二極體高了5倍由1x10-11(cm3/s)提升到5x10-11(cm3/s)。本論文除了成功開發高品質的綠光奈米金字塔二極體以外，利用垂直整合兩種發光層於同一奈米發光二極體中，成功開發波長為640nm的高品質的單色紅光奈米二極體，並且展示可同時發出兩種波長的光源。在操作電流密度170A/cm2下可得到色溫4489k, CIE x,y座標分別為0.385與0.5的暖白光源。因此奈米金字塔發光二極體在長波長光源上創造新的可能性。 論文第二部分，延伸compliance growth的概念，挑戰III-nitride材料中成長最困難的氮化銦材料。將奈米柱之間的走道以及側壁用二氧化矽覆蓋後,氮化銦只會成長在某些氮化鎵奈米柱的平面交界處。以COMSOL模擬應力分布顯示這些平面交界處有一個共通點，就是底下的氮化鎵晶格有較大的自由度可以移動,進而將氮化銦/氮化鎵這個界面的應力透過水平或垂直的方向做晶格伸張釋放掉。雖然有此規則存在，不過二氧化矽上的原子會在氮化鎵/二氧化矽的界面上堆積造成優先成長,因應力選擇產生的優先成長以及原子堆積造成的修先成長形成了競爭。將氮化鎵奈米柱形成特定的晶格面,例如: {10-10 }M平面與{10-11}S平面,並且把奈米柱直徑稍微擴大，這兩個做法能規則化原本圓柱狀的氮化鎵奈米柱。在同樣的成長條件下，氮化銦變的更不容易成長在氮化鎵表面上，必須將溫度降低才能順利成長，表示之前的晶體是靠著粗糙的表面才能順利成長出來，所以是屬於不可預期的規則.。 將奈米柱變成奈米鉛筆後，氮化銦會規則的成長在側壁M平面與頂端S平面的交界，以及側壁M平面間的交界。以COMSOL模擬氮化銦長在奈米鉛筆的氮化鎵晶體上，果然行為很類似，在所有平面的交界處都是應變能較低的位置。以平均厚度10nm的氮化銦成長在氮化鎵表面上，系統中應變能最小為17 J/m2相較於成長在S平面(102 J/m2)與M平面(151 J/m2)的氮化鎵晶體上有6~9倍的下降，因此形成了一個自我對準(self-assembly)的有序成長。這種自我對準的現象在氮化鎵奈米柱直徑越寬的時候會越明顯。在TEM的影像下可清楚看幾乎沒有缺陷的氮化鎵奈米鉛筆晶體，以及與氮化鎵具有同一個晶體方向的單晶氮化銦晶體。室低溫的光機發光譜顯示出氮化銦的內部量子效率，較成長於一般平面式的氮化鎵晶體上由3%提升到6%，有相對性100%的發光效率提升。 本論文利用compliance growth概念延伸Nano-epitaxy的方法,成功發展出以氮化銦鎵為主動層的綠光，黃光以及紅光奈米金字塔發光二極體，並克服高度晶個格不匹配的氮化銦與氮化鎵,成功成長出單晶的氮化銦晶體。高度晶格不匹配的成長會有很明顯的應力選擇現象，使晶體自我對準的長在特定的地方。此種Nano-epitaxy的方法一方面有機會解決III-V族半導體的綠光效率能隙，成為未來固態照明重要的解答。另一方面還能運用自我對準成長的概念發展氮化銦鎵量子結構，例如量子點、量子線等，進而開發室溫單光子源(single photon source)以及先進奈米元件的新應用。

Based on the concept of compliance growth, we modify the GaN nano-pillar from cylinder into nano-pyramid (NP) and nano-pencil shape, in principle has higher tolerance of stress, try to solve the problem of highly mismatch epitaxy. We focus on the development of novel long emission wavelength nano-LED and high quality high indium content InGaN thin film. In the first part of this thesis, we successfully demonstrate the green, yellow and red NP LED with emission wavelength 524nm, 585nm and 640nm respectively. The green NP LED shows a external quantum efficiency (EQE) 2.4% under current density 30A/cm2. The EQE would sustain at the maximum status once reaching the peak value till high current density 170A/cm2, which means the efficiency droop is 0% from peak value. This is due to the B coefficient of NP LED is 5 times higher than conventional planar c-plane LED, enhancement from 1x10-11(cm3/s) to 5x10-11(cm3/s). Besides the demonstration of high quality green NP LED, by integrating two emission colors vertically in one NP LED, we not only successfully develop a high quality single color red NP LED, but also demonstrate two emission peaks in single NP LED. Under current density 170A/cm2, we achieve a warm white light source with correlated color temperature (CCT) 4489k, and its coordinates CIE x and CIE y at 0.385 and 0.5 respectively. As a result, the NP LED demonstrates new possibilities for long emission wavelength light source. In the second part of thesis, according to the concept of compliance growth, we try to create new solutions to develop high quality InN crystal, which is the most difficult material of III-Nitride family. By covering the bottom area in between GaN nano-pillars with SiO2, the substrate became the sidewall passivated GaN nano-pillar (SPGP) template; the InN crystal would only grow on certain positions of intersections of different plane. By utilizing the COMSOL simulator, from the results of stress distribution, there is a common phenomenon of these intersection areas. The underlying GaN has higher degree of freedom to accommodate the lattice vertically or horizontally that generates from InN/GaN interface. Although there is a strain selected growth behavior, nevertheless, the loading effect facilitates the growth to happen at the edge of SiO2 mask that becomes a competing position with strain selected growth during the nucleation. In order to create some regular rules, we intentionally change the nano-pillar from cylinder shape into pencil like appearance with specific facets such as {10-10} M-plane and {10-11} S plane, and enlarge the diameter a little bit to eliminate certain un-controllable growth position. It becomes more difficulty for InN to grow on GaN nano-pencil. We must lower the growth temperature to obtain similar amount of InN crystals., which means that the regulation on cylinder shape nano-pillar is promoted by the rough surface not a predictable result. After changing the cylinder shape into pencil appearance, an interesting phenomenon has been observed while growing InN crystal on to SPGP template. The six corners of GaN nano-pencil seem to be the favored positions for InN to grow first, which correspond to the intersection of sidewall m-plane and the intersection of m-plane and s-plane. The simulation results show a good agreement with what we observed from SEM images; the strain energy of the corner position is nearly eight times lower than other surface on the nano-pencil. From the results of COMSOL simulator, the mechanism is very similar of InN on nano-pillar and nano-pencil; the intersection area is the lowest strain energy position in the system. Taking InN with an average thickness 10nm as an example, the lowest strain energy is 17 J/m2 which is 6~9 times lower than grown on S-plane (102 J/m2) and M-plane (151 J/m2). As a result, there is a self-assembly regular growth behavior occurred and would be more pronounced if the diameter of nano-pillar becoming wider. The GaN nano-pencil is almost defect free under TEM image and the single crystal InN is observed from selective area diffraction pattern which has the same orientation with GaN nano-pencil. The internal quantum efficiency (IQE) of InN grown on flat c-plane GaN and on SPGP is 3% and 6% respectively, and the IQE has a relatively 100% enhancement while InN grown on GaN nano-pencil. In short, by extending the concept of compliance growth, we successfully demonstrate InGaN based green, yellow and red NP LED, and successfully grow single crystalline InN overcoming the highly mismatch between InN and GaN by nano-epitaxy. There is a pronounced self-assembly growth phenomena at specific locality on GaN nano-pencil, which is assisted by the highly mismatch between epi-layer and substrate. Therefore, nano-epitaxy on one hand is a promising solution to solve the efficiency green gap as well as solid-state lighting. On the other hand, the strain-assisted growth behavior is potential method to grow InGaN based low dimensional structure, such as quantum dot and quantum wire, to explore more possibility of room temperature single photon source and advanced nano-device in the future.