在氧化鋅磊晶薄膜物理特性中晶體缺陷結構的角色

Abstract

我們利用雷射濺鍍法於不同的基板上成長高品質(0001) c軸方向的氧化鋅(ZnO)磊晶薄膜;使用的基板包括c軸方向的藍寶石(α-Al2O3)基板和使用奈米厚的γ相氧化鋁(γ-Al2O3)或氧化釔(Y2O3)做為緩衝層的(111)方向的矽基板。針對氧化鋅長在藍寶石基板的系統，X光繞射結果指出相對於c軸方向藍寶石基板的晶格，c軸方向的氧化鋅其晶格沿著樣品表面法線方向旋轉30度，亦即兩者的橫向磊晶關係式為[1010][1120]sapphireZnO&#1048614; 和 [1120][0110]sapphireZnO&#1048614;。在沿著薄膜表面法線及水平方向氧化鋅X光繞射峰寬度呈現巨大的差異，揭示出特定貫穿式差排(threading dislocation)類型的幾何關係；從X光繞射及穿透式電子顯微鏡實驗數據計算出的貫穿式差排密度顯示出大部分差排是刃(edge)差排類型。結合散射及顯微術量測結果，證實貫穿式差排並非均勻分布在氧化鋅薄膜內，氧化鋅薄膜是由柱狀的磊晶核芯周圍環繞高密度的貫穿式刃差排的磊晶晶粒所組成。藉由掃描式電容顯微鏡及導電式原子力顯微鏡針對貫穿式差排聚集處做電性量測，發現其平帶電壓平移及電位勢障提高，這歸因於高密度的貫穿式刃差排存在而造成的界面捕獲電荷密度。另一方面，由於貫穿式螺旋(screw)差排密度遠小於貫穿式刃差排，因此我們無法確認貫穿式螺旋差排的位置及其電性。 對c軸方向氧化鋅磊晶薄膜成長於使用超薄γ相氧化鋁為緩衝層的(111)矽基板的結構分析結果顯示(111)方向的γ相氧化鋁磊晶緩衝層在沿磊晶緩衝層水平方向上存在兩個相互旋轉60度的晶畴，由此可以歸納三者的磊晶關係式為{1010}||{224} {422}||{224}ZnOAlOAlOSiorγγ−−。藉由X光繞射及光激發光譜實驗對氧化鋅磊晶層的結晶品質及光學性質研究，明確地將能帶邊緣輻射與深層缺陷輻射的強度比值與偏離垂直基板表面方向X光繞射峰φ 掃描的訊號寬度聯繫起來; 並且能帶邊緣輻射的寬度與氧化鋅(0002)繞射峰 θ−rocking curve的寬度表現出很強的相依性；這些現象證明，能帶邊緣輻射與深層缺陷輻射強度的比值主要受貫穿式刃差排影響，而能帶邊緣輻射的寬度與貫穿式螺旋差排有關。 X光繞射、光激發光譜及穿透式電子顯微鏡實驗證實，使用奈米厚、高介電質(high-k)材料氧化釔為緩衝層的(111)矽基板上可以成長同時具有高品質結晶及光學特性的氧化鋅磊晶薄膜。奈米厚氧化釔不僅可以提供成長完美氧化鋅磊晶薄膜的緩衝層，更可以成為氧化鋅及矽基板間的絕緣層。藉由X光繞射及穿透式電子顯微鏡量測，氧化鋅與氧化釔間的磊晶關係式遵循23(0001)2110||(111)101ZnOYO<><>關係。 氧化鋅晶格與氧化釔中的氧六方形次晶格(sub-lattice)具同向排列，二者之間的界面結構可以妥善地用7或8倍氧化鋅{0211}的面距匹配6或7倍氧化釔{044}的面距的晶畴匹配磊晶模型描述；如此大的晶格不匹配可以藉由錯配差排(misfit dislocation)在界面上以6或7倍氧化釔{044}面距週期性地排列做調節，使得殘留應力明顯降低。即使是厚度只有0.21 μm 的氧化鋅薄膜也展現優秀的光激發輻射特性。我們的實驗結果證明了氧化釔可以成為整合氧化鋅光電元件與以矽為主體的積體電路於一體的模板。 最後，從X光繞射、穿透式電子顯微鏡實驗數據，分別計算生長於上述三種基板的氧化鋅磊晶薄膜的貫穿式差排密度，結果顯示刃差排型式都是主要的結構缺陷；氧化鋅晶格總是與藍寶石基板，γ 相氧化鋁及氧化釔中氧的六方形次晶格呈同向排列，其所對應的二維次晶格晶格常數分別為2.75、2.8、3.75 Å，與氧化鋅的晶格常數(3.249 Å)比較，預期沿水平方向的應力狀態對於氧化鋅成長於c軸方向的藍寶石、γ 相氧化鋁應為壓縮應力；相反地，成長於氧化釔上的氧化鋅應為拉伸應力。然而，僅有成長在藍寶石上的氧化鋅被觀察到其沿水平方向呈壓縮應力；成長在γ 相氧化鋁及氧化釔上的氧化鋅都是受到拉伸應力；事實上，對氧化鋅成長於使用其他氧化物為緩衝層的(111)矽基板上，包含氧化釓(Gd2O3)和參雜氧化釔的氧化鉿(Y2O3-doped HfO2)等系統，所有氧化鋅磊晶薄膜都在水平方向受到拉伸應力；此外，存在於氧化鋅及氧化物緩衝層界面上的高密度錯配差排，調節了絕大部分晶格匹配所造成的應力。由於氧化鋅的熱膨脹係數(α~4-6.5 × 10-6 K-1)小於藍寶石基板(8 × 10-6 K-1)，但大於矽基板的(8 × 10-6 K-1)熱膨脹係數，這個趨勢與我們所觀察的氧化鋅成長於藍寶石基板及矽基板上的應力態相符，顯示氧化鋅磊晶層的應力態主要是由成長完成後冷卻過程中因磊晶薄膜與基板間熱膨脹係數差所導致的熱應力所支配；因為使用的氧化物緩衝層厚度為只有數奈米厚，因此在這些使用的氧化物緩衝層案例中，來自緩衝層熱應力的影響是可以忽略的。

High-quality c-oriented ZnO film has been epiaxially grown by utilizing PLD on the sapphire (0001), and Si (111) substrates with a nano-thick γ-Al2O3 or Y2O3 buffer layer, respectively. XRD results show a 30° offset between the {2020} reflections of ZnO and sapphire verifies the in-plane epitaxial relationship of [01-10] sapphire || [10-10] ZnO and [11-20]sapphire || [01-10]ZnO; the great disparity of X-ray diffraction line widths between the normal and in-plane reflections reveals the specific threading dislocation (TD) geometry of ZnO. The calculated TDs densities from XRD and TEM indicate most TDs are pure edge dislocations. From a combination of scattering and microscopic results, it is found that the TDs are not uniformly distributed in the ZnO films, but the ZnO films consist of columnar epitaxial coressurrounded by annular regions of edge threading dislocations at a large density. The shift of flatband voltage and the raise of potential barrier at the aggregation of TDs observed by scanning capacitance microscope and conduction atomic force microscope were attributed to the interface trap densities caused by the existence of high-density edge threading dislocations. On the other hand, because the distribution of the screw TDs is much less than that of the edge TDs, we cannot identify the location of the screw TDs and their electrical properties. The structural analysis of c-oriented ZnO epitaxial films on Si(111) substrates with a thin γ-Al2O3 buffer layer erveals that epitaxial γ-Al2O3 buffer layer consists of two (111) oriented domains rotated 60° from each other against the surface normal and the in-plane epitaxial relationship among ZnO layer, γ-Al2O3 buffer and Si buffer follows (1010)||{224}or {422}||{224}. Studies on the crystalline quality and optical properties of ZnO epi-layers by XRD and PL measurements clearly indicate the intensity ratio of deep-level emission (DLE) to near-band edge emission (NBE) of ZnO films correlates with the width of φ−scan across off-normal reflection and the NBE linewidth is strongly dependent on the width of ZnO (0002) rocking curve. These observations manifest that the (IDLE/INBE) ratio is dominantly affected by edge TDs and the line width of NBE emission is mainly related to screw TDs. Both high-quality structural and optical properties of ZnO epi-film on Si (111) substrates using a nano-thick high-k oxide Y2O3 buffer layer was verified by XRD, TEM, and PL measurements. The nano-thick Y2O3 epi-layer serves not only as a buffer layer to ensure the growth of ZnO epi-film of high structural perfection but also as an insulator layer between ZnO and Si. Determined by XRD and TEM the epitaxial relationship between ZnO and Y2O3 follows 32110)111(||0112)0001(OYZnO><><. ZnO lattice aligns with the hexagonal oxygen (O) sub-lattice in Y2O3 and the interfacial structure can be well described by domain matching epitaxy with 7 or 8 ZnO {0211} planes matching 6 or 7 {044} planes of Y2O3; the large lattice mismatch is thus accommodated by the misfit dislocations (MDs) localized at the interface with a periodicity of 6(7) times of )044(OY inter-planar spacing, leading to a significant reduction of residual strain. Superior photoluminescence were obtained even for ZnO-films as thin as 0.21 μm. Our results demonstrate that the Y2O3 layer well serves as a template for integrating ZnO based optoelectronic devices with Si substrate. Finally, the calculated TDs densities from XRD and TEM indicate most TDs are pure edge dislocations for ZnO epi-films on sapphire (0001), γ-Al2O3/Si(111), or Y2O3 /Si(111) substrates. The lattice of ZnO is always aligned with the hexagonal O sub-lattice in the oxide layer underneath. The lattice constant ao of 2D hexagonaloxygen sub-lattice are 2.75, 2.80, 3.75 Å for sapphire, γ-Al2O3□and Y2O3, respectively. As compared with the lattice constant a of ZnO (3.249 Å), compressive strain along in-plane direction is expected for ZnO epi-film grown on sapphire (0001) and□ γ−Al2O3(111) . In contrast, the expected lateral strain is tensive for ZnO epi-film on Y2O3(111). However, compressive lateral strain is only observed for ZnO epi-layers grown on sapphire. On both γ-Al2O3 and Y2O3 buffer layers, ZnO epi-films bear tensile strain. In fact, for ZnO epi-film grown on Si(111) using other oxide buffer layers, including Gd2O3, and Y2O3 doped HfO2, all ZnO epi-film suffers tensile strain along in-plane direction. Moreover, high density of MDs at ZnO/oxide-buffer interface should accommodate most of the strain caused by lattice mismatch. It is noted that the thermal expansion coefficient of ZnO (α ~4-6.5 × 10-6 K-1) is less than that of sapphire (8 × 10-6 K-1) but larger than that of Si (2.6-3.6 × 10-6 K-1). The trend agrees with the observed strain state of ZnO layer grown on sapphire and Si. This observation strongly suggests that the strain of the ZnO-epi layers is dictated by the thermal stress built up during the post-growth cooling. Because of the nano-thickness of the employed buffer layers, the influence coming from the buffer is negligible in these cases.