雷射濺鍍法成長單晶非極性氧化鋅薄膜在藍寶石基板之光學與晶體結構相關特性研究

Abstract

我們利用準分子雷射濺鍍的方法分別成長非極性的氧化鋅 與 的磊晶薄膜在 與 的藍寶石基板上，而這些磊晶面的關係分別如下所示： 與 。 非極性 氧化鋅的磊晶薄膜成長在 面的藍寶石基板上，會受到非等向性的應力作用造成晶格扭曲進而改變晶體的對稱性：由本來纖鋅礦C6V的對稱性變成一個斜方晶系C2V的對稱性。 從X光繞射分析可得知此非極性 氧化鋅的c軸是受到一個壓縮的應力，而另兩個方向都是拉伸的應力(y軸延表面方向，x軸是與c軸方向垂直且在in-plane上)。 當x與y軸的應力大小差距超過了0.6%，這樣非等向性應力大小足夠讓晶體結構從原本的纖鋅礦C6V的對稱性變成一個斜方晶系C2V的對稱性。利用不同偏極化的拉曼光譜、光學反射與螢光光譜，我們都可以觀察到晶體對稱性由纖鋅礦C6V改變到斜方晶系C2V的結果。 拉曼光譜上特定的振動模態並不滿足C6V的特性卻恰恰符合C2V對稱性下的振動模態；而光學反射與螢光光譜發現特別的躍遷能階E1與 E2，這個有別於纖鋅礦氧化鋅的躍遷是由於晶體結構因為非等向性應力作用而變形成斜方晶系C2V所造成的。 而成長在 藍寶石基板的 氧化鋅利用了額外的晶相 ，來降低應力對其磊晶薄膜造成的影響。 這些雜相 的a軸與主結構 氧化鋅的a軸是完全重疊在一起，像兩個原本 氧化鋅的晶體對應其 a軸往兩側各自旋轉約59°並沿著 氧化鋅的c軸排列；因此我們可以得這些不同晶相之間的磊晶面關係: 。 此外我們也發現這些雜相的多寡與螢光光譜上的寬頻譜有關，這個躍遷能量大約在3.17電子伏特。 這個躍遷是由於不同晶相間的界面雜質或者是缺陷束縛激子所造成的，我們把它稱作是表面束縛激子。 當雜相的數量增加的同時也增加的界面的數量，進而產生出更多的表面束縛激子在這些邊界上。 為了解決雜相帶來的表面束縛激子發光，我們利用兩階段成長的方式來消除這些雜相的產生。當低溫成長緩衝層厚度在47到67奈米之間，這些緩衝層能夠分攤足夠的應力而讓再成長於高溫下的氧化鋅會是單純的 晶面，但是如果低溫緩衝層的厚度超過或者是不足都會有雜相的產生。 利用兩階段成長所長出來的 氧化鋅只要在低溫時沒有觀察到雜相的出現都能有很好的表面平整度，我們也發現雜相的數量增加會增加薄膜的表面粗糙度，所以表面粗糙度也可以用來初步判斷是否有雜相的形成。 低溫光學頻譜下我們發現了除了一般氧化鋅的發光特性外還有一個很強的基面堆疊缺陷發光，這是由於在兩階段成長的 氧化鋅有較多的基面堆疊缺陷，而大量的基面堆疊缺陷分散了應力造成的影響並降低氧化鋅的應力作用。 所以兩階段成長確實能消除表面束縛激子發光與雜相的形成，並且能有效的降低應力對磊晶薄膜的影響也有較好的表面平整度。 我們也製作了非極性的量子結構，並證實不會有量子史塔克效應產生，光學特性主要都是受量子效應影響；這樣的結構也證實是比極性結構更適合應用在光學元件上。

The nonpolar - and -oriented ZnO films have been epitaxially grown by pulsed laser deposition (PLD) on the sapphire and substrates. The epitaxial relationship of nonpolar a- and m-plane ZnO on r- and m-sapphires are and , respectively. Crystal symmetry breaking of wurtzite C6V to orthorhombic C2V due to in-plane anisotropic strain was investigated for nonpolar ZnO epi-films grown on the r-sapphire. X-ray diffraction (XRD) results reveal the epi-layer is subjected to a compressive strain along the polar c-axis and tensile strain along both y- surface normal and in-plane x- axis. The strain difference between y- and x-axes is larger than 0.6% that introduces enough anisotropic strain to break the crystal symmetry from wurtzite C6V to orthorhombic C2V. The polarized Raman spectra of modes reveal violation of the C6V selection rules; oppositely, the C2V configuration satisfies the selection rules for the Raman modes. The observed E1 and E2 bands in polarized optical reflection and photoluminescence (PL) spectra, which are different from the typical ZnO for wurtzite structure, confirm the anisotropic strain causes the structure change to the orthorhombic one. In the m-plane ZnO films grown on m-sapphire, small amount of domains were found providing strain relaxation of the m-ZnO matrix. And the a-axes of both the domains and the m-ZnO matrix are aligned with the c-axis of the m-Al2O3 substrate. The c-axis of the domains rotates about □59° against the common a-axis of the m-ZnO. From this result, we found the epitaxial relationship of . Through carefully correlating low-temperature polarized PL spectra with the XRD peak intensity ratio of of the samples grown at different temperature and after thermal treatment, we found that the broad-band emission around 3.17 eV may result from the interface defects trapped excitons at the boundaries between the domains and the m-ZnO matrix. The more domains in the m-ZnO layer cause the more surface boundary that makes the stronger surface-bound-exciton emission. To eliminate the extra domain, we used the low-temperature (LT) grown buffer of m-ZnO to investigate the optical and crystalline properties. Examined by XRD, we found when the thickness of LT-buffer layer is below 67 nm it contain no any extra domains, however, there exist a lot of extra domains for the thickness above 156 nm. The amount of extra domains increases with decreasing the buffer thickness. The optimal thickness of LT-buffer is from 47 to 67 nm, in which no observable extra -domains present in the two-step m-ZnO epilayers. The AFM measurement also shows the lower surface roughness for the two-step growth m-ZnO than those without buffers grown at the same temperature. This characteristics benefit for fabricating quantum-well (QW) structures. The LT-PL spectra show the three emission peaks around 3.364, 3.328 and 3.263 eV, which are attributed to the emissions of donor-bound excitons, basal plane stacking faults (BSFs) and free electron bound to acceptor emissions, respectively. The BSFs emission due to high BSFs density of ~2x106 cm-1 by TEM measurement, this value is larger than the m-ZnO without LT-buffer. The high BSFs density should provide the way to relax the lattice strain. In addition, the LT-PL spectra indicate absence of the broad-band emission at 3.17 eV result from the domain boundary trapping between the m-ZnO and extra domains which is dominant in the m-ZnO without LT-buffer. Finally, 5 pairs of nonpolar m-plane ZnO/MgxZn1-xO quantum well structures were successfully grown on m-sapphire with LT m-ZnO LT-buffer. The results demonstrate these QW structures possess quantum confinement without experiencing the quantum confined Stark effect due to their nonpolar nature.