III族氮化物奈米粒成長與光學特性研究

全文

(1)國立交通大學電子物理研究所博士論文 III 族氮化物奈米粒成長與光學特性研究 The growth and optical studies of group III-nitride nanodots. 研究生：柯文政指導教授：陳衛國. 中華民國九十五年七月.

(2) III 族氮化物奈米粒成長與光學特性研究 The growth and optical studies of group III-nitride nanodots. 研究生：柯文政. Student：Wen-Cheng Ke. 指導教授：陳衛國. Advisor：Wei-Kuo Chen. 國立交通大學電子物理研究所博士論文. A Thesis Submitted to Institute of Electrophysics College of Science National Chiao Tung University in partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy in Electrophysics July 2006 Hsinchu, Taiwan, Republic of China. 中華民國九十五年七月. 2.

(3) 3.

(4) 4.

(5) III 族氮化物奈米粒成長與光學特性研究研究生：柯文政. 指導教授：陳衛國國立交通大學電子物理所. 摘要本論文初期我們透過微螢光激發光譜(μ-PL)分析氮化鋁鎵薄膜表面直徑約 6 μm 的 hillock 微結構光學特性。實驗結果顯示相關於 hillock 的發光譜峰為 351 nm (IH)，而在其周圍之氮化鋁鎵薄膜發光譜峰約為 341 nm (Imatrix)；IH 的強度隨著量測位置由邊緣往 hillock 中心移動時其強度明顯增強，半高寬由 76 meV 降低至 53 meV，此量測結果顯示 hillock 為一種高發光強度之微結構。從變溫 μ-PL 的量測顯示 IH 譜峰與溫度關係呈現 S 型之變化，其轉換溫度約為 120 K，此外紅位移亦較 Imatrix 來的小，這些結果均顯示出 hillock 相較於周圍之氮化鋁鎵薄膜而言，具有較低之鋁組成含量；另一方面，我們亦發現 hillock 之尺寸變大時(從 6μm 增加到 11μm)，其鋁組成約降低~2%。本論文提出一種特別的「流量調制磊晶法」製程技術用以成長 III 族氮化物奈米粒結構。首先，我們利用此技術成功地成長 GaN 奈米粒在低晶格不匹配度之 Al0.15Ga0.85N 緩衝層上，由奈米粒密度與成長溫度關係圖中，我們發現使用該方法成長之奈米粒在較低與較高成長溫度區間主要分別由反應物於薄膜表面之擴散機制與吸附機制主導；因為該方法在通入 TMGa 反應氣體時，是先形成 Ga 金屬，而 Ga 金屬與 Al0.15Ga0.85N 緩衝層晶格長數分別為 4.51 與 3.18Å，晶格不匹配度高達 41.8%，故此方法成長 GaN 奈米粒應為 Volmer-Weber 成長模式。另一方面，我們進一步深入研究 GaN 奈米粒之光學特性，針對平均高度分別為 6.5, 7, 8.5 nm 之奈米粒進行變溫螢光光譜研究，在 PL 峰點能量與溫度倒數關係圖中，使用 Varshini 方程式模擬實驗資料，我們得到低溫區之侷限能量隨著奈米粒尺寸縮小而減小；此結果與使用 Arrhenius 方程式模擬 PL 積分強度與溫度倒數關係圖所獲得之低溫區活化能結果一致；而高溫區之活化能亦隨著奈米粒尺寸縮小而減小，此高溫區活化能代表著奈米粒內之載子因溫度升高躍遷至氮化鋁鎵能障層氮空缺能階所需之能量。另一方面，我們亦成功地在 GaN 緩衝層上成長 InN 奈米粒，在覆蓋 GaN 披覆. i.

(6) 層後進行 PL 光學分析，當 InN 奈米粒高度降低至 6.5 nm 時，其 PL 峰點能量藍位移至 1.07 eV；在溫度相依之 PL 光譜量測中，我們發現到 PL 譜峰位置與量測溫度之關係似乎不依循 Varshini 方程式之預測曲線，相較於 InN 塊材而言，其發光波長具有較高的溫度穩定性。而在 PL 積分強度與溫度倒數關係圖得知，InN 奈米粒之高溫區活化能為 73 meV 明顯高於 InN 塊材之 43 meV，顯示奈米粒之 thermal quench 程度相較於塊材而言較弱，意即 InN 奈米粒具有更佳之光學特性。最後，我們分別比較使用流率調制磊晶(FME)與 MOCVD 兩種成長方法成長之 InN 奈米粒結構與光學特性，實驗結果顯示在 FME 成長方法中，In adatom 之擴散活化能明顯低於 MOCVD 成長方法；另外我們發現到使用 FME 成長 InN 奈米粒時，在 In-rich 的條件下成長之 InN 奈米粒擁有較佳的光學特性，當 NH3 流率調低至 500 sccm 時，在未覆蓋 GaN 披覆層下，已可以獲得 PL 光譜半高寬達 63 meV 之 InN 奈米粒。. ii.

(7) The growth and optical studies of group III-nitride nanodots Student : Wen-Cheng Ke. Advisor : Wei-Kuo Chen. Institute of Electrophysics National Chiao Tung University. Abstract In this thesis, the spatial variation of the optical properties of hillocks in Al0.11Ga0.89N films has been studied by using microphotoluminescences (μ-PL) microscopy. Theμ-PL spectrum revealed a strong emission peak (IH) at 351 nm from the hillock, besides the near-band-edge peak emission (Imatrix) at 341 nm. Moreover, the IH intensity increases significantly and its full width at half maximum (FWHM) decreases from 76 to 53 meV by probing across the hillock center. These indicated that the hillock is a strong emission structure. The temperature-dependent μ-PL measurements showed that the IH also has the S-shape behavior with a transition temperature of ~120 K which is lower than that of Imatrix. The redshift of IH is also smaller than Imatrix. Both indicated that the Al composition in hillocks is lower than the surrounding area. Moreover, we also observed that the Al composition decreased ~ 2% as the diameter of hillock increased from 6μm to 11μm. Otherwise, we proposed a new technique for fabrication III-nitride nanoparticles in flow rate modulation epitaxy (FME). Firstly, the self-organized GaN dot structure is successfully grown on a slightly lattice-mismatched Al0.11Ga0.89N epilayer using FME growth technique. From the variation of dot density with growth temperature, we can observe that the GaN dot growth is controlled predominately by the surface diffusion of Ga adatoms at substrate temperatures below 915 ℃. and by. re-evaporation at higher temperatures. Because of the special alternating gas supply feature in FME, during the Ga source step, it is the Ga metal that is deposited on the underlying Al0.11Ga0.89N layer. This is because of the large lattice mismatch of 41.8% between the Ga metal (4.51 Å) and Al0.11Ga0.89N (3.18 Å). We consider that the GaN dot growth in this study is mainly through the Volmer-Weber growth mode. Moreover, the temperature dependent PL studies showed that at low temperature the localization energy, which accounts for de-trapping of excitons, decreases with the iii.

(8) reducing dot size. The decrease in emission efficiency at high temperature is attributed to the activation of carriers from the GaN dot to the nitrogen vacancy (VN) state of the Al0.11Ga0.89N barrier layer. The activation energy decreases with reducing dot size. Secondly, the self-organized InN dots successfully grown on GaN epilayer using pulsed-growth mode growth technique. The PL properties of InN dots embedded in GaN were also investigated. We observed a systematic blueshift in the emission energy as the average dot height was reduced. The widely size-tunable emission energy can be ascribed to the size quantization effect. Temperature-dependent PL measurements show that the emission peak energies of the dots are insensitive to temperature, as compared with that of bulk film, indicating the localization of carriers in the dots. A reduced quenching of the PL from the InN dots was also observed, implying superior emission properties for the embedded InN dot structures. Finally, FME was also employed to synthesize self-assembly InN dots on GaN/sapphire substrate. Experimental results clearly indicate the adatom diffusion activation energy in FME is greatly reduced as compared to that in conventional growth method. We also demonstrate that InN dots prepared by FME under In-rich growth conditions possess much better optical quality than under N-rich growth conditions. Consequently, relatively high PL intensity with linewidth as narrow as 63 meV was realized for InN dots grown by FME at a NH3 background flow rate of 500 sccm, without any encapsulating layer.. iv.

(9) 誌謝 (Acknowledgements). 漫長的博士研究生涯終於劃下了句點，回首這六年的點滴，首先最要感謝的人是我的指導教授陳衛國老師，在他的諄諄教誨下，讓我學會處理事情所需要的企圖心、嚴謹態度與永不放棄的毅力；更感謝他在我面臨放棄博士學位時，適時給予我鼓勵與協助，讓我有繼續完成學業的勇氣與信心。也要感謝他在實驗研究與論文寫作上長期耐心的指導，不斷地啟發我深入發掘問題與解決問題，才使得本論文得以順利完成。另外，也要感謝陳文雄老師、李明知老師、周武清老師及張文豪老師在每一次的博士班會議中不厭其煩地提供各種不同的意見，啟發我對問題思考的全面性。也要感謝我的啟蒙老師黃柏仁教授在碩士班期間的提攜與指導。更要感謝晶元光電公司周銘俊副總於口試時提供許多寶貴意見。亦感謝已畢業學長歐震、潘永中、徐宸科、黃懷瑩、李文雄..等人，由於您們的關心與協助，才能讓我順利走完博士生涯。也要感謝博士班古慶順、李寧、陳京玉、蔡伩哲、傅振邦..等人及所有畢業與在學的碩士班學弟妹們，有您們的幫忙與討論，所有的實驗研究才得以順利完成。而我也要特別感謝中科院五所副組長林文仁博士，提供完善的 MOCVD 磊晶設備，在他身上我學到待人處事的包容心。也感謝小組長程ㄧ誠博士及劉書史、張國仁及李大青..等大哥們在院內的協助幫忙。特別要感謝藍文厚老師、姜崇義博士在我博士班期間的鼓勵與幫助。也要感謝林科均(林大哥)、林家慶對於 MOCVD 系統的維護與實驗上的協助，也感謝翁仁斌先生協助 SEM 拍攝。亦要感謝柯誌欣學長、陳文瑞學長、陳一塵與官大明在中科院內的幫忙與討論，能夠認識您們讓我在中科院的生活更加豐富。最後，我要感謝我的父母與家人這幾年來的關懷、包容與支持，讓我可以在無後顧之憂的環境下完成博士學位，特別要感謝妻子碧芳對我的鼓勵與諒解，這幾年來跟著我吃了很多苦，也要感謝已故的張榮富老師，教導我很多的生命觀念，謝謝您們大家!!. v.

(10) CONTENTS Abstract (Chinese). i. Abstract (English). iii. Acknowledgments. v. Contents. vi. List of Tables. viii. List of Figures. ix. Chapter 1 Introduction. 1. 1-1. Recent Researches of III-Nitride Micro-Structure. 1. 1-2. Low Dimensional Structures. 4. Chapter 2 Theoretical Backgrounds. 11. 2-1. Quantum Dots Growth Mechanisms. 12. 2-2. Photoluminescence (PL), Temperature Dependent of PL Spectra. 16. Chapter 3 Hillocks in Al0.11Ga0.89N Films. 25. 3-1. Experimental Details. 26. 3-2. Microstructure in AlGaN Films. 27. 3-3. Micro-Photoluminescence (μ-PL) Spectra of AlGaN Hillock. 28. 3-4. Temperature Dependent of μ-PL Spectra of AlGaN Hillock. 29. 3-5. Size Dependent of μ-PL Spectra of AlGaN Hillock. 31. 3-6. Conclusions of AlGaN Hillock. 32. Chapter 4 GaN Nanodots Growth. 44. 4-1. Experimental Details. 46. 4-2. Growth Temperature Effect. 49. 4-3. Size Control. 51. 4-4. Growth Mode. 52. 4-5. Conclusions of GaN Nanodots Growth. 54. Chapter 5 Optical Properties of GaN Nanodots. 63. 5-1. 65. Experimental Details. vi.

(11) 5-2. GaN QDs growth mechanism. 68. 5-3. Micro-Photoluminescence (μ-PL) Spectra of GaN Nanodots. 72. 5-4. Temperature Dependent of μ-PL Spectra of GaN Nanodots. 74. 5-5. Conclusions of Optical Properties of GaN Nonodots. 77. Chapter 6 Optical Properties of InN Nanodots. 93. 6-1. Experimental Details. 95. 6-2. Growth temperature of InN dots. 97. 6-3. Photoluminescence (PL) Spectra of InN Nanodots. 101. 6-4. Temperature Dependent of PL Spectra of InN Nanodots. 103. 6-5. Conclusions of Optical Properties of InN Nonodots. 105. Chapter 7 Enhanced photoluminescence of InN Nanodots by FME. 119. 7-1. Experimental Details. 121. 7-2. Growth temperature. 122. 7-3. NH3 flow rate in TMIn flow period. 127. 7-4. Conclusions of InN Nanodots growth by FME. 131 141. Chapter 8 Conclusions. vii.

(12) Table List Tab. 1-1. Summary of three major techniques to form nitride quantum dots.. 8. Tab. 2-1. Surface energy conditions of VW and SK growth modes.. 12. Tab. 2-2. Values of the parameters in Varshini’s equantion.. 18. Tab. 3-1. The detail growth conditions of AlGaN films.. 26. Tab. 4-1. The growth conditions of GaN dots on AlGaN films.. 48. Tab. 5-1. The growth condition of GaN QDs on AlGaN films.. 67. Tab. 5-2. Sample list of GaN QDs on AlGaN films. Tab. 5-3. The calculated surface energy and wetting layer thickness of GaN. 69 70. QDs on different Al composition of AlGaN films. Tab. 6-1. The detail growth conditions of InN dots on GaN films. Tab. 6-2. The quantitative structural properties and emission energies of InN. 96 114. QDs and bulk samples. Tab. 7-1. The detail growth conditions of InN dots grown by FME technique.. Tab. 7-2. The average height and diameter of InN dots grown at different 135 growth temperatures from 550 to 730℃.. viii. 122.

(13) Figure List Fig. 1-1. Carrier density at transparent condition plotted as a function of the. 10. ratio of effective mass of holes and that of electrons for quantum well (QW) lasers and quantum dot (QD) lasers. Upper curve: me = 0.22 m0, lower curve: me = 0.067 m0. Fig. 2-1. Schematic diagram of typical films growth Frank-van-der Merwe. 20. (FvdM) mode, Volmer-Weber (VW) mode and Stranski-Krastanow (SK) mode. Fig. 2-2. (a) schematic geometries of a InGaN strained film on GaN substrate. 21. for FvdM and SK mode (b) schematic geometries of a InGaN strained film on GaN substrate for FvdM and VW mode. Fig. 2-3. Model of the interface energy.. 22. Fig. 3-1. The optical microscope (OM) image of hillock in AlGaN films. The. 34. inset shows the 3D AFM images of hillock. Fig. 3-2. The room temperature µ-PL spectra taken at different locations on. 35. OM image. Fig. 3-3. The spatial µ-PL intensity distribution of Inbe and IH.. 36. Fig. 3-4. The spatial µ-PL full-width-half-maximum (FWHM) distribution of. 37. Inbe and IH. Fig. 3-5. Temperature dependent µ-PL measurements of AlGaN hillock from. 38. 10 to 300K. Fig. 3-6. Integrated PL intensity as a function of temperature for Inbe and IH.. 39. Fig. 3-7. Temperature dependence of the emission peak of Inbe and IH. Arrows. 40. indicate the transition temperature TC. Fig. 3-8. The optical microscope (OM) image of the different size hillock in. 41. AlGaN films. Fig. 3-9. The room temperature µ-PL spectra taken from different size of. 42. hillock. Fig. 3-10 The relation between Al composition and hillock size.. 43. Fig. 4-1. 57. A scheme showing the principle of the periodic flow-rate modulation epitaxy growth GaN dots on AlGaN films.. ix.

(14) Fig. 4-2. A time chart showing the modulation of reactant molar flow rate in. 58. the periodic flow-rate modulation epitaxy. Fig. 4-3. AFM images of GaN dots grown at (a) 840, (b) 870, (c) 900, (d). 59. 930, (e) 940 and (f) 960 ℃. Fig. 4-4. Dependence of average diameter and height of GaN dots on growth. 60. temperature. The inset shows island density as a function of reciprocal temperature. Fig. 4-5. AFM images of GaN dots grown by FME technique. The exposure. 61. times of TMGa are (a) 20, (b), 15 (c), 10 (d), 7 (e), 5 and (f) 0 s per cycle. Fig. 4-6. Dependence of average diameter and height of GaN dots on. 62. exposure time of TMGa. Fig. 5-1. Plane view image of a 5μm × 5μm AFM scan on (a) the. 80. Al0.11Ga0.89N buffer layer, the GaN QD samples grown on Al0.11Ga0.89N buffer layers with different FME cycles (b) 10, (c) 20 and (d) 30 cycles. Fig. 5-2. Room temperature μ-PL spectra of (a) Al0.11Ga0.89N epilayer, the. 81. GaN QD samples grown on Al0.11Ga0.89N buffer layers with different FME cycles (b) 10, (c) 20 and (d) 30 cycles. Fig. 5-3. The cross-section SEM images of (a) Al0.11Ga0.89N epilayer, the. 82. GaN with 100 FME cycles grown on Al0.11Ga0.89N buffer layers. Fig. 5-4. Plane view image of a 2μm × 2μm AFM scan on (a) the. 83. Al0.11Ga0.89N buffer layer, the GaN QD samples grown on Al0.11Ga0.89N buffer layers with different FME cycles (b) 10, (c) 7 and (d) 3 cycles. Fig. 5-5. Plane view image of a 5μm × 5μm AFM scan on (a) the. 84. Al0.11Ga0.89N buffer layer, the GaN QD samples grown on Al0.11Ga0.89N buffer layers with different NH3 flow rate during NH3 flow period of (b) 2.68×10-2, (c) 1.34×10-2, and (d) 4.46×10-3 mol/min. The TMGa flow rate was 1.99×10-5 mol/min. Fig.5-6. Plane view image of a 5μm × 5μm AFM scan on (a) the Al0.11Ga0.89N buffer layer, the GaN QD samples grown on Al0.11Ga0.89N buffer layers with different GaN coverage of (a) 2, (b) x. 85.

(15) Fig. 5-7 Fig. 5-8. Fig. 5-9. 5.5, (c) 7.3, (d) 8.2, (e) 9.1, (f) 10.9, and (g) 13.6 MLs. Dot density versus average coverage plot of the GaN QDs. The solid curve is a fit using equation (5-1). 3D image of a 5μm × 5μm AFM scan of the GaN QD samples grown on Al0.11Ga0.89N buffer layers with different height (a) 6.5 nm, (b) 7 nm and (c) 8.5 nm. Normalized μ-PL spectra at 10 K of (a) Al0.11Ga0.89N epilayer, and. 86 87. 88. the GaN QD samples grown on Al0.11Ga0.89N buffer layers with different GaN coverage of (b) 9.1 MLs, (c) 10.9 MLs, and (d) 13.6 MLs. Fig.5-10 Fig.5-11. Comparison with the PL experimental data and calculation derived from the disk-like rectangular box model. PL peak energy as a function of temperature for the 6.5 nm. 89 90. (triangles), 7.0 nm (circles), 8.5 nm (squares) GaN dot and the GaN epilayer (rhombus). The dashed lines are theoretical fits using Vashini’s equation. Fig.5-12. PL integrated intensity as a function of temperature for the dot. 91. height of 6.5 nm, 7.0 nm, 8.5 nm and the GaN epilayer. The dashed lines represent the fitting by using the Arrhenius equation. Fig. 5-13 Schematic band diagrams of GaN QDs with height of 8.5 nm (a) 10. 92. K and (b) >100 K. Fig. 6-1. AFM images of InN dots grown at (a)600, (b)625, (c)650, (d)700, 107 (e) 720 and (f) 750 ℃.. Fig. 6-2 Fig. 6-3 Fig. 6-4 Fig. 6-5 Fig. 6-6. Fig. 6-7. AFM images of InN dots grown by conventional MOCVD at (a)600, (b)650, (c)700, (d)715, (e) 730 and (f) 750 ℃. The QDs density depicted as a function of reciprocal temperature.. 108. Dependence of (a)average height and (b)average diameter of InN dots on growth temperature (by pulsed growth mode). Dependence of (a)average height and (b)average diameter of InN dots on growth temperature (by conventional MOCVD). The 17 K PL spectra of 600, 625, 650 and 700 ℃ growth. 110. 109. 111 112. temperature of InN QDs. Top insert shows the relation between FWHM and growth temperature, and down insert shows the relation between PL peak energy and dots height. Dependence of average diameter and height of InN dots on exposure 113 time of TMIn. xi.

(16) Fig. 6-8. AFM images of InN dots with height of (a) 32.4 nm (b) 24.2 nm (c). 115. 18.5 nm (d) 12.4 nm (e) 6.5 nm. Fig. 6-9. 17 K PL spectra of InN bulk and InN dots with height of (a) 32.4 nm (b) 24.2 nm (c) 18.5 nm (d) 12.4 nm (e) 6.5 nm. The insert shows the relation of peak energy and dots height. The electron effective mass of 0.042m0 (solid line) and 0.07m0 (dotted line) were assumed in this calculation. Fig. 6-10 Temperature dependence of the PL peak energy of InN dots and. 116. 117. bulk samples. Fig. 6-11 Temperature dependence of the PL integrated intensity of InN dots. 118. and bulk samples. Fig. 7-1. The typical gas flow sequence of the formation of InN dots for FME 134 and conventional MOCVD.. Fig. 7-2. 136. Fig. 7-3. The InN dots density depicted as a function of reciprocal temperature. The 10 K PL spectra of InN dots grown by (a) conventional MOCVD and (b) FME growth technique at growth temperature of 550 to 730℃, respectively.. 137. Fig. 7-4. The relations between density and aspect ratio of InN dots and NH3. 138. Fig. 7-5. Fig. 7-6. flow rate (r0) during TMIn period. 139 The 10 K PL spectra of InN dots grown by conventional MOCVD and FME technique with different NH3 flow-rate (r0) during TMIn flow period. (a)-(f) show typical surfaces of InN dots grown under the r0 values 140 of 0, 500, 1000, 5000 and 10000 sccm and conventional MOCVD growth technique, respectively.. xii.

(17) Chapter 1 Introduction 1-1 III-nitride micro-structures GaN and its alloys with Al and In recently became the fundamentally important materials for optoelectronics. This is due mainly to their direct bandgap feature which not only covers the whole visible spectrum, but also extends into both UV and infrared regions. At present, high brightness blue and green light-emitting diodes (LEDs) and low-power blue laser diodes (LDs) are commercially available. Nevertheless, the development of GaN-based technology was and still is, suffering from the lack of high quality, lattice-matched substrates, such as GaN, AlN or InN etc. Apparently, GaN and its alloy are much more ’tolerant’ of the presence of structural defects than the conventional III-V systems, stemming from low minority carriers’ diffusion length and immobility of structural defects in these strongly bonded crystals. Despite the remarkable progress being made in the last decade in growing morphologically good hetero-epitaxial GaN layers, the as grown nitride film still shows a very high density of diverse defects, in particular, dislocations, inversion domains (IDs), nano-pipes, stacking faults and V-shaped pits are of continued concern to crystal growers [1-6]. For high-power optoelectronic and elevated- temperature device applications, the present structural characteristics of hetero-epitaxial GaN seem to be insufficient. It has been shown that dislocations in GaN epitaxial layers do indeed have nonradiative and recombinative properties [7-10]. For example, there are three different types of threading dislocations, edge, mixed, and screw types. It is reported that threading dislocations having a screw-component burgers vector act as strong nonradiative centers in GaN epitaxial layers, whereas edge dislocations, which are the majority, do not act as nonradiative centers. For the V-shaped defects, it was found that the mechanism of V-defects -1-.

(18) formation strongly depends on the In and Al compositions in InxGa1-xN and AlxGa1-xN layers, respectively [11]. By increasing the In composition, the origin of V-defects is changed from the vertex of threading dislocations to the stacking mismatch boundaries induced by stacking faults and the three-dimensional island growth at the initial stage due to the large lattice mismatch. By increasing the Al composition, the origin of the V-defects also varied from the surface undulation due to the elastic misfit strain to the vertex of threading dislocations. Further study reveals that the V-defect is correlated with the localized excitonic recombination centers which giving rise to a long-wavelength shoulder in photoluminescence (PL) and CL spectra. Beside the dislocations and V-shaped defects studies, one of the structural features often reported in microscopic observations of the surface of GaN epitaxial films is the presence of growth hillocks with a wide range of sizes, typically several microns, with a hexagonal shape. Qian et al. [12] have reported observations relating the hexagonal hillocks to the presence of nanopipes emerging at their centers and concluded that nanopipes are the open cores of screw dislocations. Middleton et al.[13] in a CL study of pyramidal hillocks found that in the center only the yellow emission is present and suggested that it originates from defects associated with inversion domain boundaries. It must be pointed that most of the recent studies of hillocks were focused on GaN material, very few results on AlGaN material. It is known that due to its wide, direct bandgap and other superior characteristics, an AlGaN system is regarded as one of the most promising materials for application in UV opto-electronics and high-temperature, high power electronics. Thus, studies of AlGaN hillocks were important for getting better understanding of AlGaN based UV opto-electronic devices. Because the spatial resolution of CL was better than PL, the optical. -2-.

(19) properties of III-nitride microstructures were usually studied by CL system. However, the temperature dependent of CL spectra can’t achieved by CL equipment. In order to obtain. temperature. dependent. optical. equipment. for. observed. III-nitride. microstructures. We needed to improve the spatial resolution of PL system. Therefore, our laboratory has developed a UV to visible micro-photoluminescence (µ-PL) system. By using UV lens, the laser spot could be focused in diameter of 1.5 μm. We are able to study the spatial dependent photoluminescence of AlGaN hillocks using μ-PL system. For AlGaN hillocks, μ-PL spectra indicted there is a strong emission peak (IH) at 351 nm for hillock itself, in addition to peak emission at 341 nm from the matrix. The hillock-associated μ-PL intensity increases significantly as the probing spot scan from peripheral to the center, accompanied with the reduction of FWHM from 76 to 53 meV. The temperature-dependent μ-PL measurements showed that the IH also has the S-shape behavior with a transition temperature of ~120 K is lower than that of Imatrix(~150K). Despite of concurrent growth of AlGaN hillocks and matrix during sample preparation, further study by EDX clearly indicates the discrepancy in emission wavelength between AlGaN hillocks and surrounding matrix is mainly due to composition different in Al component. That is the composition of Al in hillocks is 6%, 6% lower than that in matrix which is beyond our expectations at the beginning. Moreover, we also observed that the Al composition decreased ~ 2% as the diameter of AlGaN hillock increased from 6μm to 11μm.. -3-.

(20) 1-2 Low-dimensional structures For the past few decades, low-dimensional quantum structures, such as quantum wells (QWs), quantum wires (QWRs), and quantum dots (QDs) have been attracting lots of interests due to their potential advantages compared with bulk materials. Among these, QDs are expected to be the most promising due to their unique electronic states, such as δ-function-like density of states, three-dimensional (3D) carrier confinement, etc. Due to their unique properties, the semiconductor laser with a QD active layer is expected to have ultra-low threshold current, reduced temperature sensitivity, narrower spectral linewidth, and high-modulation bandwidth, etc [14]. Furthermore, the semiconductor photodetector with QDs are also expected to have the sensitivity for the normally incident light, enhanced photoexcited carrier lifetime, reduced dark current, and higher electric gain. It was in 1982 that the concept of QDs was proposed by Arakawa and Sasaki for the first time as artificial atoms for semiconductor laser application [15]. Since then, there has been lots of research devoted to the realization of predicted potential advantages of QDs. However, it took about 10 years to realize the fabrication of the practical QD structures. In the 1990s, both selective growth and self-assembled growth technique without the formation of nonradiative defects were well developed. Particularly. The Stranski-Krastanow (SK) growth mode was very successful for InGaAs/GaAs systems [16]. As a result, lasers, detectors for both inter-band transitions have been successfully demonstrated using the InGaAs/GaAs QDs [17]. From the late 1990s this material system has been extended to other material system. Among those material systems, nitride semiconductors have recently received the most attention for applications to blue and ultraviolet (UV) light emitting devices, especially, due to the great impact of QDs in GaN-based LDs [18]. The improvement -4-.

(21) of threshold current due to the QDs can be more enhanced in wider bandgap semiconductors owing to the following reasons. Generally, threshold current of the QW lasers is increased if the effective mass of electrons me or the effective mass of holes mh to that of electrons me is larger. Unfortunately, in GaN-based semiconductors these are larger compared to GaAs-based semiconductors. To see this, we calculated the carrier transparent density ntr at conditions under which the material becomes transparent. The threshold current is described by the following equation: I th = γeV ntr τ. (1-1). where γ is a constant, the value of which is in the range of 1.2-1.5, V is the total volume of the active region, ntr is the carrier density at transparent condition, and τ is the carrier lifetime. Fig. 1-1 shows the calculated ntr as a function of mh/me for two effective masses of electrons (me=0.22m0 for GaN and 0.067m0 for GaAs) [19]. As indicated in this figure, ntr increases monotonically with the increase of mh/me. Moreover, ntr is also larger when mc becomes larger. This difference in ntr leads to a difference in the minimum threshold current density Jth; Jth of GaN-based QW lasers is ~ 1kA/ cm2, while that of GaAs-based QW lasers is ~100 A/cm2. On the other hand, it was suggested that, if QDs are used in the active region and the size of the QDs is small enough that the population of carriers in the higher subband can be ignored, the achievable threshold current Ith in both GaAs-based and GaN-based LDs is almost the same, about 100nA-1μA. Therefore, with the use of QDs, the threshold current density is reduced by a factor of 100 in GaN-based LDs, which value can compare to that in GaAs-based LDs, suggesting that the impact of QDs is much bigger in GaN-based LDs than in GaAs-LDs. The promising aspect in the applications of the nitride QDs is also confirmed in the emission mechanism of the InGaN active layer. InGaN active layers for blue and -5-.

(22) green LEDs are known to have excellent optical properties despite the presence of high-density defects. It is widely accepted that their high-luminescence efficiency is due to the carrier localization induced by the presence of compositional fluctuation and/or QD-like features usually in high In-content InGaN layers [20]. Achieving confinement in all three dimensions requires lateral control on the same 1-100 nm length scale, and is much more difficult. For example, standard UV photolithography cannot yet produce such small features. While more exotic lithographic techniques, such as x-ray lithography or electron beam writing, do have adequate resolution to pattern QDs, they are comparatively expensive and not yet commonly used. Since the late 1990s, several methods to form QDs using nitride semiconductors were suggested. They can be categorized mainly into three methods: (1) using SK growth method, (2) using “anti-surfactant”, (3) using selective epitaxy. One of the most attractive method for defect-free QD formation is the SK growth in lattice-mismatched semiconductor systems, widely used in the QD fabrication of many material systems, such as InGaAs/GaAs, InAs/InP, InP/GaAs, GaSb/GaAs, InSb/InP, SiGe/Si, ZnCdSe/ZnSe, etc. In the SK growth mode, the mismatched epitaxy is initially accommodated by biaxial compression in a layer-by-layer (2D) growth region, traditionally called as “wetting layer”. After the deposition of a few monolayer, the strain energy builds up. Then, the evolution to 3D islands becomes more favorable than the continued, strained planar growth. Such islands are referred to as self-assembled QDs. Nitride semiconductors can be epitaxially grown to form a strained heterostructure, which is indispensable for the SK growth mode. The lattice mismatch between AlGaN and GaN ranges from 0 to 2.4%, which is only to have two growth modes (the strained layer-by-layer 2D growth mode, the SK growth mode). However,. -6-.

(23) the lattice mismatch between InGaN and GaN ranges from 0 to 11.1%, which is sufficiently large to have all growth modes (the strained layer-by-layer 2D growth mode, the SK growth mode, and the relaxed 3D growth mode). By the proper combination of QD and substrate materials, the extent of strain in the film and subsequence QD growth behavior can be controlled. The nitride QDs were also demonstrated by using “anti-surfactant” [21]. The pretreatment of the growth surface by anti-surfactant, such as tetraethylsilane (TESi), were found to result in the 3D growth mode of the subsequent layer, even in the case with little lattice-mismatched. The presence of the wetting layer was not confirmed in the QDs grown by this method. Since this growth mode does not require lattice mismatch, the restriction in the choice of film/substrate combination can be somewhat relaxed. Finally, selective epitaxy is also employed to fabricate QDs in nitride materials [22]. Selective epitaxy attempted in growing nitride QDs is quite similar to the method of patterning QWs, that generally used in other III-V compound semiconductors. In this case, at first, pyramidal structures are grown on a patterned substrate using selective epitaxy. QDs are then grown on top of those pyramidal structures. The use of selective epitaxy in growing GaN QDs has several advantages overwhelming the method of patterning QWs, such as (1) the possibility of arbitrary lateral shape, size, and position realization, (2) general compatibility with the modern ULSI technologies in the patterning process of this method. However, there are some drawbacks, such as defects or damages induced by etching process. Besides the techniques described above, other novel techniques were also attempted to fabricate the nitride QDs, such as the nitridation of metallic droplets [23], the surface passivation and/or pretreatment [24], the laser ablation [25], the ion. -7-.

(24) implantation [26], the colloidal synthesis [27], etc. Table 1-1 summarizes the three major techniques for nitride QD formation and material systems explored.. Table 1-1: Summary of three major techniques to form nitride quantum dots. Technique. Research Group. Author. Growth system. Material system. SK mode. CEA-Grenoble. Daudin et al.. MBE. GaN/AlN, InGaN/GaN. CNRS. Damilano et al. MBE. Tokyo Univ.. Tachibana et al. MOCVD GaN/AlN/6H-SiC; InGaN/GaN. Seoul National Univ.. Kim et al.. MOCVD In-rich InGaN/GaN. Montpellier II Univ.. Briot et al.. MOCVD InN/GaN. Anti-surfactant RIKEN. Tanaka et al.. MOCVD GaN/AlGaN/6H-SiC. Selective growth. Tokyo Univ.. Tachibana et al. MOCVD InGaN/GaN; GaN/AlGaN. Tokushima Univ.. Wang et al.. GaN/AlN/Si; InGaN/GaN. MOCVD InGaN/GaN. In this thesis, we used a growth technique, namely “flow-rate modulation epitaxy growth (FME)” to synthesize self-assembled QDs. In this process, nanodots can be grown on buffer layers of almost any materials, giving a great versatility in material selectivity. In other words, by using the FME method, we are able to prepare self-assembled nanodots on any degree of lattice mismatched substrate, even on slightly mismatched or completely matched substrate. Consequently, due to the maskless feature of FME method, the fabrication cost of self-assembled nanodots can be effectively reduced. -8-.

(25) The FME included following procedures: at the first process where turned on the purge gas and modulated the first precursor to the lower first flow rate. Then the second precursor is supplied to the buffer layer to form a metal and metal-rich island on the buffer layer for the second process. A third process, where turn on purge gas again and modulated the first precursor to the higher second flow rate onto the buffer layer on which the metal-rich island is formed in the second process. The metal-rich island is formed in the III-nitride nanodots. In chapter 4, for the GaN nanodots, we have demonstrated that the GaN dots can be grown on a slightly lattice-mismatched Al0.11Ga0.89N epilayer using FME. Because of the alternating gas supply nature in FME, we consider that the dot growth studied here. is. mainly. via. the. Volmer-Weber. growth. mode,. not. through. the. Stranski-Kranstanow growth mode. Our results indicate that the FME growth technique is a very promising tool for preparing self-organized quantum dot structures for most practical devices due to the release of requirement of large lattice mismatch between the grown dot structure and substrate. In addition, in chapter 6, FME technique was also used for growth of self-assembed InN dots on GaN films. The aspect ratio (i.e. height/diameter) of InN dots is found to be decreased greatly by the use of FME technique, as compared with conventional MOCVD deposition due to enhanced migration of In species and hence reduced defects in the dots structure. Photoluminescence measurement also supports that the quality of InN dots grown by the FME method is superior to that of layers grown by continuous growth mode.. -9-.

(26) mh/me. Fig. 1.1 Carrier density at transparent condition plotted as a function of the ratio of effective mass of holes and that of electrons for quantum well (QW) lasers and quantum dot (QD) lasers. Upper curve: me = 0.22 m0, lower curve: me = 0.067 m0.. - 10 -.

(27) Chapter 2 Theoretical Backgrounds Historically, the growth of thin films has been categorized into three types：(1) Frank-Van der Merwe (FM) (2D layer-by-layer), (ii) Volmer-Weber (VW) (3D islands), and (3) Stranski-Krastanow (SK) (2D layer followed by 3D islands). In Fig. 2-1, we show three fundamental growth features in growing thin films materials. These modes are deduced from equilibrium considerations of the surface and interface energies of lattice matched systems. In most practical applications the epitaxy of semiconductors profits from the existence of conditions where the layer-by-layer deposition mechanism (FM) occurs. This mechanism is typically realized for nearly lattice-matched combinations (i.e. <1%) with high interfacial bond energies, low supersaturation to suppress 3D nucleation, such a deposition manner are particularly favorite at high temperature on a slightly lattice-mismatched substrate. On the other hand, the SK growth mode operates in relatively higher mismatched systems (i.e. ~2-10%) in which strained films can grow in registry with the substrate until reaching a critical thickness, tc. At this thickness the accumulated elastic strain energy initiates the formation of dislocations and the strain can be partially relaxed by the formation of a dislocation network and/or relieved partially through the formation of 3D islands. Combinations of highly mismatched (>10%) and dissimilar materials, Au/NaCl, on the other hand, preferentially crystallize in the Volmer-Weber mode, forming islands or clusters on the bare unwetted surface.. - 11 -.

(28) 2-1 Quantum Dots Growth Mechanisms The primary factors that determine the island growth manner of deposition of epitaxial film on substrate are the surface free energy of the substrate (σsubstrate), the surface free energy of the deposited film (σfilm), and the interface strain energy (σinterface). The interface strain energy includes the interface energy (σif) due to lattice mismatch between substrate and deposited film and the strain energy (σst) due to induced strain caused by wetting layer and island film itself, which is increased with the increasing thickness of wetting layer. Table 2-1 list the required surface energy conditions for the SK and the VW growth modes. For both island growth modes, the sum of the surface free energy of the deposited film and their interface strain energy has to be greater than the surface free energy of substrate.. Table 2-1: Surface energy conditions of VW and SK growth modes Surface energy conditions VW mode. σsubstrate<σfilm + σinterface, (σinterface=σif+σst(t), t<one lattice layer). SK mode. σsubstrate<σfilm + σinterface, (σinterface=σif+σst(t), t>one lattice layer). As shown in Fig. 2-2, the primary difference between these two modes is the thickness of wetting layer. If the required wetting layer thickness to produce island growth is greater than one lattice layer, the associated island growth mode is categorized into SK growth mode, if not, VW mode.. - 12 -.

(29) Surface energy On the surface between the vapor and solid phases, the surface energy per unit area γ s can be given approximately by：. γ s = (1 − w u)ΔH vo N 02 3. (2-1). where u is the number of nearest neighbors of an atom in the bulk of the crystal and w is the number of nearest neighbors in the crystal of an atom on the surface in question. So, w/u means the number of bonds which connect a surface atom to atoms in the crystal, and (1-w/u) means the number of dangling bonds of an atom on the surface.. ΔH v 0 is the enthalpy of evaporation of the material, and N0 is the number of atoms per unit volume. The argument used is that the surface energy is the energy to break all of the nearest-neighbor bonds across a given plane. The number of atoms per unit surface area Ns can be related to N0 as follows: N s = N 02 3. (2-2). for a III-V wurtzite-type compound, Ns is given by: Ns =. 2 3a 2. (2-3). for the (0001) face, where a is the lattice constant of the III-V binary compound. For growth from vapor, ΔH v 0 is given by the enthalpy of evaporation per mole ΔH as follows:. ΔH v 0 =. ΔH 2N A. (2-4). where NA is Avogadro’s number (NA = 6.023 × 1023). ΔH is estimated using Stringfellow’s model written by :. ΔH = Ka −2.5. (2-5). a is the lattice constant of the binary compound and the value of K is 1.15×107 - 13 -.

(30) cal/mole-Å2.5. Therefore, the surface energy σ can be written as follows:. σ = (1 − α )Asγ s =. (1 − α )As ΔH. (2-6). 4 3a 2 N A. for the (0001) face where w is 3 and u is 4, As is the surface area of unit cell (here QD), and α is the reconstruction ratio of dangling bonds on the surface. A part of dangling bonds on the surface makes bonds with each other on the surface. The number of dangling bonds decreases as α increases.. Interface energy In order to calculate roughly the energy σif of the interface between the film and the substrate (FM and SK modes) or that between the cluster and the substrate (the VW mode), the bonding ratio at the interface was calculated. When the lattice constant of the film and cluster a is larger than that of the substrate asub (a>asub), a is given by a = kasub ,. k = Δa a + 1 = ( a − asub ) / asub + 1. ( k ≥ 1). (2-7). where Δa / a means the lattice misfit between the film (or cluster) and the substrate. In this calculation, as shown in Fig. 2-3, we assumed that at the interface all dangling bonds on the film and cluster side are bonding with dangling bond on the substrate side. At the interface, assuming the bonding ratio β 1 on the upper film or cluster side is given by. β1 = 1 , and the bonding ratio β 2 on the substrate side can be expressed by. β2 =. 1 , k. - 14 -.

(31) where this means that the fraction 1/k of the dangling bonds are satisfied. The interface energy per unit area γ i can be given by. ⎛. 1. ⎞. γ i = (1 − β 1 )γ s + (1 − β 2 )γ sub = ⎜⎜ 1 − ⎟γ sub Δa a + 1 ⎟⎠ ⎝. (2-8). where γ sub is the surface energy per unit are of the substrate. Therefore, the interface energy σif can be written as follows:. σ if = Aiγ i = Ai (1 −. 1 ΔH ) Δa / a + 1 4 3a 2 N A. (2-9). Strain energy For the calculation of the strain energy, the method developed by Nakajima et al., is used in order to calculate the precise stress distribution of an island on top of a substrate. In this method, the island is divided into many imaginary thin layers with coherent interfaces. Shear lag analysis is then used to calculate the longitudinal stress distribution over the imaginary thin layers in the FM, SK, and VW mode structures. The total strain energy of each structure σst which includes the strain energy of the layer, cluster, buffer layer, and substrate is given by m. m. i =1. i =1. σ st = ∑ U i = ∑. Ai d iσ i2 2Ei. (2-10). where Ui is the elastic strain energy in the ith imaginary thin layer, m is total number of thin layers which constitute each structure, and σ i , E i , Ai and d i are the stress (N/m2), Young modulus (GPa), surface area (m2), and thickness (m) of the ith thin layer. Thus, the growth mode dependent part of the free energy σ f can be given as:. σ f = σ + σ if + σ st . All the surface energy calculation will be described in chapter 4. - 15 -.

(32) 2-2 Photoluminescence (PL), Temperature Dependent of PL Spectra The photoluminescence (PL) was used for measurement of the optical properties of GaN and InN nanodots in this thesis. It was known that the PL is a powerful and non-destructive technique to probe the optical emission properties of materials, especially in luminescent semiconductors. The auto mapping PL system is also widely used in the industry to monitor the quality od semi-finished devices on wafer. By analyzing the PL spectra, a set of characteristic spectral features can identify the impurity types, the band gap energy, one can estimate the contents in compound semiconductors. PL analysis can also survey buried interfaces of heterostructures which are difficult to be probed by direct physical and electrical contacts. However, it is difficult to find the correlation between the spectral line intensity and concentration of the specific impurity, due to variation of non-radiative recombination through deep-levels or surface recombination centers. The luminescence process typically involves three steps: excitation, thermalization, recombination. The electron-hole pairs generated by incident light, through quasi-thermal equilibrium distribution, will recombine and produce photos. The impurities and defects can form various energy levels in the band gap, and the corresponding energies will be estimated by radiative recombination process or absorbed by non-radiative recombination process. The transition rates of these impurities are different due to various matrix elements and density of states at respective energy levels. The luminescence of semiconductors can be divided into three regions: the excitonic edge emission, the donor-acceptor pair emission, and deep level relate emission.. A relation for the variation of the energy gap (Eg) with temperature (T) in semiconductors is proposed [28]: - 16 -.

(33) E g = E 0 − αT 2 /( β + T ). (2-11). where Eg is the energy gap which may be direct or indirect, E0 is its value at 0 K, α and β are constants. Most of the variation in the energy gap with temperature is believed to arise from the following two mechanisms: (1) A shift in the relative position of the conduction and valence bands due to the temperature-dependent dilatation of the lattice. Theoretical calculations show that the effect is linear with temperature at high temperatures. In that region this effect accounts for only a fraction (about 0.25) of the total variation of the energy gap with temperature. At low temperature the thermal expansion coefficient is nonlinear with T; indeed for a number of diamond structure solids it even becomes negative over a certain temperature interval. Correspondingly the dilatation effect on the energy gap is also nonlinear. (2) The major contribution comes from a shift in the relative position of the conduction and valence bands due to a temperature-dependent electron lattice interaction. Theoretical treatments show that this leads to a temperature dependence of the following form:. T << θ ,. ΔE g ∝ T 2. T >> θ ,. ΔE g ∝ T. where θ is the Debye temperature. Eq. (1) is consistent with the theoretical results if we assume that β ≈ θ . The constants in eq. (1) were evaluated from the experimental data for a member of semiconductors and are recorded in Table 2-2.. - 17 -.

(34) Table 2-2: Values of the parameters in Varshini’s equantion Substance. Values of the parameters in Varshini’ equation α (×10-4) E0 (eV) β. Debye θ. Diamond. 5.4125. -1.979. -1437. 2220. Si. 1.1557. 7.021. 1108. 645. Ge. 0.7412. 4.561. 210. 374. SiC. 3.024. -0.3055. -311. ~1150. GaAs. 1.5216. 8.871. 572. 344. InP. 1.4206. 4.906. 327. 301. InAs. 0.426. 3.158. 93. 248. AlN. 6.23. 1.799. 1462. 800. GaN. 3.507. 0.909. 830. 570. InN. 0.69. 0.414. 454. 370. ( K) 0. Generally, bandgap energies of semiconductor decreases with increasing temperature, namely, E 0 (T ) = E 0 − αT 2 /( β + T ) , where α and β are known as Varshini’s fitting parameters. For wurzite GaN, it has been reported that α=0.909 meV and β=830K for the temperature variation of A-exciton transition. This shows that the temperature-induced red shift of transition energies is about 60 meV between 0K and 300K. The temperature dependence of the PL integrated intensity of films was generally sued to identify the mechanism of the PL quenching. In the bulk GaN (D0,X) process, the thermal dissociation of donor bound excitons involves two activation energies, namely the localization energy and exciton binding energy [29-30]. At low temperature, the free excitons tend to localize at the neutral donors (localization energy) so that there is interplay between the localization and ionization of neutral donors to reduce the number of available neutral donors. In this study, we also use the following formula to fit our results,. - 18 -.

(35) I (T ) =. I ( 0) Ea E loc ⎤ ⎡ ⎢1 + C1 exp( − kT ) + C 2 exp( − kT )⎥ ⎣ ⎦. (2-12). where I(T) and I(0) are the integrated intensity at temperatures T and 0 K. C1 and C2 are fitting parameters. Ea and Eloc are the activation energies at the high and low temperature regime, respectively. The localization state may be one of below cases: (1) Interface fluctuations particularly in narrow QWs, produce localization states for excitons and that such localization states act like QD. (i.e. localization of excitons in either QDs or potential minima in QWs). (2) Localization of the free exciton at neutral donors. (3) Detrapping of excitons from interface roughness fluctuation. (4) Potential fluctuations can be realized, for example, by CdSe islands with size fluctuations etc. For AlGaN/GaN MQWs case, From N Grandjean et al. report [31], taking into account the competitive non-radiative channel introduced by the dislocations, the PL intensity is given by Eq (2-12). The best fit of the experimental data is obtained with Ea1 = 125 meV and Ea2 = 12 meV. The 12 meV activation energy is consistent with the localization energy and accounts for the detrapping of excitons from interface roughness fluctuation. The activation energy of 125 meV corresponds to the carrier thermal escape out the well.. - 19 -.

(36) Frank- van der Merwe. Volmer-Weber. Stranski-Krastanow. d. - Layer by layer growth. - Direct island growth. - Lattice matched. - Large lattice mismatched. - GaN on AlGaN. - GaN on sapphire. - Layer by layer followed by island nucleation - Dissimilar lattice spacing. Fig. 2-1 Schematic diagram of typical films growth Frank-van-der Merwe (FvdM) mode, Volmer-Weber (VW) mode and Stranski-Krastanow (SK) mode.. - 20 -.

(37) (a). (b). 1 lattice layer. Fig. 2-2. (a) schematic geometries of a strained film on substrate for (a) SK mode. and (b) VW mode.. - 21 -.

(38) Film or cluster side a β1=1 Interface Dangling bond. β2=1/k asub. Fig. 2.3. Substrate side. Model of the interface energy.. - 22 -.

(39) References [1] B. Pěcz and Zs. Makkai, Appl. Phys. Lett., 78, 1529 (2001) [2] L. T. Romano and T. H. Myers, Appl. Phys. Lett., 71, 3486 (1997) [3] J.L. Weyher, P.D. Brown, A.R.A. Zauner, S.MuKller, C.B. Boothroyd, D.T. Foord, P.R. Hageman, C.J. Humphreys, P.K. Larsen, I. Grzegory, S. Porowski, J. Cryst. Growth, 204, 419 (1999) [4] W. Qian, G. S. Rohrer, and M. Skowronski, Appl. Phys. Lett., 67, 2284 (1995) [5] R. Liu, A. Bell, F. A. Ponce, C. Q. Chen, J. W. Yang, and M. A. Khan, Appl. Phys. Lett., 86, 21908 (2005) [6] S. K. Hong, T. Yao, B. J. Kim, S. Y. Yoon, and T. I. Kim, Appl. Phys. Lett., 77, 82 (2000) [7] Sergey Yu. Karpov and Yuri N. Makarov, Appl. Phys. Lett., 81, 4721 (2002) [8] S. F. Chichibu, M. Sugiyama, T. Onuma, T. Kitamura, H. Nakanishi, T. Kuroda, A. Tackeuchi, T. Sota, Y. Ishida, and H. Okumura, Appl. Phys. Lett., 79, 4319 (2001) [9] S. J. Rosner, E. C. Carr, M. J. Ludowise, G. Girolami, and H. I. Erikson, Appl. Phys. Lett., 70, 420 (1997) [10] E. F. Schubert, I. D. Goepfert, W. Grieshaber, and J. M. Redwing, Appl. Phys. Lett., 71, 921 (1997) [11] H. K. Cho and J. Y. Lee, Appl. Phys. Lett., 80, 1370 (2002) [12] W. Qian, G. S. Rohrer, and M. Skowronski, Appl. Phys. Lett., 67, 2284 (1995) [13] P. G. Middleton, C. Trager-Cowan, A. Mohammed, K. P. O’Donnell, W. Van Der Stricht, I. Moerman, and P. Demeester, Mater. Res. Soc. Symp. Proc. 449, 471 (1997). [14] P. Caroff, C. Paranthoen,a_ C. Platz, O. Dehaese, H. Folliot, N. Bertru, C. Labbé, R. Piron, E. Homeyer, A. Le Corre, and S. Loualiche, Appl. Phys. Lett. 87, 243107 (2005). [15] Y. Arakawa and H. Sakaki, Appl. Phys. Lett. 40, 939 (1982). [16] G. Saint-Girons, G. Patriarche, L. Largeau, J. Coelho, A. Mereuta, J. M. Moison, J. M. Gérard, and I. Sagnes, Appl. Phys. Lett. 79, 2157 (2001). [17] Kazunari Ozasa, Yoshinobu Aoyagi, Young Ju Park, and Lars Samuelson, Appl. Phys. Lett. 71, 797 (1997). [18] Yukio Narukawa, Yoichi Kawakami, Mitsuru Funato, Shizuo Fujita, Shigeo Fujita, and Shuji Nakamura, Appl. Phys. Lett. 70, 981 (1997). - 23 -.

(40) [19] T. O. Poehler, Appl. Phys. Lett. 20, 69 (1972). [20] Yen-Lin Lai, Chuan-Pu Liu, and Zheng-Quan Chen, Appl. Phys. Lett. 86, 121915 (2005). [21] Satoru Tanaka, Sohachi Iwai, and Yoshinobu Aoyagi, Appl. Phys. Lett. 69, 4096 (1996). [22] Tachibana, K.; Someya, T.; Ishida, S.; Arakawa, Y., J. Cryst. Growth, 237, 1312 (2002) [23] C.-W. Hu, A. Bell, F. A. Ponce, D. J. Smith, and I. S. T. Tsong, Appl. Phys. Lett. 81, 3236 (2002). [24] J. Zhang, M. Hao, P. Li and S. J. Chua, Appl. Phys. Lett. 80, 485 (2002). [25] Timothy J. Goodwin, Valerie J. Leppert, Subhash H. Risbud, Ian M. Kennedy and Howard W. H. Lee, Appl. Phys. Lett. 70, 3122 (1997). [26] E. Borsella, M. A. Garcia, G. Mattei, C. Maurizio, and P. Mazzoldi, J. Appl. Phys. 90, 4467 (2001). [27] O. I. Mi i , S. P. Ahrenkiel, D. Bertram, and A. J. Nozik, Appl. Phys. Lett. 75, 478 (1999). [28] Varshni Y P, Physica 34, 149 (1967) [29] K. O’Donnell et al., in Group III Nitride Semiconductor Compounds: Physics and Applications, edited by B. Gil (Clarendon Pr, New Zealand, 1998), pp. 247-248. [30] A. Bimberg and M. Sondergeld, Phys. Rev. B4, 3451 (1971). [31] N Grandjean1, Semicond. Sci. Technol. 16, 358 (2001).. - 24 -.

(41) Chapter 3 Hillocks in Al0.11Ga0.89N Films AlGaN is a wide bandgap semiconductor with many important applications including high-temperature and high-power electronics, solar-blind photodetectors, blue and ultraviolet light emitting and laser diodes [1-3]. However, AlGaN grown on sapphire substrate often contains a large amount of dislocations due to lattice mismatch. Even with recent improvements, as due to introduction of low temperature AlN or GaN buffers, AlGaN films grown in this fashion still consist of a mosaic of slightly misoriented domains of micro size with lots of microstructure like dislocations, grain boundaries, V-defect, hillocks, pores… etc. [4-5]. Thus, the detailed analysis of the spectra and spatial distribution of microstructure is very important to improve material quality. Nevertheless, microstructures sometimes show particular optoelectronic properties. For example, the hexagonal hillocks with dimensions of several microns are often observed on the surface of GaN layers. They are characterized by a high intensity of the band gap luminescence, while the boundaries of the hillocks are enriched with defects responsible for yellow luminescence at ~2.2 eV [6-7]. So far, the light emission from microstructures is commonly studied by using cathodoluminescence (CL), because the nano-probe produces secondary electron images in CL with high spatial resolution, as well as point analysis on specific areas. However, no report about micro-photoluminescence (µ-PL) studies of the microstructures of III-V nitride semiconductor appeared up to now. In this Letter, the spatial variation of μ-PL was studied to characterize the optical properties of AlGaN hillocks. The temperature dependent µ-PL spectra from 10K to 300K were obtained to show the interesting emission behaviors of the hillock.. - 25 -.

(42) 3-1 Experimental details The AlGaN films were grown on AlN/sapphire (0001) substrates by low-pressure metalorganic vapor phase epitaxy (MOVPE) system in a horizontal reactor. Trimethylgallium, trimethylaluminum, and ammonia were used as the source precursors for Ga, Al, and N, respectively. Hydrogen was used as the carrier gas. Prior to material growth, the sapphire substrate was annealed to remove any residual impurities on the surface in a H2 ambient at 1120 ℃ for 10 min. A nominal 25 nm thick AlN nucleation layer was deposited at 650℃. The substrate temperature was then raised to 1120℃ to grow a 0.8 μm Al0.11Ga0.89N layer. Other details growth conditions of AlGaN films were described in Table 3-1. For μ-PL measurements, a He-Cd laser (Omnichrome 2074) operating at 325 nm was used for above bandgap excitation and was focused to a spot size of 1.5 μm by a microscope objective (100×, 0.5 numerical aperture). The signals were collected by the same objective lens into a monochromator (ARC-500) with both the entrance and exit slits opened to about 50 μm so that the spectral resolution is about 0.2 nm.. Table 3-1: The detail growth conditions of AlGaN films. Time. Temperature. Pressure. NH3. TMGa. TMAl. (min). (℃). (mbar). (mol/min). (mol/min). (mol/min). Desorption. 10. 1120. 200. -. -. -. Nucleation. 3. 650. 100. 8.93×10-2. -. 3.05×10-5. Annealing. 1. 1120. 50. 8.93×10-2. -. -. AlGaN. 60. 1120. 50. 8.93×10-2. 4.42×10-5. 6.98×10-6. - 26 -.

(43) 3-2 Microstructure in AlGaN Films Optical examination of the Al0.11Ga0.89N layers revealed particular hillocks with a density of 1×106 cm-2 covering the substrate surface in Fig. 3-1. The base size of the hexagonal hillock is about 6 µm after 1 hour growth. From the three-dimensional (3D) atomic force microscopy (AFM) image (see inset), the hillock has the shape of a regular point-topped pyramid with a height about 200nm.. 3-3 Micro-Photoluminescence (μ-PL) Spectra of AlGaN Hillock A series of µ-PL spectra were taken at different locations along a dihedral direction across the hillock as shown in Fig. 3-2. The position label indicates the approximate distance from the hillock center. The spectra are dominated by the near-band-edge emission (Imatrix) at 341 nm as the probe spot is far away from the hillock. When it is focused on the hillock, the most significant change in the µ-PL spectra is the appearance of an extra peak (IH) at 351 nm. Obviously, this strong and prominent emission is related only to the hillock. Note that, although the Inbe is still present at 341 nm, it is so weak that is submerged in the strong and broad 351 nm band. Hoffmann et al. [8] reported that CL spectra showed the band-gap gradient from the base to the top of the selective growth GaN pyramids of 5 µm width and 10 µm height. In his report, the emission peak is 355.6 nm at the top of the pyramid and strongly red-shifted to 360.6 nm at the pyramid base. The different emission energies reveal the gradual relaxation of strain along the pyramid. However, no peak shift of IH from the hillock edge to the center was observed in our study. Thus, it suggested that the stress was considerably small in AlGaN hillock which has a rather flat top region - 27 -.

(44) in contrast to the steep pyramid. The spatial µ-PL intensity and full-width-half-maximum (FWHM) distribution of the Imatrix (341 nm) and IH (351 nm) are shown in Figs. 3-3 and 3-4, respectively. The IH intensity at the hillock center is about five times larger than the Imatrix intensity far away from the hillock. We noticed that the FWHM of Imatrix obtained far away from the hillock is ~77 meV that is close to the report of Kim et al. [9] Moreover, the FWHM of IH decreased from ~76 meV at the hillock edge to ~53 meV at the hillock center. The high intensity and narrow FWHM of IH indicated that the hillock structure is an efficient emission center. From the 3D AFM image, a nipple structure appeared on the top of the AlGaN hillock (see the inset of Fig. 1a). Since the quantum dot (QD)-like structure was observed on the top of AlGaN/GaN selectively grown pyramid [10], it suggested that the QD-like structure is likely formed on the hillock top.. - 28 -.

(45) 3-4 Temperature Dependent of μ-PL Spectra of AlGaN Hillock. To further examine optical properties of the AlGaN hillock, we also carried out temperature dependent µ-PL measurements from 10K to 300K as shown in Fig. 3-5. One clearly sees that the Imatrix photon energy not only decreases but also increases with temperature and forms a “S-shape” curve; so does the IH. It was known that the “S-shape” is due to localization phenomena induced by alloy inhomogeneous in the AlGaN films [11]. From the Arrhenius plot for the luminescence intensity decrease of the donor bound exciton (D0,X) of GaN film with increasing temperature [12]. The thermal dissociation of donor bound excitons involve two activation energies, namely the localization energy and donor binding energy. At low temperature, the free excitons tend to localize at the neutral donors (localization energy). However, at elevated temperature there is an interplay between the localization and ionization of neutral donor to reduce the number of available neutral donors. In contract, one activation energy was estimated from the PL emission intensity as a function of temperature of AlGaN films [11]. It was indicated that this activation energy implies impurity binding energies or carrier/exciton localization energies. Moreover, Cho et al. [13] also indicated that an activation energy estimated from the relationship I(T)=I(0)/[1+A exp(-Ea/kT)] in the low transition temperature region corresponds to the magnitude of effective potential fluctuations due to alloy inhomogeneous. Therefore, in order to understand the localization phenomena in AlGaN hillocks, we also finished the Arrhenius plots of the integrated PL intensities over the temperature range of 10-300 K shown in Fig. 3-6. The following formula was used to fit our results,. I (0) ⎛ ΔE a 1 ⎞ ⎛ ΔE a 2 ⎞ = 1 + C1 exp⎜ − ⎟ + C2 exp⎜ − ⎟ I (T ) ⎝ kT ⎠ ⎝ kT ⎠ - 29 -. (3-1).

(46) where ΔEa1 is the donor binding energy, ΔEa2 is the localization energy , C1 and C2 are fitting constant. From the slope, ΔEa1 is calculated to be 33.2 meV and ΔEa2 is 15meV for Imatrix, and ΔEa1 is 43.6 meV and ΔEa2 is 11.4 meV for IH, respectively. The temperature dependence of the emission energy is then shown in Fig. 3-7, for analyzing the localization energy. The expected temperature dependence (dashed lines) was calculated by using the Varshni’s equation [14], E g = E0 −. αT 2 β +T. (3-2). with α=6.31×10-4 eV/K and β=2584 K for Inbe, and α=21.2×10-4 eV/K, β=606 K for IH. The observed μ-PL temperature dependence follows Eq.(1) at high temperatures and deviates from it below a transition temperature Tc. The transition temperatures are 150 and 120 K for Imatrix and IH, respectively, which are both higher than another report for the Al0.11Ga0.89N films [9]. Nevertheless, the transition temperatures of 150 and 120 K are close to the localization energy of 15 and 11.4 meV. They reflect that the localization effect is strong in our sample. Moreover, the transition temperature of Imatrix line was higher than that of the IH line. The red-shift of the Imatrix (~ 35.8 meV) at low temperature relative to the dashed line is larger than that of IH (~17.6 meV). It has been reported that the increase of transition temperature (TC) and red-shift are due to the localization energy which increases with the Al content [9]. Our experimental results also support that the Al composition in the hillock is smaller than surrounding area in the Al0.11Ga0.89N films. The recent x-ray energy dispersive spectroscopy observations (not shown) also confirm this. Thus, the hillock microstructure has lower Al composition than the regions free of hillocks in the AlGaN films.. - 30 -.

(47) 3-5 Size Dependent of μ-PL Spectra of AlGaN Hillock. The OM images of different diameter of hillocks which ranges from 6 to 11μm are shown in Fig. 3-8. Their μ-PL spectra at the hillocks apex were shown in Fig. 3-9. It was observed that the μ-PL peak energy red-shifted from 3.537 eV to 3.509 eV as the diameter increased from 6 to 11 μm. The Al composition estimated from μ-PL spectra for various diameter hillock is also shown in Fig. 3-10. The Al composition of AlGaN films was calculated as below: From μ-PL spectra, the peak position can be converted into photon energy as : E g ( eV ) = 1240 λ ( nm ) . By substituting it into the following empirical formula, one can obtain Al composition ‘x’:. E gAl xGa1− x N ( x ) = E gGaN (1 − x ) + E gAlN x − bx (1 − x ) where. the. following. constants. were. used. for. calculation.. E gGaN = 3.42eV , E gAlN = 6.2eV and bowing parameter b is 1.3. We also shown the Al. composition of 10 μm hillock which measured from EDX. The result was agreed with the μ-PL measurement. Since Inbe peak energy is 3.625 eV around the hillock matrix region, the Al composition is ~11%. By increasing the hillock diameter from 6 μm to 11 μm, the red shift is from 88 meV to 116 meV. It was indicated that the Al composition decreases progressively while the hillock size increase. The Al composition variation is about ~2% in hillock. - 31 -.

(48) 3-6 Conclusions of AlGaN Hillock In summary, we have measured the optical properties of hillocks in Al0.11Ga0.89N film by using the µ-PL microscopy. The large intensity and narrow FWHM of IH in the hillock structure indicated that it is a strong emission center. The temperature dependent µ-PL spectra showed that the IH has the S-shape behavior with a transition temperature of 120 K reflecting the strong localization in the hillock. The lower transition temperature and smaller red-shift of IH than that of Imatrix suggest that the Al composition is lower in hillock than in other parts of AlGaN film. Moreover, by increasing the hillock diameter from 6 μm to 11 μm, the red shift is from 88 meV to 116 meV. It was indicated that the Al composition decreases progressively while the hillock size increase. The Al composition variation is about ~2% in hillock.. - 32 -.

(49) References [1] S. Makamura, M. Senoh, S. Nagahama, N. Iwasa, T. Yamada, T. Matsushita, Y. Sugimoto, and H. Kiyoku, Appl. Phys. Lett. 69, 4056 (1996). [2] M. Asif Khan, X. Hu, A. Tarakij, G. Simin, J, Yang, R. Gaska, and M. S. Shur, Appl. Phys. Lett. 77, 1339 (2000). [3] E. Monroy, M. Hamilton, D. Walker, P. Kung, F. J. Sánchez, and M.Razeghi, Appl. Phys. Lett. 74, 1171 (1999). [4] L. Chang, S. K. Lai, F. R. Chen, and J. J. Kai, Appl. Phys. Lett. 79, 928 (2001). [5] B. Pécz, Zs. Makkai, M. A. di Forte-Poisson, F. Huet, and R. E. Dunin-Borkowski, Appl. Phys. Lett. 78, 1529 (2001). [6] F. A. Ponce, D. P. Bour, W. Götz, and P. J. Wright, Appl. Phys. Lett. 68, 57 (1996). [7] A. Y. Polyakov, A. V. Govorkov, N. B. Smirnov, M. G. Miuvidskw, J. M. Redwing, M. Shin, M. Skowronski, and D. W. Greve, solid state Electron. 42, 637 (1998). [8] A. Hoffmann, H. Siegle, A. Kaschner, L. Eckey, C. Thomsen, J. Christen, F. Bertram, M. Schmidt, K. Hiramatsu, S. Kitamura, and N. Sawaki, J. Cryst. Growth 189/190, 630 (1998). [9] H. S. Kim, R. A. Mair, J. Li, J. Y. Lin, and H. X. Jiang, Appl. Phys. Lett. 76, 1252 (2000). [10] K. Tachibana, T. Someya, S. Ishida, Y. Arakawa, J. Cryst. Growth 237, 1312 (2002). [11] J. Li, K. B. Nam, J. Y. Lin, and H. X. Jiang, Appl. Phys. Lett. 79, 3245 (2001). [12] K. O’Donnell et al., in Group III Nitride Semiconductor Compounds: Physics and Applications, edited by B. Gil (Clarendon Pr, New Zealand, 1998), pp. 247-248. [13] Yong-Hoon Cho, G. H. Gainer, J. B. Lam, J. J. Song, W. Yang, and W. Jhe, A. Bimberg and M. Sondergeld, Phys. Rev. B 61, 7203 (2000). [14] B. Monemar, Phys. Rev. B 10, 676 (1974).. - 33 -.

(50) 2 µm. Fig. 3-1 The optical microscope (OM) image of hillock in AlGaN films. The inset shows the 3D AFM images of hillock.. - 34 -.

(51) IH. - 7 μm. μ-PL Intensity (arb. units). Imatrix. - 4 μm - 3 μm - 2 μm - 1 μm 0 μm + 1 μm + 2 μm + 3 μm + 4 μm + 7 μm. 330 340 350 360 370 380. Wavelength (nm). Fig. 3-2. The room temperature µ-PL spectra taken at different locations on OM image.. - 35 -.

(52) μ-PL Intensity (arb. units). Imatrix. -8 -6 -4 -2 0. IH. 2. 4. 6. Spatial Position (μm). Fig. 3-3 The spatial µ-PL intensity distribution of Imatrix and IH.. - 36 -. 8.

(53) 100. Imatrix. IH. LineWidth (meV). 90 80 70 60 50 -8 -6 -4 -2 0. 2. 4. 6. 8. Spatial position (μm). Fig. 3-4 The spatial µ-PL full-width-half-maximum (FWHM) distribution of Imatrix and IH.. - 37 -.

(54) Imatrix IH. μ-PL Intensity (a.u). 300 K 250 K 210 K 180 K 150 K 120 K 90 K 70 K 50 K 30 K 10 K. 330. 340. 350. 360. Wavelength (nm) Fig. 3-5 300K.. Temperature dependent µ-PL measurements of AlGaN hillock from 10 to. - 38 -.

(55) Intensity (arb.units). Ea2=11.4 meV. Ea1=43.6 meV Ea2=15.0 meV Imatrix data IH data Imatrix fitting IH fitting. Ea1=33.2 meV. 0.00. 0.03. 0.06. 0.09. 0.12. -1. 1/T (K ) Fig. 3-6 Integrated PL intensity as a function of temperature for Imatrix and IH.. - 39 -.

(56) Imatrixdata IH data. Peak Energy (eV). 3.72. Imatrixfitting IH fitting. 3.68 3.64. Tc = 150 K. 3.60 3.56 Tc = 120 K. 3.52 0. 50 100 150 200 250 300. Temperature (K). Fig. 3-7 Temperature dependence of the emission peak of Imatrix and IH. Arrows indicate the transition temperature TC.. - 40 -.

(57) 6 μm. 7 μm. 8 μm. 9 μm. 10 μm. 11 μm. 2 μm. Fig. 3-8 The optical microscope (OM) image of the different size hillock in AlGaN films.. - 41 -.

(58) Normalized μ-PL intensity (a.u.). μ-PL peak energy (eV). RT μ-PL. Imatrix. 3.65 3.54 3.53 3.52 3.51 3.50 5. 6. 7. 8. 9. 10. Hillock size (μm). 11. 11 μm 10 μm 9 μm 8 μm 7 μm 6 μm Matrix. 3.45. 3.50. 3.55. 3.60. 3.65. 3.70. PL peak energy (eV) Fig. 3-9. The room temperature µ-PL spectra taken from different size of hillock.. - 42 -.

(59) 8. μ-PL. EDX. Al composition (%). 6. 4. 2. 0. 6. 7. 8. 9. 10. Hillock size (μm). Fig. 3-10 The relation between Al composition and hillock size.. - 43 -. 11.

(60) Chapter 4 GaN Nanodots Growth Because of the lack of lattice-matched substrates, the growths of GaN, InGaN and AlGaN materials on sapphire substrates are known to contain numerous defects, such as threading dislocations, stacking faults and inversion domain boundary in the epilayer, accompanied generally by a high concentration of nonradiative recombination centers. The existence of such defects seems not to affect significantly the efficiency of band-edge luminescence in InGaN/GaN blue/green light-emitting diodes. It is ascribed to the formation of self-assembled In-rich islands during InGaN film growth, which form dotlike states and lead to a marked gain enhancement in their optical process [1]. However, this is not the case for ultraviolet AlGaN materials. To date, no evidence has shown that Al segregation exists. Even though Al-rich nanoislands indeed occur, because of their higher band-gap energy feature, it would not improve the carrier confinement and hence luminescence efficiency in AlGaN films. From published ports on AlGaN ternary [2,3], we consider that the presence of a high concentration of threading dislocations and the absence of self-assembled lower energy dotlike structures are the two most detrimental factors that cause the poor quantum efficiency of AlGaN-based UV-light-emitting devices. Thus, the successful fabrication of GaN or AlGaN dotlike structures operating in the ultraviolet range is an essential step for the implementation of high-brightness UV-LEDs. Despite the numerous studies on InGaN dots, the published reports on GaN dots, particularly on specific sample preparation procedure, are still quite limited. The growth of GaN dots on AlN using the commonly used Stranski-Krastanow (S-K) growth mode was not reported until 1997 by Daudin et al.[4] using molecular beam epitaxy (MBE), and more recently by Miyamura et al. [5] using metalorganic vapor phase epitaxy (MOVPE) by maximizing the advantages of the driving force induced - 44 -.

(61) by the lattice mismatch between GaN and AlN. Probably because of the insufficient lattice mismatch provided by the underlying layer, very few updated reports have been published recently on the GaN island growth on AlGaN ternary. That imposes a strict restriction for its practical use in UV-light-emitting devices. To overcome this problem, an interesting approach, called antisurfacant method, has been used to grow GaN dots on AlGaN ternary [6]. By supplying a small amount of Si antisurfacants on an AlxGa1-xN surface, the GaN growth is found to change from the step-flow growth feature to the three-dimensional island growth, resulting in the formation of nanoscale GaN dot structures on the AlGaN surface. In this letter, we present another feasible method for preparing GaN dots on the AlGaN surface. Preliminary results indicate that by alternating the source precursors during the MOVPE epitaxial growth, a dotlike GaN structure can be obtained on an Al0.11Ga0.89N epilayer. This method is proved to be a simple and yet effective way for preparing dotlike structures in the GaN material system and may find potential use in fabricating the ultraviolet GaN-based light-emitting devices.. - 45 -.