應用於光子積體晶粒之N型調變摻雜砷化鋁鎵銦應變平衡多重量子井

全文

(1)♁ 國立中山大學光電工程研究所博士論文. 應用於光子積體晶粒之 N 型調變摻雜砷化鋁鎵銦應變平衡多重量子井 N-type Modulation-Doped InGaAlAs/InP Strain-Balanced Multiple Quantum Wells for Photonic Integrated Circuits. 研究生：馮瑞陽撰指導教授：賴聰賢博士. 中華民國九十七年七月.

(2) 博碩士論文授權書 (國科會科學技術資料中心版本，93.2.6) 本授權書所授權之論文為本人在＿__國立中山＿＿大學(學院)＿光電工程＿系所＿＿＿＿＿＿＿組＿＿九十六＿＿學年度第＿二＿學期取得＿博＿士學位之論文。論文名稱：應用於光子積體晶粒之 N 型調變摻雜砷化鋁鎵銦應變平衡多重量子井 ■同意. □不同意本人具有著作財產權之論文全文資料，授予行政院國家科學委員會科學技術資料中心(或其改制後之機構)、國家圖書館及本人畢業學校圖書館，得不限地域、時間與次數以微縮、光碟或數位化等各種方式重製後散布發行或上載網路。本論文為本人向經濟部智慧財產局申請專利（未申請者本條款請不予理會）的附件之一，申請文號為：＿＿＿＿＿＿，註明文號者請將全文資料延後半年後再公開。. --------------------------------------------------------------------------------------------------------------------■同意 □不同意本人具有著作財產權之論文全文資料，授予教育部指定送繳之圖書館及本人畢業學校圖書館，為學術研究之目的以各種方法重製，或為上述目的再授權他人以各種方法重製，不限地域與時間，惟每人以一份為限。上述授權內容均無須訂立讓與及授權契約書。依本授權之發行權為非專屬性發行權利。依本授權所為之收錄、重製、發行及學術研發利用均為無償。上述同意與不同意之欄位若未鉤選，本人同意視同授權。指導教授姓名：賴聰賢博士研究生簽名：. 學號：9035807 （務必填寫）. (親筆正楷）日期：民國 1. 2.. 九十七. 年. 七. 月. 十六. 日. 本授權書 ( 得自 http://sticnet.stic.gov.tw/sticweb/html/theses/authorize.html 下載或至http://www.stic.gov.tw首頁右下方下載) 請以黑筆撰寫並影印裝訂於書名頁之次頁。授權第一項者，請確認學校是否代收，若無者，請個別再寄論文一本至台北市(106)和平東路二段 106 號 1702 室國科會科學技術資料中心黃善平小姐。 (電話:02-27377606 傳真：02-27377689).

(3) 國立中山大學研究生學位論文審定書本校光電工程研究所博士班研究生馮瑞陽 (學號：9035807) 所提論文應用於光子積體晶粒之 N 型調變摻雜砷化鋁鎵銦應變平衡多重量子井. 經本委員會審查並舉行口試，符合博士學位論文標準。學位考試委員簽章：. 中華民國九十七年七月.

(4) Institute of Electro-Optical Engineering National Sun Yat-sen University Kaohsiung, Taiwan, R. O. C. Date: July 16, 2008. N-type Modulation-Doped InGaAlAs/InP Strain-Balanced Multiple Quantum Wells for Photonic Integrated Circuits by David Jui-Yang Feng A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy.

(5) 致謝在中山大學的學習與研究生涯中，首先要感謝敬愛的指導教授賴聰賢博士與張道源博士，茲茲不倦於課業及研究上的指導，且提供良好的機會，使學生能夠經歷從實驗室的規劃建置、機台的安裝、制度的建立，進而成為【南部光電卓越研究中心】的一員。如此薰陶下，學生在治學態度與待人處事上獲益良多。更謝謝百忙中出席學生博士論文口試的委員們，您耐心的指導與建議，使得本研究更加地完整。特別謝謝邱建良同學在製程能力建置上的貢獻及林猶穎同學在光學量測系統上之設置與維持，還有歷屆【量子元件暨分子束磊晶實驗室】同學夥伴們的付出，本論文乃大家相互合作的結果。最後，感謝我最親愛的媽媽與妹妹，雖然父親與弟弟在我求學的過程中，相繼地離開了人世，你們的鼓勵與支持使我能順利完成博士學業。緊湊的研究所生涯中，更要感謝多年來伴我一起成長的女友秉芳，謝謝妳的體諒與生活上的幫助。. Dear Prof. Tao-Yuan Chang, When you acted as a teacher, you taught me such as a tutor. When you were in Lab, you demonstrated experimental skills like as a father. In private, you shared your personal life. When I left message to you, your reply were always fast and detailed. This moment, I know you are smiling happily in the heaven for my achievement, same as the encouragement you gave before.. Dear father, In spite of being silent as usual, I can feel that you always take care of me in the sky. I really appreciate you give me a warm family and memorable boyhood. I want to tell you that I got PhD degree already. I do love you. This thesis is for you. 馮瑞陽于西子灣 2008/07.

(6) 摘要本論文的主題乃以磷化銦為基板，使用 N 型調變摻雜砷化鋁鎵銦應變平衡多重量子井，設計雷射/半導體光放大器之結構，且利用分子束磊晶成長技術製備材料，進而完成元件之製作。此一量子井之組成，乃由一晶格匹配之砷化鎵銦為其核心，填入一壓縮應變之砷化鎵銦在其核心之兩旁，配合一伸張應變之砷化鋁鎵銦做為與位障之間的緩衝隔離區。兩種具有相似結構，但設計在不同之基本發光波長(1.55 微米與 1.48 微米)的晶片，將分別地被準備用來研究: (1)順偏壓時，光之放大特性。(2)逆偏壓時，電致吸收的特性。在實驗上，以固態源分子束磊晶系統成長高品質之砷化鋁鎵銦的技術已經被建立，且利用雙晶 X 光繞射儀、穿透式電子顯微鏡、電致螢光與光致螢光光譜量測技術，所成長之塊材與多重量子井晶片的特性也更進一步被檢測確定。針對波長為 1.55 微米的晶片，使用一新發展之多次濕蝕刻技術完成脊狀波導雷射的製備，包含有 Fabry-Perot (FP)與 tilted-end-facet (TEF)兩種形式。電致螢光光譜顯示，當注入電流密度大於 20 (A/cm2)，此量子井之第一電子能階至第二電洞能階的光增益(e1-hh2，波長為 1460 奈米)大於第一電子能階至第一電洞能階的躍遷(e1-hh1，波長為 1550 奈米)。FP-laser 操作在其臨限電流時，雷射出的波長為 1514 奈米。隨著注入電流的增加，將依序觀察到兩個額外的雷射發光波長(1528 奈米與 1545 奈米)。然而，在 TEF-laser 僅發現唯一雷射發光波長在 1511 奈米。我們也進一步的發現，這些 TE 極化的雷射發光波長與光電流光譜中發現之δ-like 吸收峰之位置呈現一致。此雷射行為與量子線或量子點的光能量躍遷極為相似，極可能工作在激發態。為了研究調變摻雜半導體光放大器之電致吸收調變的行為，波長為 1.48 微米的晶片，將被用來量測其在電場下之穿透係數變化量，進而求得其微分吸收頻譜。比較六片不同的樣品，其中，在一結合電洞阻擋位障、具有 N 型調變摻雜且載子薄片密度為 3.5 × 1011 (cm-2 per QW)之量子井結構，發現此一結構工作在反向偏壓下，具有最大的揪譜參數。因此，此結構將提供一絕佳的平台讓我們有機會去實現，那些需要較大的折射率變化量，但希望有較小的吸收係數變化量之電致折射元件。此外，在波導元件的設計上，利用一漸變寬度波導的概念，針對耦和係數為 0.15 與 0.28 的兩元件，我們已經成功地使得傳統上固定寬度之多模干涉器之長度縮短超過 32%。進一步延伸此一設計概念，我們展示了一個利用串接兩段長度較短之多模干涉器，形成具有新的能量分配比(7%, 64%, 80% 與 93%) 之 2x2 波導耦合器。最後，在兩段長度較短之多模干涉器之中，插入一對不等寬的波導作為相位調整之用，我們更加實現了一任意分光比之波導耦合器。這些耦合器具有簡單的幾何形狀與低損耗的特性，進而在爾後的光積體晶粒之設計上，提供了一新穎元件可供選擇。.

(7) Abstract In this work, we have reported the design, MBE-growth and fabrication of strain-balanced n-type modulation-doped (MD) InGaAlAs/InGaAs multiple quantum wells laser/SOAs on InP. The quantum well contains a lattice-matched InGaAs core, a compressive-strained InGaAs padding, and a tensile-strained InGaAlAs spacer. Two kinds of samples having similar structure but different fundamental transition wavelength of 1.55 μm and 1.48 μm are separately prepared for investigating their characteristics in optical amplification under forward bias and electro-absorption under reversed bias. Also, the technique of growing high-quality InGaAlAs with solid-source molecular beam epitaxy has been established and the resulting InGaAlAs bulk and QWs samples are extensively characterized by double-crystal X-ray diffraction, transmission electron microscopy, electroluminescence, and photoluminescence measurements. For λ = 1.55 μm samples, ridge-waveguide lasers of Fabry-Perot (FP) type and tilted-end-facet (TEF) type were fabricated by a new developed multi-step wet-etching process. When injection current density > 20A/cm2, electroluminescence spectra show higher optical gain for the quantum well e1-hh2 transition at λ = 1460 nm than the e1-hh1 transition at λ = 1550 nm. The FP laser shows a lasing peak of λ = 1514 nm at threshold. Additional lasing wavelength at λ =1528 nm and 1545 nm were observed sequentially as the injection current increased. However, for the TEF laser, only the emission at λ = 1511 nm was observed. These TE-polarized lasing wavelengths are consistent with the δ-like absorption peaks in photocurrent spectra. The lasing performance is possible attributed to optical transitions within quantum dots/wires which are formed by the strain-field profile and alloy segregation/migration. For λ = 1.48 μm samples, the differential absorption spectroscopy, which measures the change of transmission (ΔT/T) in the presence of electric field, is used to study the.

(8) electro-absorption. modulation. behavior. of. MD-SOA’s.. A. sample. with. n-type. modulation-doping amounting to a sheet density of 3.5 × 1011 cm-2 per QW and combining with a hole-stopping barrier represents the largest chirp parameter (Δn/Δk) under reversed bias, which offers an excellent platform to realize electro-refractive devices with larger refractive index changes (Δn) but lower differential absorption (Δα) near λ = 1.55 μm, which is also our interested region of operation. In addition, we have succeeded in reducing the length of conventional constant-width multimode interference (MMI) coupler of K = 0.15 and 0.28 more than 32% by a novel stepped-width design concept. By extending the stepped-with idea, we show that it is possible to obtain 2×2 waveguide couplers with new power splitting ratios of 7%, 64%, 80% and 93% for cross coupling by cascading two short MMI sections. We further realize freely chosen power splitting ratio by interconnecting a pair of unequal-width waveguides as the phase-tuning section into the middle of two short MMI sections. These compact and low loss MMI-based devices use only rectangular geometry without any bent, curved, and tapered waveguides. They offer valuable new possibilities for designing waveguide-based photonic integrated circuits..

(9) Contents. Chapter 1.. Introduction ............................................................................. 1. 1.1. Benefits of InGaAlAs, strain QWs, and n-type modulation doping ................3 1.1.1 Advantage of InGaAlAs ............................................................................3 1.1.2 Compressively Strained QWs ....................................................................5 1.1.3 N-type modulation doping .........................................................................6 1.2 Outline of the thesis .........................................................................................7 Reference .........................................................................................................10. Chapter 2. 2.1 2.2 2.3. Chapter 3.. Design of N-type Modulation-Doped InGaAlAs/InP Strained-Balanced MQWs Laser/SOA’s .............................. 12 Important issues for TE-polarized laser/SOA’s QW ........................................12 Laser/SOA’s QW design for active devices .....................................................15 p-i-n laser/SOA’s structure ...............................................................................19 Reference .........................................................................................................26. MBE Growth of InGa(Al)As Materials and Laser/SOA’s Structures................................................................................. 27. 3.1. Calibration of cell flux and growth rate ...........................................................28 3.1.1 RHEED intensity oscillations ....................................................................29 3.1.2 Flux measurements ....................................................................................31 3.2 Experiment: MBE-growth of p-i-n laser/SOA’s structures ..............................42 Reference .........................................................................................................46. Chapter 4.. InGaAlAs/InP Strain-Balanced MQWs Laser/SOA’s ......... 48. 4.1 4.2. Mesa diode .......................................................................................................49 Ridge waveguide fabrication ...........................................................................49 4.2.1 Multi-step wet-etching process ..................................................................49 4.2.2 Double-layer photoresist method ...............................................................51 4.3 Devices: experimental results and discussions ................................................54 Reference .........................................................................................................63. Chapter 5.. Electro-Absorption Characteristics in N-type Modulation-Doped Strain-balanced MQWs ........................ 64 I.

(10) 5.1 5.2 5.3. Six blue-shifted samples (λ = 1.48 μm) ...........................................................64 Photocurrent and electroluminescence (EL) spectra........................................66 Transmission and Absorption...........................................................................68 5.3.1 Kramers-Kronig Transform (KKT)............................................................69 5.3.2 Experimental results and discussions.........................................................69 Reference .........................................................................................................85. Chapter 6. 6.1. Compact Multimode Interference Couplers with Arbitrary Power Splitting Ratio.............................................................. 86. Simulation methods .........................................................................................94 6.1.1 Multi-Mode Interference, MMI .................................................................94 6.1.2 An approximate 2-D Waveguide for representing a real 3-D waveguide. .94 6.1.3 Approximations in Guided-Mode Propagation Analysis (MPA). ..............95. Conventional 2 x 2 MMIs (K ൒ 0.5) .............................................................99 6.2.1 Transfer functions of MMI’s ......................................................................101 6.3 Cascaded 2 x 2 MMI couplers .........................................................................105 6.3.1 Transfer Matrix of Cascaded 2 x 2 MMI Couplers ....................................105 6.3.2 Possible K-values of Cascading MMI’s.....................................................106 6.3.3 Wavelength sensitivity of MMI’s...............................................................109 6.3.4 Applications ...............................................................................................110 6.4 Design of 2 x 2 Couplers with Arbitrary Power Splitting Ratio ......................117 6.5 Transfer functions of half MMI-D ...................................................................123 Reference .........................................................................................................126 6.2. Chapter 7.. Summary .................................................................................. 129. APPENDICES A.. Band Lineup and Effective Mass of Strained Layers ...................................133. B.. Achievable Epitaxial Materials .......................................................................150. C.. Riber Compact 21T MBE System ..................................................................155. D.. MBE Growth Process and Conditions ...........................................................160. E.. Ring-Resonator Loop-Mirror Laser ...............................................................167. Publication List ....................................................................................................................174. II.

(11) List of Tables [Table 2-1]. Details of one fundamental period structure for 1.55-μm strain-balanced QW’s. ------------------------------------------------------------------------------------- 19. [Table 2-2]. Structure details of MD3QW. --------------------------------------------------------- 22. [Table 3-1]. Growth rate of possible pseudomorphic strained alloys based on 1eV InGaAlAs. -------------------------------------------------------------------------------- 43. [Table 5-1]. Six blue-shift samples. ----------------------------------------------------------------- 66. [Table 6-1]. Transfer functions of MMI-A, MMI-B, MMI-C and MMI-D. -------------------- 105. [Table 6-2]. Transfer functions of MMI-hD. ------------------------------------------------------- 125. [Table A-1]. Important bandgap structure parameters for the binary III-V material systems: InAs, GaAs, AlAs, InP, GaP, and AlP [A2], [A3]. ----------------------- 144. [Table A-2]. Bowing parameters of ternary and quaternary compounds. ----------------------- 145. [Table B-1]. Possible composition of In1-x-yGaxAlyAs alloys. --------------------------------- 151. [Table B-2]. Calculated parameters of possible In1-x-yGaxAlyAs alloys based on the lattice-matched 1eV InGaAlAs. ------------------------------------------------------- 152,153. [Table C-1]. The valuable information for the effusion cells and selected source materials like as cell-adapter size/angle, cell-model, material-source supplier/purity, what size and how many quantities we chosen for materials loading. ----------------------------------------------------------------------- 159. [Table D-1]. Cells temperatures for three different situations.------------------------------------ 166. III.

(12) List of Figures. [Fig. 1.1]. Diagram of Bandgap vs. lattice constant for various III-V semiconductor at room temperature (after Tien, 1988) ---------------------------------------------------- 3. [Fig. 1.2]. For 1.55 μm wavelength transition, an illustration shows the advantages of lattice-matched InGaAs QW with 1eV InGaAlAs barriers vs. 1eV InGaAsP barriers. ------------------------------------------------------------------------------------- 5. [Fig. 1.3]. For 1.55 μm wavelength transition, an illustration shows the advantages of 1% compressive-strained InGaAs QW vs. lattice-matched one with 1eV InGaAlAs barriers. ------------------------------------------------------------------------ 6. [Fig. 1.4]. A diagram for showing the outline of this thesis including four parts: QW design, materials, device characteristics, and waveguide coupler design. --------- 9. [Fig. 2.1]. Strain dependent bandgap parameters for the strained InGaAs single QW with lattice-matched InGaAlAs (Eg = 1eV) barrier and the well width is set such that the energy transition wavelength of e1-hh1 (ε|| < 0) or e1-lh1 (ε|| > 0) is 1.55 μm: (a) quantized energy position (e1, hh1, and lh1), (b) QW thickness and wavefunction overlap integral squared, (c) the inverse out-of-plane effective mass, and (d) the inverse in-plane effective mass. --------- 14. [Fig. 2.2]. Wavefunction and the band diagram of six different QW designs for 1.55 μm laser/SOA’s. -------------------------------------------------------------------------------- 18. [Fig. 2.3]. Schematic diagram for (a) p-i-n laser/SOA’s structure and (b) band diagram ---- 20. [Fig. 2.4]. Calculated critical thickness (soild-line) and the strained-layer thickness presented in structure of MD3QW (circle) as shown in Table 2-2 at different in-plain strain. ----------------------------------------------------------------------------- 24. [Fig. 2.5]. The refractive index distribution of MD3QW in the transverse direction (y, perpendicular to epi-layers) and its E-field expansion of fundamental slab TE-mode. ----------------------------------------------------------------------------------- 25. [Fig. 2.6]. Ridge waveguide laser with width = 2 μm and etching depth = 1.79 μm; (a) the near-field map of fundamental cross-section TE-mode, and (b) its. IV.

(13) far-field pattern; (c) the divergence of far-field mode pattern in the lateral direction, and (d) in the transverse direction.------------------------------------------ 26 [Fig. 3.1]. The measured As-BEP and its background pressure (PG) at different As-valve positions; the insert diagram shows the relation of As-BEP vs. PG. ---------------- 30. [Fig. 3.2]. Illustration of the mechanism for RHEED spot oscillations during growth of a monolayer. ------------------------------------------------------------------------------- 31. [Fig. 3.3]. The Arrhenius plot of measured BEP (a) and flux (b) of In1-cell at different cell temperatures. ------------------------------------------------------------------------- 33. [Fig. 3.4]. The Arrhenius plot of measured BEP (a) and fluxes (b) of Ga1- and Ga2-cells at different cell temperatures.------------------------------------------------------------ 34. [Fig. 3.5]. The Arrhenius plot of measured BEP (a) and fluxes (b) of Al1- and Al2-cells at different cell temperatures.------------------------------------------------------------ 35. [Fig. 3.6]. The basic structure for determining the doping cell fluxes of Be and Si. ---------- 36. [Fig. 3.7]. Si-cell: the Arrhenius plot for growth-rate-related doping capacity. --------------- 37. [Fig. 3.8]. Be-cell: the Arrhenius plot for growth-rate-related doping capacity. -------------- 37. [Fig. 3.9]. The relation of carrier concentration vs. Hall mobility on Si-doped GaAs. ------- 38. [Fig. 3.10] The X-ray diffraction in rocking angle for three basic lattice-matched materials, InGaAs, InAlAs, and 1eV InGaAlAs, in comparison with InP substrate. ----------------------------------------------------------------------------------- 40 [Fig. 3.11] The room-temperature PL result from lattice-matched InGaAlAs. ----------------- 41 [Fig. 3.12] The schematic diagram for explaining mechanism of pseudomorphic layer (a) compressive strain, (b) tensile strain. --------------------------------------------------- 43 [Fig. 3.13] The X-ray diffraction rocking measurement and simulation result of TE-1.55 μm SOA/laser structure (MD3QW). --------------------------------------------------- 44 [Fig. 3.14] The cross-section HR-TEM pictures of TE-1.55 μm SOA/laser MBE sample at crystal facet of (0-11) and (011); (a), (b): 430,000X; (c), (d): 1450,000X; (e), (f): 2850,000X ------------------------------------------------------------------------ 45. V.

(14) [Fig. 4.1]. The top view picture of a 250-μm diameter mesa diode. ---------------------------- 48. [Fig. 4.2]. Schematically diagram for explaining multi-step wet-etching process.------------ 50. [Fig. 4.3]. The cross-section SEM pictures: (a) sidewall profile after four times etching, (b) after smoothing process. ------------------------------------------------------------- 51. [Fig. 4.4]. The cross-section SEM picture after planarization process. ------------------------- 52. [Fig. 4.5]. The top-view optical microscope pictures for ridge waveguides: (a) Fabry-Perot type, (b) Tilted-end-facet type. ------------------------------------------- 53. [Fig. 4.6]. (a) PL spectra, and (b) EL spectra for sample “MD3QW” at room temperature. -------------------------------------------------------------------------------- 58. [Fig. 4.7]. The characteristics of current density vs. applied forward biases (J-V). ----------- 59. [Fig. 4.8]. FP-type ridge laser: (a) L-I characteristics, and (b) lasing spectra. ----------------- 60. [Fig. 4.9]. TEF-type ridge laser: (a) L-I characteristics, and (b) lasing spectra. --------------- 61. [Fig. 4.10] Photocurrent spectrum measured at V = 0. -------------------------------------------- 62 [Fig. 5.1]. Band diagram and subband positions of e1 and hh1 for blue-shifted samples. --- 65. [Fig. 5.2]. Sample 1: (a) photocurrent spectra and (b) EL spectra at T = 300K. --------------- 72. [Fig. 5.3]. Sample 2: (a) photocurrent spectra and (b) EL spectra at T = 300K. --------------- 73. [Fig. 5.4]. Sample 3: (a) photocurrent spectra and (b) EL spectra at T = 300K. --------------- 74. [Fig. 5.5]. Sample 4: (a) photocurrent spectra and (b) EL spectra at T = 300K. --------------- 75. [Fig. 5.6]. Sample 5: (a) photocurrent spectra and (b) EL spectra at T = 300K. --------------- 76. [Fig. 5.7]. Sample 6: (a) photocurrent spectra and (b) EL spectra at T = 300K. --------------- 77. [Fig. 5.8]. Sample 1: (a) Δn spectra and (b) Δα spectra at T = 300K. -------------------------- 78. [Fig. 5.9]. Sample 2: (a) Δn spectra and (b) Δα spectra at T = 300K. -------------------------- 79. [Fig. 5.10] Sample 3: (a) Δn spectra and (b) Δα spectra at T = 300K. -------------------------- 80 [Fig. 5.11] Sample 4: (a) Δn spectra and (b) Δα spectra at T = 300K. -------------------------- 81. VI.

(15) [Fig. 5.12] Sample 5: (a) Δn spectra and (b) Δα spectra at T = 300K. -------------------------- 82 [Fig. 5.13] Sample 6: (a) Δn spectra and (b) Δα spectra at T = 300K. -------------------------- 83 [Fig. 5.14] The chirp parameters at λ = 1.55 μm for samples of S1 – S6. ---------------------- 84 [Fig. 6.1]. Loop-mirror: (a) schematic diagram and (b) the calculated reflectance and transmittance as function of K-value. -------------------------------------------------- 90. [Fig. 6.2]. Symmetric single-ring resonator: (a) schematic diagram; the simulated characteristics for (b) K = 0.5 and (c) K = 0.15. -------------------------------------- 91. [Fig. 6.3]. Double-ring resonator: (a) schematic diagram with two 3-dB MMI couplers in both-side; the simulated characteristics when the middle one coupler (b) K = 0.15, (c) 0.5, and (d) 0.85. ------------------------------------------------------------- 92. [Fig. 6.4]. Benefit of shorter coupler in double-ring resonator: (a) the conventional constant-width MMI coupler and (c) the shorten one. The simulated characteristics referring to (a) and (c) are separately shown in (b) and (d). ------- 93. [Fig. 6.5]. Schematic diagrams of a generic 2x2 MMI coupler: (a) in plan view showing the notations used in this study, and (b) in cross-section showing the basic layer structure in the waveguide ridge; (c) the calculated effective refractive index for 3-D ridge waveguide, 2-D slab representation, and Eq. (6.2b); (d) the refractive index profile in the transverse direction “y” and the simulated transverse Ex distribution of the TE fundamental slab mode (TE0) at the vacuum wavelength (λ0) = 1.55 μm. --------------------------------------------------- 98. [Fig. 6.6]. (a) General N x N MMI coupler with access waveguides for arbitrary N. (b) At its limits of a = 0 or a = W/N, images merge in pairs. An overlapping-image MMI coupler is formed.------------------------------------------- 103. [Fig. 6.7]. The simulated 2-D field maps for four conventional short K ൒ 0.5 MMIs used as the basic building blocks in this study: (a) MMI-A (N = 2 and i, j = 1, 2), (b) MMI-B (N = 6 and i, j = 2, 4), (c) MMI-C (N = 4 and i, j = 1, 3), (d) MMI-D (N = 5 and i, j = 1, 3). ---------------------------------------------------------- 104. [Fig. 6.8]. The simulated 2-D field maps for K = 0 MMI devices: (a) Conventional constant-width MMI including four times sections of MMI-A, (b) Cascaded MMI formed by MMI-A + MMI-B. ---------------------------------------------------- 112. VII.

(16) [Fig. 6.9]. The simulated 2-D field maps for K = 0.15 MMI devices: (a) Conventional constant-width MMI including three times sections of MMI-C, (b) Cascaded MMI formed by MMI-A + MMI-C. ---------------------------------------------------- 112. [Fig. 6.10] The simulated 2-D field maps for K = 0.28 MMI devices: Conventional constant-width MMI including (a) four, and (b) three times sections of MMI-D for the access waveguides locating at both ends symmetrically and oppositely, respectively, (c) Cascaded MMI formed by MMI-hD + MMI-E, (d) same as (c) but the light propagating in the opposition direction. -------------- 113 [Fig. 6.11] The simulated 2-D field maps for four new cascaded MMI couplers with K = 0.07 (a), 0.93 (b), 0.64 (c), and 0.80 (d). The values for total transmittance are obtained by 2-D BPM simulations.------------------------------------------------- 114 [Fig. 6.12] K vs. length for all possible shortest 2x2 MMI couplers made up by simple rectangular geometry. Number by Italic: the device length in unit of As2. (A+D): MMI-A+MMI-D, (2A): MMI-A+MMI-A, hD: half of MMI-D. ---------- 115 [Fig. 6.13] (a) The dependence of the total transmittances on the deviation of the input wavelength from the design wavelength (1.55 μm). (b) The corresponding variation of the K values. (Note: 3C: triple MMI-C, hD: half of MMI-D) -------- 116 [Fig. 6.14] Schematic diagrams of a 2x2 coupler based on two short MMI sections interconnected by a pair of phase-shifter waveguides in their middle. ------------- 120 [Fig. 6.15] The mode coupling efficiency between two different-width ridge waveguides, one has a width = “w”, the other has a width = “w + δw”. The result is obtained from a 3-D EigenMode Expansion simulation tool, named “Fimmprop”. ------------------------------------------------------------------------------- 120 [Fig. 6.16] The simulated 2-D field maps for seven different cascaded MMI couplers with (a) K = 1 (A+A), (b) K= 0 (A+B), (c) K = 0.15 (A+C), (d) K = 0.07 (A+D), (e) K = 0.93 (B+D), (f) K = 0.64 (C+D), and (g) K = 0.80 (D+D). (0.96 Total): 96% transmittance in total. The values for total transmittance and K are obtained by 3-D BPM. ------------------------------------------------------- 121 [Fig. 6.17] The MPA analyzed by Eq. (6.11) and 2-D EME simulated results for K vs. total length (LU + LPS + LV) for the 2x2 couplers with phase shifter of (a) βa > βb (Δ : δa/δb = 0.2/0 μm; o : δa /δb = 0.1/0 μm), and (b) βa < βb (Δ : δa/δb =. VIII.

(17) 0/0.2 μm; o : δa/δb = 0/0.1 μm). --------------------------------------------------------- 122 [Fig. 6.18] The simulated 2-D field maps for half of MMI-D (MMI-hD): (a) when i = 2, power splitter ratio = 14 : 36 : 36 : 14, (b) when i = 6, power splitter ratio = 36 : 14 : 14 : 36. --------------------------------------------------------------------------- 124 [Fig. A.1]. The energy-band structure in the momentum space for a bulk In1-xGaxAs under (a) compressive, (b) lattice-matched, and (c) tensile conditions which is adopted from [A3]. --------------------------------------------------------------------- 138. [Fig. A.2]. (a) Unstrained bandgap energy, (b) strained band edge of conduction-band (Ec, blue) and valance-band of heavy-hole (Ev(hh), red), and light-hole (Ev(lh), green) vs. Ga (x) and Al (y) mole fractions for In1-x-yGaxAlyAs material system. -------------------------------------------------------------------------------------- 146. [Fig. A.3]. (a) Bandgap, (b) Band lineup, (c) ΔEc, ΔEv, and ΔEc/ΔEg on In0.523Al0.477As (M), and (d) |ΔEc|/ΔEg on InP vs. Al (y) mole fractions for lattice-matched In1-x-yGaxAlyAs material system. -------------------------------------------------------- 147. [Fig. A.4]. (a) A 3D diagram for inverse in-plane effective mass of electron and hole (heavy- and light-hole) vs. Ga (x) and Al (y) mole fractions, and (b) its side-view in y direction when in-plane strain (ε||) = -2%, -1%, 0%, 1%, and 2%. ------------------------------------------------------------------------------------------ 148. [Fig. B.1]. Band alignment for possible In1-x-yGaxAlyAs alloys on InP based on the lattice-matched 1eV InGaAlAs by Harrison’s model. -------------------------------- 154. [Fig. C.1]. A schematic drawing of the MBE machine used in this study. ---------------------- 158. [Fig. C.2]. The detailed position-distribution for effusion cells and source materials used in the present study. ----------------------------------------------------------------------- 159. [Fig. E.1]. Schematic longitudinal view of the SG-DBR device. -------------------------------- 168. [Fig. E.2]. The schematic diagrams for (a) loop-mirror laser and (b) ring-resonator loop-mirror laser. -------------------------------------------------------------------------- 170. [Fig. E.3]. Top view of ring-resonator loop-mirror laser. ----------------------------------------- 171. IX.

(18) Chapter 1 Introduction In order for semiconductor photonic devices to provide more complex functionalities in λ = 1.55 μm optical communication system economically, it is increasingly important to have a reasonably simple method to integrate different devices on one chip such as lasers, optical amplifiers, ring-resonator filters, and Mach-Zehnder interferometer (MZI) [1.1]-[1.4]. However, achieving a complex photonic integrated circuit (PIC) based on a p-i-n semiconductor optical amplifier (SOA) structure [1.5]-[1.7] is still a huge challenge. Considering active devices or passive waveguides are simultaneously operated on one chip, first concern is how to keep the absorption low for unbiased passive waveguide and reverse-biased MZI type device. A basic means to achieve this purpose is to blue shift the transition wavelength of material for passive devices out of the operating wavelength region. Fortunately, the material outside the active regions can be blue shift by quantum well intermixing (QWI) techniques [1.2], [1.8] to reduce optical absorption in the passive waveguide devices. This is a very attractive method for the fabrication of PIC’s, especially if the blue shifted material also shows a strong electrorefraction effect to allow us to obtain clean refractive index change (Δn) with very little induced electroabsorption at λ = 1.55 μm under reverse bias. It is because large Δn with low driving bias is very desirable for MZI-type devices and switch-type waveguides such as Y- and X-junction couplers [1.9], [1.10]. Therefore, we firstly intend to develop a SOA-based epitaxial wafer that can not only provide optical gain for active devices operating under forward current injection, but also have low optical losses with high refractive index change (Δn) at zero/reverse bias operation. In order to realize this goal, the technique of growing high-quality InGaAlAs with. 1.

(19) SSMBE is first established and the resulting InGaAlAs bulk and QWs samples are extensively characterized by double-crystal X-ray diffraction (DCXRD), transmission electron microscopy (TEM), electroluminescence (EL), and photoluminescence (PL) measurements. Two kinds of samples having similar structure but different fundamental transition wavelength of 1.55 μm and 1.48 μm are separately prepared for investigating their characteristics in optical amplification under forward bias and electro-absorption under reversed bias. For 1.55 μm samples, ridge-waveguide lasers of Fabry-Perot (FP) type and tilted-end-facet (TEF) type were fabricated. FP-laser operating at excited transitions has been exposed. For 1.48 μm samples, the differential absorption (Δα) spectroscopy, which measures the change of transmission (ΔT/T) in the presence of electric field, is used to study the electro-absorption modulation behavior of modulation-doped (MD) SOA’s. In addition, we also aim to design various compact waveguide couplers based on this SOA-based epitaxial wafer. We have shown that it is possible to obtain 2×2 waveguide couplers with new power splitting ratios of 7%, 64%, 80% and 93% for cross coupling by cascading two short multimode interference (MMI) sections. We further realize freely chosen power splitting ratio by interconnecting a pair of unequal-width waveguides as the phase-tuning section into the middle of two short MMI sections. These couplers have simple geometry and low loss. They offer valuable new possibilities for designing waveguide-based photonic integrated circuits. Finally, a developing integrated device, named as ring-resonator loop-mirror laser, combing with our attempt is also present. In brief, this work desires that the multi-function wafer can be processed into different devices with good performance by using reasonably simple techniques. Moreover, to achieve our purpose, three relevant issues for improving wafer performance are discussed in section 1.1.. 2.

(20) Fig. 1.1. Diagram of Bandgap vs. lattice constant for various III-V semiconductor at room temperature (after Tien, 1988). 1.1. Benefits of InGaAlAs, strained QWs, and n-type modulation doping. 1.1.1. Advantage of InGaAlAs. In the current fiber-optic technology, the wavelength near to 1.55 μm is highly useful due to its loss in fiber is minimum. Alloys of InGaAsP and InGaAlAs are still the two main material systems used to fabricate long-wavelength semiconductor lasers. As shown in Fig. 1.1, the composition of the quaternary III-V semiconductor compound InGaAsP or InGaAlAs which are lattice matched to InP can be adjusted to separately cover the bandgap range from 1.353eV (λ = 0.92 μm, for binary InP) or 1.443eV (λ = 0.86 μm, for ternary In0.523Al0.477As). 3.

(21) to 0.749eV (λ = 1.65 μm, for ternary In0.532Ga0.468As). Although, InGaAsP/InP system has been the most widely used material system as a candidate for high-performance lasers [1.11], [1.12], peoples recently pay more attentions to the InGaAlAs/InAlAs system for interesting in its large conduction band offset of ΔEc/ΔEg ~ 0.7 compared to a value of ΔEc/ΔEg ~ 0.4 for InGaAsP/InP. The larger conduction band offset had been predicted to result in better electron confinement in conduction band as illustrated in Fig.1.2, hence, combining with higher temperature stability [1.9]. In addition, electrons with InGaAlAs barrier, has a higher value of wavefunction overlap-integral- squared (OIS) between electrons and holes due to the electron wave-function being more symmetric to hole wave-function. Other benefits from InGaAlAs quantum wells (QWs) in comparing to InGaAsP ones has also been reported, including higher peak material gain (also higher peak net modal gain), less Auger recombination, lower transparency carrier densities, low-threshold current, and much potential for high-speed modulation [1.13], [1.14]. One more reason why one might consider InGaAlAs instead of InGaAsP is that it is difficult to have good control over precise alloy composition if more than one group-V elements are involved, especially alloys grown by solid-source molecular beam epitaxy (SSMBE). In addition, phosphorus (P) combining with its high vapor pressure is not easy to handle for SSMBE applications. Furthermore, burning-trouble from the natural of white-phosphorus exposed to air will also bring many disadvantages out in the maintenance process of SSMBE chambers. On the other way, one might wonder why we do epitaxy with SSMBE rather than metal-organic chemical vapor deposition (MOCVD). The major reason is SSMBE can provide much better thickness and composition control which is particularly important for quantum wells. Also, using SSMBE in this study is due to the constraint that only SSMBE is accessible to us. The goal is to see how far we can go within this given restriction.. 4.

(22) Fig. 1.2. For 1.55 μm wavelength transition, an illustration shows the advantages of lattice-matched InGaAs QW with 1eV InGaAlAs barriers vs. 1eV InGaAsP barriers.. 1.1.2. Compressively Strained QWs. According to Fig. 1.2, even lattice-matched InGaAs/InGaAlAs (QWs/barriers) material system has a higher value of OIS between subband transitions of electron-1 (e1) and havey-hole-1 (hh1) for 1.55 μm wavelength emission, however, its wider well thickness which limit us having more QW periods will also limit saturation gain of material. It has been well known compressive-strained QW can succeed in reducing QW thickness due to its larger conduction- and valance-band offset in comparison with lattice-matched one as demonstrated in Fig. 1.3. Additionally, using strained-QWs can reduce the needed carrier density for population inversion. Furthermore, the splitting of the hole subbands reduces the nonradiative Auger recombination and intervalence band absorption [1.13]. These effects combining with. 5.

(23) the increased conduction-band discontinuity from the compressively strained layer will lower the lasing threshold current and improve the quantum efficiency in strained-MQW lasers.. Fig. 1.3. For 1.55 μm wavelength transition, an illustration shows the advantages of 1% compressive-strained InGaAs QW vs. lattice-matched one with 1eV InGaAlAs barriers.. 1.1.3. N-type modulation doping Besides active devices, for SOA-based photonic integrated circuit applications, it is. important to keep the absorption low for unbiased passive waveguide components and reverse-biased Mach-Zehnder type devices. A basic means to accomplish this is to blue shift the QW’s transition wavelength out of the operating region. At photon energies higher than bandgap, a negative change in refractive index (Δn) from band-filling of the conduction band dominates over the opposite effect of bandgap shrinkage due to bandgap renormalization [1.15], [1.16] and other purely field-induced Δn such as the electrooptic effect [1.17], [1.18].. 6.

(24) Modulation doping of QW’s, therefore, allows one to obtain strong electrorefraction through carrier depletion [1.18]. In addition, it also benefits an SOA to reduce transparency current density and noise figure [1.19], [1.20] and helps to shorten the radiative lifetime (spontaneous lifetime) to increase the saturation intensity. However, considering which type of modulation doping will much benefit the device’s performance, QW with n-type modulation doping was found that turn-on delay time is shorter than that of QW with p-type modulation doping [1.21] because the necessary threshold current in n-type QW laser is lower than that of p-type laser.. 1.2. Outline of the thesis This thesis is mainly presented by four parts as a diagram shown in Fig. 1.4, including QW design, materials, device characteristics, and waveguide coupler design. For QW design, the principle to determine important material quantities for the band lineup and effective mass of strained layer is presented in appendix-A, along with an introduction of achievable materials in appendix-B, which both appendices can be regarded as the preparation for QW design. In order to prevent those basic parameters confusing reader, this thesis directly looks in more detail at the design of 1.55 μm transition wavelength InGaAlAs/InP strained-balanced QW structure, along with whole p-i-n laser/SOA’s ones in Chapter 2. Additional, in Chapter 6, six samples having similar structure to λ = 1.55 μm ones but different MD distribution inside barriers with reduced QW layer thickness for blue shifting transition wavelength to 1.48 μm are prepared for investigating the relations between Δn, Δα, and MD distribution. For materials, after introducing the method for calibrating flux of source materials, those analytic results of MBE-grown epitaxial wafers including DCXRD, PL and TEM are discussed in chapter 3. Besides, a brief description for molecular beam epitaxy (MBE) machine using in this study and the growth process combing with growth conditions are separately exposed in appendix C and appendix D.. 7.

(25) For device characteristics, the lasing properties of ridge waveguide and electro-absorption characteristics of mesa diode, including current vs. voltage (I-V), light vs. current (L-I), lasing spectrum, photocurrent, EL, and EA measurements are separately discussed in Chapter 4 and 5 as well as their fabrication process. Moreover, based on the same laser/SOA’s structure, a novel new design concept to reduce the device length more than 32% for conventional multi-mode interference (MMI) couplers, to have couplers with new power splitting ratio of 0.07, 0.64, 0.80, and 0.93, even to realize couplers with freely chosen power splitting is also presented in Chapter 6. Finally, several conclusions with a developing integrated device exposed as the future work are given in Chapter-7.. 8.

(26) Fig. 1.4. A diagram for showing the outline of this thesis including four parts: QW design, materials, device characteristics, and waveguide coupler design.. 9.

(27) Reference [1.1]. T. L. Koch and U. Koren, “Semiconductor Photonic Integrated Circuits,” IEEE J. Quantum Electron., vol. 27, pp. 641- 653, 1991.. [1.2]. E. J. Skogen, J. S. Barton, S. P. Denbaars and L. A. Coldren, “A quantum-well-intermixing process for wavelength-agile photonic integrated circuits,” IEEE J. Sel. Topics. Quantum Electron., vol. 8, pp. 863- 869, 2002.. [1.3]. M. Kohtoku, S. Oku, Y. Kadota, Y. Shibata, and Y. Yoshikuni, “200-GHz FSR Periodic Multi/Demultiplexer with Flattened Transmission and Rejection Band by using a Mach-Zehnder Interferometer with a Ring Resonator ,” IEEE Photon. Technol. Lett., vol. 12, pp. 1174-1176, 2000.. [1.4]. B. Liu, A. Shakouri, and J. E. Bowers, “Wide Tunable Double Ring Resonator Coupled Lasers,” IEEE Photon. Technol. Lett., vol. 14, pp. 600-602, 2002.. [1.5]. P. Jayavel, T. Kita, O. Wada, H. Ebe, M. Sugawara, Y. Arakawa, Y. Nakata and T. Akiyama, “Optical Polarization Properties of InAs/GaAs Quantum Dot Semiconductor Optical Amplifier ,” Jpn. J. Appl. Phys. Vol. 44, pp. 2528- 2530, 2005.. [1.6]. T. Akiyama, N. Hatori, Y. Nakata, H. Ebe and M. Sugawara, “Pattern-effect-free semiconductor optical amplifier achieved using quantum dots,” Electron. Lett., vol. 38, pp. 1139- 1140, 2002.. [1.7]. T. Akiyama, H. Kuwatsuka, T. Simoyama, Y. Nakata, K Mukai, M. Sugawara and O. Wada, “Nonlinear gain dynamics in quantum-dot optical amplifiers and its application to optical communication devices,” IEEE J. Quantum Electron., vol. 37, pp. 10591065, 2001.. [1.8]. S. D. McDougall, O. P. Kowalski, C. J. Hamilton, F. Camacho, B. Qiu, M. Ke, R. M. De La Rue, A. C. Bryce and J. H. Marsh, ” Monolithic integration via a universal damage enhanced quantum-well-intermixing technique,” IEEE J. Sel. Topics. Quantum Electron., vol. 40, pp. 636- 646, 1998.. [1.9]. M. N. Khan, J. E. Zucker, T. Y. Chang, N. J. Sauer, M. D. Divino, T. L. Coch, C. A. Burrus, and H. M. Presby, “Design and Demonstration of Weighted-Coupling InGaAs/InGaAlAs Electron Transfer Waveguides,” J. Lightwave Technology, vol. 12, pp. 2032-2039, 1994.. [1.10] Y. Siberberg, P. Perlmutter, and J. E. Baran, “Digital Optical Switch,” Appl. Phys. Lett., vol. 51, pp. 1230-1232, 1987. [1.11] P. J. A. Thijs, L. F. Tiemijer, P. I. Kuindersma, J. J. M. Binsma, and T. Van Dongen,. 10.

(28) “High performance of 1.5 μm wavelength InGaAs-InGaAsP strained quantum-well lasers and amplifiers,” IEEE J. Quantum Electron., vol. 27, pp. 1426-1438, 1991. [1.12] P. J. A. Thijs, L. F. Tiemijer, J. J. M. Binsma, and T. Van Dongen, “Progress in long-wavelength strained-layer InGaAs(P) quantum-well semiconductor lasers and amplifiers,” IEEE J. Quantum Electron., vol. 30, pp. 477-499, 1994. [1.13] J. Minch, S. H. Park, T. Keating, and S. L. Chuang, “Theory and Experiment of In1-xGaxAsyP1-y and In1-x-yGaxAlyAs Long-Wavelength Strained Quantum-Well Lasers,” IEEE J. Quantum Electron., vol. 35, pp. 771-782, 1999. [1.14] J. C. L. Yong, J. M. Rorison, and I. H. White, “1.3-μm Quantum-Well InGaAsP, AlGaInAs, and InGaAsN Laser Material Gain: A Theoretical Study,” IEEE J. Quantum Electron., vol. 38, pp. 1553-1564, 2002. [1.15] L. A. Coldren and S. W. Corzine: Diode Laser and Photonic Integrated Circuits, (John Wiley & Sons, New York, 1995), P. 137. [1.16] H. Haug and S. W. Koch: Quantum theory of the optical and electronic properties of semiconductors, (World Scientific, Singapore, 1994) 3rd ed., p.250. [1.17] A. Yariv: Optical Electronics in Modern Communications, (Oxford University Press, Oxford, 1997) 5th ed., p. 328. [1.18] J. E. Zucker, T. Y. Chang, M. Wegener, N. J. Sauer, K. L. Jones, and D. S. Chemla, “Large refractive index changes in tunable-electron-density InGaAs/InAlAs quantum wells,” IEEE Photonics Technol. Lett., vol. 2, pp. 29-31, 1990. [1.19] T. Mukai, Y. Yamamoto and T. Kimura: Semiconductors and Semimetals, ed. W. T. Tsang (Academic Press, New York, 1985) Vol. 22, Part E. [1.20] K. J. Vahala and C. E. Zah, “Effect of doping on the optical gain and the spontaneous noise enhancement factor in quantum well amplifiers and lasers studied by simple analytical expressions,” Appl. Phys. Lett., vol. 52, pp.1945-1947, 1988. [1.21] A. Niwa, T. Ohtoshi, K. Uomi, and K. Nakahara, “Doping-type dependence of turn-on delay time in 1.3 μm InGaAsP-InP modulation-doped strained quantum-well lasers,” IEEE Photon. Technol. Lett., vol. 8, pp. 328-330, 1996.. 11.

(29) Chapter 2 Design of N-type Modulation-Doped InGaAlAs/InP Strained-Balanced MQWs Laser/SOA’s In this thesis, we present the designs and MBE growth of epitaxial structures on InP containing n-type modulation-doped (MD) QW’s. They are designed to incorporate all the attractive features discussed in chapter 1. In section 2.1, important issues for designing TE-polarized laser/SOA’s wafer with lower transparency current density and higher differential gain are discussed by considering how to do a trade-off to have a narrow QW but still keep the wavefunction overlap integral squared (OIS) between conduction- and valance-band high for strained QW layers. After understanding what we concern given in section 2.1, a compact strained-QW with a high value OIS of 0.94 is exposed in section 2.2 with an introduction how to strain balance the strained-QW totally. In order to have a ridge waveguide with low loss, more circular intensity distribution in far field, and larger mode confinement factor, a frame for making our p-i-n laser/SOA’s structures completely is detailed in section 2.3; also, its n-type modulation doping distribution is involved in this section.. 2.1 Important issues for TE-polarized laser/SOA’s QW Considering a ternary InGaAs single QW with 1eV InGaAlAs (M) barriers for 1.55 μm wavelength emission, the calculated quantized energy states (e1, hh1, and lh1) and wavefunction overlap integral squared related to the needed QW thickness in different in-plane strain are separately indicated in Fig. 2.1(a) and (b). As mentioned previously, strain effect causes a separation between band edges of heavy- and light-hole. Compressive. 12.

(30) strain moves up the heavy-hole band edge and moves down the conduction band edge; it will benefit the pure TE-polarized light emission because the energy transition of e1-hh1 will be more exposed by enlarging conduction- and valance-band offset. The larger band offsets also help to reduce the necessary QW width for 1.55 μm operation. Additionally, in order to make QWs having low-transparency carrier density (for low lasing threshold) and high differential gain (for high-speed modulation), two major issues for multiple-QWs (MQWs) design should be considered: (1) the first one is how to reduce in-plane effective mass inside QWs as low as possible that is because the step-like density of states are directly proportional to its effective mass, in other word, keeping the in-plane effective mass low will follow a lower density of states; (2) the other is how to have a closely matched density of states between the valance and conduction bands. See dash-line in Fig. 2.1(c), because the out-off-plane electron effective mass is about 8 times smaller than heavy-hole one, it usually exists only one electron subband but not for heavy hole one. A narrower QW width further enlarges the separation between the lower hole subbands (i.e. hh1-hh2). And, a larger separation between these subbands has a positive influence on the gain [2.1]. In addition, according to Fig. 2.1(d), the another advantage of a closely matched density of states between conduction and valance bands from a compressive strain QW is contributed to which in-plane heavy-hole effective mass is reduced significantly and more closed to electron one. However, a QW with thinner thickness cause another problem: more asymmetric wavefunction distribution and a lower value of OIS between electron and hole. In brief, in order to lower transparency current density and have a higher differential gain, trade-off to have a narrow QW but still keep the OIS high for a compressive strained-QW is the very important issue in the design of TE-polarized laser/SOA’s QWs structures.. 13.

(31) Fig. 2.1. Strain dependent bandgap parameters for the strained InGaAs single QW with lattice-matched InGaAlAs (Eg = 1eV) barrier and the well width is set such that the energy transition wavelength of e1-hh1 (ε|| < 0) or e1-lh1 (ε|| > 0) is 1.55 μm: (a) quantized energy position (e1, hh1, and lh1), (b) QW thickness and wavefunction overlap integral squared, (c) the inverse out-of-plane effective mass, and (d) the inverse in-plane effective mass.. 14.

(32) 2.2. Laser/SOA’s QW design for active devices In this thesis, our QW’s design is based on those achievable materials mentioned in. appendix-B. Fig. 2.2 shows the wavefunction and the band diagram (conduction and heavy-hole valance bands) of six different QW designs for 1.55-μm TE-polarized laser/SOA’s. The simple QW as shown in Fig. 2.2(a) uses a lattice-matched InGaAs in the QW core and lattice-matched InGaAlAs (1eV) in the barriers. Although its ΔEc/ΔEg = 0.65 and the value of wavefunction OIS between e1 and hh1 states is quite high at 0.95, the required QW thickness of 9-nm is rather large to introduce the living of undesired first excited electron state (e2), which is not indicated inside Fig. 2.2(a). In Fig. 2.2(b), we change the QW core from lattice-matched InGaAs to compressive-strained In0.67Ga0.33As (ε|| = - 0.94 %). In this case of ΔEc/ΔEg = 0.57, the 3.5-nm QW is much more compact, but the wavefunction OIS decreases to 0.84. In Fig. 2.2(c), we use tensile-strained In0.438Ga0.386Al0.176As (ε|| = 0.63 %) as the barriers to compensate the strained stress between the QW core and barriers. In this case of ΔEc/ΔEg = 0.49, the 3.5-nm thick QW is still compact but the wavefunction OIS decreases significantly to 0.78. This is because the electron wavefunction extends much wider than the heavy-hole wavefunction. In Fig. 2.2(d), we obtain a compromise by using a thin 1.8-nm lattice-matched InGaAs as the QW core and 1.5-nm compressive-strained In0.67Ga0.33As as the QW padding beside the QW core. The wavefunction OIS improves to 0.87 with a little extended QW thickness of 4.8-nm. In Fig. 2.2(e), the design is optimized by using a thin 2.2-nm lattice-matched InGaAs as the QW core and 1.5-nm compressive-strained In0.67Ga0.33As as the QW padding on both sides of the QW core. We further use 1.6-nm wavefunction-adjustment layers of compressive-strained In0.714Al0.286As (ε|| = - 1.28 %) and thin 1.2-nm tensile-strained In0.438Ga0.386Al0.176As spacer layers to form the barrier structure, to balance the strain, and to enhance the electron confinement. With this design, we are able to achieve a high value OIS of 0.94 for a 5.2-nm-thick compact QW.. 15.

(33) Considering how to design a strain-balanced QW by strain-compensated skills, we know two strained layers will be strain compensated if one is compressively strained with Nc monolayers (ML’s) while the other one is tensile strained with Nt ML’s and simultaneously matched to following relation N. a. N. a. N. N. a (2.1). where a. and a. are the out-of-plane lattice constant for compressively and tensile. strained layers, respectively. a. is the lattice constant of substrate. Eq. (2.1) can be. transferred as d. d. d a. d a. a (2.2). where dc and dt are the thicknesses for compressively and tensile strained layers, respectively. By using Eq. (2.2) and assuming the thickness of the most outside barriers in case of Fig. 2(e) is “d”, a symmetric strain-compensated QW structure with thicknesses of “d(T)-1.6(C)-1.2(T)-1.5(C)-2.2(M)-1.5-1.2-1.6-d (nm)” can be calculated as d. 1.6. 1.2. 1.5. d 0.57960. 1.6 0.60222. 1.2 0.57960. 1.5 0.59825. 0.58687. to obtain d = 4.3-nm, which out-of-plane lattice constant of strained layers are according to Table B-2. Additionally, n-type modulation doping (Si: 1x1018cm-3) is incorporated inside the half thickness of tensile-strained In0.438Ga0.386Al0.176As barrier layers and amounting to a sheet density of 4.3 x 1011 cm-2 per QW. The purpose of modulation doping is to help low the transparency current, to enhance the spontaneous-emission factor [2.2], [2.3] and to provide large Δn under reverse bias. This is the basis of our triple-QW’s laser/SOA’s structure which is labeled MD3QW. Its internal one-period structure for this strain-balanced. 16.

(34) QW structure is detailed in Table 2-1. Moreover, it has a 3-nm-thick tensile-strained In0.305Ga0.417Al0.278As (ε|| = 1.55 % and Egu = 1.33 eV) as hole-stopping barrier, incorporated at the end of the n-region of the p-i-n layer structure immediately next to the InAlAs wavefunction-adjustment layer of the first QW as indicated in Fig. 2.2(f). The hole-stopping barrier is away from the first QW with 1.2-nm spacer layer and 1.6-nm wavefunction-adjustment layer. According to our simulation, it does not affect the value of wavefunction OIS between e1 and hh1. Also, it is expected to reduce the penetration of holes into the n-layer and to enhance the injected hole density inside the QW’s. On the other hand, trade-off between threshold current, differential gain and refractive index change (Δn) is another important issue at SOA-based devices. Usually, we can get the lowest threshold current by single QW design but higher differential gain will carry out from more QW’s. The Δn simulation relative to the number of QW’s by Chin et al. [2.4] points out that Δn does not have significant increase even at large reverse bias if the number of QW’s is more than three. In this study, we use triple-QW’s design as a compromise between threshold current, differential gain and Δn.. 17.

(35) Fig. 2.2. Wavefunction and the band diagram of six different QW designs for 1.55 μm laser/SOA’s.. 18.

(36) Table 2-1 Details of one fundamental period structure for 1.55-μm strain-balanced QW’s Function. Material Composition. Barrier Wavefunction-adjustment Spacer QW padding QW core QW padding Spacer Wavefunction-adjustment Barrier. In0.438Ga0.386Al0.176As (T) In0.714Al0.286As (C) In0.438Ga0.386Al0.176As (T) In0.67Ga0.33As (C) In0.532Ga0.468As (M) In0.67Ga0.33As (C) In0.438Ga0.386Al0.176As (T) In0.714Al0.286As (C) In0.438Ga0.386Al0.176As (T). (nm) # 4.3 1.6 1.2 1.5 2.2 1.5 1.2 1.6 4.3. g f g d a d g f g. ε|| 0.63 % - 1.28 % 0.63 % - 0.94 % 0.00 % - 0.94 % 0.63 % - 1.28 % 0.63 %. (C : compressive, T : tensile, M: match). 2.3 p-i-n laser/SOA’s structure Besides designing the active MQW’s structure, a frame for making p-i-n laser/SOA’s structure completely is also introduced with a schematic diagram incorporated with its band diagram of conduction- and valance-band as shown in Fig. 2.3(a) and (b); simultaneously detailed layer by layer in Table 2-2. Because growing InP is not possible presented in our MBE system, a 20-nm quaternary InGaAlAs (M) is used immediately as the smoothing layer in the beginning of epitaxy growth. Then, a following lower cladding layer is formed by a 100-nm InAlAs (M) sandwiched in between two pair of thin strain-balanced layers. These two strain-balanced layers are composed of a 2.6-nm In0.714Al0.286As (C) and a 4.4-nm In0.416Ga0.205Al0.379As (T) and functioned as the conduction band grading steps. A 40-nm InGaAlAs (M) in the n-side and a 49.5-nm one in the p-side are applied beside the MQW’s to form the separate-confinement-heterostructure (SCH). In the p-side, we use a 1.82 μm InAlAs (M) as the upper cladding layer, a 30-nm InGaAlAs (M) as the p-contact grading step and a 60-nm InGaAs (M) as the final p-contact layer. In order to avoid p-type impurities penetrating into the MQW’s and ease to achieve the ohmic contact, the. 19.

(37) Be-doping concentration in the p-side is stepped distributed and increased from 1 × 1018 cm-3 to 8 × 1018 cm-3; incorporated with a 75-nm undoped region. Further, the left 4.5-nm p-side and whole 40-nm n-side SCH layers are Si-doped with the same density as used inside MD-MQW’s barriers, 1 × 1018 cm-3; besides, the other layers in the n-layer is kept in a uniform doping distribution of 2 × 1018 cm-3.. Fig. 2.3. Schematic diagram for (a) p-i-n laser/SOA’s structure and (b) band diagram.. According to B.W. Wessels [2.5], we can estimate critical thickness (hc) at different in-plain strain (ε||) as indicated in Fig. 2.4. With this design, which strained-layer thicknesses presented in the structure of “MD3QW” shown in Table 2-2 are all controlled below the predicted critical layer thickness. Because the refractive index (n) of InGaAlAs material system, special for strained ones, is less discussed in early; therefore, which refractive index values of quaternary and ternary. 20.

(38) alloys listed in Table 2-2 for λ = 1.55 μm are deduced from Mondry et al [2.6] and estimated by assuming that each layer is like as an unstrained bulk material and its bandgap is corresponding to one of possible lattice-matched InGa(Al)As material. But the refractive index of InGaAs (M) and n+-InP are referred to Nojima et al [2.7] and Martin et al [2.8], respectively. In order to make the optical mode calculation more easily in a p-i-n laser/SOA structure and considering the electric field of lasing mode is almost parallel to the epitaxial layer (TE-polarized mode), a method to obtain an effective average refractive index for active MQW region is given by nav(||) = ∑ d n. d [2.9], where d is the layer thickness. in the QW’s. Thus, for structure of “MD3QW”, the resulted value of nav(||) = 3.421 is applied to the following optical mode simulation. In Fig. 2.5, we show the refractive index distribution of MD3QW in the transverse direction (y, perpendicular to epi-layers) combining with its related E-field expansion of fundamental slab TE-polarized mode (TE0); the effective index of slab (neff) TE0 is 3.2125. Considering the single-mode-operation ridge laser is desirable in application, a 2-μm wide ridge is simulated to exist only one cross-section (xy) fundamental TE-mode when the etching-depth is 1.79 μm, which is very closed to the InAlAs-InGaAlAs interface; its ridge cross-section profile with the TE0-mode (neff = 3.196) near-field map is illustrated in Fig. 2.6(a). By using fast Fourier transform (FFT) method to transfer the near-field mode pattern, a far-field mode pattern is obtained and shown in Fig. 2.6(b). The simulated result indicates the far-field pattern is still kept in a quite symmetric angle divergence with a FWHM = 35.11o in the lateral direction (x) and a FWHM = 34.4o in the transverse direction (y) as shown in Fig. 2.6(c) and (d), respectively.. 21.

(39) Table 2-2 Structure details of MD3QW Functions. Doping. Composition. -3. (cm ) p-contact layer p-contact grading step layer. p-upper cladding layer 3 p-upper cladding layer 2 p-upper cladding layer 1 i-upper cladding i-region SCH/etch-stop n-side depletion layer Barrier(N-type MD) Barrier Wave-function adjustment Spacer. QW ×2. QW padding QW core. Barrier Wave-function adjustment Spacer. ×1. QW padding QW core QW padding Spacer Wave-function adjustment Hole stopping barrier Lower SCH layer. 30. 3.361. In0.523Al10.477As (M). Be 3 × 1018. 1. In0.523Al. 0.477As. In0.523Al. 0.477As. (M). 1. In0.523Al. 0.477As 2. 2. 2. 0.26Al 0.212As. In0.528Ga. 0.26Al 0.212As. (T). 1. 2. (T). 0.386Al 0.176As. In0.438Ga. 0.386Al 0.176As. Lower cladding. 1. In0.438Ga. 2. 0.386Al 0.176As. 2. In0.67Ga. 0.33As. In0.532Ga. 0.468As. Substrate. 100. 3.200. Be 1 × 10. 18. 100. 3.200. 30. 3.200. Si Si. 45. 3.361. 1 × 10. 18. 4.5. 3.361. 1 × 10. 18. 2.15. 3.344. 2.15. 3.344. 1.6. 3.360. 1.2. 3.344. 1.5. 3.639. In0.438Ga. 2.2. 0.386Al 0.176As. 2. In0.714Al. 0.286As. 1. (T). (C) 2. In0.438Ga. (T). (M). 2. 0.386Al 0.176As. (T). In0.438Ga10.386Al20.176As (T) 1. 2. (T). 1. 2. (T). In0.438Ga. 0.386Al 0.176As. In0.438Ga. 0.386Al 0.176As. 2. In0.714Al. 0.286As. 2. In0.67Ga. 0.33As. In0.532Ga. 3.639. 1.2. 3.344. 1.6. 3.360. 2.15. 3.344. 2.15. 3.344. Si. 18. 2.15. 3.344. 2.15. 3.344. 1.6. 3.360. 1.2. 3.344. 1.5. 3.639. 1 × 10. (C). 0.468As. 1.5. 1 × 1018. (C). 1. [2.6]. 3.594. Si. In0.438Ga10.386Al20.176As (T) (M). 2.2. [2.6]. 3.594. In0.67Ga20.33As (C). 1.5. 3.639. In0.438Ga10.386Al20.176As (T). 1.2. 3.344. 2. In0.714Al. 0.286As. 3. (C). In0.305Ga. 1. 0.417Al 0.278As. 2. In0.528Ga. 2. 0.26Al 0.212As. 0.286As. 2. In0.416Ga. (C). 0.477As. (T). (M) 1. (T). In0.714Al20.286As (C) In0.528Ga. 2. 0.26Al 0.212As. S-doped InP. 22. 3.360. 1 × 10. 3. 3.272. 1 × 10. 18. 40. 3.361. 2 × 10. 18. 2.6. 3.360. 2 × 10. 18. 4.4. 3.236. 2 × 10. 18. 100. 3.200. Si. 2 × 10. 18. 4.4. 3.236. Si. 2 × 1018. 2.6. 3.360. Si. 2 × 10. 18. 20. 3.361. 4 × 10. 18. Si Si Si Si. 0.205Al 0.379As. 2. 1.6 18. Si. 0.205Al 0.379As. 1. In0.523Al. (T). (M). 1. Strain-balanced grading steps In0.416Ga Smoothing layer. Be 2 × 10. (C). 1. 2. for conduction band. 3.200. In0.714Al20.286As (C). Strain-balanced grading steps In0.714Al N-layer. (M). 2. In0.438Ga. 1500. (M). 1. 3.594. 18. (M). 2. In0.528Ga. Be 8 × 10. (M). 1. 2. for conduction band. [2.6]. Be 5 × 1018. (M). Spacer. Barrier (N-type MD). (λ=1.55μm). In0.528Ga20.26Al20.212As (M). 0.468As. 1. Barrier (N-type MD). (nm) 60. In0.532Ga. In0.67Ga20.33As (C). Barrier. n@. 18. QW padding Wave-function adjustment. QW. 1. Thickness. (M). S. [2.7]. 3.156.

(40) Fig. 2.4. Calculated critical thickness (soild-line) and the strained-layer thickness presented in structure of MD3QW (circle) as shown in Table 2-2 at different in-plain strain.. 23.

(41) Fig. 2.5. The refractive index distribution of MD3QW in the transverse direction (y, perpendicular to epi-layers) and its E-field expansion of fundamental slab TE-mode.. 24.

(42) (a). (b). (c). (d). Fig. 2.6. Ridge waveguide laser with width = 2 μm and etching depth = 1.79 μm; (a) the near-field map of fundamental cross-section TE-mode, and (b) its far-field pattern; (c) the divergence of far-field mode pattern in the lateral direction, and (d) in the transverse direction.. 25.

(43) Reference [2.1]. J. C. L. Yong, J. M. Rorison, and I. H. White, “1.3- m Quantum-Well InGaAsP, AlGaInAs, and InGaAsN Laser Material Gain: A Theorectical Study,” IEEE J. Quantum Electron., vol. 38, pp. 1553-1564, 2002.. [2.2]. T. Mukai, Y. Yamamoto and T. Kimura: Semiconductors and Semimetals, ed. W. T. Tsang (Academic Press, New York, 1985) vol. 22, Part E.. [2.3]. K. J. Vahala and C. E. Zah, “Effect of doping on the optical gain and the spontaneous noise enhancement factor in quantum well amplifiers and lasers studied by simple analytical expressions,” Appl. Phys. Lett., vol. 52, pp.1945-1947, 1988.. [2.4]. M. K. Chin, T. Y. Chang and W. S. Chang, “Generalized blockaded reservoir and quantum-well electron-transfer structures (BRAQWETS): modeling and design considerations for high performance waveguide phase modulators,” IEEE J. Quantum Electron., vol. 28, pp. 2596-2611, 1992.. [2.5]. B. W. Wessels, “Morphological stability of strained–layer semiconductors,” J. Vac. Sci. Technol. B, vol. 15, pp. 1056- 1058, 1997.. [2.6]. M. J. Mondry, D. I. Babic, J. E. Bowers, and L. A. Coldern, “Refractive index of (Al, Ga, In)As Epilayers on InP for optoelectronic applications”, IEEE Photon. Technol. Lett., vol. 4, pp. 627-630, 1992.. [2.7]. S. Nojima and H. Asahi, “Refractive index of InGaAs/InAlAs multi-quantum-well layers grown by molecular beam epitaxy”, J. Appl. Phys., vol. 63, pp. 479-483, 1998.. [2.8]. P. Martin, E. M. Skour, L. Chusseau, C. Alibert, and H. Bissessur, “Accurate refractive index measurement of doped and undoped InP by a grating coupling technique”, Appl. Phys. Lett., vol. 67, pp. 881-883, 1995.. [2.9]. P. Bhattacharya, Semiconductor Optoelectronic Devices (2nd, Prentice-Hall, New Jersey, 1994), Appendix 18.. 26.

(44) Chapter 3 MBE Growth of InGa(Al)As Materials and Laser/SOA’s Structures The most important thing toward achieving our complicated QW’s designs is based on high quality MBE-grown materials. Fortunately, the reliable MBE techniques nowadays with helping by various powerful materials characterization skills provide us more opportunities to realize our designs than before. In order to introduce MBE-growth skills to newcomer interesting in this topic, the relevant issues will be focused on the growth-procedures, temperature conditions, molecular beam flux controller, and calibration method of alloys composition, etc. A brief description for MBE machine used in this study will be given in Appendix-C. In reason of high aluminum-content InGa(Al)As and InAlAs alloys being specially difficult materials to grow by MBE system, an optimum MBE growth process is presented in Appendix-D for best-quality alloys. To introduce the detailed cell-flux calibration method for III-V element with routinely materials characteristic analysis, such as double-crystal x-ray diffraction (DCXRD) and photoluminescence (PL) spectra, three basic lattice-matched alloys (InGaAs, InAlAs, and 1eV InGaAlAs) can be firstly obtained and discussed in Section 3.1. Once the growth conditions of the fundamental alloys are ready, various strained InGa(Al)As alloys can be easily achieved by just controlling the statues of cell shutters (Appendix-B). In Section 3.2, depending on in-depth DCXRD analysis and high resolution transmission electron microscopy (HR-TEM) pictures, we further show the designed multi-layer 1.55-μm laser/SOA’s structure (MD3QW), which is chosen as the demonstrative one among those grown samples, is consistent with our original design.. 27.

(45) 3.1. Calibration of cell flux and growth rate Calibration of the growth rates in MBE system is mainly distinguished into two. different categories, in-situ and ex-situ measurements. In field of in-situ way, many tools based on optical technique have been presented for real-time growth rates monitoring, such as pryometric interferometry [3.1], optical flux monitoring [3.2], ellipsometry [3.3], [3.4], and normal incidence reflectometry [3.5], [3.6]. In consideration of their relative complicated setup and high frequently happened problem from the materials coated on the viewport, except for a long-time growth like as Distributed Bragg Reflector (DBR) mirrors, most research people still use to directly measure accurate molecular fluxes by RHEED intensity oscillations. Besides RHEED oscillations, the Beam Equivalent Pressure (BEP) from BEP gauge is another popular in-situ tool. Owing to the BEP measurement is not a precise method for molecular fluxes measuring, so some other ex-situ technique such as SEM, DCXRD, ellipsometer, and PL measurement, etc., must be applied to relate the BEP to a real flux. The biggest difficulty of BEP, however, is that while measuring the flux of a group-III component, the gauge gets coated, causing a rapid loss of sensitivity in the gauge. Although the gauge sensitivity will eventually recover if left running under an As-overpressure, but the measurement is thus uncertain and potentially time-consuming if repeatability is desired. In practical, using in-situ RHEED oscillations and BEP results with helping by routinely ex-situ materials characteristic analysis from DCXRD and PL measurement have been become a very useful method to us for calibrating cell fluxes and can control reproducible layer thickness within a couple of percent. Before growing epitaxial layers, the most important thing is to know how much arsenic vapor pressure can be provided. Also, the relation between arsenic BEP (As-BEP) and its relative background growth chamber pressure (PG) at different arsenic cracker (As-cracker) valve positions is valuable. As Fig. 3.1 indicates, both of As-BEP and PG are exponential increased with a same trend as the As-valve position is linearly increased. Once we obtain. 28.

(46) the relation of As-BEP vs. PG as shown in the inset of Fig. 3.1, it is very advantageous for fine adjusting As-pressure in a reproducible condition by just referring to PG without involving practical As-BEP measurement for each time MBE growth.. 3.1.1. RHEED intensity oscillations When MBE technology is adopted to grow group III-V materials, the epitaxy speed is. generally determined by observing the oscillation of RHEED pattern according to the variation of intensity, so as to determine the epitaxy speed. The oscillation frequency corresponds to a monolayer (ML) growth rate [3.7], where a ML is the thickness of one full layer of group-III atoms and one full layer of group-V ones. The oscillations can be explained by a layer-by-layer growth mode as demonstrated in Fig. 3.2. When the thickness of epitaxy reaches a complete monolayer (ML), it has the highest diffraction intensity. Accordingly, when epitaxy reaches thickness between layer (N) and layer (N+1/2), its diffraction intensity will gradually decrease; when the thickness of epitaxy is within layer (N+1/2), the diffraction intensity reaches minimum; when epitaxy reaches the complete layer (N+1), the signal intensity will return to the highest value, which indicates that oscillation of signal intensity exists as the continuous increase of thickness of epitaxy. Therefore, the growing thickness can be precisely controlled according to the RHEED oscillation curve, with the precision of thickness reaching 0.1 ML [3.8].. 29.

(47) Fig. 3.1. The measured As-BEP and its background pressure (PG) at different As-valve positions; the insert diagram shows the relation of As-BEP vs. PG.. 30.