摘要:本計畫的主要目的在研製由分子束磊晶成長(MBE)或金屬有機化學汽相沉積(MOCVD)成長之「垂直共振腔電晶體雷射」(Vertical-Cavity Transistor laser, VCTL),一種全新的三端半導體光電元件,同時具備電晶體和雷射的兩種特性,能夠同時輸出「電」和「光」的訊號,並達成在室溫下以「電壓調變」的方式,提供超高速的電、光直接調變訊號,提供低能耗、低成本,高傳輸量、高資料密度的傳輸光源,以利用在下一代的高速電腦運算系統。「電晶體雷射」提供了快速的載子復合放光生命週期(~ps),理論上其高速直接調變的頻寬可達100 GHz,這是過去傳統二極體雷射,垂直共振腔面射型雷射(Vertical Cavity Surface Emitting Laser, VCSEL)所無法達成的。垂直共振腔電晶體雷射的原理,是利用在異質接面雙極性電晶體(HBT)的基極(base)中至入量子井(Quantum well)來增進其放光性的電子電洞復合,再利用設計好的垂直共振腔(半導體布拉格反射鏡面,或外部介電材質鏡面),達到光場的侷限,使其達到雷射共振輸出,和「邊射型電晶體雷射」(Edge-Emitting Transistor Laser)相比,將具有較低的雷射臨界電流,較高的光場增益,及較高的調變頻寬。在過去五十年,所有的相關研究都將載子的生命週期認定在1 ~ 2 ns,而發光二極體(LED)的自發性放光(spontaneous emission)的調變速度始終在幾百個MHz。二極體雷射的調變響應,也因為載子的復合生命週期太長,造成元件有很大的共振頻率響應(resonance peak),造成光訊號的失真和傳輸頻寬的下降。「電晶體雷射」提供了另一種「電壓調變」的機制,更適用在未來的光積體電路中的電壓操作。過去的研究顯示,在電晶體雷射中由於載子復合的生命週期在幾個ps範圍,所以直接調變的響應並沒有共振峰值的現象,頻寬不會受到共振頻率的限制。過去我們已有「邊射型電晶體雷射」成功製作的經驗,以長度400微米,單一量子井的結構,可同時達到40 Gb/s的光、電訊號的輸出。因此,我們希望藉由此研究計畫,縮小雷射的共振腔,變成「垂直共振腔」的結構,成功製作出第一個室溫操作的「電壓調變」的高速垂直共振腔面射型電晶體雷射,預期傳輸速度可以超過40 Gb/s的光、電訊號頻寬,並持續微小化,朝微米、奈米雷射前進,開發具有潛力的光積體電路元件,提供光通訊系統中所需要的更快速、穩定、便宜的光源。
Abstract: The proposed research topic is to develop and characterize a novel semiconductor optoelectronic device grown by MBE and MOCVD, the vertical cavity transistor laser (VCTL), which is a three-port device with simultaneous electrical and optical output. The goal is to establish the physical foundation and conduct rigorous design verification of the first room temperature operation of voltage-modulated vertical-cavity surface-emitting laser (VCSEL) source based on the novel transistor laser structure, to provide a direct current-modulated solution that offers low power and low cost high-speed performance towards realizing the next generation data communication for high-throughput, high-density interconnects in high-end supercomputing systems. The transistor laser offers the advantages of fast carrier recombination lifetime and fast optical modulation approaching 100 GHz; features that are not available to diode VCSELs. The transistor laser employs a novel strategy of inserting a quantum-well in the transistor base in order to enhance radiative electron-hole recombinations, and a suitable vertical photon resonator cavity for surface-emitting laser emission. The material structure will employ a heterojunction bipolar transistor (HBT) design. This structure has been shown to achieve world-record high-speed operation (ft >800 GHz). The HBLET, inherited from the heterojunction bipolar transistor (HBT), not only functions as a transistor with a family curve (I-V), Gummel-plot, and electrical current gain (=IC/IB), but also emits significant optical output controlled by the base recombination current IB to form a three-port optoelectronic device. To control the base recombination (i.e. light output) is therefore the most important part to study the LET performance. Considered in a conventional n-p-n transistor, carriers (here are electrons) injected from emitter side (IE) will transport through base region and be collected by collector and from the collector current (IC) due to the reverse bias between base-collector junction. Part of the carriers (electrons) within the base transit time (in ps range) will recombine with holes in the base region and form the base current (IB), but most of the carriers will transport to collector side (IC) resulting in a current gain much greater than one (=IC/IB ~100). Here the essential concept of a light-emitting transistor is to enhance base recombination “radiatively” so that we can utilize the optical signal as an additional channel for communication and interconnect. We employ an undoped quantum-well (QW) inside the heavily-doped base region since a QW is considered an effective recombination center. In this way, we will have two collectors in our LET system: one is the “electrical” collector providing electrical output (IC), the other one is the “optical” collector providing optical output (L). The “optical” collector only collects “fast-recombining” carriers because the base transit time is only in ps range, leading to a fast recombination lifetime, which is considered about ~ 1 ns for the past 50 years in the light-emitting diodes and diode lasers. So the LED modulation bandwidth is limited in few hundred MHz, and all the optical emitter is in the form of diode lasers. However, the long recombination lifetime causes the modulation response to create a resonance peak easily greater than 10 dB, resulting in a signal distortion and limited bandwidth. Additional low-pass filters or external modulators are necessary to maintain the transmission quality at the cost of large energy dissipation. Besides, the diode laser can only be modulated through current modulation. Transistor lasers provide another voltage-modulation mechanism, which is suitable for voltage operation in the future photonic integrated circuits. From the previous experimental results, the edge-emitting TL has demonstrated an almost resonance-free modulation bandwidth. Without any additional coating, a 400 m TL with single-QW has shown the capability of 40 Gb/s electrical and optical data rate output. Therefore, we hope through this proposed investigation, we can further shrink the resonance cavity from few hundred micro meter to few nano meter and form a vertical cavity transistor laser operating in room temperature with simultaneous larger than 40 Gb/s modulation bandwidth. In this way we can explore the possibility for future photonic integrated circuits and a cheaper and faster light source in fiber communication system.