邊射型雷射光纖耦光效果與聚焦離子束蝕刻反射鏡之研究

Abstract

在積體晶片的時代，積體光學已成為突破摩爾定律縮放極限的重要途徑。在這些微型光路上，各類波導與耦合器之間的光機問題，如對準誤差，構成了重大的障礙。本文嘗試直接以雷射光作為訊號光源，使用 Ansys 的商用軟體 Lumerical FDTD，將雷射訊號耦合進各種情境並檢驗對準容忍度，同時以自製元件進行耦合實驗，並將結果與模擬相互比對。 基於本實驗室設計的 C-band 邊緣放射型 Fabry–Pérot 雷射，並借助 Ansys Lumerical MODE，我們可從磊晶結構與元件結構出發建立模型，用以研究脊型波導對雷射模態與等效折射率的影響，並將雷射模態匯出以應用於不同的模擬場景。 接著參考 ITRI 的邊緣耦合器位移容忍度實驗，我們將入射光源與光纖模態進行 mode expansion ，直接計算光源在上下左右位移時，有多少比例的光功率可轉換成光纖內的模態。利用此方法模擬出的結果與 ITRI 的實驗數據高度吻合，並進一步應用到雷射與光纖耦合的位移容忍度測試。由於模擬的耦合效率下降比實驗結果低，我們藉由修正雷射光的模態，並考慮了光纖有角度偏移，藉由修正模擬的方式，嘗試瞭解實驗用的雷射模態遭遇的狀況。 在完成上述兩項研究後，我們確認各類耦合器都面臨對準誤差的挑戰。我們嘗試利用聚焦離子束（FIB）在雷射的脊型波導上蝕刻分佈式布拉格反射器（DBR）以提升輸出功率。而結果無法達到模擬的預期。首先透過SEM照片我們發現FIB 向下蝕刻時難以形成完美垂直的鏡面。藉由Ansys Lumerical FDTD模擬錐狀的光柵，發現反射模態會偏移。上述因素使 DBR 的實際效能不理想。經過模擬實驗得出理想的DBR鏡面須完全切穿基板，但是要面對使用聚焦離子束切穿量子井則會嚴重傷害元件，因此在權衡之下，我們最終放棄了這個方法。

In the era of integrated chips, integrated photonics has emerged as a principal avenue for overcoming the scaling limits of Moore’s law. On these miniature photonic circuits, opto-mechanical issues—such as alignment errors between various waveguides and couplers—pose significant obstacles. This work utilizes the laser itself as the signal source and, using Ansys’s commercial Lumerical FDTD, couples the laser signal into representative scenarios to evaluate alignment tolerance. It also conducts coupling experiments with in–house–fabricated devices and benchmarks the measurements against simulations. Building on a laboratory-designed C-band edge-emitting Fabry–Pérot laser and leveraging Ansys Lumerical MODE, we construct models of the epitaxial stack and device geometry to investigate how the ridge waveguide influences the laser mode and effective refractive index. We then export the laser mode for use across diverse simulation contexts. Using the mode expansion method validated against ITRI's edge coupler data, we simulated the laser-to-fiber coupling tolerance. While the initial model showed good baseline agreement, the experimental coupling efficiency exhibited a steeper decline under displacement. To reconcile this, we adjusted the simulation to include laser mode variations and fiber angular misalignment, thereby providing insight into the realistic modal conditions of the laser used in the experiment. After completing these two lines of study, we confirm that alignment errors are a pervasive challenge across coupler types. We therefore attempted to enhance output power by milling distributed Bragg reflectors (DBRs) into the laser ridge using focused ion beam (FIB) processing; however, the results fell short of simulation-based expectations. Scanning electron micrographs reveal that downward FIB etching struggles to produce perfectly vertical facets. Complementary Ansys Lumerical FDTD simulations of tapered gratings show that the reflected mode becomes displaced, further degrading performance. Collectively, these factors render the DBR implementation non-ideal. Although an ideal DBR mirror would require entirely cutting through the substrate, ion-beam penetration through the quantum wells causes severe device damage. Balancing performance against reliability, we ultimately abandoned this approach.