Abstract as per original application
(English/Chinese): |
Millions of silicon photonics optical transceivers, mostly 100GbE today but migrating to 400GbE over the next five years, are being deployed in data centers which serve more than 90% of the global internet traffic. Waveguide grating couplers (WGC) are widely used in silicon photonics for coupling light between optical fibers and silicon waveguides. WGC have the advantages of not being limited in placement to the edge of the silicon chip, compatibility with wafer scale testing for selecting dies before the expensive process of photonic packaging, do not require (potentially low yield) facet polishing, have lower alignment precision requirement than edge couplers and, in the future, they may be used with spatial division multiplexing using multicore or multimode fibers. Current WGC are unsuited for use with wideband wavelength division multiplexing (WDM) because their optical bandwidth is insufficient, with only about 29nm 1dB optical bandwidth being typically offered by foundries. Edge couplers have over 100nm optical bandwidth, but being limited to the one dimensional edge of a planar waveguide, are unsuited for direct interface with the multicore fibers which can enable future increases in bandwidth density. 100GbE transceivers today can use four wavelengths in the Local Area Network (LAN) WDM grid (channel spacing of about 4.5nm) which are supported by current WGC technology. But more wavelengths, or the use of the ITU-defined 20nm CWDM channel spacings, will require WGC to have larger optical bandwidths. CWDM distributed feedback (DFB) lasers do not need precise temperature control nor precise wavelength registration, and therefore are low-cost and more suitable than the dense WDM DFB lasers developed for telecommunications. Low cost is essential for massive deployments in short reach intra-datacenter interconnects. For example, for 400GbE, eight WDM channels (instead of 4 WDM channels in 100GbE) is needed in one variant of the 400GbE standard. Further advances towards future 1.6Tb/s ethernet transceivers may need even more wavelength channels. In this proposal we aim to develop a new type of WGC which offers 1dB optical bandwidths of over 110nm. Such unprecedented wide optical bandwidths WGC will find applications in future wideband optical transceivers. In this project we propose the novel concept of Resonance-enhanced Waveguide Grating Coupler (REWGC) which we shall study in detail theoretically and experimentally. The REWGC will offer unprecedented optical bandwidth for use with wideband CWDM optical transceivers for future ultrahigh capacity data transmission.
數以百萬計的矽光收發器（目前主要是 100GbE，但在未來五年內將遷移到400GbE）正在數據中心部署，這些數據中心為全球 90%以上的互聯網流量提供服務。波導光柵耦合器 (WGC) 廣泛用於矽光子學中，用於在光纖和矽波導之間進行耦合。與邊緣耦合器相比，WGC 的優點是不受耦合位置的限制，與晶圓級測試兼容，可以在昂貴的光子封裝工藝之前，用於選擇芯片，不需要端面拋光（潛在的低良率），對耦合對準精度要求比較低，並且在未來，WGC可與多芯或多模光纖一起使用，進行空分複用。由於帶寬不足，當前的WGC不適合用來寬帶波分複用 (WDM)，代工廠通常只提供1dB帶寬大約為29nm的WGC。邊緣耦合器具有超過100nm的帶寬，但僅限用於平面波導的邊緣，不適合與實現帶寬密度提升的多芯光纖直接耦合。現在的 100GbE 收發器可以在局域網 (LAN) WDM 中使用四種波長（大約 4.5nm 的信道間隔），當前 WGC 支持這些波長。但是更多的波長，或使用 ITU 定義的 20nm CWDM 信道間隔，將需要 WGC 具有更大的帶寬。與為電信開發的密集 WDM 分佈式反饋 (DFB) 激光器相比， CWDM DFB激光器不需要精確的溫度控制和精確的波長配準，因此成本更低。低成本對於短距離數據中心內部互連的大規模部署至關重要。例如，在 400GbE 標準的一種版本中400GbE需要 8 個 WDM 通道（而不是 100GbE 中的 4 個 WDM 通道）。未來 1.6Tb/s 以太網收發器的進一步發展可能需要更多波長通道。在這個項目中，我們的目標是開發一種新型寬帶WGC，可以提供超過 110nm 的 1dB 帶寬。這種前所未有的寬帶 WGC 將在未來的寬帶光收發器中發揮重要應用。我們提出了共振增強波導光柵耦合器 (REWGC) 的新概念，我們將在理論上和實驗上詳細進行研究。REWGC 將提供前所未有的帶寬，用於未來超高容量數據傳輸的寬帶CWDM 光收發器。
|
