Compact lightwave components that provide very narrow frequency passbands are integral elements of photonic integrated circuits (PICs) for use in inter- and intra-chip networks. These high bandwidth optical interconnections, which rely on dense wavelength division multiplexing (WDM) in order to fully capitalize on the large optical domain transmission capacity, require filters and switches with narrow bandwidths in order to route optical signals to their appropriate destinations. This routing functionality can be achieved by leveraging the sharp spectral features of resonator devices. Microring resonators, in particular, have produced exceptionally narrow bandwidths, corresponding to ultra-high Q factors [1]-[3]. When this is combined with the high confinement provided by the silicon-on-insulator (SOI) platform [3], very high spatial and spectral density can be achieved. As the data capacity scales in optical networks, both the density of wavelength channels and the single-channel data rate must increase, causing the spectral width of each channel to broaden and necessitating the inter-channel wavelength spacing to decrease. For these systems to maximize their data capacity, the bandwidth of the photonic components used to route individual channels must approach the channel's modulation bandwidth. However, as this occurs, the high-speed optical signal degrades, sometimes severely. We have previously shown experimentally that, as high speed non-return-to-zero (NRZ) on-off-keying (OOK) optical data signals pass through high-Q silicon microring resonators, they are distorted by the non-uniform attenuation of the modulation sidebands [4],[5]. In addition, we have experimentally verified a numerical model, which predicts the degree of signal distortion incurred by microring resonators as quantified by the power penalty [4],[5].
展开▼
