Silicon- and plasmonics-based nanophotonics for telecom and interconnects

Silicon nanophotonics became recently a very promising solution for high-speed signal transmission and processing with high bandwidth and very compact size, even for multifunctional integrated structures. With sub-micron waveguide cross section and bending radius down to 2 μm with acceptable losses (0.04 dB/90º bend) these structures give promising perspectives for electronic — photonic integration on a common Si platform since they can be realized by conventional planar CMOS techniques. Many research groups, both in academia and the electronics industry, have demonstrated different passive and active components, as well as integrated functional devices. However for more complex optical architectures, for example in optical interconnects for inter- and intra-core data communication, even higher integration density and modal field confinement below the diffraction limit of light, are necessary. For downscaling of photonic components, different solutions based on surface plasmon waveguiding along metal — dielectric interfaces have been proposed including strip-line, slotline [1] and v-grove [2] structures, as well as some more sophisticated multi-layered systems [3] and quantum dot arrays [4]. Unfortunately, most of them suffer from the high losses always associated with maximum light confinement, resulting in short propagation lengths of the order of several micrometers. Hybrid plasmonic structures, which consist of a low refractive index slot sandwiched between e.g. a gold layer and a silicon material, appear to be a good solution [5-7]. In such structures, light is partly localized in the low-index dielectric and partly in the high-index Si giving relaxed conditions for propagation of the hybrid mode, but still with high confinement. Losses as low as 0.01 dB/μm (propagation length over 400 μm) have been obtained experimentally [8]. This is still orders of magnitude higher than those of conventional photonic waveguides, but acceptable for many functional components, not only to decrease the overall size of the structure, but also due to some specific advantages plasmonic components can bring forward, including temperature and production error tolerances, energy efficiency, large Purcell factor and others. So the best choice will probably be to use some kind of hybrid nanoplasmonic-photonic structures with conventional photonic waveguiding and functional plasmonic components. Hybrid plasmonic microring modulators, for example, with sub-micron radius have an intrinsically low quality factor Q, which on one side causes low channel bandwidth of these devices, but simultaneously allows for temperature and production error tolerances. The Q value of a 500 nm radius microring is anyway 8 times higher than the Q of similar Si slot microrings [9]. According to our calculations done for modulators using electrooptic polymer (EOP), even with a moderate electrooptic coefficient of 80 pm/V, due to the small footprint, the device capacitance is very small allowing for RC-limited modulation speed of 190 GHz (modulation speed limited by the cavity photon lifetime is larger than 300 GHz) and giving the average power consumption of only 5 fJ/bit. Similarly for lasers, due to the small cavity volume we should be able to get lower threshold power for lasing than is the case for traditional laser designs.