Nonlinear silicon photonics
The phrase ‘silicon photonics’ has captured the excitement and dynamic growth of a powerful, new integrated-optics technology.1 Use of silicon (Si) and silicon dioxide (SiO2) as an integrated photonics platform employs the high-refractive-index contrast between both materials to dramatically reduce the dimensions of an integrated circuit as well as the tools and advances of modern CMOS to fabricate complex but scatter-free shapes. These advances have, until recently, focused largely on linear Si photonics. However, equally exciting progress can be realized with nonlinear-optical Si photonics. Much work in this area is based on nonlinear effects in so-called Si-photonic wires.2–5 These have ‘wire’ cross sections of <0.1μm2 (hence the common reference to ‘nanophotonic wires’) and are formed through lithographic patterning on a thin, single-crystal layer on SiO2. Typical losses are ~1–3dB/cm.
Nonlinear effects in Si wires are striking because of the extremely large third-order nonlinear-optical susceptibility of single-crystal Si (approximately 3–4 orders of magnitude larger than that of silica). This large cubic nonlinearity, combined with strong optical confinement, leads to further enhancement of the effective optical nonlinearity. This results in a relatively low optical-power requirement (threshold) for achieving strong nonlinear-optical effects, as well as very short devices with lengths on the order of a few hundred microns to millimeters. In addition, because of the submicron dimensions of Si-photonic wires, their dispersion properties are markedly different from those of either standard optical fibers or even Si waveguides that have cross-sectional dimensions of a few microns. In particular, because of their ultrasmall dimensions, their dispersion can be controlled by the exact geometry of their cross-sectional area or by application of a thin overlayer. This control enables tailoring their basic dispersion characteristics (such as the group-velocity dispersion), which is important to control nonlinear-optical functionalities (because it affects interaction time and phase matching).3,5 Because of this combination of controllable dispersion and low linear loss, these wires are referred to as ‘fiber on a chip.’
Conventional fiber optics is an enabling technology for light transport as well as a means or device to manipulate the phase, timing, envelope, and wavelength of optical pulses. Using recent technology, these fiber-optic functions can be accomplished entirely with Si-photonic wires. Our group recently carried out basic systems-link performance measurements of multiwavelength data streams transmitted onto a single Si link. In particular, we explored the use of these small waveguides as miniature chip-scale ‘optical fibers’ for transmission of very high bit-rate data signals (>300Gb/s) to communicate among on-chip components. In fact, using wavelength-division multiplexing techniques in conjunction with high bit rates at each wavelength, we showed that these high data rates can be obtained at excellent bit error rates (BERs). Our experiments also show clearly that Si-photonic wires have the potential to enable nonlinear-optical control of on-chip functions or integrated photonic circuits.
In fiber-optical communications systems, two formats are used in intensity-modulated on-off keying (OOK) direct-detection systems, return-to-zero (RZ) format, in which a power is transmitted for a fraction of the bit slot, and non-return-to-zero (NRZ) format, for which a constant power is transmitted during the entire bit slot. The RZ format is attractive in long-haul networks because of its tolerance to fiber nonlinearities, while the NRZ format is common in metropolitan-area networks. We recently demonstrated format conversion between 10Gb/s NRZ-OOK and RZ-OOK in a compact (4mm-long), passive Si-photonic wire (see Figure 1) using cross-phase modulation. We achieved a 3dB, 10−9 BER receiver sensitivity enhancement for converted RZ-OOK relative to NRZ-OOK (see Figure 1, inset).
Finally, one important limitation for some nanowire applications is the presence of two-photon absorption in Si, which generates free carriers and can increase insertion losses in the wires. This limitation can be ameliorated by shifting to wavelengths of ~2μm. Our recent experiments show that two-photon absorption can be reduced at these IR wavelengths and increase nonlinear performance in waveguides. Thus, use of 2μm wavelengths represents a new horizon for nonlinear on-chip sources and their potential applications, which we will focus on next.
Richard Osgood is a professor of electrical engineering and applied physics. His research areas include condensed matter physics, physical chemistry, and optical physics. He was associate director of Brookhaven Laboratory and acting director of its Nanocenter from 2000 to 2002. In 1991, he received the Optical Society of America's Wood Prize.
