Silicon is of extreme importance to the electronics industry, In recent years, interest in its photonic characteristics has grown. In particular, photonic devices based on the silicon-on-insulator (SOI) material system have advantages because they can be integrated with silicon-based electronic components and fabricated using high-precision mass-manufacturing technologies. In addition to realizing ‘passive’ optical functionalities such as guiding light in SOI waveguides, this system is also suitable for making ‘active’ photonic devices, many of which have a working mechanism that relies on Raman scattering in silicon. In SOI-based Raman lasers, for example, incident pump photons are scattered off of a thermal-lattice vibration in the silicon to generate ‘Stokes lasing photons’ with a longer wavelength. This is referred to as stimulated-Stokes-Raman scattering (SSRS) and is also responsible for the light amplification in SOI-based Raman amplifiers. To fully exploit the tremendous application potential of these active SOI-based Raman devices in, e.g., telecom and computing, dense integration of these components on silicon chips is required. However, this will be feasible only if their heat dissipation can be kept low, which is a challenging goal.
Figure 1. Schematic representation of coherent anti-Stokes Raman scattering (CARS), (a) converting a Stokes photon and a pump photon to an anti-Stokes photon and a pump photon, while annihilating two phonons in the medium, and (b) converting an anti-Stokes photon and a pump photon to a Stokes photon and a pump photon, while creating two phonons in the medium. |i> and |f> represent the ground energy level and the material energy level involved in the Raman interactions, respectively, while ωa, ωp,and ωs are the anti-Stokes, pump, and Stokes photon frequencies, respectively.
We have found that coherent anti-Stokes Raman scattering (CARS) can provide a viable solution to this important challenge. CARS is a four-wave mixing process that—as opposed to SSRS—can also generate short-wavelength anti-Stokes photons. It represents the basic operation mechanism of Raman wavelength converters that transform a Stokes input signal to an anti-Stokes output signal, or vice versa.1 However, we have learned that CARS can also extract vibrational quanta or phonons (i.e., heat) from the medium. Therefore, when CARS is induced in an SOI Raman laser/amplifier medium in addition to SSRS, one can reduce the heat dissipation in the device.2 This CARS-based heat-mitigation technique for Raman lasers and amplifiers represents an optical, intrinsic heat-reduction technique that—as opposed to conventional external cooling approaches—does not inhibit extensive device miniaturization.
We derived the ability of CARS to extract heat directly from the well-established Raman propagation equations, which indicate that—when working at exact Raman resonance (the most common working point in Raman lasers and amplifiers)—the CARS photon and phonon balance obeys the energy scheme of either Figure 1(a) or 1(b). We have also found that when CARS is quasi-perfectly phase matched, it will continuously convert Stokes photons to anti-Stokes photons, while extracting phonons from the medium, as shown in Figure 1(a).3 Thus, if CARS is triggered under such conditions, heat dissipation in the Raman device can be suppressed. This heat generation is mainly due to SSRS, which converts high-energy pump photons to low-energy Stokes photons, and is thus accompanied by phonon creation.
To verify the viability of this heat-mitigation technique for Raman lasers, we numerically calculated the attainable heat-mitigation efficiency for SOI-based Raman lasers operating in the mid-IR, using the iterative resonator method that we developed to model Raman lasers. We have found that, after applying certain efficiency-enhancement methods,4 heat dissipation could be reduced by as much as 35%.5 For mid-IR SOI-based Raman amplifiers, one can expect a heat-mitigation efficiency on the same order of magnitude.
In conclusion, we have introduced a new approach for mitigating the heat dissipation of SOI-based Raman lasers and amplifiers in which CARS plays a crucial role. We are currently investigating our heat-mitigation technique experimentally to construct a proof-of-concept demonstrator, as this could accelerate adoption of SOI Raman devices in, e.g., telecommunication systems and network-on-a-chip applications. This would bring silicon photonics one step closer to becoming a comprehensive, low-cost, energy-efficient, and practical technology platform with applications in everyday life.
Nathalie Vermeulen, Christof Debaes, Hugo Thienpont
Department of Applied Physics and Photonics
Vrije Universiteit Brussel (VUB)
Nathalie Vermeulen obtained her PhD in 2008 from the Applied Physics and Photonics Department at the VUB. She currently works at the VUB as a postdoctoral researcher funded by the Fonds Wetenschappelijk Onderzoek (FWO). Her research efforts focus on modeling lasers, amplifiers, and converters based on Raman scattering and other nonlinear optical processes, and on the development of mid-IR laser sources.
Christof Debaes received his PhD from the Applied Physics and Photonics Department at the VUB and the Ginzton Laboratory, Stanford University, directed by D. A. B. Miller. He currently works as an FWO-supported postdoc on optical interconnects, Raman lasers, architectural studies for reconfigurable optical interconnects, and the development of deep-proton writing for microoptical modules.
Hugo Thienpont became professor in 1994 at the Faculty of Applied Sciences, with teaching responsibilities in photonics. In 2004, he was elected chair of the Applied Physics and Photonics Department. Currently, he is coordinator of several basic research and networking projects such as the European Network of Excellence in Micro-Optics. In addition, he manages microphotonics-related industrial projects with several companies.