Slow light in optical circuits: Harnessing on-chip-stimulated Brillouin scattering
Slow light deals with the proposition that the speed of a light pulse, also called group velocity, can be substantially reduced compared to its speed in vacuum and even controlled by using the large dispersion associated with a material or structural resonance. The field of slow and fast light has captured the imagination of physicists for more than a decade because of its importance in fundamental physics1 and novel applications. Since the breakthrough demonstration in a cold atomic gas,2 a number of slow- and fast-light techniques have been developed for applications such as tunable delay, nonlinearity enhancement, and microwave signal processing.3 From an applications point of view, realizing slow light in a chip-scale device is important for all-optical integration and compactness, and it may lead to novel photo-phononic devices.
On-chip control of light pulse speed entails generating a large group index (ng) in a short length to achieve appreciable group delay (Td= Lng/c). Stimulated Brillouin scattering (SBS)-based slow and fast light has been at the forefront of techniques used for controlling light pulse speed due to its wavelength independence, room temperature operation, delay tunability, and wide signal bandwidth range (MHz-GHz).4–6 The scattering occurs when light interacts with time-dependent optical density variations. The small Brillouin scattering cross-section and large optical mode area of silica optical fiber, which is the commonly used platform for the SBS slow and fast light, has prevented large group-index generation at chip scale without incurring large power consumption (tens of Watts).
Chalcogenide (As2S3) glass, with its large nonlinear index of refraction (n2) resulting from its large refractive index (n), has been shown to be a competent platform for on-chip, ultrafast, all-optical signal processing.7–9 The large n of chalcogenide glass results in a large Brillouin scattering cross-section due to its strong dependence on the refractive index (∝n8).10 However, having large n alone is not sufficient for strong light-sound interaction. Further restrictions are imposed by the need for strong optical-acoustic overlap, which requires both the optical and acoustic modes to be strongly confined in the same structure, and the smaller optical mode area for efficient operation.
For a chalcogenide chip fabricated on a silica substrate, the large index contrast with substrate facilitates fabrication of a device with a smaller optical mode area and strong optical confinement. The smaller sound speed in chalcogenide glass (va∼2550m/s) compared to silica (va∼5600m/s) results in strong acoustic mode confinement (see Figure 1), and thus strong light-sound interaction, making it the material of choice for chip-scale SBS slow and fast light.
In our recent work, we exploited the large SBS cross-section and small-optical mode area of our 7cm long chalcogenide photonic chip to realize strong SBS10 and SBS-based slow and fast light.11 Figures 2(a-b) show the measured group-index change (Δng) for the slow and fast light regimes, respectively, demonstrating a large Δng∼68 for the slow light, measured at a gain of ∼11.3dB, and a negative group-index of -44 for the fast light.
In our demonstration, pulse delay was measured using a quasi continuous wave (CW) pump to increase the pump power while maintaining low average power. A maximum delay of 23ns was achieved for a Gaussian pulse with full-width at half-maximum (FWHM) width of 100ns, which results in a large Δng∼130, the largest Δng ever achieved for SBS slow light. The corresponding light speed of 2307km/s was achieved in these experiments. The short device length, ∼30 times smaller than the length used in an earlier device with record ng,12 and low power (∼300mW) consumption makes it a compact, efficient device for all-optical tunable delay and microwave signal processing. Using SBS slow and fast light, pulse delay can be continuously tuned by varying the pump power (see Figure 3). For a pulse with FWHM of 25ns, we achieved a maximum delay of 22ns, which is nearly one pulse width (see Figure 3).
Many of the existing slow-light schemes are limited in terms of wavelength of operation, delay tunability, and signal bandwidth. SBS-based slow light provides tunable delay and wavelength-independent operation. In addition, the signal bandwidth can be varied by tailoring the pump spectrum, so it can accommodate signals with different bandwidths. In recent work,13 we performed a proof-of-concept experiment where the 3dB bandwidth (f3dB) of the SBS profile was doubled by tailoring the pump profile.
Figure 4(a, c) shows that by tailoring the spectrum from a single-pump to dual-pump configuration, the overall SBS gain profile (solid, see Figure 4c), which results from two individual gain profiles (dashed, see Figure 4c), has a flat top shape and larger f3B. Figure 4(b, d) shows the measured gain profiles for a single CW pump and two CW pumps (separated by 18MHz), respectively. Using two pumps, a nearly flat-top SBS gain profile is achieved with f3dB improved from ∼20MHz to ∼40MHz, demonstrating that the bandwidth of the SBS gain spectrum—and thus of a SBS slow and fast light system—can be increased.
Harnessing SBS slow and fast light in a chip-scale device enables on-chip tunable delay lines, which have application in microwave signal processing as tap delay lines for microwave photonic filters.14 Exploiting SBS in a chip-scale device, in general, opens up a number of applications ranging from Brillouin lasers15–17 and microwave signal processing13, 14,18 to tailoring light-sound interaction19–21 for enhanced opto-mechanical interaction.21
In the future, we plan to exploit SBS slow light-based tunable delay in our device to realize on-chip microwave photonic filters and true-time delay for a phase-array antenna. In general, on-chip SBS will be exploited for chip-scale Brillouin lasers and for generating frequency combs.
This work was funded by the Australian Research Council (ARC) through its Discovery grant (DP1096838), Federation fellowship (FF0776056), and Center of Excellence: Centre for Ultrahigh bandwidth Devices for Optical Systems (CUDOS) grant #CE110001018); and by the U.S. Department of Defense through Air Force Office of Scientific Research/Asian Office of Aerospace Research and Development (grant #FA23861114030). We acknowledge the contributions of Adam Byrnes, Enbang Li, and our collaborators Christopher Poulton at the University of Technology, Sydney, and Duk-Yong Choi, Steve Madden, and Barry Luther-Davies at the Australia National University.
School of Physics
University of Sydney
Ravi Pant is an Australian Research Council (ARC) postdoctoral fellow at CUDOS, University of Sydney. His research interests include developing novel devices using nonlinear photonics and via tailoring photon-phonon interaction in nanostructures.
Benjamin J. Eggleton is an ARC Federation fellow, director of CUDOS, and professor at the School of Physics, University of Sydney. He is a fellow of the Optical Society of America and IEEE.