Quantum photonics: Photon generation using silicon
In 2012, we demonstrated telecommunications-band heralded single photons from a CMOS-compatible silicon nanophotonic chip [APL 2012]. The chip operated at room temperature (uncooled) and is optically pumped with a few milliwatts of pump power at 1550 nm, easily available from laser diodes. More recently, we have extended this work to improve CAR significantly, and demonstrate thermo-optic tunability of the wavelength of the generated photon pairs [Opt Lett 2013]. The devices we have used provide large (20X) resonant enhancement of silicon's chi-(3) nonlinearity.
In [OpEx 2013], we described an analytical and computational method for calculating the pair-generation properties of such devices, including the joint spectral intensity (JSI) of the Stokes and anti-Stokes photons. In [Nature Comm 2014] we experimentally demonstrated different JSI's generated by a silicon chip, and in [Opt. Express 2015], verified that they have different degrees of entanglement as measured by unfolded and folded Franson interferometry.
Recently, we have shown pair generation from Si microring resonators, including the use of a small amount (microwatts) of pump power [Opt. Express 2016], and monitoring the microring resonance using an in-built Si p-i-n diode which enables microring pair generation over 40°C range of temperature variation [APL 2015].
Nonlinear silicon photonics
A new generation of deployable optical signal processing and communications devices can be made if we can demonstrate CMOS-compatible chip-scale devices which can perform ultrafast wavelength conversion of 1-1000 GHz bandwidth signals using as little pump power as possible, e.g., directly from an unamplified laser diode. This research involves optical dispersion engineering, electronic carrier management, and in some cases, thermo-optic tuning.
Using reverse-biased silicon microrings, CW four-wave mixing conversion efficiency of -13.4 dB is achieved using 20 um radius micro-ring resonators with only 2.5 mW pump power [IEEE PTL 2013]. Silicon rib waveguides (passive loss of 0.74 dB/cm) with free-carrier lifetime reduction via reverse biased PIN diodes show CW four-wave mixing efficiency of -8.2 dB (-4.4 dB if both signal and idler are measured at the output) with 160 mW pump power (about 2X reduction compared to other reports in literature) [CLEO 2013].
We have improved the wavelength conversion efficiency of slow-light enhanced FWM in silicon Coupled-Resonator Optical Waveguides by 20 dB from our earlier report in [Opt. Lett. 2011] to our recent report in [Opt. Lett. 2014], showing 10 Gbps NRZ eyes for the first time.
Tunable higher-order microring filters were integrated on the same chip as a carrier-swept ring mixer, to separate the generated idler wavelength from the residual pump and unconverted signal [Opt Lett 2014].
Silicon waveguides can be used as a compact source of longer wavelength (2.4 um) infrared light via four-wave mixing in conjunction with fiber-optical seed sources [Nat Photon 2010].
Disorder in silicon photonics
Nanoscale disorder has a large impact on the performance of in high-index contrast silicon-on-insulator (SOI) photonic resonators, and in particular, on slow wave structures [Opt Lett 2007]. We reported that disorder-induced localization of light in compact SOI CROWs leads to spatially concentrated and locally trapped light in a quasi-one-dimensional waveguide at wavelengths near the band edge [Nat Photon 2008]. More recently, we have demonstrated favorable statistical intensity and group delay distributions of extremely long microring CROWs (coupled-resonator optical waveguides) [Opt Express 2010] and investigated how many hundreds of silicon microrings can be coupled before disorder-induced bandwidth collapse occurs [Opt Lett, 2011].
[OFC11-PDP] At the OFC 2011 post-deadline session, we reported a set-and-forget method for tuning silicon nanophotonic microrings and Mach-Zehnder interferometers, achieving 0.002 nm (0.2 GHz) precision in aligning resonances, and zero power consumption after modification to "hold" the tuned state. The method was a scanning-probe lithography technique, based on nano-oxidation of silicon induced by electrically-biased metallic AFM tip, guided by real-time measurements of the optical transmission through the device as it is being trimmed.
Infrared imaging of silicon photonic circuits
We developed a quantitative method for extracting microring loss, ring-to-waveguide coupling coefficients, finesse and quality factor etc. from infrared microscope images of guided light. This work, published in [Opt Lett 2010] was highlighted as an OSA Spotlight in Optics. Two years later, we enhanced this method by using high dynamic range (HDR) "bracketing", i.e., a computational image composition method based on a sequence of acquired images at different exposure settings. HDR de-noising overcomes one of the main limitations of conventional imaging infrared cameras at 1550 nm: their limited usable dynamic range (typically, 20 dB) compared to fiber-based single-pixel detectors (usually 50 dB) [Opt Lett 2012].
Silicon photonics in data centers
Using a compact (0.03 mm^2) silicon photonic thermo-optic 1x2 switch, based on cascaded broadband Mach-Zehnder interferometers (designed and fabricated by Sandia), we demonstrated microsecond cross-bar switching of twenty wavelength channels of the [MORDIA] microsecond-scale circuit-switched data-center ring network. Each of the wavelengths carried 10 Gbit/second server-generated data. We used a digital driver for the heater, in which pulse width was controlled, rather than dc amplitude [Optics Express 2014].
In a post-deadline paper at the [2014 IEEE Photonics Conference], we reported on the design and demonstrate a compact silicon photonic chip (1.30 mm x 0.52 mm) that integrates electrically-tunable 4-channel add, drop and power control functionalities with low power consumption for a research-category WDM data-center network.
Work done in collaboration with Sandia National Labs (A. Lentine, C. DeRose, D. C. Trotter, A. Pomerene, A. Starbuck); funded by NSF CIAN and ONR.