Highly efficient discrete band mid-infrared to near-infrared wavelength conversion relying on Si1-xGex alloys

Summary form only given. The use of group IV materials becomes particularly promising for the transmission and processing of optical signals lying at the mid-infrared (MIR) wavelength region which is ideal for sensing, absorption-spectroscopy and free-space communications. One optical function provided by Si-based nano-waveguides is the frequency translation between near-IR (NIR) and MIR regions exploiting the parametric amplification mechanisms [1].Here, we report on the properties of waveguide structures relying on Si1-xGex alloys, with x being the Ge (%) concentration, which enhance the nonlinear potentials of Si [2]. A waveguide and nonlinear propagation analysis reveals the design rules that should be followed in order to achieve high conversion efficiency (>-5 dB) from the band of interest for gas-sensing applications (4.1-4.8 μm) to the NIR region (1.3-1.6 μm) The nonlinear wavelength conversion from the targeted MIR to the NIR region is based on the four-wave mixing (FWM) process pumped by a tunable and continuous wave source covering the 2-2.5 μm wavelength region [3], which is where the zero-dispersion wavelength (ZDW) of the waveguide needs to be. The structure is based on SOI and consists of a Si ridge, on top of which a SiGe section is formed. The waveguide was designed to favour TM polarization, to match that of a QCL source. Calculations were performed using a Finite Elements Method (FEM) solver by taking into account material dispersion. In order to reassure that the generated from the FWM process NIR wavelength will experience low losses for all x values, we chose to use the wavelengths 1.3, 1.4 and 1.6 μm as the lower bounds for the generated idler for pure Si, Si0.8Ge0.2 and Si0.6Ge0.4 respectively, which in turn leads to desirable ZDW at 2, 2.2 and 2.3 μm. In figure 1a, the dispersion curves for waveguide configurations with a width (W) of 600 nm and the three optimal values of height (H) at 2.0, 2.4 a- d 2.9 μm are shown for x ranging from 0 to 40%. It is noteworthy that by varying x, dispersion at the MIR wavelengths flattens which is indicative of a relaxed need for tunability at the pump. A numerical model which involves the propagation of the pump, signal and idler was developed according to [3], with 2 dB/cm losses for pump/MIR and 1-5 dB/cm for NIR depending on x. After analyzing the nonlinear properties of Si1-xGex alloys for x=0, 0.2 and 0.4, we conducted an analysis of the conversion efficiency (CE) defined as the ratio of the power generated at the idler NIR wavelength over the initial MIR signal power for the three waveguide structures of fig. 1a. Fig. 1b shows the conversion efficiency at its maximum value and the corresponding pump wavelength as a function of the MIR wavelength. By properly tuning the pump wavelength, CE values exceeding -5 dB are possible for the entire 4.1-4.8 μm band. The three structures exhibit similar conversion capabilities. The higher nonlinearity of Ge is compromised by the need for higher waveguides with larger EMA. Optimized design could indicate proper SiGe waveguides with smaller H, W dimensions. The efficiency could exceed 0 dB if higher pump power levels and/or longer waveguides were considered. Work supported by the EU 7th Framework Programme FP/2007-2013, Grant 288304 (STREP CLARITY).