Efficient Frequency Conversion in a Degenerate ${\chi }^{(2)}$ Microresonator

Microresonators on a photonic chip could enhance nonlinear optics effects and thus are promising for realizing scalable high-efficiency frequency conversion devices. However, fulfilling phase matching conditions among multiple wavelengths remains a significant challenge. Here, we present a feasible scheme for degenerate sum-frequency conversion that only requires the two-mode phase matching condition. When the drive and the signal are both near resonance to the same telecom mode, an on-chip photon-number conversion efficiency up to 42% is achieved, showing a broad tuning bandwidth over 250 GHz. Furthermore, cascaded Pockels and Kerr nonlinear optical effects are observed, enabling the parametric amplification of the optical signal to distinct wavelengths in a single device. The scheme demonstrated in this Letter provides an alternative approach to realizing high-efficiency frequency conversion and is promising for future studies on communications, atom clocks, sensing, and imaging.

Figure 1

Schematic illustration of the frequency conversion via nondegenerate sum-frequency generation and degenerate sum-frequency generation models. Comparing the two schemes, the degenerate case only requires the phase matching condition between two modes and produces a second-harmonic generation signal of the drive that nears resonance with the signal mode.

Figure 2

(a) Experimental setup. Optical BS, optical beam splitter; FPC, fiber polarization controller; AOM, acoustic-optical modulator; WDM, wave division multiplexing. Insert: schematic and SEM picture of our device. (b) Visible output spectra for degenerate SFG. The strong peak represents the SHG of drive light, and the right small peak corresponds to the converted signal. Inset: the converted signal probed by the detector with high gain. (c) The relationship between input and output signal on-chip powers. The blue dots are experimental values, and the red line is a linear fit.

Figure 3

Frequency tuning. (a) The schematic of the temperature controlling for the on-chip microring resonators. The environment temperature $(T_e)$ could be controlled by a heater, and the device temperature $(T_{\mathrm{ring}})$ could also be adjusted by drive laser. (b) The coarse tuning of frequency conversion by $T_e$, with on-chip input drive and signal powers of about 23 mW and 0.28 mW, respectively. (c) Fine tuning via $T_{\mathrm{ring}}$ by changing the drive wavelength. In both (b) and (c), the dots are the experimental data, and the lines are the calculated results based on device parameters. (d) The prediction of SFG frequency conversion efficiency for different $T_e$ and drive detuning. The zero frequency detuning corresponding to the optimal conversion ($\sim 40\%$), and the input on-chip drive power is 30 mW. The pink line presents 20% efficiency, which indicates an FWHM of 257 GHz. The red and yellow lines represent the typical results for fine tuning under different $T_e$ (16 K and $-16\mathrm{K}$ away from optimal temperature, respectively).

Figure 4

Cascaded nonlinear effect in frequency conversion. (a) The conversion efficiency of SHG and SFG. The dots are experimental data, and the lines are theoretical predictions. The black line represents the SHG efficiency, the blue dash line is based on the SFG model without cascaded nonlinear effect modification, the purple dash line represents the prediction by including the ${\chi }^{(2)}-{\chi }^{(2)}$ processes, and the red solid line represents the theoretical results for the full model that includes both ${\chi }^{(2)}-{\chi }^{(2)}$ and intrinsic ${\chi }^{(3)}$ effects. (b) Our nonlinear process model with cascaded effect and the detailed transmission spectrum at infrared band for degenerate SFG with drive power of 32 mW, which corresponds to the shadow region in (a). The small arrows represent the gain via the cascaded process. (c) The prediction of the required drive power for achieving 100% conversion efficiency for different $g_2$, with $g_3\propto g_2^2$ correspondingly and all other parameters are from the actual device. (d) The parameter dependence between the conversion efficiency and the external coupling rates ${\kappa }_{a1},{\kappa }_{b1}$, with all other parameters from the actual device. The drive power is 50 mW.