Quadratic Supercontinuum Generation from UV to Mid-IR in Lithium Niobate Nanophotonics

Abstract

Supercontinuum light sources are widely used for applications ranging from imaging to sensing and frequency comb stabilization. The most common mechanisms for their generation rely on cubic nonlinearities, for instance in crystals, optical fibers, and integrated photonics. However, quadratic supercontinuum generation (QSCG) offers potential for enhanced energy efficiency and broader spectral coverage because of the typically much stronger nonlinearity and ability to achieve both coherent up- and down-conversion via three-wave mixing processes. Despite such potentials, demonstrations of QSCG in integrated photonic waveguides have been sparse and have barely surpassed their cubic counterparts in terms of spectral coverage and energy-efficiency. Here, we introduce a new dispersion engineering principle and experimentally demonstrate purely quadratic supercontinuum generation in lithium niobate nano-waveguides substantially outperforming previous demonstrations in integrated photonics. In one device, by engineering a near-zero dispersion profile and using a single poling period for quasi-phase matched saturated second-harmonic generation, we achieve robust and energy efficient multi-octave QSCG with only femtojoules of pump pulse energy. In another device, we use a flat dispersion profile with two distant zero crossings of group velocity dispersion (GVD) to achieve broadband difference-frequency generation (DFG) for extending the spectral coverage further into the mid-IR and cover the entire transparency window of lithium niobate from 350 nm to 5000 nm. Our results showcase how DFG-assisted QSCG can access hard-to-access spectral regions in an energy-efficient fashion by properly utilizing dispersion engineering and quasi-phase matching.

Figures (3)

• Figure 1: QSCG design for efficient broadband SHG (left column) and DFG (right column). (a) Schematic of the spectral broadening mechanism for a quadratic supercontinuum generation (QSCG) waveguide designed for efficient saturated second harmonic generation (SHG) and sum-frequency-generation (SFG). The saturation and back-conversion is enabled by optical parametric amplification (OPA). (b) Schematic of the spectral broadening mechanism for a QSCG waveguide designed for difference frequency generation (DFG) after some initial spectral broadening through saturated SHG. Note that intra-pulse DFG can also occur between the spectral components of the input comb, where the resulting mid-IR comb would have zero $f_{ceo}$. (c) Group velocity dispersion (GVD) for the SHG waveguide. (d) GVD for the DFG waveguide. Compared to the GVD for the SHG waveguide, the zero-crossings are further separated in wavelength. (e) Group velocity mismatch (GVM) between each wavelength and its corresponding SH wavelength for the SHG waveguide. The GVM is near zero at the pump wavelength (2090 nm). (f) GVM between each wavelength and its corresponding SH wavelength for the DFG waveguide. (g),(h) Nonlinear propagation simulation based on the single envelope equation, for the SHG waveguide and DFG waveguide, respectively.
• Figure 2: TFLN on silica QSCG designs and experimental results. (a) Group velocity dispersion (GVD, blue) and group velocity mismatch (GVM, dashed) for Chip 1, designed for efficient QSCG through SHG. Inset: waveguide geometry for Chip 1. (b) GVD (blue) and GVM (dashed) for Chip 2, designed for mid-IR QSCG through DFG. Inset: waveguide geometry for Chip 2. (c) DFG phase-matching profile for Chip 1. $\lambda_1$ (overlaid contour in dashed while lines) and $\lambda_2$ (plotted on y-axis) denotes the two input wavelengths for generating output DFG wavelengths, $\lambda_{DFG}$ (plotted on x-axis). The color map is exp(-$|\Delta$k$|$), where $\Delta$k is the DFG phase-mismatch in rad/mm for different $\lambda_2$ and $\lambda_{DFG}$ pairs. (d) Same phase-matching profile plot as c for Chip 2. (e) Chip 1 DFG figure of merit (FoM) representing the product between walk-off length and DFG phase-matching (details see Supplementary). (f) Chip 2 DFG FoM. (g) Photo of Chip 2 in the experimental setup. (h) Experimental saturated-SHG QSCG spectra (Chip 1). (i) Experimental DFG-assisted QSCG spectra (Chip 2).
• Figure 3: TFLN on sapphire QSCG. (a) Group velocity dispersion (GVD, blue) and group velocity mismatch (GVM, dashed) curves for Chip 3, designed for efficient mid-IR QSCG through SHG. Inset: waveguide geometry parameters for Chip 3. (b) DFG phase-matching profile for Chip 3, $\lambda_1$ (overlaid contour in dashed while lines) and $\lambda_2$ (plotted on y-axis) denote the two input wavelengths for generating output DFG wavelengths, $\lambda_{DFG}$ (plotted on x-axis). The color map is exp(-$|\Delta$k$|$), where $\Delta$k is the DFG phase-mismatch in rad/mm for different $\lambda_2$ and $\lambda_{DFG}$ pairs. (c) Chip 3 DFG figure of merit (FoM) representing the product between walk-off length and DFG phase-matching (details see Supplementary). (d) Nonlinear propagation simulations based on the single envelope equation for Chip 3, with $\chi_{eff}^{(2)}$ divided by 4 (compared to Chip 1 and Chip 2) to account for lower poling quality. (e) Photo of Chip 3 on the experimental setup. (f) Top: combined transmission spectra over 50 cm (corresponding to the free-space optical path between the waveguide output facet and free-space to fiber collimator) for atmospheric CO$_2$, N$_2$, O$_2$, and H$_2$O HITRAN. Bottom: experimental QSCG spectra for Chip 3, with 177 pJ of on chip pump pulse energy.