Ultrashort-pulse-pumped, single-mode type-0 squeezers in lithium niobate nanophotonics

TL;DR

This work develops dispersion-engineered, ultrashort-pulse pumped, type-0 traveling-wave OPAs to generate spectrally pure degenerate squeezed light in thin-film lithium niobate on insulator. By matching group velocities and exploiting dispersion, the authors achieve near-single-mode operation with Schmidt numbers approaching unity ($K \approx 1.115$ in SPG and $K \approx 1.02$ at high gain) and >15 dB squeezing over a 5 THz bandwidth in cm-scale devices at 2090/1045 nm. They demonstrate robust suppression of parasitic interactions through phase-mismatch engineering, and show all-optical, loss-tolerant measurements using a SOPA followed by a high-gain MOPA with fidelities $F(f_T,f_{MOPA}) \gtrsim 0.99$. Extensions to C- and L-bands discuss geometries and pump requirements, revealing trade-offs in dispersion engineering and the need for chirp compensation. The results establish a practical path for ultrafast quantum information processing and sensing on the TFLN platform with potential for large-scale, broadband quantum circuits.

Abstract

We present design principles for ultrashort-pulse, type-0 phase-matched optical parametric amplifiers to generate and measure spectrally pure degenerate squeezed light. We consider a fundamental signal (second-harmonic) mode at 2090 (1045) nm and show that our proposed design achieves a Schmidt number of $K \approx 1.02$ with squeezing levels greater than 15 dB on a single temporal mode spanning over $5$ THz in bandwidth with cm-scale devices on thin-film lithium niobate (TFLN) on insulator platform. Our work opens up promising avenues for large-scale circuits for ultrafast quantum information processing and quantum sensing applications on the rapidly advancing TFLN platform with already demonstrated linear components and photodetection capabilities.

Figures (6)

• Figure 1: (a) Pump spectral amplitude for a 50 fs pulse. (b) Phase-matching function. (c) JSI in the SPG regime. (d) JSI in the high-gain limit for 18 dB of gain.
• Figure 2: (a) Periodically-poled, x-cut thin-film ($700~\mathrm{nm}$) lithium niobate on insulator ridge waveguide with top width $w$ and etch depth $d$, supported on a $2~\mu\mathrm{m}$$SiO_2$ layer and a $500~\mu\mathrm{m}$ Si substrate. (b) Transverse mode profiles for SH and FH modes, showing the tight mode confinement.
• Figure 3: (a) Schmidt number as a function of gain, determined by the average number of signal photons ($\langle N_{S}\rangle$, c.f. Eq. \ref{['eq:averageN']}). Vertical lines represent different levels of gain in dB. The Schmidt number varies from $K\approx 1.115$ (SPG regime) to $K\approx 1.02$ (18 dB gain limit). (b) Mode occupation fraction (c.f. Eq. \ref{['eq:mode_frac']}) for the first five modes in the 18 dB gain limit (inset: SPG regime). We find a first/second mode occupation ratio of 0.989/0.007 (0.95/0.02) in the 18 dB gain limit (SPG regime). (c) Schmidt number in the SPG regime as a function of deviations from the ideal geometry. The waveguide width and etch depth are varied by $\pm 10~\mathrm{nm}$ while keeping the film thickness and ridge angle fixed. The Schmidt number increases slightly away from the ideal geometry but remains close to unity, showing the fabrication tolerance of our proposed designs.
• Figure 4: (a) Minimum phase mismatch, $\Delta \bar{k}$ from the unwanted interactions between higher-order spatial modes, evaluated across a range of waveguide geometries and modes above the set threshold for guided modes. The top width and etch depth are varied in 10 nm steps within $\pm20$ nm of the ideal design, while the film thickness and sidewall angle are held constant. (b) Minimum phase mismatch, $\Delta \bar{k}$ when the signal mode (i.e., desired mode for squeezing) is fixed as the fundamental TE mode while pump and idler modes are swept. (c) Impact of phase mismatch on squeezing for the ideal geometry, shown as a function of parametric gain (set by pump power) for different values of effective mismatch $\Delta \bar{k}L$. All geometries within the $\pm 20$ nm from ideal achieve $\Delta \bar{k} \ell > 20$, ensuring robustness to fabrication errors.
• Figure 5: (a) Schematic of the all-optical, loss-tolerant measurement protocol. A squeezed vacuum (Sq. vac.) is generated in the SOPA using pump 1, and subsequently injected into the MOPA along with pump 2. The combined output is an amplified squeezed vacuum (Amp. Sq. vac.), which is then characterized through homodyne detection. (b) Fidelity [Eq. \ref{['eq:fidelity']}] between the overall output temporal mode of the system ($f_{T}$) and the MOPA’s output temporal mode ($f_{MOPA}$), plotted as a function of the SOPA gain, quantified by the average number of signal photons $\langle N^{SOPA}_{S}\rangle$. Different curves correspond to varying MOPA gain levels. The red vertical line denotes the SOPA operating point at 18 dB gain.