Wafer-Scale Squeezed-Light Chips (2509.10445v1)

Abstract: Squeezed-light generation in photonic integrated circuits (PICs) is essential for scalable continuous-variable (CV) quantum information processing. By suppressing quantum fluctuations below the shot-noise limit, squeezed states enable quantum-enhanced sensing and serve as a standard resource for CV quantum information processing. While chip-level squeezed-light sources have been demonstrated, extending this capability to the wafer level with reproducible strong squeezing to bolster large-scale quantum-enhanced sensing and information processing has been hindered by squeezed light's extreme susceptibility to device imperfections. Here, we report wafer-scale fabrication, generation, and characterization of two-mode squeezed-vacuum states on a fully complementary metal-oxide-semiconductor (CMOS)-compatible silicon nitride (Si$_3$N$_4$) PIC platform. Across a 4-inch wafer, 8 dies yield 2.9-3.1 dB directly measured quadrature squeezing with $< 0.2$ dB variation, demonstrating excellent uniformity. This performance is enabled by co-integrating ultralow-loss, strongly overcoupled high-$Q$ microresonators, cascaded pump-rejection filters, and low-loss inverse-tapered edge couplers. The measurements agree with a first-principles theoretical model parameterized solely by independently extracted device parameters and experimental settings. The measured squeezing level can be further improved by enhancing the efficiencies of off-chip detection and chip-to-fiber coupling. These results establish a reproducible, wafer-scale route to nonclassical-light generation in integrated photonics and lay the groundwork for scalable CV processors, multiplexed entanglement sources, and quantum-enhanced sensing.

• The paper demonstrates wafer-scale generation of two-mode squeezed vacuum states using Si₃N₄ PICs, achieving measured squeezing levels between 2.9 dB and 3.1 dB.
• It employs an integrated architecture with high-Q microresonators, cascaded pump-rejection filters, and low-loss edge couplers to ensure robust, reproducible performance.
• Experimental results, supported by theoretical modeling, underscore the platform’s potential for scalable continuous-variable quantum photonics in sensing and computing.

Wafer-Scale Squeezed-Light Chips: Uniform, CMOS-Compatible Quantum Photonic Integration

Introduction

The paper presents a comprehensive demonstration of wafer-scale, CMOS-compatible silicon nitride (Si$_3$N$_4$) photonic integrated circuits (PICs) for the generation of two-mode squeezed vacuum (TMSV) states. Squeezed light, characterized by quantum noise suppression below the shot-noise limit in one quadrature, is a critical resource for continuous-variable (CV) quantum information processing, quantum-enhanced sensing, and secure quantum communication. While chip-level squeezed-light sources have been previously realized, the transition to reproducible wafer-scale fabrication with strong squeezing has been impeded by the extreme sensitivity of squeezed states to device imperfections and loss. This work addresses these challenges by integrating ultralow-loss, high-$Q$ microresonators, high-extinction pump-rejection filters, and low-loss edge couplers on a single Si$_3$N$_4$ chip, achieving uniform squeezing performance across an entire 4-inch wafer.

Device Architecture and Squeezing Mechanism

The PIC architecture centers on a strongly overcoupled Si$_3$N$_4$ microring resonator, which serves as the nonlinear element for squeezed-light generation via non-degenerate four-wave mixing (FWM). A continuous-wave pump laser, operated below the parametric oscillation threshold, is coupled into the resonator, where two pump photons are annihilated to produce quantum-correlated signal and idler photons in adjacent resonances. The output signal and idler modes form a TMSV state, exhibiting correlated amplitude quadratures and anti-correlated phase quadratures.

Figure 1: Schematic of the Si$_3$N$_4$ PIC for TMSV generation and conceptual illustration of non-degenerate FWM in the microring resonator.

Efficient suppression of residual pump light is achieved using two cascaded add-drop filters with high extinction ratios, minimizing insertion loss and enabling broadband resonance tunability. The filtered squeezed light is collected via an inverse-tapered waveguide edge coupler, facilitating low-loss fiber interfacing for off-chip characterization. Thermo-optic microheaters are integrated for resonance tuning and stabilization, ensuring robust operation.

Wafer-Scale Fabrication and Uniformity

The fabrication process employs a subtractive a-Si hardmask dry etching technique, yielding 800-nm-thick Si$_3$N$_4$ waveguides with engineered anomalous dispersion. Eight dies per wafer are dedicated to TMSV generation, each integrating four squeezed-light circuits with varying filter parameters. The chips are wire-bonded to custom PCBs for thermo-optic control and mounted on temperature-stabilized stages.

Figure 2: Wafer-scale images, measured squeezing levels, quadrature noise spectra, and micrographs of core functional elements across the 4-inch Si$_3$N$_4$ wafer.

Direct measurements across all dies reveal squeezing levels ranging from 2.9 dB to 3.1 dB, with a device-to-device variation of less than 0.2 dB. This uniformity is attributed to the co-integration of ultralow-loss microresonators ($Q_{\rm i} > 10^7$), robust pump-rejection filters (extinction $>$30 dB), and efficient edge couplers (chip–fiber coupling $>$75\%).

Classical and Quantum Characterization

Classical transmission spectra confirm the presence of strongly overcoupled squeezer resonances and critically coupled pump-rejection filters. Lorentzian fitting yields loaded quality factors $Q_{\rm L} \sim 0.83 \times 10^6$ and intrinsic quality factors $Q_{\rm i} \sim 10.1 \times 10^6$, corresponding to escape efficiencies $\eta > 91\%$. Statistical analysis across the wafer demonstrates consistent optical loss and coupling, with most probable values $Q_{\rm i} = 9.2 \times 10^6$, $Q_{\rm L} = 0.9 \times 10^6$, and $\eta \approx 91\%$.

Figure 3: Transmission spectra, resonance fitting, and statistical distributions of $Q_{\rm i}$, $Q_{\rm L}$, and escape efficiency $\eta$ for squeezer resonances.

Quantum characterization is performed via balanced homodyne detection, using a bi-tone local oscillator derived from the same CW pump. The measured quadrature noise variance exhibits raw squeezing of 3.0 dB and anti-squeezing of 6.0 dB at 50 mW on-chip pump power. Squeezing and anti-squeezing levels increase with pump power, with experimental data closely matching theoretical predictions based on independently extracted device parameters.

Figure 4: Direct measurement of quadrature noise as LO phase is scanned, and squeezing/anti-squeezing levels as functions of on-chip pump power.

The observed squeezing is primarily limited by the total collection efficiency ($\sim$60%), comprising waveguide-to-fiber coupling (75%), propagation (95%), interference visibility (98%), and photodiode quantum efficiency (88%). The integrated filter design improves measured squeezing by more than 2 dB compared to previous off-chip filtering approaches.

Theoretical Modeling

The squeezing mechanism is modeled using quantum coupled-mode equations for the cavity modes, incorporating device-specific parameters such as resonance linewidths, dispersion, and nonlinear coefficients. Below threshold, the pump mode maintains a classical mean field, while side modes exhibit quantum fluctuations governed by cross-phase modulation and FWM. The escape efficiency $\eta = 1 - Q_{\rm L}/Q_{\rm i}$ sets the upper bound for extractable squeezing, with high $\eta$ enabling strong on-chip squeezing. Losses between the chip and detector are consolidated into an overall quantum efficiency $\eta_{\rm total}$, which determines the measured squeezing level.

The model predicts that, for the observed escape efficiency ($\sim$91%), the maximum achievable on-chip squeezing exceeds 10 dB, with the measured value limited by extrinsic losses. The theoretical framework is validated by the close agreement between experimental data and model predictions for squeezing and anti-squeezing as functions of pump power.

Implications and Future Directions

The demonstration of reproducible, wafer-scale squeezed-light generation on a CMOS-compatible Si$_3$N$_4$ platform establishes a new benchmark for integrated quantum photonics. The uniformity and robustness of the devices enable scalable deployment of CV quantum processors, multiplexed entanglement sources, and quantum-enhanced sensors. The platform supports additional functionalities, including reconfigurable interferometers, long delay lines for time-domain cluster-state generation, and co-integration with modulators and detectors.

The current limitation in measured squeezing arises from off-chip detection inefficiencies, suggesting that further improvements are feasible through enhanced chip–fiber interfaces and photodetector quantum efficiency. The architecture is compatible with broadband quantum microcomb generation, enabling access to multiple squeezed mode pairs for high-dimensional entanglement and cluster-state protocols.

The results have direct implications for fault-tolerant CV quantum computing, where squeezing thresholds of 4.5 dB (for cluster-state generation) and 10 dB (for error correction) are required. The demonstrated escape efficiency and device uniformity indicate that these thresholds are within reach with further optimization of extrinsic losses.

Conclusion

This work demonstrates wafer-scale, CMOS-compatible Si$_3$N$_4$ PICs for uniform, strong squeezed-light generation, achieving 2.9–3.1 dB of measured squeezing across an entire 4-inch wafer. The integration of ultralow-loss microresonators, high-extinction filters, and efficient edge couplers enables robust, reproducible performance, validated by both experimental data and theoretical modeling. The platform provides a scalable route to integrated CV quantum photonics, with clear pathways for further improvement and expansion to high-volume manufacturing of quantum processing circuits.