🔬 optics

An ultralow-loss integrated photonic platform for discrete-variable quantum information processing

This paper presents a monolithic, ultralow-loss silicon nitride integrated photonic platform that overcomes the scaling limitations of silicon photonics by achieving record-high fidelities in EPR state preparation and four-photon GHZ state synthesis, while delivering a fourfold count rate more than two orders of magnitude higher than previous implementations.

Original authors: Yi-Han Luo, Ruiyang Chen, Zeying Zhong, Sanli Huang, Sicheng Zeng, Zhenyuan Shang, Yue Hu, Zhen Chen, Yuan Chen, Xue Bai, Junqiu Liu

Published 2026-06-26

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Yi-Han Luo, Ruiyang Chen, Zeying Zhong, Sanli Huang, Sicheng Zeng, Zhenyuan Shang, Yue Hu, Zhen Chen, Yuan Chen, Xue Bai, Junqiu Liu

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are trying to build a complex machine out of tiny, fragile glass marbles. In the world of quantum computing, these "marbles" are photons (particles of light), and the machine is a circuit that processes information.

For a long time, scientists have been able to make these glass marbles dance together on small chips, but there was a major problem: the marbles kept getting lost.

Every time a photon traveled through a wire on the chip, or bounced off a mirror, a few would vanish. If you tried to build a machine with just a few marbles, you'd lose them all before the machine finished its job. It was like trying to run a relay race where every runner drops the baton before reaching the next person. This "loss" meant that as the computers got bigger and more complex, the chance of them actually working dropped to near zero.

The Breakthrough: A Super-Highway for Light

This paper introduces a new "highway" made of a special material called Silicon Nitride (Si3N4). Think of this material as a perfectly smooth, frictionless road where photons can travel without getting lost.

Here is what the researchers achieved, explained simply:

1. The Perfect Factory (The Source)

First, they built a factory on the chip that creates pairs of these light marbles.

The Analogy: Imagine a machine that shoots out two identical twins every time it fires.
The Result: The researchers made these twins so perfectly identical that they are indistinguishable. If you tried to tell them apart, you couldn't. This is crucial because if the marbles aren't identical, they can't "talk" to each other to do quantum math. Their factory produces these twins with near-perfect accuracy (99% indistinguishability).

2. The Magic Trick (Fusion)

Next, they needed to combine these pairs to create a bigger, more powerful group.

The Analogy: Imagine you have two pairs of twins. You take one twin from the first pair and one from the second pair and make them "hold hands" in a special way. This magical handshake fuses them into a single, four-person team.
The Result: The chip successfully fused two pairs of photons to create a four-photon team (called a GHZ state). This is a complex quantum state that is very hard to make.

3. The Record-Breaking Speed

In previous attempts using similar technology (Silicon), the "marbles" got lost so often that the team only managed to form this four-person group about once every 100 seconds (or even less).

The Result: Because their new "highway" (the Silicon Nitride chip) is so smooth and low-loss, they can form this four-person team 27 times every second. That is more than 100 times faster than before. It's like going from a snail's pace to a sprint.

4. The High Fidelity (Quality Control)

In quantum computing, "fidelity" is like a quality score. It measures how close the result is to the perfect, theoretical ideal.

The Result: Their four-person team had a quality score of 0.943 (out of 1.0). This is the highest score ever recorded for this specific task on a chip. It means the team is almost perfectly formed, with very few errors.

Why This Matters (According to the Paper)

The paper argues that this isn't just a small improvement; it solves the biggest bottleneck holding back quantum computers.

Scalability: Because the chips are made using standard factory methods (the same kind used to make your phone's processor), they can be mass-produced.
Reliability: By keeping the loss so low, they proved that you can build bigger and bigger quantum circuits without the photons disappearing.

In Summary:
The researchers built a super-smooth, low-loss chip that acts like a perfect highway for light. They used it to create a factory that makes identical light particles, fused them together into a complex four-particle team, and did it 100 times faster and with higher quality than anyone has ever done before on a chip. This proves that we can now manufacture the "engines" needed for large-scale quantum computers.

Technical Summary: An Ultralow-Loss Integrated Photonic Platform for Discrete-Variable Quantum Information Processing

Problem Statement
While photonic integrated circuits (PICs) offer a scalable route for discrete-variable (DV) quantum information processing (QIP) by consolidating sources and linear optical networks onto wafer-manufacturable chips, scaling these systems beyond proof-of-principle demonstrations faces a critical bottleneck: optical loss. In DV encodings, optical loss manifests as photon erasure. As circuit depth and complexity increase, compound propagation and component losses cause multiphoton generation probabilities to plummet exponentially. Previous silicon photonic implementations have struggled with this rate–loss trade-off, where increasing photon flux to boost count rates often degrades interference visibility due to multi-pair noise, while maintaining high fidelity results in negligible count rates.

Methodology
The authors present a monolithic, ultralow-loss silicon nitride (Si $_3$ N $_4$ ) integrated photonic platform engineered specifically for high-performance DV-QIP. The architecture integrates three functional stages on a single chip fabricated using a standard CMOS-compatible process on 150-mm-diameter wafers:

Source Generation: Four microresonator-based sources utilize cavity-enhanced spontaneous four-wave mixing (SFWM) to generate narrowband photon pairs. The microresonators are designed with anomalous group-velocity dispersion and operate in an over-coupled regime to maximize photon extraction efficiency (~80%).
Qubit Fusion: The platform employs unbalanced Mach–Zehnder interferometers (UMZIs) for spectral demultiplexing and path-exchange networks to construct path-encoded Einstein–Podolsky–Rosen (EPR) states. A fusion circuit then executes a path-exchange operation on specific qubits from two independent EPR states, projecting them into a four-photon Greenberger–Horne–Zeilinger (GHZ) state via post-selection.
State Analysis: Reconfigurable Mach–Zehnder interferometers (MZIs) with thermally tunable phase shifters and optical delay lines allow for measurement in various bases (computational and superposition) and local phase control.

The system is pumped by a 100 MHz train of 0.9 ns optical pulses centered at 193.5 THz. The design prioritizes high fabrication uniformity to ensure precise spectral alignment of resonances across the wafer, which is critical for maintaining photon indistinguishability.

Key Results
The platform demonstrates record-breaking performance metrics for integrated photonic DV-QIP:

Photon Indistinguishability: The on-chip sources exhibit near-unity photon indistinguishability, achieving a heralded Hong–Ou–Mandel (HOM) interference visibility of 0.990(6) at a low generation probability ( $p=0.006$ ). Even at a higher flux operating point ( $p=0.025$ ), the visibility remains above 0.90 while the count rate increases significantly.
EPR State Fidelity: The system prepares path-encoded EPR states with a fidelity of 0.9875(3), enabled by the high uniformity of the Si $_3$ N $_4$ fabrication process which ensures precise spectral alignment of the signal and idler resonances.
Four-Photon GHZ State: By fusing two EPR states on-chip, the authors synthesize a four-photon GHZ state with a record fidelity of 0.943(8). This exceeds the genuine entanglement threshold by more than 55 standard deviations.
Count Rates: The fourfold count rate for the GHZ state generation is 27 Hz at the high-fidelity operating point. Under higher pump power, the rate reaches 134 Hz while maintaining a fidelity of 0.860(10). These rates represent an improvement of more than two orders of magnitude over previous silicon-photonic implementations.
Linewidth and Loss: The sources produce photons with a narrow linewidth of approximately 160 MHz (FWHM 1.6 ns), suitable for interfacing with light–matter quantum memories. The passive circuits exhibit exceptionally low insertion loss (e.g., 0.05 dB per beam splitter).

Significance and Claims
The paper claims that this work establishes ultralow-loss Si $_3$ N $_4$ integrated photonics as a "definitive, manufacturable platform" for deployable, large-scale quantum information processors. The primary significance lies in overcoming the rate–loss barrier that has historically constrained probabilistic linear optical systems. By simultaneously optimizing narrowband photon-pair generation, near-unity indistinguishability, and ultralow propagation loss within a monolithic, foundry-grade architecture, the platform preserves practical count rates at low pair-generation probabilities, thereby suppressing multi-pair emission noise at its physical source.

The authors assert that this architecture validates a complete, monolithic source–circuit–measurement chain on a single chip, bridging the gap between laboratory few-photon physics and scalable industrial manufacturing. They highlight that the platform's narrow linewidths and low loss make it exceptionally well-suited for fiber-based qubit distribution and future coherent light–matter interfaces. While the current work focuses on four-photon states, the authors note that the geometric uniformity of the Si $_3$ N $_4$ framework provides a clear path toward scaling to large-scale, highly entangled cluster and graph states through spatial multiplexing and cascaded fusion gates.