Sustaining high-fidelity quantum logic in neutral-atom circuits via mid-circuit operations

This paper demonstrates a sustainable neutral-atom quantum computing framework that maintains high-fidelity two-qubit gates (~99.8%) across deep, repetitive circuits by integrating mid-circuit operations, including non-destructive erasure detection and in-situ cooling, to actively manage motional heating and atom loss for continuous error correction.

The Gist

Imagine you are trying to build a massive, incredibly complex tower out of Jenga blocks. But there's a catch: every time you place a block, the tower gets slightly warmer, the blocks start to vibrate, and occasionally, a block just falls out of the tower entirely.

In the world of quantum computing, this "tower" is a quantum circuit, and the "blocks" are atoms acting as qubits (the basic units of information). For a long time, scientists could build these towers, but only for a few layers. As soon as they tried to go deeper (perform more calculations), the heat and vibrations caused the blocks to wobble so much that the information became garbled, or the blocks would disappear. This is known as the "depth barrier."

This paper from the University of Science and Technology of China (USTC) introduces a revolutionary new way to keep the tower standing tall, no matter how high it gets. Here is the story of how they did it, broken down into simple concepts.

1. The Problem: The "Overheating" Computer

Think of a neutral-atom quantum computer like a dance floor where atoms are dancers. To make them "talk" to each other (perform a logic gate), scientists use a special laser trick that makes the atoms jump into a super-excited state called a Rydberg state.

However, every time they dance, they get a little bit of "sweat" (heat) and sometimes they trip and fall off the floor (atom loss). In previous experiments, if you tried to make the atoms dance 100 times in a row, the floor would get so hot and messy by the 10th dance that the 11th one would fail. The only solution used to be to stop the whole show, cool the dancers down, and start over from scratch. This is like pausing a movie every 5 minutes to reset the theater—it's too slow for real computing.

2. The Solution: The "Mid-Circuit" Refresh

The team realized they didn't need to stop the show. Instead, they built a refresh station right onto the dance floor. They created a system that can pause the dance for a split second, clean up the sweat, fix the fallen dancers, and get them back in perfect formation without breaking the flow of the performance.

They achieved this using three main "tools":

A. The "Magic Eye" (Non-Destructive Measurement)

First, they needed to know who was still dancing and who had fallen. Usually, looking at a quantum atom destroys its state (like shining a bright light that scares the dancer away).

• The Analogy: Imagine a security camera that can tell if a dancer is wearing a red shirt or a blue shirt, and if they are still on the floor, without blinding them or knocking them over.
• The Result: They developed a two-stage imaging system that acts like this magic eye. It can spot if an atom is in state "0", state "1", or if it has vanished. Crucially, it does this without destroying the remaining atoms.

B. The "Eraser" (Turning Mistakes into Known Errors)

Once they know an atom is missing, they don't panic. In quantum error correction, it's actually easier to fix a problem if you know exactly where it is.

• The Analogy: If you are playing a game of cards and you lose a card, it's a disaster. But if you know exactly which card you lost, you can just mark that spot as "empty" and keep playing.
• The Result: By detecting the missing atoms, they turned "mysterious errors" into "erasure errors" (known missing spots). This boosted their success rate from 99.60% to 99.81%.

C. The "AC Unit" (Mid-Circuit Cooling)

This is the big breakthrough. Even if the atom is still there, it might be vibrating too much (too hot) to dance correctly.

• The Analogy: Imagine the dancers are sweating and shivering. Instead of stopping the concert, the team installed a targeted air conditioner that blows cool air only on the dancers while they are still on the floor, instantly bringing them back to a calm, steady temperature.
• The Result: They used a technique called Raman Sideband Cooling. It's a sophisticated laser trick that sucks the heat (vibrations) out of the atoms and resets them to their "ground state" (perfectly calm) right in the middle of the calculation.

3. The Result: A Never-Ending Dance

With these tools, the team ran a test where they repeated the same quantum logic gate over and over again.

• Without the refresh: The performance dropped quickly after the second round (the tower started to wobble).
• With the refresh: The performance stayed rock-solid at ~99.8% even after five rounds, and theoretically, they could keep going forever.

Why This Matters

This is a massive step toward Fault-Tolerant Quantum Computing.
Think of building a skyscraper. If you can only build 10 floors before the foundation cracks, you can't build a city. But if you can build a system that repairs its own foundation while it's being built, you can build a skyscraper that reaches the clouds.

This paper proves that we can now keep a quantum computer "fresh" and "cool" while it's working. This is the key to running the massive, complex algorithms needed to solve problems like designing new medicines, creating new materials, or breaking complex codes—tasks that today's supercomputers can't touch.

In short: They figured out how to fix the computer while it's running, turning a fragile, short-lived experiment into a robust, sustainable machine ready for the future.

Technical Summary

1. Problem Statement

The realization of fault-tolerant quantum computing requires executing deep quantum circuits while maintaining gate fidelities above error-correction thresholds. While neutral-atom arrays have demonstrated high-fidelity two-qubit gates and early logical processors, they face a critical "depth barrier."

• Entropy Accumulation: In deep circuits, repetitive gate operations and photon scattering cause cumulative motional heating and atom loss.
• Thermal Decay: This entropy buildup acts as a decay mechanism, rapidly degrading gate fidelity as the circuit progresses.
• Current Limitations: Existing demonstrations are largely limited to shallow circuits or require terminating the process to reload the entire array and perform out-of-circuit cooling to reset entropy. This prevents continuous, long-term quantum operation necessary for large-scale error correction.

2. Methodology

The authors propose and demonstrate a sustainable, refreshable neutral-atom framework that integrates a suite of hardware-efficient mid-circuit operations directly into the functional gate sequence. The system utilizes single $^{87}\text{Rb}$ atoms in optical tweezers with physical qubits encoded in hyperfine clock states ($|0\rangle = |F=1, m_F=0\rangle$ and $|1\rangle = |F=2, m_F=0\rangle$).

The core methodology involves three integrated mid-circuit capabilities:

A. Loss-Resolved Non-Destructive Measurement

To convert atom loss into manageable erasure errors, the team developed a two-stage imaging scheme compatible with mid-circuit operation:

1. State-Selective Imaging: Uses a closed transition ($|F=2, m_F=2\rangle \to |F'=3, m_F'=3\rangle$) with $\sigma^+$ polarized light. Atoms in $|1\rangle$ scatter photons (bright), while $|0\rangle$ atoms remain dark. To prevent heating/loss, the tweezer and imaging beams are strobed out of phase.
2. State-Independent Imaging: Uses a modified Polarization Gradient Cooling (PGC) configuration with counter-propagating $\sigma^+$ and $\sigma^-$ beams with relative detuning. This re-establishes effective cooling and imaging under finite magnetic fields, allowing the detection of atom survival regardless of internal state.

• Result: This resolves $|0\rangle$, $|1\rangle$, and atom loss with 98.9% state-resolved fidelity and 99.7% atomic survival probability.

B. In-Circuit Raman Sideband Cooling (RSC)

To counteract motional heating, the authors implemented a mid-circuit RSC protocol:

• Mechanism: Drives Raman sideband transitions directly between the clock states $|1,0\rangle$ and $|2,0\rangle$.
• Robustness: This specific configuration is inherently robust against magnetic field inhomogeneities and vector light shifts.
• Reset: The cycle is closed by Raman-assisted pumping (RAP) back to $|1,0\rangle$, simultaneously cooling the motional state and re-initializing the qubit.

C. High-Fidelity Gate Execution

• Entanglement: Mediated by Rydberg interactions (excitation to $70S_{1/2}$ via a two-photon transition).
• Gate Control: A time-optimal, parameterized pulse with sinusoidal phase modulation is used for the Controlled-Z (CZ) gate.

3. Key Contributions

1. Mid-Circuit Refresh Architecture: The first demonstration of a neutral-atom processor that actively manages internal and motional entropy during circuit execution, rather than requiring a full system reset.
2. Loss-to-Erasure Conversion: A non-destructive measurement protocol that identifies atom loss events, converting them into known-location erasure errors (which are easier to correct in Quantum Error Correction).
3. Robust Mid-Circuit Cooling: A novel Raman sideband cooling scheme operating directly between clock states, enabling uniform cooling across the array without disrupting the qubit encoding.
4. Sustained High Fidelity: Proof that gate fidelities can be maintained over multiple operational rounds without degradation.

4. Key Results

• Raw Gate Fidelity: Achieved a two-qubit CZ gate fidelity of 99.60(1)% using global echoed randomized benchmarking.
• Loss-Corrected Fidelity: By post-selecting trials where atoms were not lost (converting loss to erasure), the fidelity increased to 99.81(1)%.
• Sustainability Test:
  • Without Cooling: Gate fidelity dropped by ~0.3% from the second round onwards due to heating.
  • With Local Gray Molasses: Similar degradation observed, indicating insufficient entropy removal.
  • With Mid-Circuit RSC: The mean phonon number ($\bar{n}$) was maintained at $\approx 1$ in both radial and axial directions across five consecutive rounds. Consequently, the loss-corrected CZ gate fidelity remained consistently at ~99.8% with no observable degradation.

5. Significance

This work addresses the primary bottleneck preventing neutral-atom systems from scaling to fault-tolerant quantum computers: circuit depth.

• Pathway to Fault Tolerance: By enabling the "refresh" of qubits in-situ, the architecture supports the repeated syndrome-extraction cycles required for topological codes (e.g., surface codes).
• Scalability: The approach is compatible with zoned layouts for independent ancilla refreshing and teleportation-based data qubit replacement.
• Future Impact: It establishes a robust operational regime for autonomous, continuous quantum error correction, moving neutral-atom processors from proof-of-principle demonstrations toward practical, large-scale algorithms.