Above 99.9% Fidelity Single-Qubit Gates, Two-Qubit Gates, and Readout in a Single Superconducting Quantum Device

This paper demonstrates a scalable path to error-corrected quantum computing by achieving simultaneous single-qubit gate, two-qubit gate, and readout fidelities exceeding 99.9% in a single superconducting device through optimized coupling parameters and a novel calibration protocol.

The Gist

Imagine you are trying to build a super-advanced calculator that uses the laws of quantum physics instead of electricity. To make this calculator work, you need to perform three specific tasks perfectly:

1. Flip a single switch (Single-Qubit Gate).
2. Make two switches talk to each other (Two-Qubit Gate).
3. Read the result (Readout).

The problem is that in the past, making the switches talk to each other usually made it harder to flip them individually or read the results accurately. It was like trying to have a loud conversation in a room; the more you shouted to be heard (entanglement), the harder it was to hear your own thoughts (individual control) or the background noise (readout errors).

This paper from IQM Quantum Computers says: "We finally figured out how to do all three at the same time with near-perfect accuracy."

Here is how they did it, explained with everyday analogies:

1. The "Goldilocks" Connection

The device uses two tiny quantum switches (called qubits) connected by a middleman (called a coupler).

• The Problem: If the connection between the switches and the middleman is too weak, they can't talk fast enough. If it's too strong, they get "confused" and mess up their individual jobs.
• The Solution: The team found the "Goldilocks" setting. They tuned the connection strength to be just right. It's strong enough to let the switches chat quickly, but not so strong that they get tangled up and make mistakes.
• The Result: They achieved 99.93% accuracy for the conversation (two-qubit gate) and 99.98% accuracy for flipping the switches (single-qubit gates).

2. The "Noise-Canceling" Calibration (PALEA)

Even with the right settings, tiny errors happen. Imagine trying to tune a radio; sometimes you get a little static.

• The Old Way: Previous methods tried to find the error by listening to the static, but the static was often mixed with other noises, making it hard to pinpoint exactly what was wrong.
• The New Way (PALEA): The team invented a new method called PALEA (Phase-Averaged Leakage Error Amplification).
  • The Analogy: Imagine you are trying to hear a specific whisper in a noisy room. Instead of just listening, you ask the room to repeat the whisper over and over, but you ask everyone to say it with a slightly different accent each time. By averaging all those different accents, the background noise cancels out, and the specific whisper becomes crystal clear.
  • The Result: This allowed them to find and fix the tiny "leaks" of information (errors) twice as effectively as before.

3. The "Safety Net" Reading

Reading the result of a quantum calculation is tricky because the act of looking at it can change the result.

• The Strategy: They used a technique called Shelving.
  • The Analogy: Imagine you have a fragile glass marble (the quantum state) that you need to weigh. If you put it directly on a scale, the vibration might break it. Instead, you gently lift the marble up to a high shelf (a higher energy state) where it is more stable, then you weigh it.
  • The Result: This allowed them to read the answer with 99.94% accuracy without breaking the fragile state. They also showed they could read it without changing the state at all (99.3% "non-destructive" reading), which is crucial for future error-checking.

The Big Picture

The team didn't just fix one part of the machine; they optimized the whole system.

• They proved that you can have fast, accurate conversations between switches AND accurate individual control AND perfect reading, all in the same device.
• They showed that this design can be scaled up. Imagine a grid of these switches; their design allows you to add more switches in a square pattern (like a chessboard) without breaking the rules they just perfected.

In short: They built a quantum processor where the switches, the conversations, and the reading are all working at a level of precision (over 99.9%) that makes building a truly powerful, error-correcting quantum computer possible. They didn't just make one thing good; they made the whole orchestra play in perfect harmony.

Technical Summary

Technical Summary: Above 99.9% Fidelity Single-Qubit Gates, Two-Qubit Gates, and Readout in a Single Superconducting Quantum Device

Problem Statement
Building scalable, fault-tolerant quantum computers requires low error rates across all core operations: single-qubit gates, two-qubit gates, and qubit readout. While individual fidelities for these operations have improved significantly in superconducting transmon systems, achieving high fidelity simultaneously in a single device remains a major challenge. Device parameters optimized for one operation often degrade others. Specifically, strong qubit-coupler coupling facilitates fast entangling gates but increases hybridization-induced crosstalk, degrading simultaneous single-qubit gate fidelities. Similarly, integrating high-fidelity readout without compromising qubit coherence or gate performance requires careful architectural trade-offs.

Methodology
The authors present a holistic optimization strategy for a superconducting device comprising two flux-tunable floating transmon qubits coupled via a tunable floating transmon coupler. The approach involves three primary pillars:

1. Optimal Qubit-Coupler Coupling: Through numerical simulations using master equations and three-level Duffing oscillator models, the authors analyzed the trade-off between single-qubit errors (dominated by hybridization crosstalk) and two-qubit errors (dominated by leakage and incoherent processes). They identified a specific qubit-coupler coupling strength ($g_{qc}$) that minimizes the total weighted Clifford gate error. The fabricated device targets a geometric average coupling of $g_{qc}/2\pi = 66$ MHz, balancing the need for fast gates against crosstalk.
2. PALEA Protocol for Gate Calibration: To address the difficulty of calibrating leakage errors in the $\{|11\rangle, |02\rangle\}$ subspace, the authors introduced the Phase-Averaged Leakage Error Amplification (PALEA) protocol. Unlike standard error amplification or MEADD protocols that require precise phase control or are susceptible to interference from other leakage modes (e.g., to the coupler), PALEA employs a dynamical decoupling layer and averages over random phase differences. This isolates the over-rotation angle ($\theta$) governing population transfer to the $|02\rangle$ state, allowing for high-contrast signal extraction and precise calibration of the conditional-Z (CZ) gate without needing to control the qubit drive phase at each repetition.
3. Readout Architecture and Optimization: The device utilizes a dedicated readout resonator and Purcell filter for each qubit to suppress spontaneous emission and off-resonant driving. High-fidelity readout is achieved using a shelving technique (transferring the $|e\rangle$ state to $|f\rangle$ via an $X_{12}^\pi$ pulse) combined with a traveling-wave parametric amplifier (TWPA). Additionally, a non-shelved readout configuration was optimized specifically for Quantum Non-Demolition (QND) performance.

Key Results
The authors report simultaneous high-fidelity performance across all core operations in a single device:

• Two-Qubit Gates: A 40-hour-averaged CZ gate fidelity of 99.93% (error $\approx 6.5 \times 10^{-4}$). The error budget indicates that incoherent errors and leakage to the second excited state are the dominant contributions, with coherent errors effectively suppressed.
• Single-Qubit Gates: Simultaneous single-qubit gate fidelities exceeding 99.98% (error $\approx 1.6 \times 10^{-4}$). The dominant error source is incoherent relaxation, with negligible hybridization crosstalk and leakage.
• Readout: Simultaneous readout fidelities over 99.94% for both qubits using the shelved configuration.
• QND Performance: A QND-optimized readout (without shelving) achieved a QNDness of 99.3%, demonstrating the ability to trade off readout fidelity for state preservation.
• System-Level Error: The total system-level error budget is estimated at $2.1 \times 10^{-3}$, approaching the thresholds required for efficient scaling of quantum error correction codes.

Significance and Claims
The paper claims to demonstrate a viable path toward scaling superconducting quantum processors while maintaining consistently high fidelities across all core operations. Key contributions include:

• Holistic Optimization: Proving that careful tuning of coupling strengths can resolve the traditional trade-offs between gate speed, crosstalk, and readout fidelity.
• Scalability: The demonstrated qubit-coupler architecture is naturally extensible to 2D square-grid layouts (necessary for surface code implementations) by coupling each qubit to four independent couplers without significantly altering device parameters. The chip layout supports dedicated Purcell filters on a single layer, avoiding complex multilayer fabrication.
• Calibration Advancement: The PALEA protocol provides a robust, efficient method for calibrating leakage errors, reducing leakage by at least a factor of two compared to standard methods and enabling precise characterization of coherent errors.
• Error Budgeting: The work provides a comprehensive error budget that disentangles incoherent, coherent, and leakage components, showing that the system is approaching the error rates necessary for fault-tolerant quantum computing.

The authors emphasize that their results represent a step toward uniformly high fidelities in scalable superconducting platforms, bringing fault-tolerant quantum computing closer to reality.