⚛️ quantum physics

Low-leakage superconducting-qubit measurement with sub-100-ns total duration

This paper demonstrates a fast, high-fidelity, and low-leakage superconducting qubit measurement protocol achieving a 97-ns total duration and 0.17% assignment error by utilizing a large resonator decay rate and an optimized dispersive shift, thereby enabling rapid readout with minimal state transitions suitable for quantum error correction.

Original authors: Peter A. Spring, Adrian L. Hesse, Shiyu Wang, Shuhei Tamate, Yasunobu Nakamura

Published 2026-08-27

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Peter A. Spring, Adrian L. Hesse, Shiyu Wang, Shuhei Tamate, Yasunobu Nakamura

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

In the world of quantum computing, the machine's brain is made of tiny circuits that can exist in two states at once, a property that allows them to solve problems impossible for ordinary computers. However, these circuits are incredibly fragile. To use them, scientists must constantly check their state, a process called measurement, to ensure they are working correctly and to correct any errors before they spread. This check must be incredibly fast and accurate, because the moment a circuit is left idle while waiting for a measurement, it begins to lose its delicate quantum information. For years, a major bottleneck in building these powerful machines has been that reading the state of a circuit is slower and more prone to mistakes than the operations that manipulate it. If the reading process is too slow, the information fades away; if it is too aggressive, the act of looking at the circuit can knock it out of its intended state entirely.

A team of researchers at the RIKEN Center for Quantum Computing in Japan has now demonstrated a way to read the state of a superconducting circuit with unprecedented speed and gentleness. They managed to complete a full measurement in less than one hundred nanoseconds, a timeframe so short it is measured in billionths of a second. More importantly, they achieved this without disturbing the circuit or causing it to leak into unwanted, higher-energy states that would ruin the calculation. By carefully tuning the interaction between the circuit and a specialized filter, the team proved that it is possible to read quantum information quickly and accurately even when the components are tuned very close to each other in frequency, a condition that was previously thought to be too risky for high-speed operations.

The researchers focused on a specific type of quantum circuit known as a transmon, which is a standard building block in many modern quantum processors. To measure the state of this circuit, they used a technique called dispersive readout. Imagine the circuit is connected to a resonator, a small chamber that can hold electromagnetic waves, much like a guitar string holds a specific note. When the circuit is in one state, it slightly changes the pitch of the resonator; when it is in the other state, the pitch changes differently. By sending a signal into this resonator and listening to how it bounces back, scientists can tell which state the circuit is in. The challenge has always been balancing speed with safety. To get a clear signal quickly, one needs a strong connection between the circuit and the resonator, but a strong connection can also cause the circuit to lose energy or jump to the wrong state.

In this new work, the team introduced a dedicated filter resonator into the setup. This filter acts as a gatekeeper, allowing the measurement signal to pass through efficiently while protecting the main circuit from unnecessary noise. They designed the system so that the resonator responds very quickly to the measurement signal, allowing them to gather enough information to make a decision in just 58 nanoseconds. Once the measurement pulse ends, the remaining energy in the resonator drains away passively in a few tens of nanoseconds, without needing any extra active pulses to clear it out. This rapid clearing means the circuit can return to normal operation almost immediately. The researchers defined the total measurement time as the period from the start of the pulse until the error introduced by the measurement drops below a tiny threshold. They found this total time to be 97 nanoseconds.

The team also rigorously tested whether this fast measurement caused the circuit to "leak" into higher energy levels that are not part of the standard two-state system. Such leakage is a common problem when measurements are too strong or the components are too close in frequency. Using a specialized, highly sensitive measurement sequence, they tracked the circuit's behavior over thousands of repeated trials. They found that the rate at which the circuit leaked into these unwanted states was extremely low, occurring only about twice as often as the natural background rate of leakage that happens even without any measurement. This leakage rate was two orders of magnitude smaller than the rate at which the measurement caused the circuit to relax or lose energy, which was the dominant source of error.

To understand why the system remained so stable despite operating in a regime where many unwanted resonances are theoretically possible, the researchers ran detailed computer simulations. They examined the complex interactions between the measurement signal and the circuit's energy levels. Their simulations showed that although the measurement pulse did pass through several points where resonances could occur, the connection between the states at these points was incredibly weak. Because the resonator drained energy so quickly, the system moved through these potential trouble spots too fast to get stuck or jump to the wrong state. The rapid passage meant the circuit crossed these boundaries without being disturbed, a process known as traversing diabatically.

The results confirm that it is possible to achieve fast, high-fidelity, and low-leakage measurements even when the circuit and the resonator are tuned very close to each other in frequency. This finding is significant because it removes a major constraint on how quantum processors can be designed. Previously, engineers might have felt forced to separate these components widely to avoid errors, which could complicate the design and reduce performance. By showing that a large decay rate combined with a carefully optimized signal can work effectively at small frequency differences, this work opens the door to more compact and efficient quantum processors. The ability to read the state of a quantum bit in under 100 nanoseconds with such high accuracy brings the practical realization of error-corrected quantum computers one step closer to reality.

Technical Summary: Low-leakage superconducting-qubit measurement with sub-100-ns total duration

Problem Statement
Fast, accurate, and low-leakage qubit measurement is a fundamental requirement for quantum error correction (QEC), as it conditions feed-forward operations and extracts stabilizer syndromes. In superconducting quantum processors, readout remains a bottleneck, often being slower and more error-prone than single- and two-qubit gate operations. A primary trade-off exists in dispersive readout: achieving a high signal-to-noise ratio (SNR) per photon typically requires a dispersive shift ( $\chi$ ) comparable to the resonator decay rate ( $\kappa$ ), which is most easily achieved at small qubit–resonator detuning. However, small detuning regimes are theoretically prone to dense multiphoton resonances that can induce measurement-induced state transitions (MIST), causing leakage out of the computational subspace. Conversely, high-detuning regimes suppress these resonances but often sacrifice SNR efficiency or require complex engineering. The challenge is to determine if fast, high-fidelity readout is compatible with low leakage in the small-detuning regime.

Methodology
The authors investigate a superconducting transmon qubit within a 64-qubit processor featuring a tileable 3D-integrated circuit architecture. The readout system utilizes a dedicated filter resonator coupled to the readout resonator, a design that suppresses qubit relaxation through the feedline while permitting a large readout bandwidth.

Key experimental strategies include:

Optimized Readout Parameters: The system is tuned to a small qubit–resonator detuning with a frequency ratio $\omega_r/\omega_q = 1.24$ . The design targets a state-averaged effective resonator decay rate of $\kappa_{\text{eff}}/2\pi = 30.8$ MHz and a dispersive shift close to the optimal SNR-per-photon condition ( $|\chi| \approx \kappa/2$ ).
Fast Measurement Pulse: A 58-ns measurement pulse is employed, featuring a high-amplitude initial section to boost early SNR and cosine ramps for edges.
Passive Depletion: The experiment relies on the large $\kappa_{\text{eff}}$ to allow residual readout photons to deplete passively within tens of nanoseconds, avoiding the need for active depletion pulses.
Benchmarking Protocols:
- Assignment Fidelity: Measured using a sequence involving state preparation, a 58-ns measurement, a delay, and a second measurement, utilizing a flux-driven impedance-matched parametric amplifier (IMPA) for high quantum efficiency.
- Total Duration: Defined as the time from the start of the measurement pulse until the measurement-induced error on a subsequent $\pi$ -pulse falls below $10^{-4}$ . This is determined by repeating measurement– $\pi$ -pulse blocks and analyzing the final state population.
- Leakage Characterization: A leakage-sensitive measurement is performed using a 192-ns pulse and a machine-learning classifier (log-signature of the heterodyne path) to distinguish between computational states ( $|0\rangle, |1\rangle$ ) and leakage states ( $|2\rangle$ to $|4\rangle$ ).
- MIST Benchmarking: A repeated-measurement sequence accumulates leakage and relaxation errors to extract per-measurement transition probabilities.
Theoretical Analysis: Floquet simulations are used to analyze multiphoton resonances and hybridization parameters at the operating point.

Key Results

Speed and Fidelity: The system achieves an assignment error of $\epsilon_F = 0.17(1)\%$ using a 58-ns measurement pulse. The total measurement duration, defined by the recovery time for high-fidelity operations, is $t_{\text{meas}} = 97(1)$ ns. This recovery time is achieved without active photon depletion, relying solely on passive decay.
Leakage Rates: The measurement-induced leakage rate is measured to be $P_L^{\text{on}} = 2.7(2) \times 10^{-5}$ per measurement. This is only twice the background leakage rate ( $1.3(3) \times 10^{-5}$ ) and two orders of magnitude lower than the measurement-induced relaxation rate ( $P_{10}^{\text{on}} = 7.55 \times 10^{-3}$ ).
Dominant Error Source: The assignment error is dominated by measurement-induced relaxation ( $|1\rangle \to |0\rangle$ ), which accounts for approximately 80% of the total error, rather than leakage.
Mechanism of Low Leakage: Despite operating in a small-detuning regime where multiphoton resonances are expected to be dense, Floquet simulations reveal that the resonances traversed by the measurement pulse are weakly coupled. The effective decay rate is sufficiently large that these resonances are traversed diabatically (rapidly), preventing significant population transfer to leakage states. The maximum leakage probability estimated from simulations is on the order of $10^{-6}$ to $10^{-7}$ , consistent with experimental observations.

Significance
The paper demonstrates that fast, high-fidelity, and low-leakage dispersive readout is achievable in the small qubit–resonator detuning regime. By combining a large state-averaged effective resonator decay rate with a dispersive shift optimized for SNR per photon, the authors show that the potential drawbacks of small detuning (dense multiphoton resonances) can be mitigated. The results establish that a large resonator decay rate enables rapid passive photon depletion, allowing qubit operations to resume within sub-100 ns, while the weak coupling of resonances at the specific operating point prevents leakage. This validates a practical pathway for high-speed readout in superconducting quantum processors without requiring the high-detuning regime or active depletion schemes.

More like this