Quantum Computer Components Remain Stable for a Year Despite Temperature Swings

Researchers investigating the long-term reliability of quantum computing hardware have revealed crucial insights into qubit stability. Cong Li, Zhaohua Yang, and Xinfang Zhang, from the College of Computer Science and Technology at the National University of Defence Technology, alongside Zhihao Wu, Shichuan Xue, and Mingtang Deng et al., systematically characterised 27 transmon qubits over a year, subjecting them to four thermal cycles. Their findings establish a clear distinction between the inherent stability of qubit fabrication and the fluctuating influence of environmental factors. The team demonstrated that while core qubit parameters remain robust, the microscopic defect landscape undergoes substantial change with each thermal cycle, effectively acting as a ‘hard reset’ for noise realisation. This work highlights the necessity for automated recalibration in scaling up quantum systems, as fabrication quality alone cannot guarantee consistent performance over time.

Intrinsic Qubit Parameters Exhibit Resilience to Thermal Cycling in Superconducting Hardware

Scientists have established a distinct hierarchy of stability within superconducting quantum hardware through comprehensive longitudinal characterisation of 27 frequency-tunable qubits. This work details a year-long study spanning four thermal cycles, revealing that intrinsic device parameters governing qubit frequency and energy relaxation times (T1) demonstrate remarkable robustness against thermal stress.

Frequency deviations were typically confined within 0.5%, with coherence baselines remaining largely unaffected by the temperature fluctuations. In contrast, environmental variables, notably background magnetic flux offsets and the microscopic landscape of two-level system (TLS) defects, underwent substantial stochastic reconfiguration following each thermal cycle.

Researchers employed frequency-dependent relaxation spectroscopy and a newly defined quantitative metric, the T1 Spectral Topography Fidelity, to demonstrate that thermal cycling effectively acts as a “hard reset” for the local defect environment. This process introduces a level of spectral randomization equivalent to the effects of thousands of hours of continuous low-temperature operation.

The study confirms that while the fundamental fabrication quality of the qubits is preserved, the specific realization of noise affecting their performance is statistically unique after each thermal cycle. This necessitates the development of automated recalibration strategies for future large-scale quantum systems.

The investigation focused on a superconducting quantum processor featuring a flip-chip architecture with 66 aluminum-based frequency-tunable transmon qubits and 110 tunable couplers. Each qubit was meticulously tracked across four thermal cycles, involving warming to room temperature and cooling back to a base temperature of approximately 20 mK.

Analysis of the sweet-spot frequency and flux bias offset revealed that maximum qubit frequencies remained highly stable, with deviations generally below 20MHz, a relative variation of less than 0.5% given typical frequencies around 4.5GHz. This suggests the core quantum hardware, including capacitor geometry and Josephson junction tunnel barriers, maintains its structural integrity throughout the thermal cycling process.

Furthermore, the research introduces the T1 Spectral Topography Fidelity as a means to evaluate correlations within the qubit relaxation landscape. Findings indicate that a single thermal cycle can reshuffle the spectral landscape to an extent comparable to prolonged low-temperature evolution, without causing observable degradation to the qubits themselves. This detailed insight into the impact of thermal cycling on superconducting quantum processors underscores the importance of adaptive recalibration protocols for building scalable and reliable quantum computing systems.

Superconducting Qubit Fabrication and Cryogenic Measurement Setup

A 72-qubit superconducting processor served as the platform for a comprehensive longitudinal characterisation of 27 frequency-tunable qubits over one year, encompassing four thermal cycles. The device comprised two stacked chips fabricated on sapphire substrates, a top chip integrating 66 aluminum-based frequency-tunable transmons and 110 tunable couplers, and a bottom chip housing readout and control circuitry electrically connected via indium bump bonds.

Each qubit was controlled using a combined line for microwave drive and flux bias, with state measurement performed through dispersive readout coupled to a Purcell filter. For systematic stability investigation, the assembled chip was wire-bonded into a gold-plated copper package mounted on the mixing chamber of a dilution refrigerator.

To mitigate external magnetic noise, the sample holder was enclosed in a magnetic shielding cover, and all input/output lines were heavily attenuated and filtered at multiple thermal stages to suppress thermal and photon shot noise. The base temperature was consistently maintained at approximately 20 mK throughout the year-long characterisation period.

A thermal cycle involved warming the system to above 290 K and subsequently cooling it back to the base temperature, with the longitudinal dataset tracking the evolution of the 27 qubits across these cycles of varying durations. Flux-dependent spectroscopy was performed for each cycle to characterise the qubit frequency response, enabling accurate determination of the sweet-spot frequency (f max 01) and its corresponding flux bias offset (Imax b).

Fitting the measured spectral arches allowed for precise quantification of these parameters. The evolution of these parameters for a subset of 11 qubits, labelled Q1 to Q11, was then analysed, displaying the maximum qubit frequency and the magnitude of deviation in the sweet-spot flux bias offset, both referenced to the cycle-1 baseline.

Observed frequency deviations |∆f max 01| remained confined within ±20MHz for all qubits across all cycles, representing a relative variation of less than 0.5% given a typical f max 01 of around 4.5GHz. In contrast to the stable qubit frequencies, the flux bias offsets exhibited significant stochastic fluctuations, reaching magnitudes as large as approximately 0.12 Φ0 for qubits Q10 and Q11 during specific cycles.

This fluctuating behaviour is indicative of dynamic flux trapping and reconfiguration processes within the device, suggesting that thermal cycling acts as a reset mechanism for the magnetic environment. Energy relaxation time, T1, was also characterised over the four thermal cycles, providing a complete picture of qubit performance under thermal stress.

Intrinsic qubit parameters demonstrate resilience to thermal cycling despite environmental instability

Frequency deviations remained within 0.5% for the intrinsic device parameters determining qubit frequency across the one-year study and four thermal cycles. Baseline energy relaxation times, denoted as T1, also exhibited high robustness against thermal stress, indicating preserved fabrication quality.

A comprehensive longitudinal characterisation was performed on 27 frequency-tunable transmon qubits to establish a distinct hierarchy of stability for superconducting hardware. The research focused on quantifying the impact of thermal cycling on device parameters, specifically examining the effects of large thermal swings between room temperature and millikelvin conditions.

In contrast to the stable intrinsic parameters, environmental variables such as background magnetic flux offsets and the microscopic landscape of two-level system (TLS) defects underwent significant stochastic reconfiguration after each thermal cycle. Frequency-dependent relaxation spectroscopy was employed alongside a quantitative metric, the T1 Spectral Topography Fidelity, to assess the impact of thermal cycling on the local defect environment.

Results demonstrated that a single thermal cycle induced a level of spectral randomization equivalent to thousands of hours of continuous low-temperature evolution. The T1 Spectral Topography Fidelity served as a measure of correlations in the qubit relaxation landscape, revealing substantial changes after each thermal cycle.

This “hard reset” of the local defect environment confirms that while fabrication quality is maintained, the specific noise realization is statistically distinct for each thermal cycle. No observable qubit degradation was attributed to the thermal cycling process itself, highlighting the resilience of the core qubit structure.

The study underscores the necessity of automated recalibration strategies for large-scale quantum systems to account for these stochastic changes in the environmental noise. These findings provide detailed insights into the long-term operational requirements and maintenance of superconducting quantum processors.

Thermal cycling induces stochastic resetting of qubit noise landscapes

Researchers have demonstrated a clear hierarchy of stability in superconducting quantum hardware through a year-long study of 27 frequency-tunable qubits across multiple thermal cycles. Intrinsic qubit parameters, such as frequency and energy relaxation times, proved remarkably robust to thermal stress, exhibiting frequency deviations of less than 0.5% and maintaining coherence baselines.

Conversely, environmental variables, specifically background magnetic flux offsets and the distribution of two-level system (TLS) defects, underwent substantial stochastic reconfiguration with each thermal cycle. Quantitative analysis, utilising a metric termed Spectral Topography Fidelity, revealed that each thermal cycle effectively acts as a “hard reset” for the local defect environment.

This reset introduces spectral randomization equivalent to thousands of hours of continuous low-temperature evolution, confirming that while fabrication quality remains consistent, the specific noise characteristics are statistically unique after each cycle. These findings underscore the need for automated recalibration strategies in large-scale quantum systems to account for this environmental randomness.

The study acknowledges that the microscopic landscape of TLS defects is fundamentally stochastic, with changes potentially arising from interactions between defects or quasiparticle behaviour. Future research could explore controlled thermal cycling as a method for probabilistically resetting problematic defect configurations. This work establishes a foundation for adaptive calibration and maintenance procedures crucial for the development of scalable quantum computing, bridging the gap between microscopic defect dynamics and practical quantum engineering.