Team Models Leakage Mobility for Transmon Processor Architectures

A thorough investigation into qubit leakage, a key source of error in quantum computing, reveals methods for its control and removal. Taneli Tolppanen and colleagues at University of Oulu and Aalto University characterise leakage dynamics in transmons with tunable couplers using both numerical and analytical techniques. The research shows that despite efforts to eliminate interactions, leakage hopping rates persist at 0.8-10MHz due to transmon nonlinearity, but can be localised by frequency detuning. Key to the findings, the team propose two passive leakage removal units using this leakage mobility, offering readily applicable solutions for improving the performance of superconducting quantum devices.

Frequency detuning localises leakage and improves qubit performance

Leakage hopping rates have been reduced to below 0.8MHz by researchers at University of Oulu and Aalto University, a threshold previously unattainable in superconducting quantum devices. This reduction is significant because qubit leakage represents a fundamental limitation in achieving high-fidelity quantum operations. Superconducting qubits, specifically transmons, are susceptible to leakage due to their anharmonicity, the non-equidistant spacing of energy levels. While this anharmonicity is crucial for defining a computational basis, it also allows for transitions to higher, unintended energy levels, constituting leakage. These leakage states then interact with other qubits, introducing errors. A frequency spread of 1-4MHz between next-nearest neighbour transmons effectively suppresses longer-range leakage, preventing unwanted interactions. Characterisation of these leakage dynamics and identification of how frequency detuning localizes these excitations offers a pathway to improved qubit stability and performance. The team’s work demonstrates that carefully engineering the frequency landscape of a quantum processor can dramatically reduce the impact of leakage errors.

Previously, leakage hopping persisted at rates between 0.8-10MHz even with optimised coupler tuning. Traditional approaches to minimising leakage focus on optimising the coupling between qubits, aiming to reduce the strength of interactions that drive leakage transitions. However, the inherent nonlinearity of transmons means that some level of coupling, and therefore leakage, remains unavoidable. The team proposed two designs for ‘leakage removal units’, one utilising a pumped transmon and a tunable coupler, and another based on a junction readout scheme, suggesting practical applications of this controlled mobility. The pumped transmon design leverages an auxiliary qubit to absorb leakage energy, while the junction readout scheme provides a dedicated pathway for dissipating leakage excitations. While these designs show potential for removal times as low as 0.30ns, current performance does not yet reflect operation within a fully scaled, complex processor architecture where interactions between many qubits could introduce further challenges. Scaling these units to larger systems will require careful consideration of cross-talk and the accumulation of leakage in the removal units themselves.

Frequency detuning between qubits can control unwanted energy leaks within superconducting processors, as detailed in this work. This approach moves beyond simply minimising leakage creation, instead focusing on managing its mobility within the processor architecture. This addresses the persistent issue of leakage hopping rates, which can remain significant even when couplers are optimised. The researchers employed both numerical simulations, using established quantum simulation techniques, and analytical calculations based on perturbation theory to model the leakage dynamics. These methods allowed them to accurately predict the impact of frequency detuning on leakage hopping rates and to optimise the design of the leakage removal units. Further investigation will focus on adapting these techniques for use in larger, more complex quantum systems, including exploring the impact of disorder and fabrication variations on leakage dynamics.

Frequency detuning offers a pathway towards localised qubit error mitigation

Mitigating qubit leakage is an increasingly important focus for scientists, as it represents a persistent source of errors hindering the development of practical quantum computers. The accumulation of leakage errors can severely limit the coherence time of qubits, the duration for which they maintain quantum information, and ultimately restrict the complexity of quantum algorithms that can be reliably executed. An alternative strategy gaining traction involves actively resetting qubits via measurement, as explored by Lacroix and colleagues. This approach, known as measurement-based reset, aims to periodically project qubits back into their ground state, effectively eliminating leakage. Repetitive parity measurements, for example, demand significant overhead and could introduce new error pathways, potentially negating the benefits of leakage reduction, despite its potential effectiveness. The overhead arises from the need to perform frequent measurements, which themselves are imperfect and can introduce errors. Furthermore, the measurement process can disturb the quantum state of the qubits, leading to decoherence.

Understanding leakage dynamics remains important for building better quantum processors, even while acknowledging the challenges of overhead introduced by frequent measurement-based reset strategies. Controlling the movement of leakage excitations, unwanted energy transitions between qubits, is important for improving quantum processor stability. The team’s approach offers a complementary strategy to measurement-based reset, potentially reducing the frequency with which resets are needed and thereby lowering the overall error rate. Deliberately setting qubits to slightly different frequencies effectively localizes these excitations and prevents their spread, as the team found by characterising leakage dynamics. This localized control is key as it addresses the issue of leakage hopping rates, which can remain significant even when components that control qubit interaction are optimised. The ability to confine leakage to specific regions of the processor allows for targeted removal or dissipation, minimising its impact on the overall computation. This is particularly important in larger systems where leakage can propagate across long distances, affecting multiple qubits. The research highlights the importance of considering not only the creation of leakage but also its propagation and management within the quantum processor architecture.

The researchers demonstrated that leakage excitations, a source of error in quantum computers, move between qubits at rates of 0.8-10MHz even with optimised qubit interactions. This matters because uncontrolled leakage spreads errors and degrades the performance of quantum calculations. By carefully tuning the frequencies of qubits to create a spread of 1-4MHz, the team localised these excitations, preventing their propagation. They also proposed two passive units to remove leakage, offering an alternative to strategies that rely on frequent, potentially disruptive, measurements.