Fault-tolerant scalability of the superconducting two-dimensional cavity grid: full error budget and threshold validation

Helmer, Mariantoni, Fowler, von Delft, Solano, and Marquardt (EPL 85, 50007 (2009); arXiv:0706.3625) proposed a two-dimensional 'cavity grid' for superconducting quantum computation: orthogonal on-chip microwave resonators with one frequency-tunable qubit at every intersection, a Sudoku-style frequency allocation that suppresses idle interactions, cavity-mediated dispersive iSWAP gates between arbitrary qubit pairs at distance-independent swap overhead, and Steane-code unit cells with placeholder qubits as a route to a scalable fault-tolerant architecture. The authors acknowledged the distance between this sketch and a validated machine: threshold-compatible gate fidelities, crosstalk in realistic multilayer packaging, residual idle couplings and refocusing overhead, faster resonant gates, and end-to-end threshold simulations of the logical circuits. On cumulative Landau–Zener switching errors from the many frequency crossings, the paper asserted that its eight-qubits-per-cavity scalable design keeps them 'under control'. The problem is whether the architecture's composed physical error budget lies below a fault-tolerance threshold at scale. The problem is open, though today it is of mainly alternative-route interest.

Background

In circuit quantum electrodynamics (circuit QED — the superconducting-circuit analogue of cavity QED, in which Josephson-junction qubits couple to microwave resonator modes; Blais et al., Phys. Rev. A 69, 062320 (2004)), a superconducting qubit such as the transmon (a capacitively shunted Josephson junction whose transition frequency is first-order insensitive to charge noise) couples to a coplanar waveguide resonator (a planar microwave transmission line forming a standing-wave mode). In the dispersive regime, where the qubit–cavity detuning $\Delta$ greatly exceeds the coupling $g$, the cavity can be adiabatically eliminated, producing an effective flip-flop (exchange) interaction of strength $J$ between any two qubits sharing a cavity, with $J=g_{\alpha}g_{\beta}(\Delta_{\alpha}+\Delta_{\beta})/2\Delta_{\alpha}\Delta_{\beta}$. Bringing two detuned qubits into mutual resonance for a time $t=\hbar\pi/2|J|$ realizes the entangling iSWAP gate; in the idle state the interaction is off-resonant and suppressed, but not to zero, leaving a residual always-on coupling of order $J^{2}/\hbar^{2}\delta\omega$, where $\delta\omega$ is the qubit frequency spacing. In the proposed cavity grid, qubits sit at the intersections of two orthogonal resonator 'buses', and a Sudoku-like frequency assignment (all frequencies distinct within any row or column) reduces the required bandwidth from $N\delta\omega$ to roughly $\sqrt{N}\delta\omega$ for $N$ qubits. Because gates are performed by tuning qubit frequencies, each frequency sweep crosses the transition frequencies of other qubits in the same cavity; a crossing traversed too slowly transfers population spuriously (Landau–Zener tunneling — nonadiabatic transitions at an avoided level crossing), while sweeping too fast leaks population to higher transmon levels. The paper estimated a per-crossing transfer probability below $10^{-2}$ for a 10 ns, $2\pi\cdot10$ GHz sweep, and proposed refocusing (dynamical-decoupling echo sequences that average out unwanted always-on couplings) against residual idle couplings. For fault tolerance, the paper tiled a 64-qubit unit cell storing two logical qubits in the seven-qubit Steane code (a distance-3 code correcting any single physical error), with ancilla qubits for syndrome extraction and 'placeholder' qubits so that a failed SWAP cannot corrupt two data qubits at once. A quantum error-correcting code is fault-tolerant below threshold when the physical error probability per operation is small enough that the logical error rate falls exponentially as the code is scaled; by 2007–2009 the surface code (a topological code with a threshold near $10^{-2}$ for realistic noise) had become the reference standard, and the threshold-compatible engineering of a concrete superconducting architecture was the central open issue (DiVincenzo, Phys. Scr. T137, 014020 (2009)). The subsequent mainline of the field — fixed-frequency transmons in nearest-neighbor two-dimensional lattices with tunable couplers (an extra, flux-tunable circuit element that switches the qubit–qubit coupling on and off without moving qubit frequencies; Yan et al., Phys. Rev. Applied 10, 054062 (2018)) — abandoned both global cavity buses and frequency-sweep gating, and on it Google's processors demonstrated surface-code logical-error suppression with code size (Nature 614, 676 (2023)) and below-threshold, beyond-break-even operation of a distance-7 surface code with real-time decoding (Nature 638, 920 (2025)).

Problem Statement

Can the two-dimensional superconducting cavity-grid architecture of Helmer et al. be validated to operate below a quantum fault-tolerance threshold at scale? Specifically: determine whether the architecture's composed physical error budget — (i) single- and two-qubit errors of the cavity-mediated dispersive gate set, including the faster resonant gates on the $1/g$ timescale whose implementation in the grid the Helmer et al. proposal left open; (ii) cumulative Landau–Zener population transfer over all qubit-frequency crossings traversed during gate and swap sequences (chains of SWAP gates exchanging the states of neighboring qubits to move quantum information across the grid) in grids of increasing size, with optimized pulse shapes; (iii) residual always-on idle couplings of order $J^{2}/\hbar^{2}\delta\omega$ and the sufficiency and time overhead of refocusing sequences in large arrays; (iv) crosstalk, spurious modes, and packaging effects in realistic multilayer hardware; and (v) decoherence at achievable values of the energy-relaxation time $T_{1}$ (the time over which an excited qubit decays) and the pure-dephasing time $T_{\varphi}$ (the time over which quantum phase coherence is lost) — is low enough that the swap-buffered Steane-code logical circuits proposed in the Helmer et al. paper operate below threshold, with logical error rates falling as the code is scaled; and identify which engineering improvements are required where it is not. These components jointly serve the single objective of validating and optimizing the cavity-grid architecture for large-scale fault-tolerant quantum computation; a threshold verdict for the stated architecture, not for a substituted one, is the completion boundary.

References

1. Helmer, F., Mariantoni, M., Fowler, A. G., von Delft, J., Solano, E., Marquardt, F., 'Cavity grid for scalable quantum computation with superconducting circuits', EPL 85, 50007 (2009), DOI 10.1209/0295-5075/85/50007, arXiv:0706.3625, https://arxiv.org/abs/0706.3625
2. Blais, A., Huang, R.-S., Wallraff, A., Girvin, S. M., Schoelkopf, R. J., 'Cavity quantum electrodynamics for superconducting electrical circuits: an architecture for quantum computation', Phys. Rev. A 69, 062320 (2004), DOI 10.1103/PhysRevA.69.062320, arXiv:cond-mat/0402216, https://arxiv.org/abs/cond-mat/0402216
3. DiVincenzo, D. P., 'Fault-tolerant architectures for superconducting qubits', Phys. Scr. T137, 014020 (2009), DOI 10.1088/0031-8949/2009/T137/014020, arXiv:0905.4839, https://arxiv.org/abs/0905.4839
4. Mariantoni, M., Wang, H., Yamamoto, T., Neeley, M., Bialczak, R. C., Chen, Y., Lenander, M., Lucero, E., O'Connell, A. D., Sank, D., Weides, M., Wenner, J., Yin, Y., Zhao, J., Korotkov, A. N., Cleland, A. N., Martinis, J. M., 'Implementing the Quantum von Neumann Architecture with Superconducting Circuits', Science 334, 61–65 (2011), DOI 10.1126/science.1208517, arXiv:1109.3743, https://arxiv.org/abs/1109.3743
5. Ghosh, J., Fowler, A. G., Geller, M. R., 'Surface code with decoherence: An analysis of three superconducting architectures', Phys. Rev. A 86, 062318 (2012), DOI 10.1103/PhysRevA.86.062318, arXiv:1210.5799, https://arxiv.org/abs/1210.5799
6. Yan, F., Krantz, P., Sung, Y., Kjaergaard, M., Campbell, D. L., Orlando, T. P., Gustavsson, S., Oliver, W. D., 'Tunable Coupling Scheme for Implementing High-Fidelity Two-Qubit Gates', Phys. Rev. Applied 10, 054062 (2018), DOI 10.1103/PhysRevApplied.10.054062, arXiv:1803.09813, https://arxiv.org/abs/1803.09813
7. Google Quantum AI (Acharya, R., et al.), 'Suppressing quantum errors by scaling a surface code logical qubit', Nature 614, 676–681 (2023), DOI 10.1038/s41586-022-05434-1, arXiv:2207.06431, https://arxiv.org/abs/2207.06431
8. Google Quantum AI (Acharya, R., et al.), 'Quantum error correction below the surface code threshold', Nature 638, 920–926 (2025), DOI 10.1038/s41586-024-08449-y, arXiv:2408.13687, https://arxiv.org/abs/2408.13687
9. Kartal, G., Simion, G., Sorée, B., 'A superconducting quantum information processor with high qubit connectivity', arXiv:2307.08101 (2023), https://arxiv.org/abs/2307.08101