Dual-rail erasure qubits can substantially improve the efficiency of quantum error correction, allowing lower error rates to be achieved with fewer qubits, but each erasure qubit requires 3× more transmons to implement compared to standard qubits. In this work, we introduce a hybrid-erasure architecture for surface code error correction where a carefully chosen subset of qubits is designated as erasure qubits while the rest remain standard. Through code-capacity analysis and circuit-level simulations, we show that a hybrid-erasure architecture can boost the performance of the surface code -- much like how a game of Minesweeper becomes easier once a few squares are revealed -- while using fewer resources than a full-erasure architecture. We study strategies for the allocation and placement of erasure qubits through analysis and simulations. We then use the hybrid-erasure architecture to explore the trade-offs between per-qubit cost and key logical performance metrics such as threshold and effective distance in surface code error correction. Our results show that the strategic introduction of dual-rail erasure qubits in a transmon architecture can enhance the logical performance of surface codes for a fixed transmon budget, particularly for near-term-relevant transmon counts and logical error rates.
@article{arxiv.2505.00066,
title = {Erasure Minesweeper: exploring hybrid-erasure surface code architectures for efficient quantum error correction},
author = {Jason D. Chadwick and Mariesa H. Teo and Joshua Viszlai and Willers Yang and Frederic T. Chong},
journal= {arXiv preprint arXiv:2505.00066},
year = {2025}
}