Hanyang University Team Designs Lattice-Patch Architecture for Surface-Code Mapping

Scientists at Hanyang University have developed a new lattice-patch architecture to address critical limitations in superconducting transmon processors, a prominent platform for building large-scale quantum computers. Chanpyo Kim and colleagues demonstrate a method of coupling four fixed-frequency transmons to a single coupler, significantly enhancing qubit connectivity and streamlining the process of mapping onto the surface-code lattice. This entirely fixed-frequency design mitigates susceptibility to external flux noise and achieves CNOT gate fidelities exceeding 0.98 across all connectivity directions within the patch, representing a considerable advancement towards robust and efficient fault-tolerant quantum computation.

High-fidelity CNOT gates enable scalable quantum error correction via lattice-patch architecture

CNOT gate fidelities now surpass 0.98 across all six connectivity directions, marking a substantial improvement over previous designs. Maintaining consistently high fidelity across multiple qubit connections has historically been problematic due to frequency crowding, a phenomenon where qubit transition frequencies become too close, leading to unwanted interactions, and inefficient mapping onto the surface code. The surface code is a leading error correction scheme requiring specific qubit connectivity to effectively detect and correct errors. At Hanyang University, the team overcame these limitations with a novel “lattice-patch” architecture, deliberately aligning it with the surface-code lattice unit to minimise the complex software adjustments typically required for logical qubit implementation. Logical qubits, formed from multiple physical qubits, are essential for performing reliable quantum computations as they provide inherent error resilience.

Multi-level numerical simulations rigorously confirmed the performance characteristics across all connections within the lattice-patch. These simulations employed sophisticated modelling techniques to predict qubit behaviour and optimise the design parameters. To calibrate residual phase accumulation, a subtle error source arising from the four-qubit interaction network, virtual Rz gates were implemented. Rz gates are single-qubit rotations used to adjust the phase of the quantum state, effectively compensating for imperfections in the coupling and interactions. Consistently achieving CNOT gate fidelities exceeding 0.98 across all six connectivity directions was demonstrably possible using this innovative “lattice-patch” architecture. This level of fidelity is crucial because the probability of error in a quantum computation scales exponentially with the number of qubits; therefore, minimising error rates at the gate level is paramount. While these results establish a strong foundational building block, practical fault-tolerant quantum computation still necessitates scaling this architecture to substantially larger qubit numbers and addressing the challenges of controlling complex interactions within the expanded system, including crosstalk and decoherence.

This design couples four fixed-frequency superconducting qubits to a single fixed-frequency coupler, fundamentally eliminating susceptibility to external flux noise, a pervasive source of instability in superconducting quantum systems. Flux noise arises from fluctuations in magnetic fields, which can disrupt qubit coherence and introduce errors. By fixing the qubit frequencies, the system becomes less sensitive to these external disturbances. The entirely fixed-frequency approach significantly simplifies logical qubit implementation for quantum error correction, reducing the computational burden and complexity of the required software adjustments. The team, originally based at Santa Barbara before relocating to Hanyang University, have created a genuinely new architecture that tackles key problems hindering the construction of larger quantum processors. However, the significance of software phase updates, while minimised, remains a factor in achieving optimal performance and must be carefully considered during system calibration and operation.

The lattice-patch architecture presents a novel approach to connecting superconducting qubits by coupling four qubits to a single component, offering a more efficient use of resources compared to traditional qubit-coupler-qubit (QCQ) architectures. Direct alignment with the surface code simplifies the complex software adjustments usually needed to implement logical qubits, reducing the overhead associated with error correction. Unintended phase shifts arising from qubit interactions necessitated calibration, accomplished using software tools to fine-tune the quantum state and compensate for systematic errors. Specifically, this design directly addresses frequency crowding and simplifies the complex mapping needed for error correction, a vital step towards reliable computation. The surface code requires qubits to be connected in a specific lattice structure; the lattice-patch architecture facilitates this mapping by providing inherent connectivity. Future research will focus on scaling the system to larger qubit numbers, investigating methods for interconnecting multiple lattice-patches, and exploring the potential for three-dimensional integration to further increase qubit density and connectivity. The long-term goal is to build a quantum computer capable of solving problems intractable for classical computers, with applications in materials science, drug discovery, and financial modelling.

The researchers demonstrated a new lattice-patch architecture for superconducting qubits, achieving CNOT gate fidelities exceeding 0.98 across six connectivity directions. This design improves qubit connectivity and aligns directly with the surface-code lattice, simplifying the implementation of logical qubits for quantum error correction. Although the architecture requires calibration to address residual phase accumulation, this is achieved through software phase updates. The lattice-patch architecture represents an efficient building block for future, more robust quantum computers, and the authors intend to scale the system to larger qubit numbers.