Scalable Method for Eliminating Residual $ZZ$ Interaction between Superconducting Qubits

TL;DR

This paper presents a scalable, experimentally validated method to eliminate residual $ZZ$ interactions in superconducting qubits using a weak microwave drive, improving quantum operation fidelity for large-scale quantum computing.

Contribution

The authors introduce a practical approach employing a microwave drive on the coupler to cancel residual $ZZ$ interactions, compatible with various coupler types and scalable to large quantum processors.

Findings

Complete $ZZ$ cancellation verified by measuring two-qubit phases and correlations.

Idling gate fidelity approaches the coherence limit, confirming effective $ZZ$ suppression.

Method is applicable to both tunable and nontunable couplers, enhancing scalability.

Abstract

Unwanted $ZZ$ interaction is a quantum-mechanical crosstalk phenomenon which correlates qubit dynamics and is ubiquitous in superconducting qubit systems. It adversely affects the quality of quantum operations and can be detrimental in scalable quantum information processing. Here we propose and experimentally demonstrate a practically extensible approach for complete cancellation of residual $ZZ$ interaction between fixed-frequency transmon qubits, which are known for long coherence and simple control. We apply to the intermediate coupler that connects the qubits a weak microwave drive at a properly chosen frequency in order to noninvasively induce an ac Stark shift for $ZZ$ cancellation. We verify the cancellation performance by measuring vanishing two-qubit entangling phases and $ZZ$ correlations. In addition, we implement a randomized benchmarking experiment to extract the idling gate fidelity which shows good agreement with the coherence limit, demonstrating the effectiveness of $ZZ$ cancellation. Our method allows independent addressability of each qubit-qubit connection, and is applicable to both nontunable and tunable couplers, promising better compatibility with future large-scale quantum processors.