Researchers at Origin Quantum Computing Company Limited, working with the University of Science and Technology of China, have developed a new method to significantly reduce signal interference in superconducting qubits, a major hurdle in building more powerful quantum computers. Magnetic flux crosstalk between qubits and couplers limits the scalability of these processors, but the team’s spin-echo-based technique effectively separates quantum and flux crosstalk for more accurate characterization. This approach, detailed in a recent publication in Physics Applied, achieves stabilization of frequency-shift fluctuations at a noise baseline of approximately 20 kHz, with crosstalk coefficient accuracy reaching an order of 10−5 after compensation. “This method provides a robust and efficient framework for mitigating crosstalk, paving the way for high-fidelity control of large-scale quantum processors,” the researchers state, offering a path toward more stable and complex quantum systems.
Spin-Echo Method Separates Quantum and Flux Crosstalk
The team, led by Peng Duan and Guo-Ping Guo, addressed these challenges with a spin-echo-based method validated through experimentation, offering a pathway to more accurate characterization of flux crosstalk. This approach integrates a learning-based algorithm with a high-parallelism measurement scheme to boost efficiency, moving beyond previous limitations in crosstalk mitigation. Crucially, the accuracy of the crosstalk coefficient reached an order of 10−5 after compensation, indicating a significant reduction in unwanted signal interference. The research, which began with manuscript submission on June 6, 2025, and concluded with acceptance on January 21, 2026, represents a key advancement in addressing a fundamental obstacle to realizing practical quantum computation.
Existing tunable coupling architectures, which rely on frequency-tunable qubits and couplers, are particularly susceptible to this interference; however, a newly validated spin-echo-based method effectively isolates quantum and flux crosstalk, allowing for more precise measurements. This separation is critical because quantum crosstalk induced by strong qubit-coupler interactions complicates traditional compensation methods. The team’s innovation extends beyond improved measurement to include a learning-based algorithm integrated with a high-parallelism measurement scheme, significantly boosting efficiency.
This approach achieves the stabilization of frequency-shift fluctuations at a noise baseline of approximately 20 kHz, with the accuracy of the crosstalk coefficient reaching an order of 10 − 5 after compensation.
