Dynamical decoupling and noise spectroscopy with a superconducting flux qubit

TL;DR

This paper demonstrates how dynamical decoupling with CPMG sequences significantly extends the coherence time of a superconducting flux qubit by mitigating low-frequency noise, enabling detailed noise spectroscopy.

Contribution

It introduces a method to use dynamical decoupling for noise mitigation and spectroscopy in superconducting qubits, achieving a 50-fold increase in T2 coherence time.

Findings

50-fold increase in T2 coherence time using CPMG sequences

Relaxation-limited T2 of 23 microseconds achieved

Successful reconstruction of environmental noise spectral density

Abstract

The characterization and mitigation of decoherence in natural and artificial two-level systems (qubits) is fundamental to quantum information science and its applications. Decoherence of a quantum superposition state arises from the interaction between the constituent system and the uncontrolled degrees of freedom in its environment. Within the standard Bloch-Redfield picture of two-level system dynamics, qubit decoherence is characterized by two rates: a longitudinal relaxation rate Gamma1 due to the exchange of energy with the environment, and a transverse relaxation rate Gamma2 = Gamma1/2 + Gamma_phi which contains the pure dephasing rate Gamma_phi. Irreversible energy relaxation can only be mitigated by reducing the amount of environmental noise, reducing the qubit's internal sensitivity to that noise, or through multi-qubit encoding and error correction protocols (which already presume ultra-low error rates). In contrast, dephasing is in principle reversible and can be refocused dynamically through the application of coherent control pulse methods. In this work we demonstrate how dynamical-decoupling techniques can moderate the dephasing effects of low-frequency noise on a superconducting qubit with energy-relaxation time T1 = 1/Gamma1 = 12 us. Using the CPMG sequence with up to 200 pi-pulses, we demonstrate a 50-fold improvement in the transverse relaxation time T2 over its baseline value. We observe relaxation-limited times T2(CPMG) = 23 us = 2 T1 resulting from CPMG-mediated Gaussian pure-dephasing times in apparent excess of 100 us. We leverage the filtering property of this sequence in conjunction with Rabi and energy relaxation measurements to facilitate the spectroscopy and reconstruction of the environmental noise power spectral density.