IEEE Transactions on Instrumentation and Measurement
A Noise Diagnostic and Suppression Methodology for Improving Measurement Stability of SERF Atomic Sensors
Jiaxin Liu, Puchen Gong, Xusheng Lei, Haoying Pang
Abstract
Achieving high measurement stability in spin exchange relaxation free (SERF) atomic sensors is a central challenge in quantum instrumentation, yet the intrinsic noise mechanisms that ultimately limit measurement performance remain insufficiently characterized. This article proposes a model-based noise diagnostic and suppression methodology that systematically links microscopic spin fluctuations to macroscopic measurement noise. The Langevin stochastic terms are introduced into the coupled Bloch equations to form a noise characterization framework that captures stochastic excitation, dissipation, and finite correlation of the coupled electronic and nuclear spin modes, enabling identification and quantification of the noise sources responsible for spectral coloring and low-frequency accumulation in the measurement output. Two previously overlooked noise amplification pathways in the measurement system are diagnosed: projection leakage caused by nonorthogonal optical readout geometry and stochastic magnetic injection introduced by electromagnetic coupling in the temperature control subsystem. For each pathway, the targeted instrumentation optimization procedures are derived and implemented. The experimental evaluation on a SERF atomic gyroscope demonstrates that joint optimization of the heating and optical subsystems reduces the amplitude spectral density at 1 Hz from$7.66 \times 10^{-6}$to$3.12 \times 10^{{-6} {\circ }} $/s$/\sqrt {\mathrm {Hz}}$, a 41.6% improvement, and suppresses the nonwhite hump in the Allan deviation over$\tau \approx 0.1$–10 s. The proposed methodology provides a reusable, physics-informed route for diagnosing, attributing, and mitigating noise in high-stability quantum measurement systems.