The comagnetometer has been one of the most sensitive devices with which to test new physics related to spin-dependent interactions, but the comagnetometers based on overlapping ensembles of multiple spin species usually suffer from systematic errors due to magnetic field gradients. Here, we propose a comagnetometer based on the Zeeman transitions of the dual hyperfine levels in ground-state ${}^{87}\mathrm{Rb}$ atoms, which shows nearly negligible sensitivity to variations of laser power and frequency, magnetic field, and magnetic field gradients. We measured the hypothetical spin-dependent gravitational energy of the proton with the comagnetometer, which is smaller than $4\times {10}^{-18}\mathrm{eV}$, comparable to the most stringent existing constraint. Through optimizing the system parameters such as cell temperature, laser power, amplitude of driving magnetic field, as well as choosing better current source, it is possible to improve the sensitivity of the comagnetometer further.

Magnetic field gradients reduce the transverse relaxation time of nuclear spins, which usually degrades the sensitivity of atomic sensors based on nuclear spins. We demonstrate improved magnetic field gradient compensation by applying first-order and second-order magnetic gradients simultaneously in a cubic vapor cell containing ⁸⁷Rb vapor and ¹²⁹Xe gas. Compared with applying only first-order magnetic gradient compensation, the transverse relaxation time of ¹²⁹Xe is up to 4.3 times longer when applying both first-order and second-order compensating magnetic gradients, which indicates that the total magnetic gradient is greatly suppressed by the joint compensation in our experiment. The magnetic gradients induced by the polarized ⁸⁷Rb spins, the static magnetic field, and the residual magnetic field are also explored. As the main sources of internal magnetic inhomogeneities, these gradients are experimentally validated to have a sizable value. Furthermore, the total internal magnetic gradient in the system could be self-compensated when the directions of these internal gradient components are appropriately set. The experimental results in this paper are important for suppressing the magnetic gradients and optimizing the gradient compensation in nuclear magnetic resonance systems.

We study the impact of the heater-induced longitudinal field on the vector rubidium atomic magnetometers detecting low-frequency transverse magnetic field, which enlightens the parameter optimization of the applied heaters. Our results, both theoretical simulation and experimental measurement included, indicate that by carefully adjusting the frequency and amplitude, the impact of the heater-induced longitudinal field can be efficiently eliminated; meanwhile, the signal amplitude can be potentially enhanced. The numerical simulation agrees well with the experimental measurement in various cases, demonstrating that our proposed model is well suitable for predicting the actual performance of atomic magnetometers by taking the heater-induced longitudinal field into account.

Optically pumped alkali metal atoms currently provide a sensitive solution for magnetic microscopic measurements. As the most practicable plan, Faraday rotation of linearly polarized light is extensively used in spin polarization measurements of alkali metal atoms. In some cases, near-resonant Faraday rotation is applied to improve the sensitivity. However, the near-resonant linearly polarized probe light is elliptically polarized after passing through optically pumped alkali metal vapor. The ellipticity of transmitted near-resonant probe light is numerically calculated and experimentally measured. In addition, we also analyze the negative impact of elliptical polarization on Faraday rotation measurements. From our theoretical estimate and experimental results, the elliptical polarization forms an inevitable error in spin polarization measurements.

The transverse relaxation time τ₂ of the built-in alkali metal magnetometer is an important parameter to characterize its performance and the working state of the nuclear magnetic resonance (NMR) oscillator system. The method we propose to measure τ₂ is to apply the frequency-scanning magnetic fields along the x-axis and obtain τ₂ from the 3 dB bandwidth of the measured x-axis amplitude–frequency response curve of the magnetometer. We give a detailed and rigorous theoretical derivation by using the perturbation-iteration method to demonstrate the rationality of this measuring method. Then, both the actual experiment and the simulation experiment results are presented to verify the validity and good accuracy of the method. All the sampled frequencies in the scanning range should be far away from both the ¹²⁹Xe and ¹³¹Xe NMR frequencies, to avoid introducing any influence on the nuclear spins; therefore, this method can provide a non-interference way for measurement of τ₂, helping us monitor the NMR oscillator system’s state under normal operating conditions.