Optimization of performance for an EIT-based vector magnetometer.

In this work we present the advance on the experimental realization of a vector atomic magnetometer based on electromagnetically induced transparency (EIT) resonances. Magnetic fields modify the separation between EIT spectrum peaks, independently of their orientation, allowing for magnetic field magnitude measurements. However, the resonance amplitudes depend on the field orientation relative to the light wave vector and polarization direction, making this system capable of vector measurements. Here we report our evaluation of the short-term stability and sensitivity of our prototype using a magnetically isolated 100mm³ vapor cell of hot ⁸⁷Rb atoms. We show an achievable scalar stability below 10 pT/rtHz in the 1Hz-100Hz bandwidth. An unsupervised machine learning technique is used to analyze the amplitude of the EIT resonances and determine the magnetic field orientation with a sensitivity better than 1°.