Fig. 4: Noise measurements.

a Time traces of the Hall voltage (offset for clarity) and b Hall voltage noise spectral density S_V^1/2 for device G1 at fixed bias current and 4.2 K. The three curves correspond to the gate voltages marked at the top of the upper panel of (d). Dashed lines in b follow the expected dependence of random telegraph noise (RTN) at high frequency (f⁻¹) and 1/f noise (f^−1/2). c Comparison of S_V^1/2 spectra at different bias currents. At each bias current, we set V_g such that R_H ≈ 7.8 kΩ T⁻¹, corresponding to n ≈ 8 × 10¹⁰ cm⁻². d IR_H and R_offset = V_H(B = 0)/I for 20 μA bias current. e S_V^1/2 and magnetic field detection limit S_B^1/2 at 1 kHz. f Bias current dependence of the minimum S_B^1/2 at 1 kHz. In panels d–f, error bars are determined considering the standard error of the linear fit for R_H and the standard deviation of S_V^1/2 in a window of width 200 Hz centered at 1 kHz.