Real Time Magnetic Field Sensing and Imaging Using a Single Spin in Diamond

The Zeeman splitting of a localized single spin can be used to construct a highly sensitive magnetometer offering almost atomic spatial resolution. While sub-$\mu \mathrm{T}$ sensitivity can be obtained in principle using pulsed techniques and long measurement times, a fast and easy method without laborious data postprocessing is desirable for a scanning-probe approach with high spatial resolution. In order to measure the resonance frequency in real time, we applied a field-frequency lock to the optically detected magnetic resonance signal of a single electron spin in a nanodiamond. We achieved a sampling rate of up to 100 readings per sec with a sensitivity of $6\mu \mathrm{T}/\sqrt{\mathrm{Hz}}$. Images of the field distribution around a magnetic wire were acquired with $\sim 30\mu \mathrm{T}$ resolution and 4096 submicron sized pixels in 10 min. The response of several spins was used to reconstruct the field orientation.

Figure 1

(a) A single NV spin responds to a frequency-modulated microwave close to resonance with a luminescence modulation proportional to the frequency offset. The luminescence is recorded via lock-in principle and used for frequency correction and local magnetic field calculation. (a) A $100\mu \mathrm{m}$ magnetized steel wire is scanned over a nanodiamond containing a NV center. The microwave is generated by an antenna $\sim 100\mu \mathrm{m}$ apart.

Figure 2

ODMR spectrum of a NV center with $E\ne 0$. (a) FM detected signal, $r_{\mathrm{mod}}=100\mathrm{kHz}$, ${\nu }_{\mathrm{mod}}=2.5\mathrm{MHz}$ (b) integral of (a) (lower trace), CW signal (upper trace) shown for comparison. The sampling time was 100 ms per point for all measurements.

Figure 3

Amplitude dependence of the modulation detected signal on the modulation rate, showing a response time ${\tau }_{3\mathrm{dB}}=1/2r_{\mathrm{mod},3\mathrm{dB}}=1.5\mu \mathrm{s}$ ($P_{\mathrm{RF}}=1\mathrm{W}$, $P_{\mathrm{Laser}}=50\mu \mathrm{W}$ at objective entry). The inset shows the mean error in the resonance frequency derived from the signal after averaging multiple modulation cycles. The results are in excellent agreement with a photon shot noise of $170\mathrm{kHz}/\sqrt{\mathrm{Hz}}$.

Figure 4

Spin transition frequencies vs external field strength measured by tracking the resonance positions during a magnetic field sweep (sampling rate 25 Hz).

Figure 5

Realtime measurement of an ac magnetic field. $f_{\mathrm{ac}}$ was 8.3 Hz corresponding to a maximum magnetic field gradient of $\sim 10\mathrm{mT}/\mathrm{s}$ (sampling rate 100 Hz).

Figure 6

Magnetic field images taken with single NV centers. (a) NV axis normal to $xy$-plane ($\theta =0°$), (b) NV axis projection on the $xy$ plane along $x$, pointing slightly downwards ($\varphi =0°$, $\theta =100°$), (c) $\varphi =-60°$ (clockwise), $\theta =45°$, (d)–(f) simulation of (a)–(c) assuming a simple dipole model [Eq. (6)], no $B_x{S}_x$ components have been taken into account, (g) simulation of the normal magnetic field showing the scan areas, the inset shows an in-plane field component.