Wide-Field Dynamic Magnetic Microscopy Using Double-Double Quantum Driving of a Diamond Defect Ensemble

Wide-field magnetometry can be realized by imaging the optically detected magnetic resonance of diamond nitrogen-vacancy ($\mathrm{N}$-$V$) center ensembles. However, $\mathrm{N}$-$V$ ensemble inhomogeneities significantly limit the magnetic field sensitivity of these measurements. We demonstrate a double-double quantum (DDQ) driving technique to facilitate wide-field magnetic imaging of dynamic magnetic fields at a micron scale. DDQ imaging employs four-tone radio-frequency pulses to suppress inhomogeneity-induced variations of the $\mathrm{N}$-$V$ resonance response. As a proof of principle, we use the DDQ technique to image the dc magnetic field produced by individual magnetic nanoparticles tethered by single DNA molecules to a diamond-sensor surface. This demonstrates the efficacy of the diamond $\mathrm{N}$-$V$ ensemble system in high-frame-rate magnetic microscopy, as well as single-molecule biophysics applications.

Figure 1

Wide-field pulsed magnetic imaging using an $\mathrm{N}$-$V$ ensemble. (a) The $\mathrm{N}$-$V$ electronic-energy-level diagram, showing the ground, excited spin states ($|m_s=0⟩$,$|m_s=\pm 1⟩$), singlet states, optical excitation (green arrow), emitted photoluminescence (red arrows), and spin-selective nonradiative intersystem crossing (gray arrows). (b) A schematic showing 50-nm ferromagnetic nanoparticles adhered to the diamond surface. The magnetic moments of the particles are oriented randomly. The 150-nm $\mathrm{N}$-$V$ layer (pink) is fabricated on top of the diamond substrate (gray) and a single $\mathrm{N}$-$V$ pointed along the (111) orientation is shown (red). (c) The $\mathrm{N}$-$V$ ground-state energy-level diagram, showing the zero-field splitting ($D$), Zeeman splitting of the $|m_s=\pm 1⟩$ states ($2{\gamma }_{\mathrm{N-}V}{\hat{z}}_{\mathrm{N-}V}\cdot \overrightarrow{B}$), and ${}^{15}\mathrm{N}$–$\mathrm{N}$-$V$ hyperfine splitting (3.05 MHz). The rf excitation (orange and blue arrows) rotates the $\mathrm{N}$-$V$ spin between the $|m_s=0⟩$ and $|m_s=\pm 1⟩$ states. Two-tone rf excitation is simultaneously driven over the two ${}^{15}\mathrm{N}$-$\mathrm{N}$–$V$ hyperfine transitions to produce a single combined resonance for each $\mathrm{N}$-$V$ electron spin state ($|m_s=\pm 1⟩$). (d) Laser and multitone rf $\pi$ pulses are repeated throughout the camera exposure to facilitate wide-field imaging. (e) A Lorentzian-shaped reduction in $\mathrm{N}$-$V$ PL is observed when a rf scan is performed through each spin-transition frequency. The resonances are Zeeman split by the external magnetic field. The outer (inner) inflection points $f_1,f_4$ ($f_2,f_3$) are denoted by black squares (circles). For the resonance curves shown, the FWHM line width $\delta \nu =300\mathrm{kHz}$ and fractional optical contrast $C=0.03$, with optical $\mathrm{pulse}=500$ ns, rf $\mathrm{pulse}=3500$ ns, photon collection $\mathrm{rate}=1.1\times {10}^7$ Hz from a $1\mu {\mathrm{m}}^3$ pixel ($0.15\mu {\mathrm{m}}^3$ voxel), and integration time per data point = 144 ms.

Figure 2

Multiple mechanisms lead to changes in $\mathrm{N}$-$V$ ODMR on a micron scale when imaging static magnetic dipole and strain fields. (a) The resonance-frequency $\nu$ shifts due to the magnetic field and the crystal strain. (b) The FWHM line width $\delta \nu$ varies due to the gradients of the magnetic and strain fields that inhomogeneously broaden the $\mathrm{N}$-$V$ ODMR in each pixel. (c) The optical contrast $C$ also varies due to inhomogeneous broadening of the $\mathrm{N}$-$V$ ODMR.

Figure 3

Competing magnetic imaging modalities. (a) The “true” static magnetic field projection map generated with the frequency-scanning technique outlined in Sec. 3 (acquisition time 12 s). (b) SQ difference imaging (2.4 s). The signal measured with the SQ DI modality is a convolution of the magnetic and strain fields, which are impossible to separate with a single measurement. (c) DQ difference imaging (2.4 s). While the DQ DI modality has reduced the impact from the fields that homogeneously shift the $\mathrm{N}$-$V$ centers, the DQ signal is more sensitive to the local contrast and line-width variations of the $\mathrm{N}$-$V$ sensing curves. (d) DDQ difference imaging (2.4 s). An inspection of the competing dynamic imaging schemes (b)–(d) reveals that both the SQ (b) and DQ (c) schemes are significantly compromised by spurious contrast caused by strain gradients and curve-shape variation, respectively, while the DDQ scheme (d) faithfully approximates the “true” magnetic field projection (a) with a decreased acquisition time.

Figure 4

DDQ imaging of the reorientation of a DNA-tethered magnetic nanoparticle under applied flow. In each panel, the observed DDQ image is compared with a fitted DDQ image (inset) to estimate the magnetic nanoparticle dipole orientation ($\theta$, $\varphi$), where the $\mathrm{N}$-$V$ ensemble symmetry axis is (${54.735}^{\circ }$, $0^{\circ }$). For all DDQ DI in this figure, a Gaussian smoothing filter with $\sigma =533\mathrm{nm}$ is applied. (a) A time-averaged DDQ DI (8 s) showing the initial magnetic nanoparticle orientation before flow. (b) Representative DDQ frames (64 ms of exposure) showing nanoparticle reorientation in response to an applied flow. (c) A time-averaged DDQ DI (8 s) showing the final magnetic nanoparticle orientation with applied flow.