Current-induced switching of thin film $\alpha \text{−}{\mathrm{Fe}}_2{\mathrm{O}}_3$ devices imaged using a scanning single-spin microscope

Electrical switching of Néel order in an antiferromagnetic insulator is desirable as a basis for memory applications. Unlike electrically driven switching of ferromagnetic order via spin-orbit torques, electrical switching of antiferromagnetic order remains poorly understood. Here we investigate the low-field magnetic properties of 30-nm-thick, $c$-axis-oriented $\alpha \text{−}{\mathrm{Fe}}_2{\mathrm{O}}_3$ Hall devices using a diamond nitrogen-vacancy center scanning microscope. Using the canted moment of $\alpha \text{−}{\mathrm{Fe}}_2{\mathrm{O}}_3$ as a magnetic handle on its Néel vector, we apply a saturating in-plane magnetic field to create a known initial state before letting the state relax in low field for magnetic imaging. We repeat this procedure for different in-plane orientations of the initialization field. We find that the magnetic field images are characterized by stronger magnetic textures for fields along $\left[\overline{1}\overline{1}20\right]$ and $\left[11\overline{2}0\right]$, suggesting that despite the expected 3-fold magnetocrystalline anisotropy, our $\alpha \text{−}{\mathrm{Fe}}_2{\mathrm{O}}_3$ thin films have an overall in-plane uniaxial anisotropy. We also study current-induced switching of the magnetic order in $\alpha \text{−}{\mathrm{Fe}}_2{\mathrm{O}}_3$. We find that the fraction of the device that switches depends on the current pulse duration, amplitude, and direction relative to the initialization field.

Figure 1

Schematic of the scanning NV center setup. We use a commercial diamond probe (QZabre LLC) with a single NV center implanted approximately 10 nm below the tip surface. Scans are obtained with a probe-to-sample separation of 100 nm. A microscope objective is used to focus the green excitation beam and collect the red photoluminescence from the NV center. Insets show optical images of device A (top) and B (bottom). The Pt-capped $\alpha \text{−}{\mathrm{Fe}}_2{\mathrm{O}}_3$ appears bright, and the bare ${\mathrm{Al}}_2{\mathrm{O}}_3$ substrate is dark. The red boxes indicate the $10\mu \mathrm{m}\times 10\mu \mathrm{m}$ scan area.

Figure 2

(a) Examples of magnetic images from device A after it was initialized with a 1 T in-plane magnetic field. The black dashed line indicates a strong domain feature that appears in multiple images. The in-plane angles of the initialization field with respect to the $y$ axis of the image are given in the bottom left corners. Autocorrelation images are calculated from the red-boxed regions. (b) Magnetic field standard deviations of the post-initialization images as a function of the initialization field angle. Peaks are visible for $0^{\circ }$ and ${180}^{\circ }$ corresponding to the $y$ and $-y$ direction. Insets are the magnetic images for $0^{\circ }$ and ${90}^{\circ }$ initialization field replotted in a diverging color map. The ${90}^{\circ }$ image is more homogeneous than the $0^{\circ }$ image. (c) The fitted autocorrelation peak angle as a function of the initialization field angle. The linear fit has a slope of 0.95 and an offset of $23.8{}^{\circ }$, with an $R^2$ value of 0.96.

Figure 3

Magnetic field images acquired using the dual-iso-B method. (1) Image before applying any current pulse. (2)–(6) Difference images obtained by subtracting the scan in (1) from the scan taken after passing 100 $\mu \mathrm{s}$ current pulses with an amplitude of (2) 3 mA, (3) 7 mA, (4) 13 mA, (5) 15 mA, and (6) 16 mA. The green arrows indicate the direction of the initialization field and the red arrows indicate the current pulse direction.