Electrical Switching of Tristate Antiferromagnetic Néel Order in $\alpha \text{−}{\mathrm{Fe}}_2{\mathrm{O}}_3$ Epitaxial Films

We demonstrate nondecaying, steplike electrical switching of tristate Néel order in $\mathrm{Pt}/\alpha \text{−}{\mathrm{Fe}}_2{\mathrm{O}}_3$ bilayers detected by the spin-Hall induced anomalous Hall effect. The as-grown $\mathrm{Pt}/\alpha \text{−}{\mathrm{Fe}}_2{\mathrm{O}}_3$ bilayers exhibit sawtooth switching behavior generated by current pulses. After annealing by a high pulse current, the Hall signals reveal single-pulse saturated, nondecaying, steplike switching. Together with control experiments, we show that the sawtooth switching is due to an artifact of Pt while the actual spin-orbit torque induced antiferromagnetic switching is steplike. Our Monte Carlo simulations explain the switching behavior of $\alpha \text{−}{\mathrm{Fe}}_2{\mathrm{O}}_3$ Néel order among three in-plane easy axes.

Figure 1

(a) Schematic of the $\alpha \text{−}{\mathrm{Fe}}_2{\mathrm{O}}_3$ hexagonal lattice with FM-aligned Fe moment in the $ab$ plane and AFM coupling between adjacent $ab$ planes (oxygen atoms not shown). (b) $2\theta /\omega$ XRD scan of a 30 nm $\alpha \text{−}{\mathrm{Fe}}_2{\mathrm{O}}_3$ epitaxial film on ${\mathrm{Al}}_2{\mathrm{O}}_3(001)$. The inset shows an enlarged region around the $\alpha \text{−}{\mathrm{Fe}}_2{\mathrm{O}}_3(006)$ peak. (c) STEM image of a $\mathrm{Pt}(2\mathrm{nm})/\alpha \text{−}{\mathrm{Fe}}_2{\mathrm{O}}_3(30\mathrm{nm})$ bilayer.

Figure 2

(a) The $ab$ plane of an $\alpha \text{−}{\mathrm{Fe}}_2{\mathrm{O}}_3$ lattice with three in-plane easy axes, [210], [120], and $[1\overline{1}0]$ labeled as $\mathbf{E}\mathbf{1}$, $\mathbf{E}\mathbf{2}$, and $\mathbf{E}\mathbf{3}$, resulting in a triaxial anisotropy, where the double arrows represent the AFM spins. (b) Optical microscopy image and (c) schematic of an eight-leg Hall cross of a $\mathrm{Pt}(2\mathrm{nm})/\alpha \text{−}{\mathrm{Fe}}_2{\mathrm{O}}_3(30\mathrm{nm})$ bilayer, where $\alpha$ is the angle between an in-plane field and the $\mathbf{E}\mathbf{2}$ direction. (d) A sequential pulse current of $I_p=9\mathrm{mA}$ is applied along one of the three easy axes (10 pulses for each segment) at 300 K and a reversible control of the tristate Hall resistance is detected by applying a 0.1 mA sensing current along $\mathbf{E}\mathbf{2}$. (e) In-plane $\alpha$ dependence of $\mathrm{\Delta }{R}_{xy}$ at $H=0.1$, 1, and 3 T, where $\mathrm{\Delta }{R}_{xy}$ saturates at $H\ge 1T$. The gray and purple solid curves are $\sin 2\alpha$ fits.

Figure 3

Evolution of $\mathrm{\Delta }{R}_{xy}$ when the pulse current is switched between E3 and E1 (10 pulses each) under (a) 0 and (b) 3 T in-plane field applied perpendicular to $\mathbf{E}\mathbf{2}$ for a $\mathrm{Pt}(2\mathrm{nm})/\alpha \text{−}{\mathrm{Fe}}_2{\mathrm{O}}_3(30\mathrm{nm})$ bilayer. Insets: semilog plots of $\mathrm{\Delta }{R}_{xy}$ vs ${\mathit{I}}_p$. The red line in inset (b) is an exponential fit, $y=(1.38\times {10}^{-11})e^{1.44x}$, and the red curve in inset (a) is given by $y=(1.38\times {10}^{-11})e^{1.44x}+(-0.0183+0.00243x)$, which is the sum of the exponential fit in inset (a) here and the linear fit in Fig. 4. Comparison of $\mathrm{\Delta }{R}_{xy}$ at 0 and 3 T with (c) $I_p=16$ and (d) $I_p=12\mathrm{mA}$ for a fresh sample. Comparison of $\mathrm{\Delta }{R}_{xy}$ for a fresh sample and the same sample after 18 mA annealing at (e) $I_p=16$ in a 3 T in-plane field ($\mathit{H}\perp \mathbf{E}\mathbf{2}$), (f) $I_p=12$ at 3 T, (g) $I_p=16$ at 0 T, and (h) $I_p=12\mathrm{mA}$ at 0 T.

Figure 4

(a) Pulse current dependence of $\mathrm{\Delta }{R}_{xy}$ for a $\mathrm{Pt}(2\mathrm{nm})/\alpha \text{−}{\mathrm{Fe}}_2{\mathrm{O}}_3(30\mathrm{nm})$ bilayer when ${\mathit{I}}_p$ is switched between $\mathbf{E}\mathbf{3}$ and $\mathbf{E}\mathbf{1}$ (10 pulses each) measured at 300 K. (b) Temperature dependence of $\mathrm{\Delta }{R}_{xy}$ (between $\mathbf{E}\mathbf{3}$ and $\mathbf{E}\mathbf{1}$) at $I_p=9\mathrm{mA}$. All measurements here are taken on a sample after 18 mA annealing. (c) $\mathrm{\Delta }{R}_{xy}$ vs ${\mathit{I}}_p$ from (a), showing a linear dependence (red fitting line: $y=-0.0183+0.00243x$). (d) Semilog plot of $\mathrm{\Delta }{R}_{xy}$ vs $T$ for $I_p=9\mathrm{mA}$ from (b), indicating an exponential dependence. (e) In-plane field dependence of $\mathrm{\Delta }{R}_{xy}$ with $\mathit{H}\perp \mathbf{E}\mathbf{3}$ [$\alpha =30°$, see Fig. 2], which tends to align $\mathit{n}\parallel \mathbf{E}\mathbf{3}$. The field is ramped from 0 to 1 T (green), then back to 0 T (red), which corresponds to a first-quadrant full hysteresis loop. In a separate scan, $H$ is ramped from 0 to 0.1 T (green), then back to 0 T (blue), corresponding to a minor hysteresis loop. (f) Monte Carlo simulations of the full and minor hysteresis loops of the component of $\mathit{n}$ along $\mathbf{E}\mathbf{3}$ ($n_{\mathbf{\text{E3}}}$) as a function of the effective magnetic field due to SOT generated by a pulse current ${\mathit{I}}_p\parallel \mathbf{E}\mathbf{3}$, which agrees with the experimental data in (e).