Anomalous Magneto-Optical Effects in an Antiferromagnet–Topological-Insulator Heterostructure

We show that large magneto-optic Kerr effects (MOKEs) emerge when an antiferromagnet (AFM) is proximately coupled to a topological insulator (TI) film—where neither the perfect collinear Néel ordered single-domain AFM nor the unmagnetized TI individually shows any MOKE. Because of the lack of macroscopic magnetization, the AFM only couples to the spin of one of the TI’s surfaces breaking time-reversal and inversion symmetry—which leads to a small microdegree MOKE signal. This small MOKE can be easily enhanced by 5 orders of magnitude, via cavity resonance, by optimizing the AFM and TI film thicknesses on the substrate. For slightly off-resonant structures, a $6^{\circ }$ Kerr rotation can be electrically switched on by varying the Fermi energy. This requires less than 20 meV, which is encouraging for low-power spintronics and magneto-optic devices. We further show that this simple structure is easily resilient to 5% material growth error.

Figure 1

MOKE from a thin-film collinear antiferromagnet on top of a thin-film topological insulator. We specifically consider a $\mathrm{Ni}\mathrm{O}$ film on a ${\mathrm{Bi}}_2{\mathrm{Se}}_3$ thin film grown on a ${\mathrm{Si}\mathrm{O}}_2$ substrate.

Figure 2

(a) Band structure and (b) density of states (DOS) for TI only ($J_H=0$, dashed blue line) and for TI/AFM ($J_H=40$ meV, solid red line). The inset in (a) shows the higher energy gap and the Rashba-type dispersion that occur with $J_H=40$ meV. (c) DOS as a function of $J_H$.

Figure 3

Real and imaginary parts of the (a) diagonal and (b) off-diagonal optical conductivity for the TI alone and a coupled TI/AFM. Note that (a),(b) share the same legend. (c) Kerr rotation and ellipticity and (d) reflectivities for TI alone and for a single AFM/TI interface as denoted in the legend.

Figure 4

The Kerr rotation (${\theta }_k$) phase diagram shown as a function of (a) $d_N$ and $\omega$ for $d_B=10$ nm, (b) $d_N$ and $\omega$ for $d_B=40$ nm, (c) $d_N$ and $\omega$ for $d_B=1.17\mu \mathrm{m}$, and (d) $d_N$ and $d_B$ at $\omega =125$ meV.

Figure 5

The Kerr rotation angle as a function of Fermi energy $E_f$ and optical frequency $\omega$ for (a) $d_B=10$ nm and $d_N=10$ nm and (b) $d_B=1.16\mu \mathrm{m}$ and $d_N=100$ nm. The insets are enlargements to highlight the sudden switching behavior as a function of $E_f$.

Figure 6

(a) Kerr rotation and ellipticity and (b) the reflectivity spectra for $d_N=d_B=10$ nm. (c) MOKE spectra and (d) reflectivity spectra for $d_N=1162$ nm and $d_B=37$ nm. The inset of (c) shows an expanded view of the smaller, higher-frequency resonances. The left (right) $y$ axis of all the panels is for $E_f=0$ ($E_f\ne 0$).

Figure 7

(left-hand axis) The maximum Kerr rotation (${\theta }_k^{\max }$) as a function of Fermi energy ($E_f$) for structures with no error and four combinations of $\pm 5$% error in the layer thickness. This thickness error is denoted by $d_{N(B)}^{\pm }=(1\pm 0.05)\times d_{N(B)}^0$, where $\{d_N^0,d_B^0\}=\{1162,37\}$ nm. (right-hand axis) The value of $\omega$ at which ${\theta }_k^{\max }$ occurs as a function of $E_f$.