Deterministic domain selection of antiferromagnets via magnetic fields

TL;DR

The paper shows that deterministic antiferromagnetic domain selection by magnetic fields is enabled by a generic surface-exchange asymmetry in layered A-type AFMs, such as MnBi$_2$Te$_4$. Through statistical mechanics, committor analysis, and extensive Monte Carlo simulations, the authors demonstrate that reduced surface exchange combined with surface spins’ enhanced susceptibility biases the field-driven evolution toward a single domain, with the outcome dependent on the ramp direction. They connect this mechanism to MOKE measurements, reproducing hysteresis and the field-dependent sign of the Kerr response, and identify two characteristic fields ($H_z^{SF}$ and $H_z^{*}$) governing the transitions. The results generalize to other layered AFMs and offer a practical path to magnetic-field control of AFM domains, with implications for vdW materials and AFM-based logic applications.

Abstract

Antiferromagnets (AFMs) hold promise for applications in digital logic. However, switching AFM domains is challenging, as magnetic fields do not couple to the bulk antiferromagnetic order parameter. Here we show that magnetic-field-driven switching of AFM domains can in many cases be enabled by a generic reduction of magnetic exchange at surfaces. We use statistical mechanics and Monte Carlo simulations to demonstrate that an inequivalence in magnetic exchange between top and bottom surface moments, combined with the enhanced magnetic susceptibility of surface spins, can enable deterministic selection of antiferromagnetic domains depending on the magnetic-field ramping direction. We further show that this mechanism provides a natural interpretation for experimental observations of hysteresis in magneto-optical response of the van der Waals AFM $\mathrm{MnBi_2Te_4}$. Our findings highlight the critical role of surface spins in responses of antiferromagnets to magnetic fields. Furthermore, our results suggest that antiferromagnetic domain selection via purely magnetic means may be a more common and experimentally accessible phenomenon than previously assumed.

Figures (12)

• Figure 1: Field response of a periodic bulk A-type AFM as a function of magnetic field strength $H_z$ applied parallel to the easy axis. $H_z^{SF}$ and $H_z^{\mathrm{sat.}}$ denote the spin flop and the saturation field, respectively.
• Figure 2: Transition with decreasing applied field for an asymmetric AFM slab from spin-flop state (left) to one or more intermediate "bilayer" phases (center) to the most probable collinear AFM domain (right) as $H_z\rightarrow0$, reflecting a ramping-down protocol. Layers 1-5 have interlayer couplings $J^{\mathrm{AFM}}$, while the top coupling is $(1+x)J^{\mathrm{AFM}}$. (a) Transition pathway for $x<0$. (b) Pathway for $x>0$.
• Figure 3: Total transition probabilities of a six-layer slab for varying surface exchange asymmetry $x$ from any BL state to either domain "mod. surf $\uparrow$" or "mod. surf $\downarrow$". (a) evaluated at $+4~\mathrm{T}$ reflects a ramping-down field and (b) at $-4~\mathrm{T}$ represents ramping-up.
• Figure 4: (a) Ensemble-averaged Néel vector $\langle L_z\rangle$ as a function of applied field based on Monte Carlo (MC) simulations for surface exchange asymmetries $x=0$ and $x=-0.4$ with ramping-down and ramping-up fields. Cartoons show BL states at $\pm5\mathrm{T}$, just below $|H_z^{SF}|$, with central spins left out. (b) Simulated RCD measurement using the MC data from $x=-0.4$ in (a). The simulation is done for the photon energy of $2.3\text{eV}$, using the same dielectric model as in Ref. Sunko2025.
• Figure S1: Crystal structure of $\mathrm{MnBi_2Te_4}$ with the ground-state A-type antiferromagnetic ordering in the primitive rhombohedral setting (left) and in the hexagonal setting (right) which we use as the input cell for our Monte Carlo simulations. We model the $(001)$ hexagonal surface cut, reflecting the experimental surface crystal orientation in Ref. Sunko2025, by using vacuum boundary conditions along the $(001)$ direction and periodic in-plane boundary conditions. The picture of the hexagonal structure shows six $\mathrm{Mn}$ layers.