Thickness-dependent magnon spin transport in antiferromagnetic insulators: Crossover from quasi-three-dimensional to quasi-two-dimensional regimes

TL;DR

This work addresses thickness-dependent magnon spin transport in antiferromagnetic insulators by studying hematite across easy-axis and easy-plane phases using finite-temperature stochastic micromagnetic simulations in a nonlocal geometry with spin Hall injection and inverse spin Hall detection. The authors reveal a pronounced quasi-3D to quasi-2D crossover at a critical thickness around four to five layers ($L_z \approx 20$--$25$ nm), accompanied by a fourfold increase in the magnon diffusion length $L_D$ in the thin regime, attributed to changes in the magnon density of states (DOS) and the emergence of standing modes along the film thickness. The crossover is rationalized through analytical DOS comparisons between 2D and 3D magnons and by identifying a characteristic frequency at which the excited magnons transition between dimensional regimes, with the phenomenon observed in both easy-axis and easy-plane phases. These results highlight the critical role of dimensionality in magnon transport and suggest ultrathin antiferromagnetic insulators as promising components for next-generation spintronic devices with reduced dissipation.

Abstract

Motivated by the recent observation of giant room-temperature magnon spin conductivity in an ultrathin ferromagnetic insulator [X.-Y. Wei et al., Nat. Mater. 21, 1352 (2022)], we investigate thickness-dependent magnon spin transport in thin antiferromagnetic insulators (AFIs). We study the prototypical AFI hematite, known for its exceptionally low magnetic damping and two distinct magnetic phases: a low-temperature uniaxial easy-axis phase and a high-temperature biaxial easy-plane phase. Using stochastic micromagnetic simulations, we investigate thickness-dependent magnon spin transport across both magnetic phases. Our results uncover a crossover from quasi-three-dimensional to quasi-two-dimensional magnon spin transport at a critical thickness, determined by the frequency or energy of the excited magnons. Below this critical thickness, we observe a pronounced enhancement in the magnon diffusion length in both magnetic phases. This rise is attributed to a change in the effective magnon density of states, reflecting the reduced phase space available for scattering in the thinner, quasi-two-dimensional regime. Understanding and controlling long-distance magnon spin transport in AFIs is crucial for developing next-generation spintronic nanodevices, especially as materials approach the two-dimensional limit.

Figures (15)

• Figure 1: Setup for micromagnetic simulations of magnon spin transport in a nonlocal transverse geometry. Magnons are injected into the AFI from the central lead via interfacial torque generated by a spin Hall–induced spin accumulation. This increases the magnon spin chemical potential, driving diffusive transport toward the left and right leads along the $x$ direction. Through spin pumping and the inverse spin Hall effect, the magnon spin signal is detected at various distances. The spatial magnon distribution along the thickness show a clear difference between quasi-2D (top) and quasi-3D (bottom) magnon spin transport regimes.
• Figure 2: Distance-dependent magnon spin signal in the easy-axis magnetic phase of hematite for varying layer thicknesses. The magnon spin accumulation or magnon chemical potential $\mu$ is normalized by the magnon accumulation at the injector $\mu_0\equiv\mu(x=0)$. The thickness of each layer is 5nm and we set $V/L_z= 12µV\per nm$.
• Figure 3: Spin diffusion length for systems in the easy-axis (green squares) and easy-plane (orange triangles) magnetic phases of hematite for varying thicknesses, as obtained by fitting the exponential decay shown in \ref{['fig:uniaxial_distance-dependent_transport', 'fig:biaxial_distance-dependent_transport']}. There is a clear enhancement of the diffusion length at a thickness of four to five layers, corresponding to $20$-25nm, marking the transition from quasi-3D to quasi-2D transport. We set $V/L_z= 12µV\per nm$. The large error bar at the critical thickness of four layers suggests a coexistence of 2D and 3D behavior in the easy-axis phase, see \ref{['fig:diffusion_length_over_voltage_4layer']}.
• Figure 4: Measured frequencies of excited magnon modes during the transport simulations in systems in the easy-axis (green squares) and easy-plane (orange triangles) magnetic phase for varying thickness. The frequency is read out from spectra that are recorded during the transport experiments, see \ref{['sec:ReadOutOccupiedStates']} for more details. The colored background shows analytically calculated regions of the 2D (red area) versus 3D (blue area) regime. The gray line shows the critical frequency dividing the two transport regimes, see \ref{['eq:criticalE']}. The data points corresponding to the critical thickness (four layers) are not shown, as the occupied frequencies exhibit substantial variation at this transition point.
• Figure A1: Distance-dependent magnon spin signal in the easy-plane magnetic phase for varying layer thicknesses. The corresponding diffusion lengths are shown in \ref{['fig:diffusion_lengths_across_thicknesses']}.