Stabilizing Magnetic Bubble Domains in Epitaxial 2D Magnet/Topological Insulator Heterostructures through Interfacial Interactions

Abstract

Epitaxial heterostructures of two-dimensional van der Waals magnets and topological insulators offer a powerful platform for probing interfacial spin interactions that govern magnetic textures in low-dimensional quantum systems, while simultaneously enabling highly efficient, atomically thin spin-orbit-torque memory and computing architectures. Despite this promise, the fundamental role of these interfacial interactions in determining magnetic domain-phase stability remain largely uncharted. Here, we perform scanning transmission X-ray microscopy to image nanoscale magnetic textures in epitaxial Fe3GeTe2 Bi2Te3 heterostructures, enabled by a thermal-release-tape dry transfer process onto X-ray transparent silicon-rich nitride membranes. Under zero-field-cooled conditions, we observe robust bubble domain phases from 75 to 165 K, and across different number of folds of the multilayer Fe3GeTe2 Bi2Te3 heterostructures. This is in stark contrast with exfoliated single-crystal Fe3GeTe2 flakes, where ZFC stripe domains are observed for flakes thicker than 20 nm and no domains have been reported for thin flakes less than 15 nm. First-principles calculations and micromagnetic simulations reveal that interfacial coupling to Bi2Te3 modifies the magnetic anisotropy and introduces interfacial Dzyaloshinskii-Moriya interaction, shifting the magnetic phase space towards bubble-domain stabilization without field-cooling. Together, our results offer a new strategy for phase-selective control of magnetic domains through interfacial engineering.

Figures (4)

• Figure 1: MBE-grown epitaxial Fe$_3$GeTe$_2$/Bi$_2$Te$_3$ heterostructures. (a) Schematic of atomic structure of the van der Waals magnet/topological insulator heterostructure. consisting of five quintuple layers (QL) of Bi$_2$Te$_3$ and eight van der Waals layers of Fe$_3$GeTe$_2$ (partial layers shown for clarity). (b) Optical micrograph of a large-area Fe$_3$GeTe$_2$/Bi$_2$Te$_3$ film transferred onto an X-ray--transparent silicon-rich nitride (SiRN) membrane using a thermal-release-tape (TRT) process (scale bar: 100 $\mu$m). (c) Optical magnetic circular dichroism (MCD) hysteresis loops measured at different temperatures, demonstrating strong perpendicular magnetic anisotropy and a Curie temperature of approximately 190 K. (d) Schematic illustration of the full-film TRT transfer process, enabling the transfer of epitaxial Fe$_3$GeTe$_2$/Bi$_2$Te$_3$ heterostructures from sapphire substrates onto SiRN membranes for XMCD-STXM measurements.
• Figure 2: X-ray magnetic circular dichroism images of bubble domains stabilized in an epitaxial Fe$_3$GeTe$_2$/Bi$_2$Te$_3$ heterostructure transferred onto a SiRN membrane. (a) Schematic of scanning transmission X-ray magnetic circular dichroism measurements of nanoscale bubble domains. All XMCD images were acquired over an 8 $\mu$m $\times$ 8 $\mu$m raster scan with a step size of 40 nm. (b) X-ray transmission averaged between right and left circularly polarized X-rays, of regions with different number of heterostructure repeats [FGT/ Bi$_2$Te$_3$]$_n$ (n = 1 to 5) from folding during the transfer process. (c-d) XMCD images of the zero-field-cooled (ZFC) state at 75 K and 120 K. e) XMCD images at different magnetic fields taken sequentially during the out-of-plane field hysteresis sweep directions: (i) negative, (ii) positive, (iii) negative following the zero-field cooling to 165 K.
• Figure 3: Characterization of magnetization and magnetic domains based on XMCD-STXM contrast imaging. The spatially averaged XMCD grayscale value is normalized between 0 and 1 and used as the effective magnetization. (a) Magnetic field sweep hysteresis for regions with [FGT/Bi$_2$Te$_3$]$_n$ heterostructure repeats from n = 1 to 5 at 165 K. Hysteresis loops of each region is scaled and artificially vertically offset for clarity. (b) Saturated XMCD switching amplitudes as a function of repeats, exhibiting linear scaling with effective magnetic thickness. (c) Representative image-processing pipeline for domain identification, including median blurring, a non-local means denoising, and threshold-based segmentation (S.I.VII; example shown for a one-fold region at 75 K). (d) Zero-field domain-size distributions extracted from the processed XMCD for fold 1-3 at 75, 120, and 160 K; box plots indicating the 5th, 25th, 50th, 75th, and 95th percentiles. (e) Thresholded images showing bubble domains in fold 1-3: (i-iii) zero-field states at 75, 120, and 165 K, and (iv-v) field-driven states at 165 K under 50 mT and –50 mT, respectively.
• Figure 4: First-principles calculations and micromagnetic simulations of bubble-domain stabilization. (a) Slab supercell used in the DFT calculations together with the corresponding atomic-layer--resolved magnetic moment (MM, red) and Dzyaloshinskii--Moriya interaction (DMI, blue), aligned along the out-of-plane direction. Bi, Te, Fe, and Ge atoms are shown in magenta, brass, brown, and purple-blue, respectively; the topmost Te layer of Bi$_2$Te$_3$ defines the zero reference. (b, c) Projected band structure of the heterostructure and a zoomed-in view near the Fermi level, showing Fe$_3$GeTe$_2$-derived (blue) and Bi$_2$Te$_3$-derived (red) states. (d) Zero-field domain morphologies across micromagnetic parameter space with DMI = 0.16 mJ/m$^2$, exchange stiffness (Aex) = 0.7–1.5 pJ/m, and uniaxial anisotropy (Ku) = 50–150 kJ/m$^3$ (i) Domain classifications: smooth stripe (green), bubble (peach), and rigid stripe (blue). (ii) Corresponding simulated domain snapshots. (e) Field-induced switching simulations for Aex = 0.9 pJ/m, Ku = 75 kJ/m$^3$, and Dind = 0.10 (purple), 0.16 (teal), and 0.22 (orange) mJ/m$^2$. (i) Evolution of the standard deviation of out-of-plane magnetization ($\sigma(M_z)$) as a function of out-of-plane (OOP) magnetic field ($B_z=0$--$200$ mT). The inset shows $M_z$ versus $B_z$. Filled symbols mark the field of maximum nucleation at $B_z$ = 100, 40, and 20 mT for DMI = 0.10, 0.16, and 0.22 mJ/m$^2$, respectively (see details in Fig. S15). (ii) Corresponding domain snapshots at maximum nucleation.