The manipulation of spin textures using spin currents is a significant area of research in both fundamental physics and technological applications. A promising material for this purpose is the two-dimensional (2D) van der Waals ferromagnet Fe3GeTe2 (FGT), recently shown to exhibit Néel-type skyrmions. The origin of these chiral spin textures is believed to be a bulk Dzyaloshinskii-Moriya interaction (DMI) arising from broken inversion symmetry in the crystal structure, caused by variations in Fe occupancy of certain crystallographic sites. Previous studies on current-induced manipulation of FGT magnetization have often relied on indirect methods like anomalous Hall resistance measurements or were limited by low domain wall velocities. This research directly images and investigates current-induced domain wall motion (CIDWM) in FGT, aiming to achieve significantly higher velocities and explore the interplay of spin transfer torques (STTs) and spin-orbit torques (SOTs) in controlling domain wall movement. The development of efficient CIDWM is crucial for the realization of high-speed, high-density, and low-energy Racetrack memory devices, which rely on the ability to precisely control domain wall movement using electric currents.

Extensive research explores spintronics, focusing on converting electric current to spin current for manipulating magnetizations via STTs and SOTs. Racetrack memory devices, utilizing CIDWM, are a leading candidate for next-generation memory technology. 2D van der Waals materials, particularly FGT, offer an attractive platform due to their metallic nature, tunable Curie temperature, and strong perpendicular magnetic anisotropy. Recent studies have observed chiral magnetic nanostructures in FGT, suggesting the presence of DMI. While interfacial DMI was initially proposed, studies now confirm a bulk DMI due to broken inversion symmetry. Prior investigations of CIDWM in FGT have shown limited success, often using indirect methods or reporting low velocities. This paper builds on this prior work by employing direct imaging techniques and aiming for substantially improved velocity.

The researchers fabricated racetrack devices from pristine FGT flakes and FGT/heavy metal (Pt or W) heterostructures. Pristine FGT flakes were exfoliated and transferred onto a Si/SiO2 wafer. Standard e-beam lithography and magnetron sputtering were employed to create racetrack devices with Ti/Au electrodes. FGT/heavy metal heterostructures were made by depositing a 3 nm layer of Pt or W directly onto the FGT flakes using magnetron sputtering in an ultra-high vacuum system to ensure a clean interface. Magneto-optical Kerr (MOKE) microscopy, with a Nikon CFI S Plan Fluor ELWD 60XC objective lens, was used to image the domain walls at various temperatures (down to 20 K). CIDWM was studied by applying sequences of current pulses with varying current densities (up to 2.5 × 10¹¹ A m⁻²), pulse widths (5-80 ns), and temperatures. The domain wall velocity (v) was determined by tracking the domain wall position over time. The dependence of v on current density, pulse width, and temperature was systematically investigated. Differential Kerr microscopy was used to image magnetization switching in heterostructures. A 1D CIDWM model, incorporating both STT and SOT, was used to model the experimental data, particularly the dependence of velocity on longitudinal magnetic field.

In pristine FGT flakes, STT-driven CIDWM was observed, with the domain wall moving opposite to the current direction. A maximum domain wall velocity of 5.68 m s⁻¹ was achieved at 20 K and a current density of 2.41 × 10¹¹ A m⁻². This velocity is an order of magnitude higher than previously reported values. The velocity was found to depend on both current density and pulse width, with shorter pulses yielding higher velocities due to Joule heating effects. As pulse width increases, Joule heating becomes significant leading to a temperature increase which decreases the saturation magnetization (M) and limits the velocity. The velocity was also temperature-dependent, decreasing monotonically with increasing temperature until it reached zero at the Curie temperature (~150 K). In FGT/Pt heterostructures, the direction of domain wall motion changed with increasing current density, reflecting the competition between STT and SOT. At low current densities, the motion was opposite to the current direction (STT-dominated), while at high current densities, the motion was parallel to the current direction (SOT-dominated). In FGT/W heterostructures, however, STT and SOT acted cooperatively, resulting in domain wall motion similar to that in pristine FGT, but with a distinct dependence on the longitudinal magnetic field. The longitudinal magnetic field dependence of the domain wall velocity for both pristine FGT and FGT/W heterostructures was consistent with the presence of a Néel-type domain wall due to the bulk DMI.

The significantly higher domain wall velocity achieved in this study demonstrates the potential of FGT for high-speed spintronic devices. The observed dependence of velocity on current density, pulse width, and temperature highlights the critical role of Joule heating in limiting performance. The different behaviors of FGT/Pt and FGT/W heterostructures emphasize the importance of considering the interplay between STT and SOT in designing spintronic devices. The observation of a Néel-type domain wall, evidenced by the non-zero field peak in the velocity-field curves, provides strong support for the presence of a bulk DMI. This contrasts with the interfacial DMI seen in some other systems. The bulk nature of the DMI in FGT is tied to Fe vacancies and self-intercalated Fe atoms. The results demonstrate the possibility of tuning the domain wall motion by carefully selecting heavy metal layers, highlighting the significance of material choices in enhancing device performance.

This work successfully demonstrates high-velocity current-induced domain wall motion in the 2D van der Waals ferromagnet FGT, surpassing previous reports by an order of magnitude. The study clearly shows the impact of Joule heating, the interplay between STT and SOT, and the presence of bulk DMI in controlling domain wall dynamics. The differing behaviors observed with Pt and W highlight the significant role of material selection in spintronic device design. Future research could explore higher Curie temperature 2D materials or novel strategies to mitigate Joule heating effects to further improve domain wall velocity.

The relatively low Curie temperature of FGT limits the maximum achievable current density and therefore the maximum domain wall velocity. Joule heating is a significant factor affecting the domain wall motion and this is an intrinsic limitation of the material in its current form. The 1D model used to fit the experimental data is a simplification, neglecting potential complexities of the domain wall structure and interactions.