Berry curvature-induced local spin polarisation in gated graphene/WTe2 heterostructures

Abstract

Experimental control of local spin-charge interconversion is of primary interest for spintronics. Van der Waals (vdW) heterostructures combining graphene with a strongly spin-orbit coupled two-dimensional (2D) material enable such functionality by design. Electric spin valve experiments have thus far provided global information on such devices, while leaving the local interplay between symmetry breaking, charge flow across the heterointerface and aspects of topology unexplored. Here, we probe the gate-tunable local spin polarisation in current-driven graphene/WTe₂ heterostructures through magneto-optical Kerr microscopy. Even for a nominal in-plane transport, substantial out-of-plane spin accumulation is induced by a corresponding out-of-plane current flow. We present a theoretical model which fully explains the gate- and bias-dependent onset and spatial distribution of the intense Kerr signal as a result of a non-linear anomalous Hall effect in the heterostructure, which is enabled by its reduced point group symmetry. Our findings unravel the potential of 2D heterostructure engineering for harnessing topological phenomena for spintronics, and constitute an important step toward nanoscale, electrical spin control.

Fig. 1. Optical KR microscopy and graphene transfer curves.

a Optical microscope image of a heterostructure comprised of graphene (black dashed line), WTe₂ (blue dashed line) and hBN capping on a back-gated Si/SiO₂ substrate. The WTe₂ crystallographic axes are indicated. The overlay shows the current-induced KR signal at the junction using a colour code as in Fig. 3, for a current applied between the contacts labelled 1 and 3. Scale bar, 5 μm. The probed area is marked by the shaded rectangle. b An oscillating current j₁₃ at frequency ω is applied along the graphene stripe. At the heterojunction, the current-induced out-of-plane spin component (red and blue arrows) is locally read-out by a polar KR measurement at frequency ω. The effect of Joule heating is detectable at twice the oscillation frequency 2ω. c Two-terminal resistance $R_{13,13}^{\omega }$ of the graphene stripe measured between contacts 1 and 3 of the device in a using an AC voltage $V_{13}^{\omega }$ that increases stepwise up to 1.5 V. The gate voltage V_g was applied to the silicon back contact. The two Dirac points reveal different doping levels (see inset) for the bare graphene leads (charge neutral) and the proximitized graphene (p-doped).

Fig. 2. Mapping of local charge current.

Current-induced polarisation rotation ${\theta }_K^{2\omega }$ (lower panel) for different bias configurations of the graphene/WTe₂ junction (upper panel). a, b Bias applied between graphene and WTe₂. Bias applied along (c) WTe₂ and (d) graphene only. In c, the colour scale was expanded by a factor of eight for better visibility. The arrows indicate the direction of the ac current j as expected from the bias configuration. The numbers indicate the contacts used for biasing. All other contacts are floating. The scale bar is 2 μm. Experimental parameters are V_g = 0 V, a V₂₃ = 3 V, b V₂₁ = 3 V, c V₂₄ = 4 V, d V₁₃ = 1.5 V.

Fig. 3. Gate-dependent KR microscopy.

Spatially-resolved KR, detected on the device in Fig. 1 under AC current injection along the graphene stripe (V₁₃ = 4 V) for a V_g = −30 V and b V_g = 30 V. Grey dashed lines highlight the metal electrodes, labelled 1 and 3. c, d Profiles of the current-induced Kerr angle ${\theta }_K^{\omega }$ and the Joule heating-induced polarisation rotation ${\theta }_K^{2\omega }$, along the horizontal dotted/dashed lines in a and b. Vertical solid lines indicate the WTe₂ ribbon.

Fig. 4. Electronic and optical interface spectroscopy.

a Differential resistance $V_{23}^{\omega }/I_{12}^{\omega }$ as a function of gate voltage V_g and DC drain-source voltage V₂₃. A vertical tunnel barrier is assumed due to the vdW gap between graphene and WTe₂. b Local photocurrent acquired at the graphene/WTe₂ junction as a function of V_g and drain-source voltage V₁₃. c Differential KR signal ${\theta }_K^{\omega }/V_{13}^{\omega }$ vs. V_g. In each case, the corresponding experimental configuration is displayed at the bottom. The horizontal dotted lines indicate the positions of the two Dirac points. The horizontal white or black arrows in the upper panels mark the Dirac point of the proximitized graphene. All data were taken at 4.2 K on a second heterostructure device.

Fig. 5. Theory of current-induced KR in WTe₂.

a Schematic drawing of the electric fields associated with the linearly polarised laser light (red), the induced non-linear Hall response (blue), and the vertical component of the ac current flow (black). b, c Sketch of the current flow profiles in the WTe₂, shown for b p-type and c n-type doping of the proximitized graphene under overall current flow j_x in positive x-direction; p is the carrier momentum. d Theoretically calculated and e experimentally measured ${\theta }_K^{\omega }$ (V₁₃ = 3 V) as a function of x and V_g. The dashed/dotted lines indicate where the spatial profiles in Fig. 3 were taken. The data are from the device in Fig. 1.