Field angle-independent high magnetoresistance and field angle-dependent coercivity in Fe₃GaTe₂/Phosphorus all-van der Waals spin valves at room temperature

Abstract

Two-dimensional (2D) spin valves are important for energy-efficient memory and computation devices but remain limited in achieving high magnetoresistance (MR) ratios at room temperature due to the lack of suitable ferromagnetic materials, optimal spin transport channel spacers, and great challenges in atom-level interface control in heterostructures. Here we report a record room temperature MR up to 103% for all-van der Waals (vdW) heterostructures by employing vdW ferromagnetic Fe₃GaTe₂ (FGaT) electrodes with vdW monoelemental black or violet phosphorus spacers. Notably, these devices exhibit magnetic field angle-independent MR while retaining anisotropic coercivity. This behavior originates from domain wall motion-driven magnetization reversal and the strong perpendicular magnetic anisotropy of the FGaT. First‑principles transport calculations reveal an intrinsic zero‑bias MR of up to 330% at the Fermi level for ideal interfaces and a monolayer spacer, supporting the high spin‑polarized transport capability of these heterostructures. The experimental room temperature MR is lower due to finite bias, temperature, spacer thickness, and realistic interface conditions. These results demonstrate the potential of 2D vdW monoelemental phosphorus as a high-efficiency spin-transport channel in room temperature all-vdW spin valves and provide a framework for designing field-orientation-resilient non-volatile spintronic memory devices.

The authors found that a room temperature magnetoresistance of 103% is achieved in all-van-der-Waals spin valves using Fe₃GaTe₂ and black phosphorus, and that the MR ratio is independent of field direction due to domain-wall-motion-driven magnetization switching dynamics in Fe₃GaTe₂.

Introduction

Electron spin dynamics and spin-polarized carrier injection represent the cornerstone of emerging quantum information storage paradigms and non-von Neumann computing architectures^1,2. This quantum mechanical property enables new approaches to information encoding, transmission, and decoding, also drives the development of spin-based nanoelectronics with low energy consumption, ultrahigh computational density, subthreshold operation, and quantum coherence preservation^3–5. In next-generation spin memory devices such as spin valves, three parameters are central to room temperature high magnetoresistance (MR): the spin polarization in ferromagnetic electrodes at room temperature, the optimization of spin-diffusion length (λ_S) in spin transport channel spacer, and atomic smooth interfaces between ferromagnetic electrodes and spacers for efficient spin injection^6–8.

Two-dimensional (2D) spin valves based on van der Waals (vdW) heterostructures have enabled unprecedented control over the spin degree of freedom through quantum-confined 2D materials^9–12. Recent progress in 2D spin valves is driven by the emergence of 2D itinerant ferromagnets with robust perpendicular magnetic anisotropy (PMA), high spin polarization near the Fermi level¹³ and high Curie temperatures (T_C) above 300 K such as Fe₃GaTe₂ (FGaT), allowing room temperature spintronic applications^14,15. The 2D transition metal dichalcogenides compound semiconductors with interfacial Rashba splitting below 10 meV^16,17 are also applied as spin transport channel spacers in 2D spin valves with room temperature MR 0.4%-85%^18–25, leveraging their gate-tunable band structures²⁶, valley-dependent Berry curvature²⁷, and suppressed Dzyaloshinskii-Moriya interactions (DMI)²⁸. The MR of 2D spin valves is much lower compared with conventional CoFeB/MgO systems²⁹. 2D spin valves remain highly limited in achieving higher MR ratios at room temperature due to the lack of optimal spin transport channel spacers with better spin diffusion and spin injection capabilities and the great challenges in atom-level interface control in heterostructures for better spin-polarized carrier injection. Furthermore, unresolved mechanisms of magnetic vector switching in atomically thin 2D ferromagnets and an incomplete understanding of the relationship between MR and magnetic vector switching in vdW heterostructures are also two challenges in achieving high room temperature MR in 2D spin valves.

Here, we report field-angle-independent, room temperature, high magnetoresistance in FGaT/monoelemental phosphorus all-vdW spin valves. Combining 2D vdW monoelemental black or violet phosphorus (BP or VP) with 2D vdW ferromagnetic FGaT yields two types of ferromagnet/spacer/ferromagnet (FM/Spacer/FM) all-vdW spin valves. The FGaT/BP/FGaT spin valves achieve a high room temperature MR ratio of 103% for all-vdW 2D spin valves, and FGaT/VP/FGaT show a MR of 29%. It is worth noting that while room temperature MR values of approximately 100% have been demonstrated in graphene³⁰, the present work achieves a comparable MR in a compact vertical tunneling geometry using all‑vdW FGaT/monoelemental phosphorus/FGaT structures. To the best of our knowledge, this is the first realization of over 100% MR in an all‑vdW vertical spin valve with a monoelemental phosphorus spacer. The high room temperature MR is attributed to ultralow spin-orbit coupling, long spin lifetimes, and diffusion length of BP spacer^31–33 and atomically smooth interfaces in heterostructures enabling excellent spin transmission. Moreover, a notable observation is the non-collinearity between the magnetic moment orientation and the applied magnetic field direction, leading to room temperature, field angle-independent magnetoresistance. These effects are attributed to magnetic switching governed by domain wall (DW) motion, as directly visualized via magneto-optical Kerr effect (MOKE) microscopy. The large room temperature MR observed in all-vdW spin valves is further supported by first-principles theoretical calculations. The demonstration of over 100% isotropic MR at room temperature in all vdW 2D heterostructures provides a strong theoretical and experimental foundation for the development of advanced all-vdW 2D spintronic devices.

Results

Atomic imaging and interface characteristics in all-vdW heterostructures

Figure 1a–c presents the cross-sectional high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) images of BP, VP, and FGaT. Combined with optical microscopy, and X-ray diffraction (XRD), the high crystallization quality is confirmed (Supplementary Fig. 1a–f). Semiconducting BP and VP serve as spacer layers, functioning as spin-polarized electron tunneling channels^32,34–38. The ferromagnetic electrodes are 2D FGaT^39–41. This combination yields a “hamburger-like” device configurations: FGaT/BP/FGaT and FGaT/VP/FGaT (Fig. 1d and Supplementary Fig. 1g).

Fig. 1. Schematic diagram and characterization of interfaces for all-vdW heterostructures of FGaT and monoelemental BP/VP.

Cross-sectional HAADF-STEM images of BP (a), VP (b), and FGaT (c) crystals along the [100], [110], and [100] zone axes, respectively. All STEM images can be precisely matched with the atomic structure model on the right-side. d Schematic diagrams of FGaT/BP/FGaT and FGaT/VP/FGaT heterostructures. The hamburger diagrams on the upper-right corner vividly illustrate the structure of the three-layer heterostructures. e–h Cross-sectional STEM images and EDS maps of the interfaces in FGaT/BP (e, f) and FGaT/VP (g, h) heterostructures. The STEM images display the planar interface between 2D vdW ferromagnetic and phosphorus layers. EDS maps show the spatial distribution of Fe, Ga, Te, and P elements in the heterostructures. The pink, red, cyan, and brownish red maps correspond to Fe, Ga, Te, and P, respectively. The scale bar is 10 nm. i Raman spectra of FGaT, BP, and FGaT/BP heterostructure, respectively.

In spintronic devices, the interfacial quality of heterojunctions determines key performance characteristics^{11,12,42–44}. HAADF-STEM imaging (Fig. 1e–h) reveals atomically sharp interfaces in all-vdW heterostructures, favoring efficient spin-polarized electron transport. Elemental mapping by energy-dispersive spectroscopy (EDS) further confirms precise chemical composition and interfacial integrity. Further Raman spectroscopy (Fig. 1i) of the BP nanosheet reveals three main peaks at ~361.4, 438.4, and 466.1 cm⁻¹, which can be attributed to the $A_{\mathrm{g}}^1$, B_2g, and $A_{\mathrm{g}}^2$ modes, respectively. The FGaT nanosheet displays two characteristic peaks, A₁ and E₂, while the FGaT/BP heterostructure contains all the Raman characteristic signals of the two materials, verifying the structural features of the all-vdW heterostructure.

Magnetoelectric transport characteristics in all-vdW spin valves

Precise adjustment of the thicknesses of the upper and lower ferromagnetic layers in FM/spacer/FM heterostructures produces a H_C difference, which enables deterministic magnetization switching between parallel and antiparallel states under both positive and negative applied magnetic fields. This process yields reversible transitions between low- and high-resistance states during magnetic field sweeps, resulting in pronounced MR in magnetoelectric transport measurements. To test magnetoelectric transport characteristics, we conducted measurements on the all-vdW heterostructures fabricated with semiconducting 2D phosphorus (Supplementary Fig. 2) and FGaT (Supplementary Fig. 3).

Figure 2a presents representative out-of-plane (OOP) magnetic field (H)-dependent MR curves at 300 K and 320 K. Both BP and VP-based spin valves exhibit clear resistance jumps during magnetic moment reversal between parallel and antiparallel states—a low-resistance state (R_P) for parallel magnetic moments in the upper and lower ferromagnetic layers, and a high-resistance state (R_AP) for antiparallel alignment (Supplementary Fig. 4). The MR ratio is defined as (R_AP-R_P)/R_P. As shown in Fig. 2b, c, the MR ratio in these devices shows a negative correlation with both bias current (I_bias) and temperature. In semiconducting spin valves or magnetic tunnel junctions, the observed inverse correlation between MR and I_bias is primarily ascribed to the semiconductor’s low Fermi level or the pronounced scattering of high-energy electrons within the localized interface states of the heterostructure^21,24,45,46. At 300 K, the FGaT/BP(4.6 nm)/FGaT heterojunction attains an MR ratio of 103%, while FGaT/VP/FGaT achieves 29%. To the best of our knowledge, this represents the first demonstration of over 100% MR in an all‑vdWs vertical spin valve employing a monoelemental phosphorus spacer at room temperature^47–49. It is instructive to compare this value with other representative vdW heterostructures that exhibit significant room temperature MR. For instance, terraced single‑layer graphene shows a large MR arising from pseudomagnetic fields induced by strain engineering⁵⁰. Graphene moiré superlattices can also achieve considerable MR at room temperature due to band structure reconstruction⁵¹. Similarly, graphene/black phosphorus heterostructures exhibit notable MR originating from charge transfer and interface‑induced band bending⁵². Very recently, an Fe₃GeTe₂/graphene heterostructure was reported to achieve a giant MR of ~9400% at 9 T and room temperature, attributed to spin‑dependent orbital coupling⁵³. Although some systems may offer comparable or even larger MR ratios under optimized conditions, the distinctive merit of our work lies not in the absolute MR value alone. Our compact all‑vdW vertical tunneling geometry with elemental phosphorus spacer achieves a similar MR magnitude while operating seamlessly at room temperature, more importantly, can be directly integrated into non‑volatile memory and information storage applications. Notably, at 10 K, these values increase to 160% and 75%, respectively. These results provide experimental validation for high-performance all-vdW heterostructures operable across a broad temperature range (Supplementary Fig. 6 and Supplementary Fig. 7). Furthermore, the increase in spacer thickness from 7.5 nm to 12 nm results in a pronounced reduction in the MR at 300 K, from 35% for the FGaT/BP(7.5 nm)/FGaT heterojunction to 0.04% for the FGaT/BP(12 nm)/FGaT heterojunction (Supplementary Figs. 8–10). Similarly, when the VP thickness increases from 3.6 nm to 11 nm, the MR of the FGaT/VP/FGaT heterojunction decreased from 29% to 0.08% at 300 K (Supplementary Fig. 11). These findings underscore the pivotal role of spacer thickness in magnetic transport, notably, its inverse correlation with MR^24,44,54. This relationship could be attributed to the increase in spin damping with thickness, which leads to more pronounced spin transport dissipation⁵⁵. The band gap of 2D BP and VP decreases with the increasing of thickness, effectively lowering the transport barrier^38,56–58. For a given spacer layer, the reduction in the potential barrier typically weakens the spin-polarized tunneling effect, resulting in a decrease in MR. These heterostructures display distinct transport characteristics: Resistance manifests two well-defined states—R_P and R_AP; MR and I_bias show an inverse relationship (Fig. 2b and Supplementary Fig. 9d). MR decreases progressively with increasing temperature (Fig. 2c and Supplementary Fig. 9e), mainly due to thermally induced reduction in spin polarization (P) of the FGaT layer via spin-wave excitation^25,59, following the characteristic scaling P ∝ (1 − αT^3/2)^59,60. The sub-10 nm thickness of BP/VP spacers (Supplementary Figs. 5 and 8) remains well below their spin diffusion lengths^36,61, facilitating robust device performance.

Fig. 2. Magneto-transport characterizations of FGaT/BP(VP)/FGaT spin valves.

Temperature-dependent measurements across 10–320 K. a OOP field-dependent magnetoresistance (MR-H) curves at 300 K and 320 K. b MR versus I_bias, extracted from field-dependent resistance (R-H) hysteresis. c Temperature dependence of MR for BP- and VP-spacer devices. d, e Magneto-transport response: R–H curves for FGaT/BP/FGaT (d) and FGaT/VP/FGaT (e) at 300 K, measured at different θ_H-I (the angle between H and spontaneous magnetization direction). The light green shaded area marks the trend of the resistance jump positions. f Schematic of heterostructure configuration and angular setup. g Polar plot of MR for the two ferromagnetic layers in FGaT/BP/FGaT and FGaT/VP/FGaT. h, i H_C versus θ_H-I for FGaT in FGaT/BP/FGaT (h) and FGaT/VP/FGaT (i). All MR ratios and H_C values are derived from R–H curve analysis.

It is noticed that certain R-H curves lack the typical rectangular hysteresis loop features observed in well-defined spin valve systems under high-resistance conditions, a deviation that frequently appears in other reports^{11,21,23,34,59,62}. This phenomenon can primarily be attributed to the degraded signal-to-noise ratio under low-current measurement conditions and the inconsistent domain inversion characteristics in thick vdW nanosheets⁶³.

The all-vdW 2D spin valves also show exceptional magnetic-field angle-independent magneto-transport characteristics. Previous studies confirm that magnetic hysteresis loops and anomalous Hall effect (AHE) signatures in 2D ferromagnets change continuously with the field-magnetization angle θ_H-I (the angle between H and spontaneous magnetization direction)^4,14. For instance, H_C shows monotonic enhancement with increasing θ_H-I^14,41,64,65, a trend independently verified by AHE measurements of FGaT Hall device (Supplementary Fig. 3). However, a big challenge is to clarify the relationships between spontaneous magnetization orientation and the direction of the applied magnetic field and between MR magnitude and magnetization configuration in all-vdW 2D spin valves.

To examine the effect of θ_H-I on device performance, we applied magnetic fields deviating from the easy-axis orientation while tracking resistance during field sweeps. Figure 2d, e shows angular-dependent resistance-field (R-H) characteristics for FGaT/BP/FGaT and FGaT/VP/FGaT spin valves under 360° rotation for field angle θ_H-I (Fig. 2f), revealing pronounced magneto-transport features. Magnetoresistance remains invariant when changing the magnetic field angle (Fig. 2g), yet the H_C reaches a maximum value at 90° and 270° (Fig. 2h, i). FGaT exhibits minimal OOP H_C, which increases monotonically from OOP to in-plane (IP) rotation, with abrupt H_C enhancement near IP orientations. The above phenomenon is also observed in the FGaT/BP(7.5 nm)/FGaT heterostructure (Supplementary Fig. 12). These results highlight two essential points: first, the angular-invariant MR demonstrates robust spin transport beyond strict magnetic confinement; second, the pronounced angular dependence of H_C shows considerable promise for magnetic field orientation sensing. Further theoretical examination is needed to elucidate these effects and mechanisms. We attribute the observed characteristics to intrinsic magneto-transport properties of FGaT, wherein AHE hysteresis broadens progressively during OOP-to-IP field rotation, while the Hall resistance (R_xy) amplitude remains constant (Supplementary Fig. 3). The abrupt R_xy transition signifies coherent magnetic moment reversal under OOP fields. Nevertheless, the dynamics of moment reorientation during arbitrary field-direction switching, including domain nucleation and propagation, require further investigation.

Domain wall motion-driven magnetization switching

Understanding magnetic moment reconfiguration requires direct investigation of domain switching mechanisms. Direct experimental evidence for domain reversal in 2D vdW magnets remains limited. Using temperature-dependent MOKE microscopy on FGaT nanosheets, we observed pronounced magnetic hysteresis with distinct switching transitions. As shown in Fig. 3a, b, sweeping the magnetic field from 1000 Oe to –1000 Oe, the Kerr signal displays an abrupt increase at approximately –120 Oe (300 K) and –700 Oe (80 K), corresponding to a dark-to-bright transition in MOKE contrast, mediated by the propagation of a moving DW boundary. Conversely, during field reversal from negative to positive saturation, a sharp signal decrease occurs at approximately 120 Oe (300 K) and 700 Oe (80 K), with a bright-to-dark transition governed by DW motion (see Supplementary Movie 1). These measurements provide the first confirmation that magnetization switching under a magnetic field occurs via DW motion in vdW materials, similar to previous observations in the nonlayered Kagome Weyl semimetal Co₃Sn₂S₂⁶⁶. On the basis of variable-temperature Kerr signals (Supplementary Fig. 13) and AHE loops (Supplementary Fig. 15) across samples of different thicknesses (Supplementary Fig. 14), we conclude that the magnetic field required to drive DW motion–induced magnetization reversal is inversely proportional to temperature and sample thickness. This finding is further supported by Supplementary Movies 2 and 3, which present the temperature-dependent MOKE images of the nanosheets in Supplementary Fig. 13b and Supplementary Fig. 13d, respectively. These visualizations, together with Supplementary Movie 1, clearly illustrate how DW dynamics vary with temperature and nanosheet thickness. These observations provide critical insights into designing 2D spintronic devices, particularly in optimizing their operational temperature range and physical dimensions.

Fig. 3. Field-orientation-independent magnetization switching in 2D FGaT.

Magneto-optical Kerr response versus applied field for FGaT nanosheets at 300 K (a) and 80 K (b). The illustrations in the upper part of Kerr signal panels are the schematic diagram of magnetic domain wall (a) and angle-dependent test configuration (b). Green and orange-red traces correspond to field sweeps from positive to negative and negative to positive, respectively. The left and right sides of the Kerr curves show the time-resolved magnetic domain imaging under negative scanning (green arrow) and positive scanning (orange-red arrow), respectively. Herein, the white dotted lines denote the position of magnetic domain wall, while the red arrow indicates the direction of domain wall motion. c Angular dependence of anomalous Hall resistance R_xy versus θ_H-I (top) and magnetization orientation angle (θ_M) as a function of θ_H-I (bottom) for FGaT device. All AHE data acquired with I_bias = 1 μA. d Simulated magnetization states (spin polarization: up or down) and DW positions in FGaT nanosheet under increasing field magnitudes at θ_H-I = 0°, 30°, 60°, and 90°, respectively. Field progression: low to high (indicated by arrows). These diagrams reveal field-direction-dependent DW motion dynamics and demonstrate monotonic increase of H_C with θ_H-I from 0° to 90°.

Through systematic investigation of magnetization reversal in all-vdW heterostructures under tilted magnetic fields, we established direct correspondence between magnetization switching behavior and DW motion as observed via MOKE imaging in FGaT, which exhibits strong PMA, consistent with previous reports. However, our measurements demonstrate nearly isotropic MR. Given that the MR signal scales with the relative magnetization orientation angle (Δθ) between ferromagnetic electrodes, this isotropic response indicates discrete switching of Δθ between 0° and 180° during reversal, suggesting a process insensitive to the applied magnetic field vector.

To clarify field-orientation-independent MR behavior, we performed AHE measurements on an individual FGaT device under varied magnetic field angles (θ_H-I). All measurements employed a fixed field amplitude of 3000 Oe, near the switching threshold in vdW heterostructures. Figure 3c presents the angular evolution of R_xy versus θ_H-I, demonstrating preferential magnetic moment alignment along the crystallographic c-axis, rather than continuous rotation toward the applied field direction (see Fig. 3c inset). For fields applied at angle θ_H-I, the actual magnetization orientation angle (θ_M) relative to the z-axis is given by ref. ⁶⁷:

$${\mathrm{\theta }}_{\mathrm{M}}=\arccos \left[\frac{{\mathrm{R}}_{\mathrm{xy}}\left({\mathrm{\theta }}_{\mathrm{H}-\mathrm{I}}\right)}{{\mathrm{R}}_{\mathrm{xy}}\left({\mathrm{\theta }}_{\mathrm{H}-\mathrm{I}}=0\right)}\right]$$ 1

The θ_M-θ_H-I correlation (Fig. 3c) indicates near-coaxial alignment of magnetization and applied field near the c-axis, whereas significant angular deviation occurs as the field approaches the ab-plane. This behavior originates from the Zeeman energy at 3000 Oe being insufficient to overcome the strong PMA and demagnetization field, thereby suppressing IP moment reorientation. Nevertheless, this field magnitude is adequate to initiate DW motion, enabling abrupt switching between bistable magnetic configurations with magnetization polarized along opposite directions of the z-axis, consistent with the vertical component of the applied field. Magnetization pinning along the crystalline z-axis fundamentally explains the observed isotropic MR dependence on θ_H-I.

Previous studies have demonstrated that boundary magnetization configurations, defect-mediated pinning, and thermal fluctuations control domain nucleation and stabilization in 2D ferromagnet^68–70. Once established, these domains propagate under minimal Zeeman energy input. This DW-mediated switching mechanism-distinct from coherent rotation—represents a promising pathway for ultralow-power spintronic memory and logic devices. Our findings establish that leveraging DW dynamics in strong-PMA 2D magnets enables energy-efficient magnetization reversal, accounting for the observed θ_H-I-independent MR. However, the microscopic origin of H_C anisotropy (H_C versus θ_H-I) remains unresolved.

The illustration in Fig. 3a displays a schematic of a Néel-type magnetic DW in the 2D vdW ferromagnet, representing a 180° DW. Under applied magnetic fields, the DW undergoes curvature-driven deformation and propagation, rather than remaining rectilinear. This behavior originates from the equivalent pressure (P_e) exerted normal to the DW interface (see the inset in Fig. 3b), supplying a portion of the energy. When the field is applied perpendicular to the nanosheet plane (parallel to the DW orientation), the pressure is given by ref. ⁷¹:

$$P_e=2I_{s}H$$ 2

where I_S represents the spontaneous magnetization and parallel to the z-axis. When the magnetic field H and I_S are oriented at angle θ_H-I, only the field component normal to the DW (Hcosθ_H-I) drive wall motion. Therefore, P_e transforms to⁷¹:

$$P_{\mathrm{e}}=2I_{\mathrm{s}}H\mid \cos {\theta }_{\mathrm{H}-\mathrm{I}}\mid$$ 3

In the first quadrant, P_e decreases with increasing θ_H-I at constant magnetic field magnitude. Therefore, larger θ_H-I values require stronger applied fields to generate sufficient pressure for DW motion (Fig. 3d). This accounts for the maximized H_C observed when the field is parallel to the device plane (see Supplementary Fig. 16a, c). From Eq. (3), the relationship P_e/I_S = 2H|cosθ_H−I| enables extraction of the 2H|cosθ_H−I| versus θ_H−I correlation from MR data (Supplementary Fig. 16b, d), demonstrating P_e stability across θ_H−I at constant I_S. The pressure minimum occurs at θ_H−I = 90° and 270°, where exceptionally high fields are required for magnetization switching. Collectively, these results explain the isotropic MR versus θ_H−I (magnetization is always perpendicular to the device), and the anisotropic relationship between the critical magnetic field-H_C and θ_H−I for all-vdW spin valves.

Theoretical framework

Compared to spin valves using other 2D material spacers, our vdWs heterostructure—based on vdW monoelemental phosphorus—achieves not only higher MR ratios and enhanced spin polarization⁴⁴, but also establishes a new class of efficient spin-polarized electron transport channels (Fig. 4a).

Fig. 4. MR comparison of different all-vdW 2D spin valves and first-principles spin-polarized transport computation of FGaT/P/FGaT heterostructure.

a Benchmark comparison of MR ratios: FGaT/2D-spacer/FGaT devices (300 K). Brown arrows indicate the data in this work. The remaining spacers, from left to right, correspond to WS₂²², WSe₂³⁴, WSe₂⁴⁶, WSe₂¹⁹, MoS₂²¹, MoSe₂²³, GaSe⁴⁹, FGaT⁷⁴, and h-BN⁷⁵, respectively. Atomic structures of the scattering region with flanking ferromagnetic electrodes for the FGaT/BP/FGaT (b) and FGaT/VP/FGaT (e) heterostructures. Spin-summed transmission coefficients under parallel (P) and antiparallel (AP) magnetic alignments for FGaT/BP/FGaT (c) and FGaT/VP/FGaT (f). MR ratio as a function of electron energy for FGaT/BP/FGaT (d) and FGaT/VP/FGaT (g), with the Fermi level aligned at zero.

To theoretically comprehend these phenomena, we evaluated spin-polarized transport properties at Fermi energy (E_F) for FGaT/BP/FGaT heterostructures using density functional theory (DFT) in combination with the nonequilibrium Green’s function (NEGF) approach, as implemented in the Nanodcal package⁷². According to the Landauer formula, the T_σ is the spin-dependent transmission coefficient with σ = ↑ or ↓ indicating the spin index, which can be calculated by:

$$T_{\sigma }={T_{\mathrm{r}}\left[{\Gamma }_{\mathrm{L}}{G}^{\mathrm{r}}{\mathrm{\Gamma }}_{\mathrm{R}}{G}^{\mathrm{a}}\right]}_{\sigma \sigma }$$ 4

where T_r represents the trace of the matrix, G^r(a) is the retarded (advanced) Green’s function and Γ is linewidth functions describing the coupling between leads and the scattering region. The calculation model is illustrated in Fig. 4b, e, wherein the MR is evaluated by comparing the spin polarizability between the parallel and antiparallel configurations of the left and right ferromagnetic layers. Using computed transmission coefficients, we determined the equilibrium MR ratio in the zero-bias limit through equation⁷³:

$$\mathrm{MR\; ratio}=\frac{G_{\mathrm{P}}-G_{\mathrm{AP}}}{G_{\mathrm{AP}}}\times 100\%=\frac{T_{\mathrm{P}}-T_{\mathrm{AP}}}{T_{\mathrm{AP}}}\times 100\%$$ 5

Where G_P (T_P) and G_AP (T_AP) denote the total conductance and total transmission for parallel (P) and antiparallel (AP) magnetic configurations, respectively. The spin-summed quantities are defined as G (T) ≡ G_↑ (T_↑) + G_↓ (T_↓). G_↑ (T_↑) and G_↓ (T_↓) represents the spin-resolved (up and down) conductance, respectively. The computational procedure first performs self-consistent calculations for the electrodes, followed by the central region; after achieving convergence, the calculation computes the electron transmission spectrum for each configuration. In fact, the Fermi level may deviate from E_F due to factors such as doping, defects, or lattice strain. The total density of states (TDOS) for the FGaT/BP/FGaT heterostructure (Supplementary Fig. 17) confirms metallic characteristics. In the parallel magnetic configuration, the spin-resolved density of states exhibits pronounced asymmetry between the spin up and spin down channels, indicating significant spin polarization at the Fermi level. Details of the computational methodology are provided in Supplementary Section 6: Theoretical calculation details.

The calculated transport properties of FGaT/BP/FGaT and FGaT/VP/FGaT heterostructures are summarized in Fig. 4c, d, f, g. We investigate transmission and MR as a function of energy (−2 eV to 2 eV). The results indicate that the spin transmission curve for parallel and antiparallel magnetization alignment configurations without overlap across the entire energy range. This leads to a difference in the corresponding resistance states in experiments under parallel and antiparallel magnetic moments. At the Fermi level, the MR of FGaT/BP/FGaT and FGaT/VP/FGaT heterostructures reaches as high as 330% and 87%, respectively, owing to the 100% spin polarization at the Fermi surface in the 2D Brillouin zone (2DBZ). We note that the calculated MR of 330% represents the intrinsic zero‑bias limit for an ideal interface and a monolayer BP spacer. In our experimental devices, the spacer thickness is 4.6 nm, and measurements are performed at finite bias and room temperature. The lower experimental value of 103% is therefore attributed to (i) reduced spin polarization in thicker spacers due to enhanced spin-orbit coupling and defect scattering, (ii) finite‑temperature electron–phonon interactions, (iii) series resistance from contacts and leads, and (iv) residual interface imperfections compared to the pristine theoretical model. Despite this discrepancy, the theoretical results confirm that the FGaT/P system possesses intrinsically high spin‑polarized transmission, and further optimization of spacer thickness and interface quality could approach the theoretical limit. When the energy deviates from E_F, the MR exhibits several peaks. For instance, an extremely high magnitude MR ratio of 2307% is observed at 0.27 eV in FGaT/BP/FGaT. Theoretical results confirm that the FGaT/BP/FGaT heterostructure delivers superior MR performance compared to the VP-based heterostructure. This enhancement is attributed to the BP spacer, which creates a more distinct difference in spin-polarized transmission between the parallel and antiparallel configurations, thereby yielding a higher MR value. We performed k-resolved transport function calculations for the FGaT/VP(1 L)/FGaT (Supplementary Fig. 18) and FGaT/BP(1 L)/FGaT (Supplementary Fig. 19) heterojunctions, allowing a comprehensive comparison of the transport properties between the two systems. The spin up transmission in the parallel state for the BP system is lower than that for the VP system. The transmission in both spin channels in the antiparallel state is also significantly lower than that in the VP system. The former is attributed to poorer band matching and weaker interface coupling between BP and the ferromagnetic electrodes, thereby lowering the tunneling probability for most spin channels in comparison to the VP system. In the antiparallel state, the BP system exhibits relatively weak interface coupling, and spin-switching-induced scattering is strongly suppressed, leading to a much lower transmission coefficient than that of the VP system. Taking into account both the parallel and antiparallel configurations, the transmission difference between the parallel and antiparallel states in the BP system is substantially larger than that in the VP system, resulting in a larger MR. For the BP system, resonance matching between spin up and spin down states occurs in the parallel configuration, which means that the spin up and spin down channels at specific k-points do not undergo strong decay due to band mismatch. Consequently, both spin channels can actively participate in tunneling near the Fermi level in the parallel state. In the antiparallel state, the interface coupling in the BP system is relatively weak, and the spin-switching-induced scattering is suppressed. As aforementioned, the FGaT/BP(1 L)/FGaT heterojunction exhibits substantial spin transport in the parallel state, while spin transport is suppressed in the antiparallel state, providing favorable conditions for achieving a large MR. These findings suggest that the FGaT/P/FGaT should possess an ultra-high MR ratio within a relatively broad energy window close to the E_F, highlighting substantial optimization potential of all-vdW FGaT/P/FGaT heterostructures in information storage devices.

Discussion

This study establishes vdW monoelemental phosphorus/FGaT-based all-vdW heterostructures as an efficient quantum platform for room temperature 2D spintronics, achieving a high MR (103% at 300 K in FGaT/BP/FGaT) through spin-polarized transport by sub-10 nm vdW BP or VP spacers. First‑principles transport calculations reveal that the intrinsic zero‑bias MR at the Fermi level for ideal monolayer spacer interfaces is as high as 330% for FGaT/BP/FGaT and 87% for FGaT/VP/FGaT, serving as an upper bound for the achievable MR in this system. The experimental room temperature MR of 103% approaches one‑third of this theoretical limit, indicating room for further improvement through interface engineering and spacer optimization. Given the resonance matching of spin up and spin down states in the parallel configuration of FGaT/BP/FGaT, the spin up and spin down transport channels at the specific k-point do not undergo significant decay due to band mismatch. Therefore, both spin up and spin down electrons in the parallel state of the BP system can participate in tunneling at the Fermi surface. This special transportation channel results in a large space for its MR ratio. Notably, we demonstrate angular-independent MR, which originates from strong PMA and stable θ_M, and is of significance in developing high-density information storage. The magnetic field-driven DW motion is observed in vdW magnetic materials. The magnetic field angle anisotropy of H_C can be verified through MOKE-microscopy-resolved DW motion. This can be accounted for by the universal magnetization switching model ($P_{\mathrm{e}}=2I_{\mathrm{s}}H\mid \cos {\theta }_{\mathrm{H}-\mathrm{I}}\mid$) during the DW motion process. These all-vdW vertical heterostructures address fundamental 2D spintronic limitations by decoupling MR magnitude from magnetic field orientation, enabling robust memory devices operable under arbitrary magnetic environments. Future research may leverage this magnetic field-orientation resilience to develop non-von Neumann spintronic memory architectures that exploit angular-independent magneto-transport phenomena.

Methods

Sample preparation and characterizations

Fe₃GaTe₂ crystals are synthesized using the self-flux method (see Supplementary Section 7: Two-dimensional ferromagnetic crystals preparation). BP, VP, and h-BN crystals are commercially obtained from Nanjing Muke Nano Technology Co., Ltd. Scanning transmission electron microscopy (STEM) imaging is performed using a double-aberration-corrected transmission electron microscope (Titan G2 Cubed Themis, FEI) operated at 200 kV with a 21.4 mrad probe convergence angle. Chemical composition analysis is conducted through EDS spectrum imaging using a Super-X detection system (four windowless silicon drift detectors) mounted on the STEM. Cross-sectional TEM lamellae are prepared by in situ lift-out using a Ga⁺ focused ion beam (Helios 5 CX, Thermo Fisher). To reduce charging effects, a 10 nm platinum protective layer is deposited before FIB processing. Atomic force microscopy images are acquired using a Dimension Icon (Bruker). Raman spectra are collected with a confocal Raman microscope (LabRAM HR-Evolution, Horiba Jobin Yvon) equipped with a 532 nm laser. X-ray diffraction measurements are performed on a diffractometer (D/Max-2400, Rigaku) using Cu K_α radiation. First-principles calculations are provided by Hongzhiwei Technology (Shanghai) Co., Ltd.

Device preparation

The fabrication of FGaT/BP(VP)/FGaT heterostructures starts with electrode patterning and sequential transfer of 2D materials. Ti/Au electrodes (5/15 nm) are deposited on SiO₂/Si substrates via photolithography and thermal evaporation. Single-crystalline FGaT is mechanically exfoliated into few-layer nanosheets using Scotch tape and mounted on a polydimethylsiloxane stamp. Flakes of suitable thickness, confirmed by optical microscopy and atomic force microscopy, are selected and transferred onto predefined electrodes using a dry-transfer method. Spacer-layer BP/VP and the top FGaT are then sequentially aligned and transferred onto the underlying layers. To prevent oxidation, the assembled heterostructures are encapsulated with h-BN capping layer. The devices are thermally annealed at 120 °C for 10 min to improve interfacial adhesion. All exfoliation and transfer steps are performed in an Ar-filled glovebox.

Device measurements

Electrical and magneto-transport measurements are carried out using a physical property measurement system (ColdTUBE™, MultiFields) over the temperature range 4.2–325 K. DC transport characterization employs a two-terminal configuration with a precision current source (Keithley 6221) and nanovoltmeter (Keithley 2182A).

Author contributions

Y.J. (Yong Jiang), Z.W., and H.C. conceived and supervised the research. Z.J., Y.T., K.Y., and G.Z. contributed equally to this work. Z.J., Y.T., K.Y., and Q.C. designed experimental methodologies, conducted data analysis, and authored the manuscript. G.Z., Y.T., M.Z., W.J., and H.W. synthesized 2D vdWs single crystals. Y.T., Z.W., L.L., Y.W., X.G., and F.Z. performed Raman, XRD, and MOKE characterization. Z.J., Y.T., Q.C., and M.Z. performed device fabrication, device characterization, and data analysis. J.Y. (Junxin Yan) and K.Y. carried out the STEM and EDS measurements. Z.J., Y.T., and Q.C. wrote and revised the manuscript. All authors provided critical feedback and participated in scientific discussions throughout the research.

Peer review

Peer review information

Nature Communications thanks Faxian Xiu and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

This work was supported by the National Key Research and Development Program of China (Award Nos. 2022YFA1204000 and 2022YFE0134600), the National Natural Science Foundation of China (Grant Nos. U24A6002, 52571263, 52302192, 52571229, and 52272152), and the Tianjin Key Research and Development Program (Award No. 24YFXTHZ00150). Additional support was provided by the Young Top-Notch Talent Program of the Hebei Education Department Science and Technology Project (Grant No. BJK2023082) and the Hebei Natural Science Foundation (Grant Nos. E2026110016 and E2026110036). Z.W. acknowledges the Fundamental and Interdisciplinary Disciplines Breakthrough Plan of the Ministry of Education of China (Grants No. JYB2025XDXM701), and is supported by Science and Technology Innovation Key R&D Program of Chongqing. We also acknowledge research support from the Tianjin Municipal Science and Technology Bureau. The Analytical & Testing Center at Tiangong University is thanked for its technical assistance.

Data availability

The Raman, magnetoelectric transport, and computational data generated in this study have been deposited in the figshare database under accession code 257deb3162177078fb5e [https://figshare.com/s/257deb3162177078fb5e].

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-75103-8.