A Space-Charge-Limited van der Waals Spin Transistor

TL;DR

The work addresses the challenge of achieving transistor-like operation with nonvolatility in 2D magnetic semiconductors by introducing a space-charge-limited spin transistor (SCST) that blends vertical and lateral transport across a monolayer-bilayer CrSBr junction. Gate-controlled electrostatic doping tunes carrier density and interlayer exchange, delivering giant gate-tunable magnetoresistance up to 3000% (TMR) in a device only two atomic layers thick. Correlated electrical transport and scanning NV center magnetometry reveal a field-training and layer-sharing mechanism that governs domain-wall motion during spin-flip transitions, enabling multilevel memristive conductance states. The results demonstrate the potential of 2D magnetic semiconductors for nonvolatile and neuromorphic computing by providing electrically controllable interlayer coupling and spintronic functionality in ultrathin platforms.

Abstract

Integrating semiconducting and magnetic materials could combine transistor-like operation with nonvolatility and enable architectures such as logic-in-memory. Here, we employ correlated electrical transport and scanning nitrogen-vacancy (NV) center magnetic imaging to elucidate a spin transistor concept that amalgamates both vertical and lateral transport in a 2D antiferromagnetic semiconductor, distinct from purely vertical tunneling devices. Our device, based on a monolayer-bilayer junction in CrSBr, displays giant, gate-tunable magnetoresistance driven by the dual action of electrostatic doping on space-charge-limited lateral conduction and interlayer exchange coupling. Moreover, we visualize a field-trainable, layer-sharing effect that selects between coherent or domain-wall reversal at the spin-flip transition, enabling multilevel, memristive conductance states. These findings open opportunities for 2D magnetic semiconductors to address limitations in contemporary computing.

Figures (6)

• Figure 1: Device overview. (a) Schematic of the SCST with laterally offset contacts on two ends of a monolayer-bilayer step in CrSBr. FLG - few layer graphene, $V_{sd}$ - bias voltage, $V_{G}$ - gate voltage. (b) Current $I_{sd}$ across the SCST versus bias $V_{sd}$ for several gates $V_G$ at zero field. (c) Schematic of a conventional sf-MTJ. (d) $I_{sd}$ versus $V_{sd}$ for the sf-MTJ at zero field. (e) $I_{sd}$ versus $V_G$ at fixed bias for the SCST and sf-MTJ. The dip in current for the sf-MTJ near $V_G = 0$ is due to the Dirac point of monolayer graphene. (f) Log-log plot of $I_{sd}$ versus $V_{sd}$ for the SCST, showing a power law scaling $I_{sd} \propto V_{sd}^m$. Data for negative $V_{sd}$ is shown. (g) Schematic of SCL conduction in the low carrier density regime. Left: Negative $V_G$ depletes the channel of free electrons. Right: Negative $V_{sd}$ injects electrons, forming a decaying distribution from the source contact. (h) Density of trap states (DOS) between the valence ($E_v$) and conduction ($E_c$) bands. Larger exponents $m$ correspond to the quasi-Fermi level $E_f$ residing in energy windows where the trap density increases more slowly. (i) Mechanism of MR in the SCST. Right: In the AFM state of the bilayer, free electrons induced by $V_G$ or injected by $V_{sd}$ (not shown) are primarily localized to the layer that each contact touches. Left: In the FM state, induced free electrons distribute more uniformly between the two layers.
• Figure 2: Gate-tunable MR. (a) $I_{sd}$ versus $V_{sd}$ for the SCST in the FM state at $B_b = 0.4$ T for select gate voltages. The $m=1$ and $m=2$ lines are guides. (b) Resistance ratio $R_{AFM}/R_{FM}$ between the AFM and FM states of the SCST versus gate $V_G$ and bias $V_{sd}$. (c) $R_{AFM}/R_{FM}$ versus $V_G$ and $V_{sd}$ for the sf-MTJ. (d) Linecuts of the magnetoresistance versus gate for the SCST and sf-MTJ at constant bias. The left and right $y$-axes denote different conventions MR and TMR, as defined in the main text. The SCST's $|\mathrm{MR}|$ increases with carrier density, while that of the sf-MTJ is independent.
• Figure 3: Training effect and domain-wall-controlled AFM-to-FM transition. (a) Hysteresis sweep of the conductance under negative field training. The sweep starts at a large negative field ($B_b = -0.4$ T) and pauses at an intermediate positive field (0.3 T) before ramping backward. The labels of the bilayer/monolayer states are verified by magnetic imaging (Appendix B). For negative field training, the AFM-to-FM transition on the forward sweep is gradual, while it is abrupt on the backward sweep. The monolayer's hysteresis loop is outlined by the purple box. (b,c,d) Magnetic imaging of the AFM-to-FM transition on the forward sweep. The transition proceeds by reversing the bottom layer of the bilayer via domain-wall translation from the monolayer interface. The flake's optical image is shown as the inset in (b). (e,f,g) Images of the AFM-to-FM transition on the backward sweep, which instead occurs by an abrupt thermally-activated, global rotation of the top layer during scanning. Note the linear defects parallel to the $a$-axis (purple arrows), which act as pinning sites. $B_s$ is the stray field along the NV axis.
• Figure 4: Electrical modulation of interlayer exchange. (a) $I_{sd}$ versus $V_G$ at several fields near the bilayer spin-flip transition. (b,c,d) Magnetic images for increasing gate voltage ($V_G = -4$, 0, and 4 V) at $B_b = 168$ mT. The FM regions of the bilayer expand with increasing electron doping. (e,f) Images as $V_G$ is decreased back to $-4$ V. The FM regions of the bilayer recede but are obstructed by pinning, introducing hysteresis. The contrast $S$ is roughly proportional to the stray field.
• Figure 5: Resolving the antiphase states of bilayer CrSBr. (a) Optical image of SCST 1, corresponding to the device presented in the main text. (b) Stray field $B_S$ image of the monolayer-bilayer interface when a domain wall exists between the monolayer and bilayer. The monolayer is identified as $\rightarrow$ by its negative $B_S$, from which we deduce that the bilayer is $\rightarrow{}\!\!\!\leftarrow{}\;$ , denoting the bilayer's top and bottom layer magnetization, respectively. The domain wall (DW) is revealed by the bumpy monolayer-bilayer edge (box) and by a perturbation at the monolayer corner (circle) where the domain wall cuts the corner (see cartoon). (c) Stray field image of SCST 1 when no domain wall is present after flipping the monolayer to $\leftarrow$. The same edge is now smooth, and the corner is unblemished. (d) Optical image of a second device, SCST 2. (e,f) Images when a domain wall is present (e) and absent (f) in SCST 2. Similar telltale features as SCST 1 are observed, enabling determination of the bilayer state as $\rightarrow{}\!\!\!\leftarrow{}\;$ .