Avances recientes en espintrónica bidimensional

Reproducido con permiso de Han et al., Nat. Nanotechnol. 9, 794 (2014). Copyright 2018 Nature Publishing Group [14]. c , d Reproducido con permiso de Mendes et al., Phys. Rev. Lett. 115, 226601 (2015). Copyright 2015 Sociedad Química Estadounidense [68]. e , f Reproducido con permiso de Shao et al., Nano Lett. 16, 7514 (2016). Copyright 2016 Sociedad Química Estadounidense [71]

Conversión de giro y carga en materiales 2D. un Inyección y detección de espín eléctrico con geometrías de medición no locales. b Inyección y detección de espín eléctrico con geometrías de medición local. c Conversión de giro a carga en grafeno en YIG, un aislante ferromagnético. La corriente de giro se genera a partir del bombeo de giro de YIG y se convierte en corriente de carga en el grafeno. d Dependencia del campo magnético del voltaje de bombeo de centrifugado medido en YIG / Grafeno e Medidas SOT para MX ₂ / CoFeB bicapa. El MX ₂ representa MoS ₂ y WSe ₂ . f La ilustración de la acumulación de espín inducida por el efecto Rashba-Edelstein (REE) en la interfaz de MX ₂ / CoFeB bajo un campo eléctrico externo. un , b

Hill y col. informó por primera vez de la inyección de espín en el grafeno mediante el uso de electrodos magnéticos suaves de NiFe [57]. Sin embargo, se estima que la eficiencia de la inyección de espín es relativamente baja, alrededor del 10%, lo que podría atribuirse al desajuste de conductancia entre el metal ferromagnético y el grafeno. Luego, algunos investigadores propusieron usar una barrera aislante como Al ₂ O ₃ o MgO como una capa para ajustar la resistividad dependiente del espín interfacial y mejorar la eficiencia de la inyección del espín [58, 59, 60], pero el crecimiento de una capa de óxido de alta calidad es un desafío importante. Se han utilizado algunos métodos para mejorar la técnica de crecimiento de la capa de óxido o cambiar a otra capa de óxido interfacial, como una capa de TiO ₂ o HfO ₂ [61, 62]. Sin embargo, la resistividad interfacial dependiente del espín sigue siendo el problema fundamental, lo que conduce a una baja eficiencia de inyección de espín. Un material de aislamiento 2D, nitruro de boro hexagonal (h-BN), tiene una estructura cristalina similar a la del grafeno. Estudios teóricos y experimentales han demostrado que el uso de h-BN como barrera de túnel puede producir una interfaz de alta calidad y mejorar en gran medida la eficiencia de inyección de espín del grafeno. El h-BN de pocas capas exhibe un mejor rendimiento de inyección de giro que el h-BN monocapa [63, 64]. Sin embargo, los resultados de estas investigaciones aún dejan un gran vacío que llenar antes de que sea posible la aplicación práctica. En última instancia, lograr una inyección de espín perfecta (100%) requiere mucha investigación, y los materiales 2D proporcionan una dirección prometedora, como heteroestructuras 2D compuestas de materiales ferromagnéticos 2D, barreras de túneles 2D y canales de transporte 2D.

El efecto (inverso) Rashba-Edelstein es un efecto de interfaz que se origina en el SOC fuerte, que se puede utilizar para lograr la conversión de espín-carga [65]. Aunque el grafeno intrínseco tiene un SOC bastante débil, puede lograr una conversión de carga de espín eficiente utilizando el SOC fuerte del material adyacente a través del efecto de proximidad [66, 67]. Como se muestra en la Fig. 4c, cuando el grafeno está adyacente al aislante ferromagnético YIG, la corriente de espín se genera en la capa YIG mediante bombeo de espín y luego se convierte en una corriente de carga en grafeno por el efecto Edelstein inverso [68]. La Figura 4d muestra las curvas de voltaje de bombeo de rotación en función del campo en el dispositivo YIG / grafeno. Los voltajes de bombeo de espín se pueden detectar en el campo magnético perpendicular al canal de grafeno. Además, cuando el campo magnético externo se gira a lo largo del canal de grafeno, no hay voltaje de bombeo de giro. Además, una puerta líquida iónica aplicada sobre la superficie del grafeno obviamente puede modular las propiedades del grafeno para cambiar la eficiencia de conversión de espín a carga de YIG / grafeno [56].

A diferencia del grafeno, los TMDC con un SOC fuerte se consideran materiales prometedores para lograr la conversión de espín-carga [69, 70]. Un gran par de giro en órbita (SOT) en TMDC monocapa (MoS ₂ o WSe ₂ ) / La estructura bicapa de CoFeB se generó a través de la acumulación de espín inducida por la corriente causada por el efecto Rashba-Edelstein (Fig. 4e, f) [71]. El par de tipo campo y el par de amortiguación se determinaron mediante una medición de segundo armónico, y los resultados muestran que los TMDC monocapa de área grande tienen aplicaciones potenciales debido a su alta eficiencia para la inversión de magnetización. Además, se ha utilizado la técnica de resonancia ferromagnética de par de giro (ST-FMR) para investigar la conversión de giro y carga en TMDC. Por ejemplo, un resultado interesante de ST-FMR muestra que el SOT se puede controlar a través de la simetría de cristal de WTe ₂ en WTe ₂ / Bicapas de permalloy. Cuando se aplica corriente a lo largo del eje de baja simetría de WTe ₂ , se puede generar un par anti-amortiguación fuera del plano [72]. La propiedad de bloqueo del momento de giro en los estados de superficie de TI es útil para lograr la inyección de corriente de giro en materiales adyacentes a través de SOT. Debido a la fuerte correlación entre la dirección de polarización del espín y la dirección de la corriente de carga, la dirección del espín puede manipularse mediante la corriente de carga en los TI. Se han utilizado diferentes técnicas de medición para investigar la conversión de carga de giro, incluida la medición de segundo armónico, bombeo de giro y ST-FMR. Estos resultados de medición demuestran que es posible generar SOT eficiente en materiales 2D como TMDC y TI.

Transporte giratorio

La clave para el transporte de espín es obtener un canal de transporte de espín favorable con una longitud de difusión de espín y un tiempo de relajación de espín largos. La relajación del espín es causada por la dispersión del momento, por lo que el grafeno con un SOC débil se considera un material ideal para el transporte del espín [14, 73]. Tombros y col. [74] realizó transporte de espín electrónico y precesión de espín en una válvula de espín de grafeno lateral simple a temperatura ambiente mediante medición no local en 2007. Como se muestra en la Fig. 5a, b, la válvula de espín no local está compuesta de cobalto ferromagnético de cuatro terminales como electrodos, una fina Al ₂ O ₃ capa de óxido como barrera y una hoja de grafeno como canal de transporte de espín. La señal de medición en la Fig. 5c muestra que si los electrodos ferromagnéticos para inyección de espín y detección de espín tienen magnetizaciones paralelas, la resistencia no local medida por los contactos 1 y 2 tiene un valor positivo. Si los electrodos ferromagnéticos para inyección de espín y detección de espín tienen magnetizaciones antiparalelas, entonces la resistencia no local muestra un valor negativo. La precesión de centrifugado de Hanle se puede utilizar para determinar la longitud de difusión del centrifugado y la vida útil del centrifugado. Como se muestra en la Fig. 5d, la vida útil del giro ( τ _sf ) y longitud de relajación de giro ( λ _sf ) son 125 ps y 1,3 μm, respectivamente, en una válvula de giro de grafeno lateral simple a temperatura ambiente. Además, la puerta se puede utilizar para mejorar la duración de la relajación del giro y la vida útil del mismo [75, 76]. La teoría predijo que la vida útil del espín en el grafeno prístino puede alcanzar 1 μs, mientras que los valores de los experimentos informados oscilan entre picosegundos y unos pocos nanosegundos.

Reproducido con permiso de Tombros et al., Nature 448, 571 (2007 ). Copyright 2007 Nature Publishing Group [74]. e - h Reproducido con permiso de Avser et al., Nat. Phys. 13, 888 (2017). Copyright 2017 Nature Publishing Group [84]

Transporte de centrifugado en válvulas de centrifugado laterales. un Geometrías de medición de transporte de espín no local. Se inyecta una corriente desde el electrodo 3 a través del Al ₂ O ₃ barrera en grafeno y se extrae en el contacto 4. b Micrografía electrónica de barrido de una válvula de giro de cuatro terminales con grafeno de capa única como canales de transporte de giro y Co como cuatro electrodos ferromagnéticos. c Señal de válvula de giro no local a 4,2 K. Las configuraciones magnéticas de los electrodos se ilustran para ambas direcciones de barrido. d Precesión de espín de Hanle en la geometría no local, medida en función del campo magnético perpendicular B _z para configuraciones paralelas. e Esquemas de una válvula giratoria de fósforo negro. El recuadro muestra los esquemas de la heteroestructura. f Imagen óptica del dispositivo. g Señal de válvula de giro no local en función del campo magnético en el plano. La magnetización relativa de los electrodos del inyector y del detector se ilustra mediante flechas verticales, y las flechas horizontales representan las direcciones de barrido del campo magnético. h Precesión de espín de Hanle en la geometría no local, medida en función del campo magnético perpendicular B _z para configuraciones paralelas y antiparalelas. El recuadro muestra la precesión de espín bajo el campo magnético aplicado. un - d

Se utilizan muchos métodos mejorados para aumentar la longitud de difusión del espín y la vida útil del espín, y algunos dispositivos ya exhiben longitudes de difusión de espín largas en el rango micrométrico [13, 77, 78]. Por ejemplo, el grafeno cultivado epitaxialmente en SiC tiene una alta movilidad, exhibiendo una eficiencia de transporte de espín de hasta un 75% y una longitud de difusión de espín superior a 100 µm [79]. La heteroestructura h-BN / grafeno / h-BN exhibe un rendimiento de transporte de espín a larga distancia, donde la longitud de difusión del espín puede alcanzar 30,5 μm a temperatura ambiente [13]. El transporte de espín en materiales 2D puede verse afectado por la difusión / deriva, que puede modularse aplicando un campo eléctrico. Ingla-Aynés y col. [80] informó una longitud de relajación de espín de hasta 90 μm en grafeno bicapa encapsulado en h-BN mediante el uso de derivación del portador. Sin embargo, el SOC débil y la banda prohibida cero en el grafeno intrínseco restringen sus perspectivas de dispositivos de espín semiconductores. El fósforo negro tiene una banda prohibida directa considerable y una movilidad a temperatura ambiente de 1000 cm ² V ⁻¹ s ⁻¹ , que lo convierten en un material espintrónico semiconductor ideal [81,82,83]. Avsar y col. [84] construyó una válvula de giro lateral basada en una hoja ultrafina de fósforo negro y midió sus propiedades de transporte de giro a temperatura ambiente a través de la geometría no local (Fig. 5e, f). El transporte de espín electrónico en la Fig. 5g muestra que cuando las direcciones de magnetización de los ferroimanes cambian, la resistencia no local tiene un cambio de Δ R ≈ 15Ω. Además de esto, la precesión de giro de Hanle muestra tiempos de relajación de giro de hasta 4 ns y longitudes de relajación de giro superiores a 6 µm (Fig. 5h). El transporte de espín en el fósforo negro está estrechamente relacionado con la concentración del portador de carga, por lo que la señal de espín se puede controlar aplicando un campo eléctrico.

Manipulación de giros

Darse cuenta de la manipulación del efecto es la clave para una funcionalización eficaz del dispositivo. La aplicación de un voltaje de puerta puede controlar la concentración de portadores en el material, que se puede utilizar para manipular las señales de giro [85, 86]. Se han investigado varios materiales 2D como canales de transporte de espín para realizar el ajuste de los parámetros de transporte de espín mediante la aplicación de un voltaje de puerta. Por ejemplo, el grafeno inducido por sesgo puede obtener una polarización de detección e inyección de espín de hasta el 100% en la heteroestructura ferromagnet / bicapa h-BN / grafeno / h-BN [64]. A gate-tunable spin valve based on black phosphorus can reach a spin relaxation time in the nanosecond range and a long spin relaxation length [84]. For a semiconducting MoS₂ channel, applying a gate voltage can still get a relatively long spin-diffusion length, larger than 200 nm [70]. However, a suitable spin field-effect device requires a clear switching ratio, which is a challenge for graphene and even for semiconducting 2D materials [87, 88].

To solve this issue, a vdW heterostructure based on atomically thin graphene and semiconducting MoS₂ has been developed to achieve a spin field-effect switch via applying a gate voltage (Fig. 6a) [89]. In this structure, the superior spin transport properties of graphene and the strong SOC of MoS₂ are combined. The applied gate voltage can change the conductivity of MoS₂ and spin absorption during the spin transport, which results in switching of the spin current between ON and OFF states in the graphene channel (Fig. 6b). Another research effort produced a similar report about the graphene/MoS₂ vdW heterostructure. In this report, an electric gate control of the spin current and spin lifetime in the graphene/MoS₂ heterostructure was achieved at room temperature [90]. Moreover, that report pointed out that the mechanism of gate tunable spin parameters stemmed from gate tuning of the Schottky barrier at the MoS₂ /graphene interface and MoS₂ channel conductivity.

source contact. b Effective B _SO en función de V _g . The inset shows the surface carrier distribution in the Cr-TI layer under V _g  = − 10 V, + 3 V, and + 10 V. a - c Reproduced with permission from Yan et al., Nat. Commun. 7, 1 (2016). Copyright 2016 Nature Publishing Group [89]. d , e Reproduced with permission from Fan et al., Nat. Mater. 13, 699 (2014). Copyright 2014 Nature Publishing Group [95]. f , g Reproduced with permission from Fan et al., Nat. Nanotechnol. 11, 352 (2016). Copyright 2016 Nature Publishing Group [96]

Spin manipulation. un Schematic illustration of a 2D spin field-effect switch based on a vdW heterostructure of graphene/MoS₂ with a typical nonlocal magnetoresistance measurement. b The nonlocal resistance R _nl switches between R _P y R _AP for parallel and antiparallel magnetization orientations of the Co electrodes. The spin signal is calculated as ΔR _nl = R _P - R _AP . c The plot with blue circles shows the gate modulation of the spin signal ΔR _nl . The solid black line represents the sheet conductivity of the MoS₂ en función de V _g . The insets show the spin current path in the OFF and ON states of MoS₂ . d Schematic illustration of SOT-induced magnetization switching in a Cr-doped TI bilayer heterostructure. The inset shows illustrations of the Hall bar device and the measurement setting. e Experimental results of SOT-induced magnetization switching by an in-plane direct current at 1.9 K while applying a constant in-plane external magnetic field B _años durante la medición. The inset shows an enlarged version of the circled part in the figure. f 3D schematic of the Hall bar structure of the Al₂ O ₃ /Cr-TI/GaAs stack with a top Au gate electrode. A gate voltage of V _g can be applied between the top gate and the

Current-induced SOT is regarded as another efficient strategy to manipulate spin. The spin current, generated by the spin Hall effect within the heavy metals or the Rashba effect at the interfaces, can exert a spin torque to ferromagnets and thereby realize magnetization switching [91,92,93]. Efficient current-induced magnetization switching via SOT may lead to innovative spintronic applications. Due to strong SOC and time-inversion symmetry breaking, magnetically doped TIs are being considered as a promising material to manipulate spin signals via SOT [94]. Wang’s group first experimentally demonstrated a magnetization switching induced by an in-plane current in an epitaxial Cr-doped TI (Bi_0.5 Sb_0. 5)₂ Te ₃ /(Cr_0.08 Bi_0.54 Sb_0.38 ) ₂ Te ₃ bilayer film (Fig. 6c) [95]. The spin Hall angle in the Cr-doped TI film, ranging from 140 to 425, is almost three orders of magnitude larger than that in heavy metal/ferromagnetic heterostructures, and the critical switching current density is below 8.9 × 104 A cm ⁻² at 1.9 K (Fig. 6d). Furthermore, this team also reported an effective electric field control of SOT in a Cr-doped (Bi_0.5 Sb_0.42 ) ₂ Te ₃ thin film epitaxially grown on GaAs substrate (Fig. 6e) [96]. The gate effect on the magnetization switching was investigated by scanning gate voltage under a constant current and an applied in-plane magnetic field in the film (Fig. 6f). The SOT intensity depends strongly on the spin-polarized surface current in the thin film, and it can be modulated within a suitable gate voltage range. The effective electric field control of SOT in the TI-based magnetic structures has potential applications in magnetic memory and logic devices.

In addition, electrical control of emerged 2D magnets has also been investigated. For example, utilizing electric fields or electrostatic doping can achieve the magnetic conversion of bilayer CrI₃ antiferromagnetic to ferromagnetic [97]. The coercivity and saturation field of few-layer Cr₂ Ge ₂ Te₆ can be modulated via ionic liquid gating [98]. In contrast to magnetic semiconductor, electrostatic doping can be used to control the carrier concentrations of the ferromagnetic metal, and the ferromagnetic transition temperature of Fe₃ GeTe₂ can be dramatically raised to room temperature via an ionic gate [99]. The emergence and research of 2D magnets provide a new platform for engineering next-generation 2D spintronic devices.

Applications of 2D Spintronics

2D materials exhibit great potential for the engineering of next-generation 2D spintronic devices. Graphene with high electron/hole mobility, long spin lifetimes, and long diffusion lengths is a promising candidate for a spin channel. Moreover, graphene can gain magnetism by introducing adatoms, or magnetic proximity effect [23, 25]. The carrier density in proximity-induced ferromagnetic graphene can be modulated by gating, allowing to observe Fermi energy dependence of the anomalous Hall effect conductivity. This result can help understand the physical origin of anomalous Hall effect in 2D Dirac fermion systems. Realizing a ferromagnetic surface state in a TI is predicted to allow several prominent phenomena to emerge, such as the interfacial magnetoelectric effect [33], and the electric field-induced image magnetic monopole [34]. However, the current technology of inducing magnetism in TI is confined to low temperatures, which restrict its potential for applications. A key requirement for useful applications is the generation of room temperature ferromagnetism in the TI. The PNR result shows that the Bi₂ Se₃ /EuS bilayer has a ferromagnetic order at the interface, and this topologically enhanced interfacial ferromagnetism can persist up to room temperature [32]. The topological magnetoelectric response in such an engineered TI could allow efficient manipulation of the magnetization dynamics by an electric field, providing an energy-efficient topological control mechanism for future spin-based technologies.

The STT, and tunnel magnetoresistance (TMR) effects offer alternative approaches for write and read-out operations. The STT effect refers to the reorientation of the magnetization of ferromagnetic materials via the transfer of spin angular momenta. Efficient current-induced magnetization switching via SOT may lead to innovative spintronic applications [71, 100]. Due to strong SOC and time-inversion symmetry breaking, magnetically doped TIs are being considered as a promising material to manipulate spin signals via SOT [93]. TMR refers to magnetization-dependent magnetoresistance behavior. A high TMR ratio is the key to achieve spintronic devices with higher sensitivity, lower energy consumption. 2D materials with high-quality crystal and sharp interfaces can achieve some new functionalities such as spin filtering. The 2D vdW MTJ consists of a 2D magnetic CrI₃ layer as a spin filtering tunnel barrier, which reaches a value of TMR up 19,000% [101]. Progress in the fabrication of graphene-based and other 2D heterostructures has led to the optimization of long-distance spin diffusion (up to tens of micrometres), as well as directional guiding of the spin current [13, 64]. Spin manipulation, electrical gating [56], electrical field induced drift [80], SOT-induced switching [95, 96], and the magnetic proximity effect [25, 32] have been explored to develop next-generation MRAM.

2D Spintronic Devices for Memory Storage and Logic Applications

Great efforts have been made to search for new 2D spintronic devices. According to the function, 2D spintronic devices can be classified as memory storage or logic devices. Here we focus on several important 2D spintronic devices, including the 2D magnetic tunnel junction (MTJ), 2D spin field-effect transistor (sFET), and 2D spin logic gate.

2D MTJ

The discovery of the GMR opens the door for 2D spintronics. However, TMR has a stronger magnetoresistance ratio than GMR, so TMR holds greater potential in magnetic storage applications. The TMR structure consists of two ferromagnetic layers and an intermediate insulating layer, which is called the MTJ. The tunneling probability is related to the density of states near the Fermi energy in the ferromagnetic layers. When the two magnetic layers are parallel, the similar density of states for each spin-state can provide more available states for tunneling, resulting in a low resistance state. On the other hand, when the layers are antiparallel, a mismatch between spin channels of the source and sink will result in a high resistance state. Some issues in traditional thin-film MTJs limit the achievement of a high TMR ratio, such as the quality of the insulation barrier and the thermal stability [102]. 2D materials with high-quality crystal and sharp interfaces may offer promising routes to address these issues and even achieve some new functionalities such as spin filtering.

Karpan et al. first explored graphene layers as the barrier in vertical MTJ by computational means in 2007 [103]. They proposed a match between the band structure of graphene and that of nickel, predicting a large spin polarization close to 100%, which can result in a large TMR up to 500%. However, the subsequent experimental results show that the MTJs based on graphene exhibit a very low TMR. Compared to monolayer or bilayer graphene, the few-layer MTJ holds the highest recorded TMR signal of up to 31% in graphene-based MTJs [11, 15]. In addition to graphene, some other 2D materials have been explored as tunneling barrier layers, including insulating h-BN and semiconducting TMDCs [104, 105]. Piquemal-Banci et al. [63] fabricated Fe/h-BN/Co junctions where the h-BN monolayer was directly grown on Fe by using the chemical vapor deposition (CVD) method, observing large spin signals of TMR and the spin polarization of P  ~ 17%. MTJs based on MoS₂ or WSe₂ were reported to have only a few percent of the TMR signal; further exploration is needed to achieve a high TMR ratio.

Emerging 2D magnetic materials exhibit many surprising properties. When the magnetizations in bilayer CrI₃ are switched to different magnetic configurations (Fig. 7a), the MTJ based on CrI₃ exhibits a giant TMR produced by the spin-filtering effect [101, 106, 107]. As demonstrated in Fig. 7b, the 2D vdW MTJ consists of a 2D magnetic CrI₃ layer as a spin filtering tunnel barrier, graphene as a contact electrode, and h-BN as an encapsulation layer to prevent device degradation. The transport result shows that the TMR is enhanced as the CrI₃ layer thickness increases, and it reaches a value of 19,000% in four-layer CrI₃ based MTJ at low temperature (Fig. 7c) [101]. Subsequently, Xu’s group also reported gate-tunable TMR in a dual-gated MTJ structure based on four-layer CrI₃ . The TMR can be modulated from 17,000 to 57,000% by varying the gate voltages in a fixed magnetic field [108, 109]. Moreover, with few-layer Fe₃ GeTe₂ serving as ferromagnetic electrodes, the TMR in Fe₃ GeTe₂ /h-BN/Fe₃ GeTe₂ heterostructures can reach 160% at low temperature [110]. More interestingly, Zhou et al. reported a theoretical investigation of a VSe₂ / MoS ₂ /VSe₂ heterojunction, where the VSe₂ monolayer acts as a room-temperature ferromagnet, and the large TMR can reach 846% at 300 K [111]. On the other hand, the strong spin Hall conductivity of MoS₂ holds potential to switch the magnetization of the VSe₂ free layer by SOT. Therefore, they put forward the concept of SOT vdW MTJ with faster reading and writing operations, which offers new opportunities for 2D spintronic devices.

Reproduced with permission from Song et al., Science (2018). Copyright 2018 The American Association for the Advancement of Science [101]. c , d Reproduced with permission from Wen et al., Phys. Rev. Appl. 5, 044003 (2016). Copyright 2016 American Chemical Society [118]

2D spintronic Devices. un Magnetic states of bilayer CrI₃ with different external magnetic fields. b Schematic illustration of a 2D spin-filter MTJ with bilayer CrI₃ sandwiched between graphene contact. c Tunneling current of a bilayer CrI₃ sf-MTJ at selected magnetic fields. The top inset shows an optical image of the device, and the bottom inset shows the schematic of the magnetic configuration in different magnetic fields. d Diagram of a proposed 2D XOR spin logic gate, where A, B, and M are ferromagnetic electrodes on top of a spin transport channel. yo _s y yo _fuera denote the injection and detection currents, respectively. The magnetizations of the electrodes are input logic 1 and 0. The detected current I _fuera serves as the logic output. e yo _fuera measured as a function of H . Vertical arrows indicate the magnetization states of A and B. The top-left inset shows the table of XOR logic operation. un , b

2D sFET

Datta and Das first proposed the idea of the sFET in 1990 [112]. The sFET consists of the source and sink ferromagnetic electrodes, and a two-dimensional electron gas (2DEG) channel which can be controlled by an electrical gate. The gate voltage can result in a spin precession and, consequently, a change in the spin polarization of the current on the channel. Since switching the current through the device requires only little energy and a short time, sFET is expected to be a 2D spintronic device with low power consumption and high computing speed.

As mentioned in the previous section, graphene with high carrier concentration and weak SOC is considered to be a promising candidate as a spin transport channel [113]. Michetti et al. [76] designed a double-gate field-effect transistor structure, where bilayer graphene acts as the transport channel. Theoretical analysis shows that the spin precession of carriers in the graphene channel can be turned on and off by the application of a differential gate voltage. Experimentally, Avsar et al. first reported a dual-gated bilayer graphene structure with h-BN as a dielectric layer, where the spin current propagation in bilayer graphene channel can be controlled by exerting a vertical electric field [114]. The transport results show that the spin-relaxation time decreases monotonically as the carrier concentration increases, and the spin signal exhibits a rapid decrease, eventually becoming undetectable close to the charge neutrality point. A suitable spin field-effect device requires a clear switching ratio, which is a challenge for graphene.

To solve this issue, a graphene/MoS₂ heterostructure has been developed to achieve a spin field-effect switch via applying a gate voltage. Two independent groups demonstrated that the applied gate voltage can change the conductivity of MoS₂ and result in spin absorption during the spin transport, which gives rise to switching the spin current between ON and OFF states in the graphene channel [89]. Due to the low spin injection efficiency and rapid spin relaxation in channels, it is a challenge to achieve a large high-to-low conductance ratio in 2D sFET device. However, the discovery of 2D magnetic crystals provides new opportunities to explore new 2D spintronic devices. Kin Fai Mak’s group reported a spin tunnel field-effect transistor (sTFET) based on a dual-gated graphene/CrI₃ /graphene heterostructure [115]. By using bilayer CrI₃ as a magnetic tunnel barrier, the applied gate voltage can switch magnetization configurations of bilayer CrI₃ from interlayer antiferromagnetic to ferromagnetic states under a constant magnetic field near the spin-flip transition. Distinct from conventional spin transistors, these devices rely on electrically controlling the magnetization configuration switching rather than the signal of spin current in the channel. This technique allows the sTFET devices to achieve a large high–low conductance ratio approaching 400%, which provides a new approach for exploring memory applications.

2D Spin Logic Gate

Dery and Sham first reported a spin logic device based on an “exclusive or” (XOR) gate [116]. The XOR logic gate structure includes a semiconductor channel and three ferromagnetic terminals. An XOR logic operation can be implemented by different spin accumulations, which is caused by different magnetization configurations of the input terminals [117]. Experimentally, the proposed three-terminal XOR logic gate achieved logical operations in a graphene spintronic device at room temperature [117,118,119]. As shown in Fig. 7c, the device includes single-layer graphene as the channel, and three ferromagnetic terminals composed of A, B, and M Co electrodes with MgO tunnel barriers. The magnetization of the electrodes A and B represents the input states 0 or 1, and the current of the electrode M acts as the output state. The magnetizations of input electrodes A and B will be switched by varying an applied external magnetic field, which results in a different spin accumulation in the M electrode, corresponding to a different output current. If the input A and B electrodes have identical contributions to the output M electrode, then the current in the output ferromagnetic terminal has a detectable value only when the magnetization of input ferromagnetic terminals are antiparallel (01 or 10). When the magnetizations of the input ferromagnetic terminals are parallel (00 or 11), the output current is almost zero. Therefore, the XOR logic operation can be achieved (Fig. 7d).

Dery et al. further designed a reconfigurable magnetologic gate with five-terminal structure combining two XOR gates-XOR (A, X) and XOR (B, Y) with a shared output terminal, M [119]. Similar to the three-terminal XOR logic gate, the different magnetic configurations of input electrodes give rise to the different spin accumulation in the output terminal M, which results in different output signals. By analogy, a finite number of these XOR gates can be used to implement any binary logic function. Subsequently, other groups extended this theoretical design to experimental studies by constructing graphene spin logic gates [120,121,122]. Various modeling, simulation, and experimental explorations of 2D spin logic gates have helped to accelerate the progress toward building practical spin logic applications. However, two key issues remain in the research of graphene spin logic gates. The first one is to balance the contributions of two input terminals to the output one. The other one is to eliminate the influence of background signals on the output.

Challenges and Opportunities in 2D Spintronics

As discussed above, much theoretical and experimental research has been carried out to explore spintronics based on 2D materials, and considerable progress has been achieved [15, 123, 124]. However, great challenges still need to be addressed for the practical application of 2D memory and logic applications. We now discuss three of these:physical mechanisms, materials science, and device engineering.

Physical Mechanisms

Due to the complexity of the experiments, the proposed theoretical research and experimental results usually have large discrepancies. For example, based on the mechanism of spin relaxation, theory predicted that the spin lifetime for pristine graphene would be up to 1 μs, whereas experimental values range from tens of picoseconds to a few nanoseconds [14, 57, 103]. Furthermore, the spin injection efficiency of graphene measured experimentally ranges from a few percent to 10%, which is far smaller than the theoretical prediction value of 60–80% [125]. These differences indicate that more in-depth physical mechanisms and accurate theoretical models need to be proposed and developed to better guide the research direction and analyze the experimental results.

Materials Science

2D materials provide an ideal platform to construct various heterostructures for spintronic applications. However, there are still many major problems in 2D materials. For example, stability is a great challenge for 2D materials. Most 2D materials of thickness close to the atomic level are susceptible to moisture, oxygenation, and temperature, especially the recently emerging 2D magnetic materials, which must be peeled off in a glove box with ultra-low water and oxygen content. Besides this, most currently available 2D magnets rely on mechanical exfoliation, and their low magnetic transition temperature is far below room temperature. These are significant limitations because stability in air, convenient wafer-scale synthesis, and operation above room temperature are prerequisites for 2D materials used in practical applications.

Device Engineering

Breakthroughs have been made in the fundamental study of 2D spintronics, such as enhanced spin injection efficiency by using 2D tunnel barriers h-BN, long spin diffusion length up to 30 μm at room temperature in graphene-based 2D heterostructures [13], and high TMR up to 19,000% by using 2D magnets as spin filter barriers [101]. Based on the study of 2D spintronic devices, it is promising to develop the low-power device applications, including advanced magnetic memories and spin logic circuits, which are compatible with the existing complementary metal-oxide semiconductor (CMOS) electronics. However, the design and application of functional 2D spintronic devices are still in the early theoretical prediction and proof-of-concept stage.

Opportunities

2D spintronics is an important scientific research field with many potential applications for future technologies. As mentioned above, considerable challenges currently remain, but there are also many opportunities. Spin valves based on graphene as the spin transport channel can exhibit a long spin diffusion length up to 30 mm at room temperature [13]. Magnetic tunnel junctions with four-layer CrI₃ as spin filter tunneling barriers show giant TMR up to 19,000% at low temperatures [101]. The magnetic transition temperature of Fe₃ GeTe₂ can reach above room temperature via an ionic liquid gate or when tailored by a TI [99, 126]. Spin-polarized current can be injected from WTe₂ into magnetic substrates by SOT switching [127]. New concepts of spin tunneling field-effect transistors based on 2D magnets CrI₃ have been proposed as well. The demonstration of giant TMR, the efficient voltage control of 2D magnetism, and the magnetization switching in 2D magnets by STT or SOT all open up opportunities for potential next-generation spintronic devices based on atomically thin vdW crystals [21, 100].

Conclusion

The study about the magnetic properties of 2D materials is of great significance to the development of 2D spintronics. The magnetic interaction in graphene and TIs has scarcely been explored, and recently discovered 2D magnets also provide an ideal platform to study 2D magnetism. Great progress has been made in 2D spintronics in recent decades, especially in graphene spintronics. However, the origin of spin relaxation in graphene is still a major open question, and further improvement in the spin lifetime and spin diffusion length remains an important research direction for graphene-based spintronic devices. The practical application of 2D spintronic devices still requires meeting great challenges, including related physical mechanisms, materials science, and device engineering. The development of technology, the improvement of theoretical models, and the exploration of new materials all provide more opportunities for new-generation 2D spintronic device applications in the future.

Disponibilidad de datos y materiales

No aplica.

Abreviaturas

2D:

Bidimensional

GMR:

Giant magnetoresistance effect

STT-MRAM:

Spin-transfer-torque magnetoresistive random-access memory

vdW:

Van der Waals

SOC:

Spin–orbit coupling

TMDCs:

Dicalcogenuros de metales de transición

TIs:

Topological insulators

SQUID:

Dispositivo superconductor de interferencia cuántica

YIG:

Yttrium iron garnet

QAHE:

Quantum anomalous Hall effect

PNR:

Spin-polarized neutron reflectivity

MOKE:

Magneto-optical Kerr effect

MBE:

Epitaxia de haz molecular

h-BN:

Nitruro de boro hexagonal

SOT:

Spin–orbit torque

ST-FMR:

Spin-torque ferromagnetic resonance

MTJ:

Magnetic tunnel junction

sFET:

Spin field-effect transistor

TMR:

Tunneling magnetoresistance

CVD:

Deposición de vapor químico

2DEG:

Gas de electrones bidimensionales

sTFET:

Spin tunnel field-effect transistor

XOR:

Exclusive or

CMOS:

Complementary metal-oxide semiconductor

Aplicación multifuncional de nanocompuesto de óxido acoplado de Zn – Fe – Mn con ayuda de PVA Crecimiento y grabado selectivo de estructuras de múltiples capas de silicio / silicio-germanio dopado con fósforo para transistores verticales Aplicación