Progress and Prospects in Metallic FexGeTe2 (3≤x≤7) Ferromagnet

1. Introduction

The Mermin-Wanger theory [1,2] asserted that thermal fluctuations occurred in 2D isotropy systems at non zero finite temperatures, which would destroy the long-range (LR) magnetic order. Specifically, only exchange interactions should not generate magnetic order in 2D systems, and magnetic anisotropy [3,4,5] was also needed. Surprisingly, it was found experimentally that low-temperature LR ferromagnetic order could exist in monolayer Cr₂Ge₂Te₆ [4] and CrI₃ [5,6]. Subsequently, a vast range of 2D magnetic systems, including metallic (Fe₃GeTe₂ (FGT) [7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23]), semiconductors (Cr₂Ge₂Te₆ [4,24,25,26,27,28,29,30,31,32], CrI₃ [5,33]), and topological insulators (MnBi₂Te₄ [34]), were successively achieved. Recently, Fe_xGeTe₂ (3≤x≤7) has received intense attentions as a metallic and high curie temperature (T_C) ferromagnet. Importantly, six synthesis methods including solid-state reaction (SSR) [35,36], chemical vapor transport (CVT) [8,13,37], the flux method [11,21,38,39,40,41,42,43,44,45], exfoliation [14,15,34,46,47,48,49,50], chemical vapor deposition (CVD) [51,52] and molecular beam epitaxy (MBE) [7,53,54,55,56,57,58], have been used to attempt to obtain wafer-scale Fe_xGeTe₂ (3≤x≤7) materials with room-temperature ferromagnetism (RTFM). More interestingly, RTFM has been mediated with ten strategies, such as Fe stoichiometry [9,39,51,59,60,61,62,63,64], strain engineering [46,48,65,66,67,68,69,70,71,72,73,74], hydrostatic pressure [75,76,77,78,79,80], light-control [53,81], electrical-control [82,83], proximity effects [56,57,84,85,86,87,88], doping engineering [14,20,38,43,44,60,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105], intercalation [106,107] or irradiation [108], twisting [109,110] and patterning [16]. Moreover, three devices have also been fabricated based on FGT, mainly including magnetic tunnel junctions (MTJ) [111,112,113,114], tunneling spin valves [18,98,115,116,117], and spin−orbit torque devices [20,118], to enrich their physical properties and develop their spintronic applications.

In this review, we introduce the developments and structures of metallic Fe_xGeTe₂ ferromagnet. Subsequently, we have summarized six experimental methods in Figure 1, and the early samples prepared through SSR, CVT and flux method were mainly bulk single crystals. Last, we have outlined the challenges and potential research directions in this field.

2. Crystal Structure and Band Structure of Ferromagnetic Fe_xGeTe₂

The Fe₃GeTe₂ monolayer [46] included five atomic layers in Figure 2(A). In details, Te atoms were located at the bottom and top layers, while Fe (I) atoms were located in the second and fourth layers. Notably, the intermediate layer was composed of Fe (II) atoms and Ge atoms. In particular, the local magnetic moments of Fe atoms for FGT monolayer were 1.723 µ_B and 1.005 µ_B through density functional theory-local density approximation (DFT-LDA) calculation, and may be related to the several partially d-band occupied passing through the Fermi level. Similar to its bulk, FGT monolayer was metallic as shown in Figure 2(B). Its band structures near the Fermi level were mainly due to the contribution of the Fe 3d orbitals. Moreover, it could be confirmed that the FGT monolayer had the itinerant FM order according to Stoner’s criterion [46,119]. More importantly, the Stoner model related to itinerant electrons could be used to better elucidate the spontaneous magnetization in most of 2D metallic ferromagnets. In addition, the number ratio of Fe³⁺ to Fe²⁺ in 2D Fe_xGeTe₂ [62] was related to the x value, as shown in Figure 2(C-D). When the value of x was 3, the ratio of Fe³⁺/Fe²⁺ was 2:1; However, when the x value was 5, only Fe³⁺ was present. As shown in Figure 2(E), the electronic band structures of all the Fe_xGeTe₂ systems were metallic, be similar to 2D FGT.

3. Synthesis of Metallic Fe_xGeTe₂ with FM

3.1. Solid-State Reaction (SSR)

Solid-state reaction (SSR) is an experimental method for preparing bulk FGT crystals. As early as 2006, Deiseroth et al. [35] successfully prepared FGT crystals with the hexagonal plates using SSR, which exhibited new air-stability and black-metallic. Through magnetic testing, it was found that below 230 K, it exhibited FM; while above 230 K, it exhibited Curie-Weiss paramagnetism behavior. After increasing annealing, black Fe_3-δGeTe₂ (0<δ<0.3) polycrystalline powders could be easily obtain with SSR. In details, Fe1 and Ge1 atoms are in different coordination environments Figure 3A-B, and two layers consist of its unit cell. In details, its lattice parameters increase monotonically with decreasing δ (Figure 3C). But when δ exceeds 0.3, FeTe₂ as an impurity phase will appear. Its magnetic phase transition temperature (Figure 3D) is about 240 K. Furthermore, its saturation behavior (inset in Figure 3D) slows down under high magnetic fields, which is different from ordinary ferromagnets.

In order to obtain large quantity and high-quality FGT single crystal , Li et al. [120] designed a new experimental method of solid-phase sintering followed by recrystallization (Figure 3E-H). As-grown plate-like sample (~10g) is a layered single crystal with a smooth and complete surface, and its size can reach up to 8.5 mm. By intercalating sodium into as-grown FGT (Figure 3I-J), Weber et al. [106] could raise its T_C to 350 K. After intercalation, the sample retained obvious layered features, with edge lengths of grain size ranging from 10-50 μm.

3.2. Chemical Vapor Transort (CVT)

One main difference from SSR was that CVT often used iodine [8,13,14,15,17,37,39,59,118] or TeCl₄ [12] as the transport agent in Figure 4. However, the samples obtained with SSR and CVT were both bulk single crystals in the inset of Figure 4A-B. Previous studies mainly focused on the magnetic microstructures of quasi-2D FGT. Inspired by the prediction that monolayer FGT could be mechanically exfoliated [46], soon after, Chu et al. [15] and Zhang et al. [14] respectively obtained monolayer FGT samples with the assistance of Au film or Al₂O₃. Interestingly, Zhang et al. [121] exfoliated monolayer FGT from the most possible cleaving plane (001), with a thickness of 1.75 nm and the nearest neighbor atomic spacing of 0.338nm, which was very consistent with the lattice constant (a= 0.399nm; c=1.63nm) of FGT crystal. However, thin layer FGT was very prone to deteriorate in air, and the device fabrication processes (Figure 4C-E) needed to be carried out in a glove-box [49]. Importantly, many novel physics effects such as patterning-induced [16], gate-tunable [14], and layer-dependent FM [15] have been recognized.

3.3. Flux Growth

Flux method [122,123,124] is a commonly used method for preparing single crystals. For example, Canfield et al [122,123] grew a wide variety of single crystal binary or ternary intermetallic compounds from molten flux solutions. However, the thickness and lateral size of the samples couldnot be accurately controlled, which still required mechanically exfoliating to obtain thinner samples when fabricating FGT devices.

Interestingly, Gong et al. [44,45] proposed a universal flux-assisted growth (FAG) method (Figure 5D-E) to synthesize the Fe_xGeTe₂ and M_yFe_5-yGeTe₂ (M = Co, Ni) nanosheets on various substrates (Figure 5F-L). More importantly, the sample thickness and lateral size (Figure 5M-Q) of FGT could be precisely controlled by growth temperature (Figure 5R) or cosolvents (Figure 5S). Although FGT samples with a thickness of 5-10 nm (Figure 5T) were prepared on various substrates, in order to obtain atomically thin materials (ATMs), a confinement environment must be provided through two substrates. Up to 80 layered and non-layered ATMs [45] have also been successfully synthesized by FAG, which provided a new strategy for preparing wafer-scale 2D materials.

3.4. Exfoliation

3.4.1. Mechanical Exfoliation

The conventional mechanical exfoliation [126,127] could cleave thin FGT flakes onto SiO₂/Si substrates, but its thinnest thickness was around 4.8nm. After depositing Au onto SiO₂/Si substrate, thinner samples could be obtained, and the Au substrate would improve the yield of various thin layers of materials, including graphene [128,129], MoS₂ [47,130,131,132,133], WSe₂ [47,130,134], Bi₂Te₃ [130] and FGT [15,47]. Nevertheless, only a small amount of material could be exfoliated to a monolayer, which hindered the development of 2D magnetic materials. Notably, an Al₂O₃-assisted exfoliation was also designed to produce monolayer FGT [14] and MnBi₂Te₄ [34] single crystals. More importantly, when the sample was thinned from bulk to a monolayer, its T_C would decrease from 180 K to 20 K.

3.4.2. Liquid-Phase Exfoliation

Although many methods including SSR [106], CVT [14,15,17,19,59,118,121,135], flux [39,43,44,45] and MBE [7,53,54,55,56,57,58,64,136], have been used to prepare 2D FGT, there was still a lack of an economical method to prepare large-scale of few or single layer FGT nanoflakes. As a typical example, Ma et al. [50] developed three-stage sonication-assisted liquid-phase exfoliation (TS-LPE) (Figure 6A) to achieve large-size semiconductive FGT nanoflakes. After ball milling, the sample size and thickness (Figure 6B-C) will be reduced by milling time (Figure 6G), exposing more boundaries. Stirring would make the interlayer spacing expand (Figure 6C-D), weakening the interlayer force to facilitate detachment and obtain high-integrity nanoflakes. In addition, XRD analysis [137] (Figure 6H-I) also reflected the evolution of interlayer spacing. The expansion of interlayer spacing would cause FGT unit cell to move away from the equilibrium state in the c-direction (Figure 6J), making them unstable and prone to spall. More interestingly, the oxidation on the surface layer altered the electronic structure of FGT system, making FGT sample semiconductive and different from the metallic FGT prepared by other methods.

3.5. Chemical Vapor Deposition (CVD)

So far, researchers mainly used CVT to prepare 2D magnetic bulk single crystals, which were then exfoliated into atomic layers to prepare devices. However, poor control of the number of layers and limited sample size have hindered the development of 2D magnets. As a typical example, Liu et al. [51] designed a confined space chemical vapor deposition (CS-CVD) to prepare 2D FGT or F5GT ferromagnets. They found that the optimal growth temperature was 570-580℃, with an optimal distance of 10 cm between Si/SiO₂ substrates and Te precursor. When the thickness of F5GT flakes changed from 4 nm to 1 nm, its T_C would decrease by 100K. Very recently, Liu et al [52] also introduced a general competitive-chemical-reaction-controlled CVD method for producing FGT crystals. The sample was a single layer with a grain size of ~50 μm.

3.6. Molecular Beam Epitaxy (MBE)

Notably, wafer-scale single crystalline FGT thin films (Figure 7) were grown on (0001) sapphire, (111) GaAs and (111) Ge substrates by the molecular beam epitaxial (MBE) [7,54,55,56,57,138] technique. Interestingly, in situ reflection high-energy electron diffraction (RHEED) exhibits the same periodicity (Figure 7B), and can complete one layer of growth in approximately 111 seconds. Similary, the same 2D growth mode (Figure 7H, I) was also observed during the FGT and FGT|Bi₂Te₃ grown. Furthermore, the layered characteristics and a interlayer spacing of 0.82 nm could be confirmed through cross-section high-resolution transmission electron microscopy (HRTEM) (Figure 7D-F, O and P). In addition, Fe, Ge, and Te elements are uniformly distributed on the cross-section, and their atomic ratios can be determined by X-ray spectroscopy (EDX) mapping (Figure 7P).

4. Controlling FM in metallic Fe_xGeTe₂ (3≤x≤7)

4.1. Fe stoichiometry

The earliest discovery was that FM in polycrystalline FGT bulk [9] was related to Fe content (Figure 8). The higher the Fe content, the larger the lattice constant of the a-axis and the smaller the lattice constant of the c-axis (Figure 8A). Single crystal samples also have similar results with polycrystalline samples. More interestingly, the T_C (Figure 8B) and M_S decreased with the decrease of Fe content. Subsequently, ferromagnetic F4GT [44,59,64] and F5GT [39,42,44,51,60,139,140,141,142,143,144,145,146] materials were also obtained in experiments.

However, most of previous reports have focused on FGT materials with a single Fe stoichiometry, and there have been few studies on Fe_xGeTe₂ materials using the same experimental method. Interestingly, theoretical calculations [61] revealed that as the Fe content increased, the interlayer gap gradually increased, and the magnetic anisotropy of its monolayer changes from out-of-plane (FGT) to in-plane (F4GT and F5GT).

Although previous reports have made significant progress in 2D Fe_xGeTe₂ system, the mediation of magnetic anisotropy and magnetic nature remain unresolved. Very recently, Liu et al.[62] brought forward a valence-dependent magnetic exchange model to explain the complex magnetic phase in Fe_xGeTe₂ system. Furthermore, the magnetic moment and MAE (Figure 9A) were almost linearly correlated with Fe²⁺/Fe³⁺. Specially, Fe³⁺ had a greater impact on magnetism, reducing the magnetic anisotropy energy in F5GT. Based on MAE and J, the T_C could be estimated with the 2D Heisenberg model. When x was greater than 4, the T_C was much higher than RT (Figure 9B).

More importantly, the results obtained through different calculation methods [63] have significant differences, especially when compared with experimental results. As x increased, its easy axis direction (shown in the black arrows) and the highest exchange interaction changed from off-plane to in-plane (Figure 9C-F). Moreover, there were significant differences in the magnitude of exchange interaction obtained by different calculation methods (GGA+DMFT, GGA and GGA+U), among which the results calculated through GGA+U were overestimated in Figure 9G-J. Interestingly, MAE (Figure 9K-M) also exhibits a similar evolution from off-plane to in-plane. However, for different F5GT (UUU or UDU) configurations (Figure 9M-N), the calculation results varied greatly. It was very obvious that the T_C calculated by GGA+DMFT (Figure 9O-R) was underestimated, while the result calculated by GGA was overestimated, compared to Figure 9B.

. Reprinted with permission from Ref.[63], Copyright 2023, Springer Nature.

4.2. Strain Engineering

Strain engineering is indeed an efficient stragety to modulate the FM of 2D materials [66,67,147]. However, previous theoretical works have focused on applying strain to FGT supercells by changing lattice constants [46,69,70,72,148], and calculating the exchange coupling, magnetic anisotropy, and magnetic moment of strain through ab initio density functional theory (DFT). Furthermore, the T_C could be estimated by mean field theory (MFT) [10,59,63,149,150,151], random phase approximation (PRA) [149,151], or Monte Carlo (MC) [150,151,152,153] simulation. Very recently, Miao et al. [48] and Yan et al. [71] loaded FGT nanoflakes to a three-points bending experimental set-up, and applies uniaxial tensile strain to the sample on a polyimide (PI) or polyvinyl alcohol (PVA)/ polyethylene terephthalate (PET) flexible polymer substrate by moving needle. Moreover, the magnitude of the applied strain could be calculated using the following formula [48,71,154]:

$$\epsilon =\frac{T}{2R}$$

(1)

Note that, T and R were the film thickness and bending radius, respectively. Surprisingly, after 0.32% strain was applied, the coercivity increases by 150% [48], far greater than the improvement in H_C of other traditional magnetic materials [155]. More importantly, its T_C could be increased to 400 K [71] through uniaxial strain, which would further promote the development of its practical applications.

4.3. Hydrostatic Pressure

Actually, tuning the exchange coupling and magneto-crystalline anisotropy by hydrostatic pressure was also a commonly used method for regulating 2D magnetism, which had been achieved in Cr₂Gr₂Te₆ [156,157], CrI₃ [33,158], and FGT [76,77,78] systems (Figure 10). The ferromagnetic evolution of FGT nanosheets under different pressures can be revealed through in situ magnetic circular dichroism (MCD) spectroscopy (Figure 10B).

Furthermore, the magnetic hysteresis loop at 30 K exhibited a rectangular shape below 7 GPa, while its loop presented an 8-shaped skewed shape above 7.3 GPa (Figure 10C-D). Moreover, Tc will increase as the pressure further decreases (Figure 10E), which may be related to the strengthening of the exchange interactions. As another typical example, Tc and the magnetic moment also increased with the decrease of hydrostatic pressure (Figure 10F-I). It was evident that increasing pressure reduced the length of Fe-Fe bond, which inhibited magnetization through modification the exchange interactions. In addition, a monotonic relationship between Tc, the magnetic moment and pressure was also found in Fe-deficient FGT sample [76], similar to FGT system. Interestingly, pressure will modify the metallic form of Fe_3-xGeTe₂ to nonmetallic.

4.4. Light Control

As a typical example, continuous modulation of monolayer layered TMDs without intrinsic magnetism, including MoS₂ [159], WS₂ [159], and WSe₂ [160], has been achieved through the optical approach. Recently, Tengdin et al. [81] demonstrated that spin polarization was transferred from Mn sublattices to Co on Heusler compound Co₂MnGe via femtosecond laser pulse, which was closely related to the wave function of an electrons before and after being excited by light. The ultrafast spin transfer caused by the instantaneous incident light on the material not only occurs in Co₂MnGe, but also was a common feature of many materials. Notably, Xu et al. [53] reported that the magnetic anisotropy energy (MAE) and T_c were mediated with a femtosecond laser pulse in Figure 11. The optical doping effect alters the electronic structure of FGT, thereby affecting exchange interactions, T_c, and MAE. According to Figure 11(B), the T_c of FGT was estimated to be ~200 K. Under the excitation of femtosecond laser, electrons transitioned from occupied state to unoccupied state, causing the Fermi level E_F to shift downwards and crossing the enhanced density of states (DOS) in Figure 11D. Furthermore, some clear magnetic hysteresis loops (Figure 11E-G) at room temperature (RT) can be observed in FGT samples with different thicknesses through MOKE measurements (Figure 11C). The T_c of FGT could be increased to above RT through light control, providing many opportunities for the development of spintronic applications for 2D magnets.

4.5 Electrical Control

Previous studies have shown that electric fields modified the magnetism of metal films [161,162,163] and Fe/MgO junctions [164] by influencing the behavior of the electrons. Recently, Wang et al. [82] calculated the effect of electric field on the magnetic anisotropy of monolayer FGT in Figure 12A-C. The effect of orbital splitting caused by electron doping on magnetic anisotropy was more pronounced; while the influence of hole doping related to orbital occupation was relatively weak. In addition, the change in magnetic anisotropy (Figure 12C) was more obvious in the single-gate configuration (Figure 12B).

Additionally, the generation of negative differential conductance (NDC) [83]can also be driven by a local electric field in FGT (Figure 12D). Furthermore, the three peaks in the Fe d orbits underwent significant shifts under the electric field, as shown by the green line in Figure 12(E-G). As the electric field was enhanced, the off-plane FM of FGT weakened, resulting in a decrease in MAE (Figure 12H). Importantly, In single-layer FGT, the electric field induces charge transfer in monolayer FGT in the field direction (Figure 12I). Therefore, applying an electric field has become an effective way to mediate 2D FM.

4.6. Proximity Effects

Proximity effects [84,165,166,167,168,169] also dominated in 2D materials. For example, by adjacent 2D magnetic materials to bulk semiconductor substrate [170] or 2D materials with strong spin-orbit coupling [171], their magnetism can be enhanced. Intriguing, zhang et al. [84] fabricated the antiferromagnetic FePS₃(FPS)/ferromagnetic Fe₃GeTe₂(FGT) heterostructures and found the enhancement of Tc and H_C through proximity coupling effects. Furthermore, FPS/FGT/FPS has slightly different modulation of Hc compared to FPS/FGT, which was related to AFM-FM coupling. After integration with a topology insulator Bi₂Te₃ [56], the Tc of FGT could be increased to 400 K by enhancing the intralayer spin interaction, mainly due to the interface exchange coupling effect. More interestingly, the long-range magnetic order induced by topology triggered by femtosecond laser pulses [57] could also be maintained at room temperature.

In addition to the aforementioned methods for mediating magnetism, additional means [7,14,33,48,53,56,57,71,75,76,77,78,156,157,158,171] were required. Directly exfoliating FGT nanosheets [87] onto different substrates could also tailor magnetism as shown in Figure 13. Intriguingly, FGT samples with different thicknesses all exhibited magnetism, while samples on different substrates exhibited different T_C, indicating that the substrate has a modulation effect on T_C. Furthermore, the lattice distortion and charge redistribution at the interface were related to substrate-induced FM, and this mechanism needed further exploration. Importantly, substrates not only affected the growth of 2D materials, but also determined their performance [172,173].

4.7. Doping Engineering

4.7.1. Doping with 3d Transition-Metal

Doping 3d-transition metal atoms was an effective strategy for controlling magnetism [65,68,89,174,175,176,177,178]. Theoretical calculations shown that almost all 3d-transition metal atoms (except for Co atom) [96] were more inclined to replace Fe1 atoms (Figure 2A). The charge transfer generated by doping atoms would weaken the magnetic moment of Fe atoms, while the weakening effect of Fe1 atomic magnetic moment was more significant. However, the magnetism increases after doping with Co atoms, which may be related to the a-axis lattice constant shrinking. In experiments, doping 3d-transition metal atoms in bulk single crystal samples in Figure 14 was usually achieved through CVT [92] or self-flux [38]. Doping Ni atoms suppressed ferromagnetic order, which rapidly decreased with the increase of doping amount. The T_C decreased from 212 K to 50 K, and after reaching 0.44, the magnetic moment remained almost constant (Figure 14A-B). Furthermore, long-range magnetic order was suppressed and subsequently transformed into a glassy magnetic phase (Figure 14C). However, doping Co atoms (Figure 14D) could cause an increase in H_C (Figure 14E) and the appearance of hard magnetic phases (Figure 14F), which was related to the movement of pinned domain walls [8].

As another typical example, bulk F5GT single crystals was also doped with Co atoms through CVT [94,95] in Figure 15. As the doping amount of Co increases, it can drive the evolution of lattice and magnetism (Figure 15B). But the nominal doping concentration (Figure 15C) was slightly different from the measured one, with only a specific concentration being more consistent. After Co atoms were doped into the lattice, resulting in the slightly increase of their interlayer spacing (Figure 15D). Indeed, a phase transition from FM to AFM occurred at high doses of doping in Figure 15(E-G). Indeed, a phase transition from FM to AFM occurred at high doses of doping in Figure 15(E-G). However, Tian et al. [95] found that doping 20% Co could increase its T_C to 337 K, and induce complex magnetic phase transitions at higher Co doping levels. More importantly, 2D Co_yFe_5−yGeTe₂ (Figure 15I) and Ni_yFe_5−yGeTe₂ (Figure 15J) nanoflakes with hexagonal shape were prepared by flux-assisted growth (Figure 15H) [44]. As shown in Figure 15H-M, various elements in the nanosheets were evenly distributed through energy dispersive spectroscopy (EDS) elements mapping, and there were significant differences in the energy spectrum of samples doped with different Co, with a doping amount of up to 66.7%. Furthermore, the doping of Co atoms caused a decrease in T_C, and as the doping amount increased, its T_C would decrease even lower (Figure 15N). In addition, the magnetic anisotropy has also undergone significant changes (Figure 15O-P). Similarly, Ni doping could also cause a decrease in T_C. In other words, the higher the content of Fe, the higher its T_C.

4.7.2. Doping with Non-Metallic Atoms

Not only could Fe atoms be substituted with Co or Ni atoms [38,44,92,94,95,96], but doping could also be achieved by replacing Ge atoms with As atoms [91,97]. The doping of As atoms caused a decrease in the a-axis lattice constant and an increase in the c-axis lattice constant, thereby reducing the density of spin states below fermi level and resulting in a decrease in T_C [91]. Furthermore, its M_S has decreased linearly with the increase of doping amount in polycrystalline Fe_3-yGe_1-xAs_xTe₂ (0≤x≤0.85). Similarly, the expansion of F5GT unit cell [97] in the c-axis direction and the contraction in the ab plane were also observed after As atoms doping. In addition, its T_C and M_S have decreased in polycrystalline Fe₅Ge_1-yAs_yTe₂ (0≤y≤1), similar to Fe_3-yGe_1-xAs_xTe₂ (0≤x≤0.85) samples. Moreover, the stacking disorder caused by doping induced local AFM coupling, thereby reducing its M_S.

4.7.3. Electron Doping

Remarkably, Deng et al. [14] found that the T_C of monolayer FGT could be raised to RT through ionic gate, providing a new idea for mediating 2D FM. Although they did not fully explain the relationship between its ferromagnetism and electron doping, this strategy was fully recognized. Not long after, gate-control was implemented to regulate magnetic resistance [98], magnetic phase [60,100], and interlayer coupling [99,101,179]. Furthermore, the T_C and H_C in FGT flakes [100] were decreased after Li⁺ doping from lithium-ion conducting glass-ceramics (LICGC). In addition, electron doping influenced the Fe-Ge plane in the middle of monolayer FGT, weakening the resistance and enhancing its T_C [104]. As a typical example, the modulation of interlayer coupling was achieved by fabricating FGT hall devices on solid-state proton conductors (Figure 16A). Interestingly, they discovered clear exchange-bias (EB) by changing the gate voltage (Figure 16B). which may be related to the presence of AFM phase at low temperatures (Figure 16C). However, a random exchange bias (Figure 16D) occurs after applying a higher gate voltage. Furthermore, the exchange bias and coercivity (Figure 16E-F) have undergone a complex evolution with the measurement times, but there have also been cases of small EB and large coercivity. More importantly, the type of AFM-FM interface coupling determines the positive and negative exchange bias (Figure 16G).

Additionally, Tan et al. [60] achieved a high electron concentration doping of F5GT through solid proton conductor (Figure 16H-I). When a positive bias voltage was applied, the transport properties does unchange; while a negative bias voltage was applied, there was a significant change in transport properties caused by proton intercalation, especially when it reached -5 V, and the anomalous hall loop disappeared (Figure 16J), accompanied by the appearance of magnetic phase transition. Moreover, different calculation methods have revealed that electron doping achieves a reversal of hall conductivity and phase transition (Figure 16L-N). Therefore, electron doping or protonic gating was indeed an efficient method of controlling magnetis phase transition. Furthermore, Tang et al. [101] also found that magnetic anisotropy in F5GT was continuously mediating with electrolyte gating. Importantly, the screening effect of itinerant electrons driven magnetic anisotropy to switching from off-plane easy axis to in-plane easy axis.

4.7.4. Hole Doping

Inspired by the gate-mediated RTFM in FGT thin flakes [14], many attempts have been implemented to control its ferromagnetism through hole [43,93,105] or electron [43,93,104] doping. In particular, the magnetic anisotropy in exfoliated Fe_2.75GeTe₂ flake (Figure 17A-B) was inhibited by hole doping, resulting in a decrease in H_C (Figure 17C). The magnetic anisotropy could undergo a 93% attenuation, but the change in magnetic moment was very small as shown in Figure 17D. Furthermore, the electronic structure of Fe_2.75GeTe₂ single crystals changes caused by hole doping driven significant changes in magnetic anisotropy. In addition, another report [93] suggested that hole doping was beneficial for maintaining long-range ferromagnetic order.

Remarkably, the intrinsic Fe vacancies [105] was probed by scanning tunneling microscopy (STM) in Figure 17E-H. The peak near 20 mV originated from the Kondo lattice [105,121], and this Fe vacancy was called Kondo hole (Figure 17G-H). Hole doping elevated the energy band, and the Fermi surface of FGT was shifted towards a lower ene-rgy level (Figure 17I-J). After the formation of Fe vacancies at the Fe (II) site, the magnetic moment near the Fe (I) site decreased (Figure 17K-L), accompanied by the appearance of higher charge density. Interestingly, the introduction of Fe vacancies reduced the magnetic moment near them, further strengthened the Kondo screening effect, and thus weakened magnetism in Figure 17M-N. Kondo hole could affect the charge distribution of their own sites (Figure 17O), and converting them into momentum space had a more significant impact (Figure 17P). In other words, hole doping weakened ferromagnetism of FGT.

4.8. Intercalation or Irradiation

Interestingly, inserting sodium into Fe_2.78GeTe₂ powders [106] could raise its T_C to ~300 K in Figure 18. After intercalating Na, more exposed edges appeared, and their layered features remained unchanged, still in a single crystal structure (Figure 18A-B). In details, Fe, Ge, and Te elements (Figure 18C) were evenly distributed in the sample, while the inserted Na was concentrated at the edge. A phase transition occured from PM (Fe_2.78GeTe₂) to FM (NaFe_2.78GeTe₂) at 200 K (Figure 18D), and the M_S was enhanced (Figure 18G). Furthermore, the magnetic hysteresis loops will also be measured at 350 K. Notably, impurities phases such as Fe or Fe_2-xGe dominated the RTFM in NaFe_2.78GeTe₂ samples. Alternatively, the T_C and exchenge bias could be mediated with Fe-intercalation [107], which induced magnetic order by reinforcing magnetic coupling. But the detected Te_Ge antisite defects had no modulation effect on T_C of different samples. Furthermore, Fe intercalation provided a novel strategy for enhancing T_C.

Inspired by pattern-induced ferromagnetism [16], yang et al. [108] also improved the T_c of FGT to 450 K through Ga irradiation. The amorphous structure on the surface caused by irradiation and single crystal FGT jointly form a magnetic vortex state. Moreover, irradiation also causes a magnetization transition.

4.9. Twisting

Twisting 2D materials could introduce some novel properties such as magnetism [180,181] and superconductivity [182], which triggerred the interactions topology with magnetism in 2D ferromagnet, resulting in the formation of skyrmions [183,184] or magnons [185,186] in twisting system. Actually, stacking-order would directly affect the magnetismof bilayer CrI₃ by changing the crystal structure [180] or interlayer magnetic coupling [180]. Surprisingly, the magnetism was obtained in a double bilayer CrI₃ [187] at small twist angles. Very recently, phase transition from AFM to FM was achieved through twist-stacking bilayer FGT [109,181], which was theoretically easy to implement but experimentally challenging [188].

4.10. Patterning

Magnetic domain patterns on FGT surfaces could be modulated with various mechanisms [8,13,189,190], one of which was the phase transition related to interlayer coupling from FM to AFM [13]. The photoemission electron microscopy (PEEM) image in Figure 19(A-H) clearly shows the magnetic domain structure of FGT nanosheets, and the stripe domain structure disappears after reaching the T_c of 230 K. After patterning the FGT sample into diamond and rectangular shapes by focused ion beam (FIB), stripe magnetic domain structures similar to those in the unpattern FGT (Figure 19B) were also observed in Figure 19I. However, the stripe domain structure did not completely disappear and significantly weakened at 230 K (Figure 19G).

More interestingly, there were only in-plane magnetic domains in patterned FGT at 240 K (Figure 19L), which disappeared at ~370 K, indicating that the T_C of bulk FGT was increased to 370 K. Notably, a novel magnetic vortex state or magnetic multidomain state was developed in patterned FGT (Figure 19M) at 300 K. More importantly, spin reorientation occurred with increasing temperature (Figure 19N). When using FIB to construct FGT patterns, Ga ions were unintentionally implanted into the sample, which may have caused an increase in T_C. But it has not been confirmed at one point, just speculation.

5. FGT-Based Devices

Based on FGT, three typical types of devices have also been fabricated, such as magnetic tunnel junctions (MTJ) [111,112,113,114,138] (Figure 20A-D), tunneling spin valves [18,98,115,116,117] (Figure 20E-H), and spin−orbit torque devices [20,118] (Figure 21), have also been fabricated, to enrich their physical properties and develop their spintronic applications. As shown in Figure 20B, a nonlinear behavior originating from tunneling characteristics [114] was exhibited in the I-V curve. Furthermore, a typical spin-valve behavior was also found in the hysteresis loops (Figure 20C). After applying a specific voltage, spin-transfer torque (STT) generated by the current caused the bottom FGT electrode to switch, which was closely related to MAE.

As another typical device, wang et al. [18] observed a tunneling spin-valve behavior (Figure 20F) in FGT/hBN/FGT heterostructure (Figure 20E). Its TMR reaches up to 160%. The spin of tunneling electrons created a nonlinear bias-dependent I-V curve related to (Figure 20G). As the bias voltage increased, TMR exhibited a very significant attenuation in Figure 20H. The inelastic tunneling channel related to bias leaded to spin relaxation, which may suppress TMR signals.

After applying voltage in FGT/Pt hybrid devices (Figure 21A) [20], a current was generated between FGT and Pt, forming a spin-orbit torques in Pt. Interestingly, a hard magnetic loop similar to a FGT device has also been observed in FGT/Pt devices. As the applied in-plane magnetic field H_x was increased, its transition current decreased (Figure 21C-F), regardless of the direction of H_x. This switching was related to the magnetic domain and domain walls. Moreover, the low switching current of monolayer FGT was beneficial for exploring more effective devices.

6. Outlook

In this review, we have introduced the developments and structures of 2D metallic Fe_xGeTe₂ ferromagnet. Then, we summarized six experimental methods, ten FM modulation strategies, and three spintronic devices of 2D Fe_xGeTe₂ materials. Last, we have outlined the challenges and potential research directions in this field.

Actually, all the samples prepared through SSR, CVT and flux method were mainly bulk single crystals. Since the successful exfoliation of mono- or few-layers FGT, 2D Fe_xGeTe₂ has only been obtained experimentally. Very recently, CVD and MBE have also been used to prepare its materials. However, the lateral dimension of samples obtained by CVD was small and the number of layers was very difficult to accurately control. Although MBE could prepare wafer-scale materials and achieve 2D mode growth, it required a high vacuum environment and was costly. So far, there was still no cost-effective method to prepare wafer-scale materials with controllable layers. In addition, new experimental methods such as the substitution reactions [191] were also worth further exploration.

Although ten strategies for regulating ferromagnetism were proposed, not all of them have been achieved experimentally. Among those, twisting was only achieved through theoretical regulation. This required developing more strategies that could be executed experimentally to obtain RFTM in 2D Fe_xGeTe₂ materials. Furthermore, it was also very valuable to obtain FGT materials with good air stability [192] through new methods, which would provide more convenient conditions for fabricating devices [193].