E.
Georgopoulou-Kotsaki
*ab,
P.
Pappas
a,
A.
Lintzeris
ac,
P.
Tsipas
a,
S.
Fragkos
ad,
A.
Markou
e,
C.
Felser
e,
E.
Longo
f,
M.
Fanciulli
g,
R.
Mantovan
f,
F.
Mahfouzi
h,
N.
Kioussis
h and
A.
Dimoulas
a
aNational Centre for Scientific Research “Demokritos”, Institute of Nanoscience and Nanotechnology, Agia Paraskevi, 15341, Athens, Greece. E-mail: e.georgopoulou@inn.demokritos.gr
bSection of Condensed Matter Physics, Department of Physics, National and Kapodistrian University of Athens, Panepistimiopolis Zografou, 15784, Athens, Greece
cDepartment of Physics, National Technical University of Athens, School of Applied Mathematical and Physical Sciences, 15780, Athens, Greece
dDepartment of Mechanical Engineering, University of West Attica, 12241 Athens, Greece
eMax Planck Institute for Chemical Physics of Solids, 01187, Dresden, Germany
fInstitute for Microelectronics and Microsystems, CNR-IMM Unit of Agrate Brianza, Via C. Olivetti 2, 20864 Agrate Brianza, Italy
gDepartment of Material Science, University of Milano Bicocca, Via R. Cozzi 55, Milan 20125, Italy
hDepartment of Physics and Astronomy, California State University Northridge, Northridge, CA 91330-8268, USA
First published on 6th January 2023
Two-dimensional (2D) van der Waals (vdW) ferromagnetic metals FexGeTe2 with x = 3–5 have raised significant interest in the scientific community. Fe5GeTe2 shows prospects for spintronic applications since the Curie temperature Tc has been reported near or higher than 300 K. In the present work, epitaxial Fe5−δGeTe2 (FGT) heterostructures were grown by Molecular Beam Epitaxy (MBE) on insulating crystalline substrates. The FGT films were combined with Bi2Te3 topological insulator (TI) aiming to investigate the possible beneficial effect of the TI on the magnetic properties of FGT. FGT/Bi2Te3 films were compared to FGT capped only with AlOx to prevent oxidation. SQUID and MOKE measurements revealed that the growth of Bi2Te3 TI on FGT films significantly enhances the saturation magnetization of FGT as well as the Tc well above room temperature (RT) reaching record values of 570 K. First-principles calculations predict a shift of the Fermi level and an associated enhancement of the majority spin (primarily) as well as the total density of states at the Fermi level suggesting that effective doping of FGT from Bi2Te3 could explain the enhancement of ferromagnetism in FGT. It is also predicted that strain induced stabilization of a high magnetic moment phase in FGT/Bi2Te3 could be an alternative explanation of magnetization and Tc enhancement. Ferromagnetic resonance measurements evidence an enhanced broadening in the FGT/Bi2Te3 heterostructure when compared to FGT. We obtain a large spin mixing conductance of g↑↓eff = 4.4 × 1020 m−2, which demonstrates the great potential of FGT/Bi2Te3 systems for spin-charge conversion applications at room temperature.
The great success with the prototypical Fe3GeTe2, motivated researchers to explore the whole family of 2D metallic FM FexGeTe2 (x = 3–5).16 The x = 5 member of this family, namely Fe5GeTe2 (FGT) has a unit cell which can be described as an assembly of three Fe–Ge slabs separated by vdW gap between the Te atoms and it is very interesting structure in many respects. It appears either in an inversion symmetric (Rm space group, no. 166) or a non-centrosymmetric structure (R3m space group, no. 160). The former is the most common crystal structure in which there are three non-equivalent Fe positions per unit cell. The Fe(1) atoms preferentially occupy the crystallographic sites above or below neighboring Ge atoms in an alternating motif while it is forbidden to occupy both sites simultaneously as is also predicted by density functional theory (DFT) calculations.17 The lower symmetry structure R3m contains five different Fe sites with the Fe(1) atoms occupying sites either above or below the Ge atoms.18 The characteristic signature of FGT structure is the √3 × √3 reconstruction which is associated with the Fe(1) site displacements in ref. 17 and 19 or Fe deficiencies17 in Fe5−δGeTe2.
As a result of the complexity of the structure and its correlation with magnetic properties, a rich spectrum of magnetic order has been reported including pure ferromagnetism,19–23 ferrimagnetism24–27 and more exotic helimagnetic phase17 that calls for further investigation of this material. In addition, by electron doping24 a transition from FM to antiferromagnetic (AFM) has been observed. It has been suggested that FGT is an easy plane magnetized material.21,22,28,29 On the other hand, weak PMA has been reported in mono and few layer flakes,23,24,26,27 leaving the question about magnetocrystalline anisotropy in FGT widely open.
Although FGT is not considered as a strong PMA material, large magnetoresistance hysteresis at a small switching field has been reported,30 indicating its potential for devices. Since the structure does not consist of just a single layer, FGT behaves more like a 3D ferromagnet;23 the large number of spin pairs enhances the exchange interactions, thus the Tc of FGT is the highest among the FexGeTe2 family. To the best of our knowledge, FGT either bulk, exfoliated or epitaxial films,17–23,25–28 present a Tc close to – or just above – RT (∼332 K).29 This makes FGT very attractive material for novel commercial spintronics.
However, operation of devices just above 300 K is marginal and there is pressing need for even higher Tc to realize high performance magnetic and spintronic devices. Moreover, most of the work on FGT has been performed using exfoliated flakes from the bulk. Real world applications though require wafer scale growth using MBE or other synthetic thin film growth methodologies.
Work in this direction has already started although still rather limited. Recent studies31,32 have revealed that the substitution of Ge in FGT with As or Co enhances ferromagnetism. Although MBE has been successfully used for the wafer scale epitaxial growth of FGT on Al2O3 substrates,19 no published data are available regarding the combination of FGT with Bi2Te3 topological insulator in analogy with the Fe3GeTe2/Bi2Te3 heterostructure which produced Fe3GeTe2 with enhanced Tc.15 In any case, the role of Bi2Te3 in enhancing ferromagnetism in 2D vdWs ferromagnets remains unclear. Moreover, data associated the spin-charge conversion, a critical functionality for spintronic devices, are lacking.
To face the aforementioned challenges in achieving higher Tc FGT by wafer scale methods, in this work we employ MBE to fabricate FGT/Bi2Te3 heterostructures with two different FGT thicknesses and compare with bare FGT. We show that Bi2Te3 increases dramatically the saturation magnetization Ms and the Curie temperature Tc of FGT. Record-breaking value of Tc ∼ 570 K is reported in this work using high temperature SQUID magnetometry. Using first principles calculations, we elucidate the possible role of Bi2Te3. More specifically, it is proposed that the proximity of Bi2Te3 to FGT increases the density of states at the Fermi level or, alternatively, induces tensile strain in FGT which stabilizes a high magnetic moment phase. Both mechanisms could be responsible for the observed enhancement of Ms. Also, using ferromagnetic resonance measurements, we report on large spin mixing conductance in FGT/Bi2Te3, which indicates that this heterostructure is prospective for charge-spin conversion applications at room temperature.
Sample | Substrate | Structure | Capping |
---|---|---|---|
FGT | Sapphire (0001)/WS2 | Fe5−δGeTe2 (17 nm) | AlOx |
FGT/Bi2Te3 | Sapphire (0001)/WS2 | Fe5−δGeTe2 (17 nm) | Bi2Te3 |
FGT | SrTiO3 | Fe5−δGeTe2 (9 nm) | AlOx |
FGT/Bi2Te3 | SrTiO3 | Fe5−δGeTe2 (9 nm) | Bi2Te3 |
The samples were characterized in situ by reflection high energy electron diffraction (RHEED). Fig. 1 shows the RHEED patterns of the sapphire (0001)/WS2 substrate along the two azimuthal directions of sapphire (top) and the deposited 17 nm thick FGT film (bottom). The FGT streaky pattern is rotationally aligned with the substrate, showing no evidence for 30 deg or 90 deg rotated domains, despite the large lattice mismatch, which indicates van der Waals epitaxial growth of FGT on the sapphire/WS2 substrate. Moreover, the RHEED pattern shows a 3 × 1 reconstruction which is compatible with a (√3 × √3) R30° superstructure as seen by directly comparing the RHEED pattern with the reciprocal lattice in Fig. 1. Similar superstructure is observed in thin (9 nm) FGT grown on STO(111) substrates (see Fig. S1†). The (√3 × √3) R30° superstructure indicates the presence of Fe-deficient Fe5−δGeTe2 structure as previously reported.17
From the RHEED pattern (Fig. S2†) the lattice constant α is estimated to be 4.063 Å and 4.049 Å for the 17 nm 9 nm thick FGT, respectively which are in good agreement with literature values.19
As mentioned above, recent reports have revealed that the tendency to form Fe deficiencies allows the synthesis of Fe5−δGeTe2 with a variety of crystal structures both centrosymmetric and noncentrosymmetric.17,18,25
Fig. 2(a) left panel shows the structure of noncentrosymmetric (R3m) Fe5−δGeTe2, whereas the right panel shows the centrosymmetric structure (Rm). The main difference between the two structures is the presence of split site Fe(1) occupying sites above and below Ge alternately (centrosymmetric) or occupying a site only above Ge in the unit cell and thus break the inversion symmetry.
X-ray diffraction (XRD) patterns are shown in Fig. 2(b) and (c). The 17 nm thick FGT (Fig. 2(b)) consists of a clear Fe5GeTe2 lattice whereas in the FGT/Bi2Te3 pattern secondary weak Bragg reflections are present which are attributed to the Fe3GeTe2 phase. The 9 nm thick FGT film (Fig. 2(c)) shows a clear contribution from Fe5GeTe2 structure and for FGT/Bi2Te3 sample there is a small peak that can be identified as the (006) reflection of Fe3GeTe2, thus suggesting the presence of F3GeTe2 as a secondary phase. The XRD results in combination with the RHEED patterns verify that our epitaxial films contain mainly the Fe5GeTe2 phase. In order to verify the stoichiometry of our samples, complementary Energy Dispersive Spectroscopy (EDS) analysis was performed for 17 nm thick FGT/Bi2Te3 and was confirmed that is in very good agreement with the nominal stoichiometry (xFe = 5) (Fig. S3†).
Table 2 displays the lattice constant c as determined by the XRD patterns, for all heterostructures which is in good agreement with literature.18,19,25,26
Hetero structures | FGT (17 nm) | FGT (17 nm)/Bi2Te3 | FGT (9 nm) | FGT (9 nm)/Bi2Te3 |
---|---|---|---|---|
c (Å) | 29.30 | 29.47 | 29.64 | 29.37 |
It can be seen that for FGT capped with a protective AlOx layer, the hysteresis loop is gradually reduced in terms of remanent magnetization and coercive field as the temperature rises and is effectively diminished after 310 K. This behavior is compatible with a Tc just above 300 K as has already been reported for this material.17,19,21,23,26,28 In contrast, the hysteresis loop of FGT/Bi2Te3 is reduced mainly in terms of coercive field as the temperature rises up to 290 K and then it shows a clear tendency to remain invariant as the temperature is further raised up to 320 K, indicating that the Tc of FGT/Bi2Te3 could be much higher than 300 K. The same behavior is observed for the thin FGT samples as shown in supplementary (Fig. S4†). In summary, the presence of Bi2Te3 in the heterostructure makes FGT ferromagnetic with a well-defined hysteresis loop well above RT. The difference in the behaviour of the MOKE M–H loops for bare FGT and FGT/Bi2Te3 is attributed to the difference in the Tc of the two materials systems. In the case of bare FGT, as the temperature approaches Tc ∼ 310 K, the magnetization drops to zero, dramatically affecting the saturation magnetization of the hysteresis loop (Fig. 3(a)). In contrast, in the case of FGT/Bi2Te3, the Tc is much higher, ∼570 K, (see section about SQUID magnetometry below) so the saturation magnetization varies slowly leaving the hysteresis loops essentially unaltered (Fig. 3(b)).
It should be noted that AlOx is deposited as a capping layer to protect FGT from oxidation. AlOx is known to be non-magnetic, it has little – or no – spin orbit coupling and has no topological insulator properties; therefore, it is not expected to affect the magnetic behaviour of FGT underneath. Therefore, the MOKE hysteresis loops of FGT/AlOx are expected to be the same as in the case of an uncovered (bare) FGT which is measured under inert atmosphere conditions to prevent oxidation.
Ferromagnetic domains are also imaged under in plane (IP) magnetic field (Fig. 4). The material is first saturated at ∼+300 mT and the domains show up as black and white contrast when the material is demagnetized at an opposite (coercive) field of ∼−1 mT. The domains appear as elongated irregular shapes on average about 100 μm long and 10–20 μm wide. It is important to note that the domains are better visible at 77 K but they persist with a clear contrast up to 300 K (Fig. 4).
Fig. 5 Comparison of in plane (IP) and out of plane (OP) component by SQUID measurements for 17 nm thick FGT/Bi2Te3 (a and b) and FGT (c and d). |
Fig. 5 shows the main M–H characteristics curves for RT and the lowest attainable temperature of 20 K. The highly nonlinear M–H curve reveals ferromagnetism in all films which persists and is quite strong up to 300 K at least. This confirms the MOKE results that indicated FM films above RT especially for the FGT/Bi2Te3 heterostructure.
The data in Fig. 5 and Fig. S5–S7† show that films are rather isotropic. In the case of FGT/Bi2Te3 a clear IP magnetic anisotropy prevails above 200 K since the Ms is reached at lower magnetic fields (Fig. 5(b)) and a very sharp loop is observed at RT in conjunction with MOKE data.
Fig. 6 shows the normalized (in units emu cm−2) saturation magnetization Ms as a function of temperature extracted from the M–H curves. The Ms for all samples increases with decreasing temperature showing typical behavior for ferromagnetic ordering.
Fig. 6 Temperature dependence of saturation magnetization Ms measured by SQUID for all the samples for the in plane (IP) component (a), out of plane (OP) component (b). |
The two FGT samples with different thickness (black and red squares in Fig. 6(a)) show very similar IP normalized Ms behavior as expected. On the other hand, the two FGT/Bi2Te3 heterostructures (black and red circles in Fig. 6(a)) show enhanced IP Ms which is attributed to the beneficial influence of the Bi2Te3 layer. It is important to note that this influence is more pronounced in the case of the thinner sample (9 nm) where the enhancement reaches 148% compared to the case of the thicker sample (17 nm) where the enhancement is estimated to be 83% at RT. It is anticipated that Bi2Te3 TI results in electron doping of FGT originating from the topological surface states. Then, the bigger influence on the thinner sample is expected the region close to the interface represents a bigger portion of the total FGT volume. Similar behavior is observed for the OP Ms presented in Fig. 6(b).
The data in Fig. 6 suggest that the FGT/Bi2Te3 heterostructures are expected to have a larger Tc compared to the FGT, which could be significantly larger than 300 K. In order to accurately determine the Tc, an evacuated furnace has been used to cover the temperature range 300–620 K and the normalized Ms − T results are shown in Fig. 7 for one of our FGT/Bi2Te3 heterostructures with FGT thickness of 17 nm.
It can be inferred that the Tc is around 570 K which, to the best of our knowledge, is the highest reported to literature values.
In summary, from our magnetic measurements can be concluded that for the 17 nm thick FGT sample the presence of the TI helps retaining the FM order at elevated temperatures shifts the Tc significantly above RT and stabilizes an IP nonzero magnetization in the high temperature range, in contrast to the FGT which exhibits a more isotropic behavior with a significantly smaller magnetization, becoming almost negligible above RT.
The calculated spin-resolved energy density of states (DOS) of FGT (Fig. S12(a)†) is compared to that of FGT/Bi2Te3 (Fig. S12(b)†) heterostructure. They have the expected DOS of itinerant ferromagnet with sharp features similar to those reported in ref. 41. When the DOS is plotted in a limited range around the Fermi energy, a twofold effect is observed (Fig. 8). First, near the Fermi level there is an increase in the overall DOS of FGT/Bi2Te3 due to the contribution of Bi and Te orbitals via proximity (Fig. S13†). Second, there is an energy shift of the Fermi level of FGT/Bi2Te3 compared to the FGT monolayer. Noting that the Fermi level is kept at E = 0 by convention, it can be seen from Fig. 8 that the Fermi level moves from just below the peak indicated by arrows in FGT to above the same peak in FGT/Bi2Te3. As a result, the majority spin (D(EF)↑) density of states for the two layers (FGT and FGT/Bi2Te3) differ significantly at the Fermi level in favor of FGT/Bi2Te3. The combined effect of the two, results in a significant increase by 28% of the FGT/Bi2Te3 total density of states D (EF) compared to FGT. Within the Stoner approximation for itinerant ferromagnets an enhancement of D (EF) is associated with an increase in magnetization as observed in the experiment (Fig. 6).
In a mean field approximation, an increase of saturation magnetization is accompanied by an increase in the Tc as observed in Fig. 7. This is in line with previous report42 that femtosecond laser pulses induced doping can modify the electronic band structure of Fe3GeTe2 enhancing Ms and Tc from 250 to 300 K. Similarly, MBE grown 4 ML Fe3GeTe2 exhibits enhanced Tc up to 400 K when it is in contact with Bi2Te3 attributed to increased carrier doping of Fe3GeTe2 induced by Bi2Te3.15 Finally, effective electrostatic doping has been considered responsible for the increase of Fe3GeTe2Tc from 225 K to RT.2
Alternatively, the significant increase of the total magnetic moment in the present of Bi2Te3 (Fig. 6), suggests the presence of multiple (two) magnetic phases in FGT, whereby, the change of chemical environment in the presence of the Bi2Te3 can favor one phase over the other. Recent ab initio calculations of the FGT monolayer predicted a significant enhancement of the total magnetic moment under an in-plane tensile strain.43 The lattice constant mismatch of more than 10% between FGT (4.05 Å) and Bi3Te2 (4.5 Å) indicates the presence of a tensile strain on the FGT layers at the interface region. We have carried out ab initio electronic structure calculations using the Vienna ab initio Simulation Package (VASP)33 to investigate the effect of strain on the magnetic properties of bulk FGT. In Fig. 9(a) we show the calculated total energy versus the constrained total magnetic moment for bulk FGT under 2% and 5% in-plane tensile strain denoted by the dashed-blue and solid-red curves, respectively. In the case of 2% tensile strain (blue dash-dotted curve), the results predict the coexistence of the two magnetic phases as they are energetically degenerate and separated by a relatively small energy barrier height of about 6 meV. As the tensile strain increases to 5% (red curve in Fig. 9(a)), the high magnetic moment (11.2μΒ) phase is stabilized showing a total energy minimum. In Fig. 9(b) we plot the total magnetic moment versus applied tensile strain which shows a phase transition at about 2.5% strain. The atom-resolved magnetic moments, shown in Fig. 9(c), demonstrate that the enhancement of the magnetic moment is primarily due to emergence of magnetization in Fe5 atom under strain. This enhancement explains the larger saturation magnetization values Ms observed in FGT/Bi2Te3 heterostructures by SQUID magnetometry.
Fig. 10 Comparison of the FMR signals for the FGT (empty black squares) and FGT/Bi2Te3 (full blue squares) samples acquired by resonant cavity FMR at 9.78 GHz. |
From the fit of the two curves, we can extract the resonance field (Hres) and the signal's linewidth (ΔH). The results are HFGTres = 2082 ± 0.8Oe and ΔHFGT = 511 ± 2Oe for the FGT sample, and and ΔHFGT/Bi2Te3 = 965 ± 12Oe for the FGT/Bi2Te3 sample. We detect a clear enhancement of the ΔH in the FGT/Bi2Te3 sample. This is associated to an enhancement of the damping parameter α, since these quantities are correlated as depicted in eqn (1).
(1) |
Indeed, according to eqn (1), by assuming ΔH0 equal for both samples, we can write .
In the framework of the spin pumping model, the enhancement of α is related to the generation of pure spin currents in a ferromagnetic layer.46,47 Indeed, these phenomena are already reported for several chalcogenide-based TIs (i.e. Sb2Te3, B2Se3, Bi2Te3).40,48,49
Our results show that Bi2Te3 acts as a good spin sinking material when in contact with FGT, with the spin current that is generated at resonance within the FGT layer being markedly absorbed by the neighboring Bi2Te3. This process can be quantified by calculating the so called spin mixing conductance G↑↓eff, an intrinsic property of a system whose real part is proportional to the amount of the spin current crossing the FGT/Bi2Te3 interface and absorbed by the spin sink layer, as regulated by eqn (2).
(2) |
Considering f = 9.78 GHz, the γ and g-factor values extracted from broadband FMR (see ESI†), and the extracted from SQUID at 300 K by assuming a FGT thickness of 9 nm (see Fig. 6), from eqn (2), we calculate g↑↓eff = 4.4 × 1020 m−2.
It is of interest to compare the g↑↓eff we obtain in the FGT/Bi2Te3 system, with those reported for the Bi2Te3 and others topological insulators in contact with traditional ferromagnets. Indeed, the Table 1 of ref. 40 shows how the latter value is higher than most of those found in literature for similar systems. More specifically, F. B. Abdulahad et al.49 used FMR to probe a NiFe/Bi2Te3 heterostructure, reporting a g↑↓eff in the 1–4 × 1019 m−2 range, being one order of magnitude lower than the g↑↓eff value we measure in FGT/Bi2Te3.
Our results demonstrate the great potential of the developed FGT/Bi2Te3 system towards its use as an efficient spin-charge converter based on the 2D ferromagnet Fe5−δGeTe2 and the 3D topological insulator Bi2Te3.
MOKE in the longitudinal mode near RT reveals that FGT/Bi2Te3 heterostructures are ferromagnetic with a clear IP M–H ferromagnetic loop which remains robust for temperatures well above 300 K indicating a high Tc. This is in contrast to the case of FGT which shows a fast-reducing M–H loop near RT indicating a Tc close to-or just above 300 K. The enhancement of Tc in FGT/Bi2Te3 is verified by high temperature SQUID measurements in a furnace for a 17 nm FGT/Bi2Te3 sample which shows Tc as high as 570 K. Moreover, a systematic investigation by SQUID shows that the FGT/Bi2Te3 has an enhanced Ms compared to FGT at all temperature range below 350 K, which becomes more pronounced in the thinner samples. First principles calculations predict a contribution to the total DOS of Bi and Te due to proximity as well as a Fermi level shift influenced by the proximity to the Bi2Te3 layer, both of which favor a larger total DOS for the FGT/Bi2Te3 compared to FGT. In the framework of a Stoner approximation for an itinerant ferromagnet, such an increase of the DOS could explain the observed enhancement in Ms and Tc and the overall strengthening of ferromagnetism in the case of FGT/Bi2Te3 in line with the experiments. Alternatively, the stabilization of a high magnetic moment state in the presence of >2% tensile strain, likely originating from the lattice mismatch between FGT and Bi2Te3, as predicted by DFT calculations, could explain, at least partly, the magnetization enhancement. Finally, the FGT and FGT/Bi2Te3 heterostructure were probed by ferromagnetic resonance, with a revealed enhancement of damping parameter for FGT/Bi2Te3 when compared to FGT, suggesting the suitability of such an heterostructure to exploit spin-charge conversion effects in spintronic devices based on 2D ferromagnets and 3D topological insulators.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d2nr04820e |
This journal is © The Royal Society of Chemistry 2023 |