Modulating above-room-temperature magnetism in Ga-implanted Fe₅GeTe₂ van der Waals magnets

The recent discovery of two-dimensional (2D) magnetism within the family of exfoliative van der Waals (vdW) materials^1,2 has attracted tremendous attention in both fundamental and applied research. In particular, there has been an upsurging enthusiasm for enhancing the Curie temperature (T_C) of vdW magnets well above room temperature (RT) and modulating their magnetic properties.³ Among the prominent bulk vdW materials used to study quasi-2D magnetism, the Fe_nGeTe₂ family has stood out due to its high T_C, stability, and large saturation magnetization, making it a promising candidate for manipulating magnetic anisotropy and spin textures. Specifically, bulk Fe₃GeTe₂ (Fe3GT) has a T_C of 220–230 K with a saturation magnetization of ∼376 emu/cm³,^4–7 and bulk Fe_nGeTe₂ (n = 4, 5) has a T_C ranging from 270 to 330 K with a saturation magnetization of 500–640 emu/cm³.^8–13 While Fe₃GT is an itinerant ferromagnet with weakly bonded Fe₃Ge layers (Fe atoms occupying two inequivalent Fe1 and Fe2) that alternate between two Te layers,^4–7 the Fe-rich Fe₅GeTe₂ (Fe5GT) has a covalently bonded Fe₅Ge slab [Fig. 1(a), Fe atoms occupying three inequivalent Fe1, Fe2, and Fe3 sites] sandwiched between two Te layers, which could introduce more complex magnetic orders.^8–12 Rich phenomena regarding different orientations of easy axis,^12,14 the temperature-dependent magnetic phases,^8,11–13 and the thickness-dependent spin reorientation transitions^15–18 in Fe5GT have been reported.

Various methods have been shown to increase the T_C and modulate the magnetic properties of Fe_nGeTe₂ vdW materials.¹⁹ The application of high pressure²⁰ and tensile strain,²¹ which could change the bond length or bond angles between ions and the lattice constants, dramatically increased the T_C and changed the magnetic anisotropy energy. The interlayer coupling or interfacial exchange coupling of a Fe3GT layer with an ferromagnetic (FM)²² or a topological insulator layer²³ also showed the enhancement of T_C to room temperature (RT). By cobalt or nickel substitution of Fe5GT, i.e., (Fe_1−xCo_x)₅GeTe₂^24–26 or (Fe_1−xNi_x)₅GeTe₂,²⁷ the magnetic ordering temperature was further increased to ∼360 and ∼480 K, respectively. More interestingly, the layered magnets showed an extremely sensitive response to the intercalation of foreign atoms into the vdW gaps, which could efficiently modulate the interlayer distance and the interlayer exchange coupling strength J. Intercalation of Li⁺ via an ionic gate in atomically thin Fe3GT²⁸ was shown to enhance T_C up to RT and to modulate magnetic anisotropy in vdW Fe5GT. Apart from the Li⁺-intercalation method for J modulation, the proton intercalation in Fe3GT and Fe5GT induced a gate-controlled FM-atomic force microscopy (AFM) phase transition¹⁴ and a zero-field cooled large exchange bias,²⁹ respectively. Meanwhile, we noticed that by intercalation of sodium in Fe3GT³⁰ or arsenic substitution for Ge in Fe5GT,³¹ the T_C was enhanced with a modified crystal structure and magnetic anisotropy. Our previous studies reported that through tiny implantation of Ga⁺ into microstructure-patterned Fe3GT flakes, the T_C was enhanced by almost 100% (from 230 to 450 K), with the magnetic anisotropy changing from out-of-plane at low temperature to in-plane at RT.^32,33 Thus, this proved to be an effective and reliable method of tuning the T_C and magnetic anisotropy of vdW magnets by controlling the fluence of Ga irradiation without the need for a special sample growth recipe^24–27,31 or particular device geometry,^28–30 as mentioned in previous reports. In this work, we report on the enhancement of T_C and the modulation of magnetic anisotropy in the newly synthesized vdW Fe5GT material through the implantation of Ga ions with site-specific capabilities using unique element-resolved photoemission electron microscopy and spatially resolved Kerr microscopy.

A flat Fe5GT single crystal was used to investigate macroscopic magnetization through vibrating sample magnetometry (VSM). The temperature dependence of the magnetization was obtained with a 0.2 T magnetic field applied along the in-plane direction after the zero-field cooling process. The magnetization decreases with increasing temperature and has a sharp transition at ∼310 K, indicating a ferromagnetic (FM) to paramagnetic (PM) phase transition (i.e., T_C ∼ 310 K) for the bulk Fe5GT crystals [Fig. 1(b)], which is further confirmed by the differential curve, as shown in the inset of Fig. 1(b). The temperature-dependent hysteresis loops of the Fe5GT crystal [Fig. 1(c)] were measured with a magnetic field along both the out-of-plane (H//c) and in-plane directions (H//ab). We found that the saturation field for the in-plane hysteresis loop is always smaller than that for the out-of-plane hysteresis loop at T < T_C, showing that the magnetic easy axis is along the in-plane direction for bulk Fe5GT crystals within the studied temperature range of 100–310 K, which is consistent with previous reports.^8,10–12

Figure 2(a) shows an image of a 90-nm-thick Fe5GT flake. Using a focused ion beam, we exposed a series of square-shaped areas (5 × 5 µm²) on this flake with Ga irradiation (30 keV, 11 pA) of different exposure times, as labeled in Fig. 2(b). Other conditions, such as the square size, sample thickness, and Ga voltage/current, were kept the same to single out the effect of Ga irradiation fluence on the modulation of magnetism in Fe5GT.³³ The high energy Ga indeed sputtered away some Fe5GT materials [Fig. 2(h)] and implanted Ga into the milled regions. We also carried out energy dispersive x-ray (EDX) spectroscopy mapping of Ga-milled Fe5GT to identify the atomic ratio of Ga implanted in the Fe5GT [Figs. 2(c)–2(f)]. The Ga⁺ ratio increased from ∼0.30% for the 15 s milled area to ∼0.85% for the 120 s milled area [Fig. 2(i)], which was consistent with the calculation of Ga⁺ implantation in the Fe5GT flake shown below. According to our calculation, the Ga dose was 4.125 × 10¹⁵ ions·cm⁻² for the 15 s milled area. Based on our TRansport of Ions in Matter (TRIM) simulation, we found that the average Ga⁺ implantation length in a Fe5GT flake was about 14 nm (see Sec. II of the supplementary material). Thus, the Ga ion’s implantation density was 2.95 × 10⁻³ ions·Å⁻³ for the 15 s milled area (assuming 100% incident Ga⁺ were implanted into Fe5GT).

To study the modulation of magnetism from Ga-implanted Fe5GT, we used spatially and element-resolved photoemission electron microscopy (PEEM), which utilizes secondary electrons to reconstruct the magnetic domain structures of the magnetic samples (see Sec. III of the supplementary material). With a circular x ray at 16° grazing incidence, PEEM is sensitive to both in-plane and out-of-plane magnetization in Fe5GT [Fig. 3(a)]. The magnetic domain images were taken by dividing the PEEM image at the Fe L₃ edge (710 eV) from that at the Fe L₂ edge (723 eV). The reversed magnetic domain contrast in Figs. 3(b) and 3(c) with opposite circular x-ray polarity confirmed the magnetic origin of the domain images. Moreover, we found that the milled Fe5GT exhibited a multidomain or magnetic vortex state for a low Ga⁺-dose area, indicating the in-plane-oriented magnetization for the Ga-implanted area. To quantitatively characterize the modulation of magnetism from the Ga-milled region, the x-ray magnetic circular dichroism (XMCD) signal was analyzed at different spots, as labeled in Fig. 3(b). Thus, we found that the un-milled Fe5GT flake showed a zero flat XMCD signal [Fig. 3(d)], indicating the PM state of the Fe5GT flake, which was consistent with the T_C value concluded from the later temperature-dependent hysteresis loop results in Fig. 4(c). Figures 3(e) and 3(f) show the typical x-ray absorption spectra (XAS) and XMCD signals from the Ga-implanted area with exposure times of 45 and 120 s. The XMCD value increased from zero to ∼9.2% with increasing Ga⁺ milling time and saturates for milling time above 60 s [Fig. 3(g)], indicating the induced stronger magnetism in Fe5GT with longer Ga exposure times.

Since the PEEM does not allow the simultaneous application of magnetic field and domain imaging, we studied the modulation of magnetic anisotropy in Ga-milled Fe5GT using spatially resolved magneto-optic Kerr microscopy. Kerr microscopy enabled hysteresis loop measurement with a magnetic field applied both along the in-plane direction in the longitudinal magneto-optic Kerr effect (MOKE) geometry [Fig. 4(a)] and the out-of-plane direction in the polar MOKE geometry (see Sec. IV of the supplementary material). Figure 4(c) shows temperature-dependent hysteresis loops from the un-milled Fe5GT flakes, where the out-of-plane hysteresis loops show larger squareness and smaller saturation fields compared with in-plane hysteresis loops, indicating that the easy axis remains along the out-of-plane c axis at various temperatures below the T_C of ∼290 K. This is different from the result of the easy axis along the in-plane direction in the bulk Fe5GT crystals shown in Fig. 1. In fact, controversies have existed regarding the easy axis direction of the Fe5GT flake, with different conclusions of in-plane¹² and out-of-plane oriented easy axes¹⁴ and thickness-^15–18 or temperature-dependent^8,10,11,15 spin reorientation transitions.

Figure 4(d) shows the in-plane hysteresis loops from the series of milled squares on the Fe5GT flake. Different from the zero flat hysteresis loop at the un-milled area, we found that all of the Ga-implanted areas show square-shaped or near-square-shaped hysteresis loops, demonstrating that the Ga⁺ implantation enhanced the T_C above RT and switched the easy axis of Fe5GT from the out-of-plane to the in-plane direction, which is consistent with the domain patterns shown in Figs. 3(b) and 3(c). The coercive field as a function of the exposure time is summarized in Fig. 4(b), which increases from 15 Oe at the 15 s milled area to ∼80 Oe at the 120 s milled area, indicating an enhanced magneto-crystalline anisotropy for Fe5GT with longer Ga exposure time. To identify the T_C enhancement with Ga⁺ implantation, we performed temperature-dependent hysteresis loop measurement from the Ga-milled area. Despite the relatively large noise level due to the limited cumulative area (5 × 5 µm²), we were able to identify that the T_C was enhanced to above 360 K for the 15 s milled area [Fig. 4(e)]. Considering that the XMCD signal was stronger in areas with longer Ga exposure times at RT, the T_C should be higher for Ga-milled areas with longer exposure times. Therefore, we conclude that by implanting Ga⁺ with >10⁻³ ions·Å⁻³ and a fluence of >10¹⁵ ions·cm⁻² in Fe5GT, T_C can be enhanced from 290 to above 360 K with a switched magnetic easy axis from an out-of-plane to an in-plane direction and a stronger magneto-crystalline energy at a higher Ga⁺ implantation level. Moreover, since the near square-shaped hysteresis loop is indispensable for the application of 2D vdW magnetic materials in spintronic devices,⁶ our result provides an avenue for spintronic devices based on vdW FM.

After confirming the enhancement of T_C and the modulation of magnetic anisotropy in Fe5GT by Ga irradiation, the next question is what the effect of Ga irradiation on the Fe5GT flake is and why the magnetism of Fe5GT changes. As clearly shown in the depth profile characterized by atomic force microscopy (AFM) in Fig. 2(h), the high energy Ga did sputter away some top Fe5GT in the milled region. For the 120 s milled area, about 40 nm Fe5GT was milled away. In addition to the sputtering effect, high energy Ga has also been reported to result in the change of surface morphology,³⁵ amorphization³⁶ or structural change, formation of defects,³⁷ or change of the valence state in the original material.³⁸ With increasing milling time, we found that the surface of the milled region became rougher, as indicated by fluctuations in the depth profile [Fig. 2(h)]. This point is further confirmed by the multiple isolated white (black) grain domains in the background of the big black (white) domain for 90 and 120 s milled area in Figs. 3(b) and 3(c). However, it is difficult to conduct a depth-profile structural analysis when the idea of the FIB-cutting process introduces more damage to the Fe5GT material or when the idea of microdiffraction characterization cannot rule out the small amount of amorphous or structurally changed Fe5GT.³³ More likely, some parts of Ga-exposed Fe5GT changed into amorphous or structurally damaged Fe5GT. Although it is difficult to quantify the amount of amorphous layer and the percentage of the remaining crystalline Fe5GT, we estimated a possible 10–30 nm amorphous layer according to the result of the TRIM simulation (see Sec. II of the supplementary material). Figure 3(h) clearly shows the chemical analysis of Fe atoms by taking element-resolved XAS at the Fe L absorption edge. In addition, we found that the height of the second peak at the Fe L₃ edge (711.4 eV) decreased with a longer milling time. Specifically, the Fe ratio had a sudden drop for milling time from 30 to 60 s [Inset of Fig. 3(h)]. This result indicates a change in the valence state of Fe atoms after Ga⁺ implantation, which is different from that of the Fe3GT flake after Ga⁺ implantation, where the Fe L₃ edge shows little change with the height of the second peak becoming slightly higher.³³ The drop of the second peak intensity at the Fe L₃ edge (711.4 eV) combined with the single peak of Fe XMCD at the L₃ and L₂ edge rules out the ferromagnetic origin from any possible Fe oxides, such as γ-Fe₂O₃ and Fe₃O₄.³⁹ The large XMCD ratio (∼9.2%) [Fig. 3(g)] from the Ga-milled area suggests that the magnetic signal is from the majority of Fe atoms in Fe5GT and, thus, rules out the magnetic origin of Fe atoms next to the Ga atoms (the latter case would lead to only ∼10% of our observed XMCD magnitude).

Although PEEM is extremely surface sensitive with a probe depth of several nm⁴⁰ and MOKE has a probe length of ∼20 nm,⁴¹ the formation of a magnetic vortex and a multidomain state in Figs. 3(b) and 3(c) indicate that the magnetic layer in Fe5GT was thicker than 50 nm considering the size and shape of the milled microstructures in this work.³² Therefore, deeper Fe5GT materials than the probed depth should have changed the magnetic properties. Although it is difficult to unambiguously identify the underlying mechanism for the enhancement of T_C and the modulation of magnetic anisotropy in Fe5GT by Ga irradiation, we offered several possibilities as shown below. Many studies have reported that a small number of foreign atoms introduced by intercalation could dramatically change the magnetic properties of vdW magnetic materials.⁴² For example, carrier doping at an electron density of 10¹⁴ cm⁻² significantly increases the T_C from 61 K to over 200 K and switches the magnetic easy axis from the out-of-plane to the in-plane in Cr₂GeTe₆.^43,44 Lithium ions intercalation at the level of 10¹⁴ cm⁻² enhanced the T_C of 2D-regime Fe3GT from 100 K to over RT.²⁸ Moreover, electrostatic doping at an electron density of 2.5 × 10¹³ cm⁻² induced the AFM-FM transition in CrI_3,⁴⁵ and organic cations doping at the level of 0.2–0.5 electrons/cell induces an AFM-ferrimagnetic transition in NiPS₃.⁴⁶ Moreover, only 2.5% and 5% arsenic substitution in Fe5GT enhanced the FM order and weakened the magnetic anisotropy.³¹ Thus, it is highly likely that the intercalation of Ga ions (at the influence of 10¹⁵ ions·cm⁻² and amount of 10⁻³ ions·Å⁻³) could be a possible mechanism for the modified magnetism in Ga-milled Fe5GT. The second possibility could be the strain effect. The sodium-intercalation in Fe3GT has been reported to have caused ∼3% in-plane tensile strain and 1% out-of-plane compressive strain.³⁰ The tiny lithium intercalation (∼0.2 Li per nm³) in WTe₂ caused up to ∼5% uniaxial in-plane strain and ∼6% out-of-plane expansion.⁴⁷ Since the strain engineering has been demonstrated to be an efficient way to manipulate magnetic properties,⁴⁸ the strain could possibly lead to the enhancement of T_C and change of the magnetic easy axis in Ga-implanted Fe5GT. Another possibility could be that the intercalated Ga atoms form new chemical bonds with or replace the host atoms, leading to modulated magnetic properties in Fe5GT. Recent studies have reported a high above-RT T_C (350–380 K) and large magnetic moment in 2D vdW Fe₃GaTe₂.⁴⁹ Obviously, more investigations into the effect of Ga⁺ implantation (structure change, defect amount, amorphization level, chemical bonding, distribution of intercalation ions, etc.) on Fe5GT are needed in the future.

In summary, we demonstrate an efficient and reliable way to enhance the T_C and change the magnetic anisotropy of the vdW Fe₅GeTe₂ material by Ga⁺ implantation. We show that the easy axis of the bulk Fe5GT (T_C = 310 K) is in plane, while the easy axis of the Fe5GT flake (T_C = 290 K) is out of plane. By implanting Ga⁺ with the amount of 10⁻³ ions·Å⁻³ and fluence of ∼10¹⁵ ions·cm⁻², the Fe5GT Curie temperature was greatly enhanced from 290 to 360 K with a switched magnetic easy axis from the out-of-plane to the in-plane direction. The room-temperature XMCD signal of 2D Fe5GT materials was enhanced from 0% to 9% by an implantation amount of 10⁻² Ga⁺·Å⁻³. Our result opens up a promising opportunity for tailoring the above-room-temperature magnetism for the future application of vdW materials in spintronic devices.

See the supplementary material for additional experimental details and introduction of FMR characterization, PEEM measurement, Kerr microscopy setup, and detailed results of TRIM simulation.

The project was primarily supported by the Fundamental Research Funds for the Central Universities (Grant No. wk2310000104), the USTC Research Funds of the Double First-Class Initiative (Grant No. YD2140002004), the National Natural Science Foundation of China (Grant Nos. 12174364, 12104003, and 12241406), Users with Excellence Program of Hefei Science Center CAS (Grant No. 2021HSC-UE003), and Open Funds of Hefei National Research Center for Physical Sciences at the Microscale (Grant No. KF2021001). This work was partially carried out at the USTC Center for Micro and Nanoscale Research and Fabrication. Z.Q.Q. acknowledges the support of the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences, Materials Sciences and Engineering Division under Contract No. DE-AC02-05CH11231 (van der Waals heterostructures program, KCWF16), Future Materials Discovery Program through the National Research Foundation of Korea (Grant No. 2015M3D1A1070467), Science Research Center Program through the National Research Foundation of Korea (Grant No. 2015R1A5A1009962), and the King Abdullah University of Science and Technology (KAUST) under Award No. ORA-CRG10-2021-4665. The PEEM measurement was performed at the XPEEM endstation at BL09U Dreamline at Shanghai Synchrotron Radiation Facility (SSRF). This research used resources of Beamlines XMCD-A and XMCD-B (Soochow Beamline for Energy Materials) at NSRL.

The authors have no conflicts to disclose.

M.Y. and Q.L. designed the experiments, analyzed data and wrote the paper. Y.N.Y., D.X.L., J.J.Y. and G.H.Z. performed the PEEM measurements. Y.N.Y., D.X.L., J.J.Y., F.F.P., S.Y.W., and J.M.G. performed the Kerr microscopy and FMR measurement. X.C. helped with the VSM characterization. R.Q.L., Z.L., and W.S.Y. helped with EDX characterization. L.W. and Z.L.L. helped with the AFM measurement. S.G.W. and Z.Q.Q. were involved in the data analysis and discussion.

Yanan Yuan: Data curation (lead); Formal analysis (lead). Daxiang Liu: Data curation (supporting). Jingjing Yu: Data curation (supporting). Guanhua Zhang: Data curation (supporting). Xiang Chen: Data curation (supporting). Ruiqi Liu: Data curation (supporting). Siyu Wang: Data curation (equal). Fangfang Pei: Data curation (supporting). Long Wei: Data curation (supporting). Zhi Li: Data curation (supporting). Junming Guo: Data curation (supporting). Shouguo Wang: Investigation (supporting). Zhaoliang Liao: Resources (supporting). Wensheng Yan: Resources (supporting). Ziqiang Qiu: Supervision (supporting). Mengmeng Yang: Conceptualization (equal); Supervision (equal); Writing – review & editing (equal). Qian Li: Conceptualization (equal); Funding acquisition (equal); Investigation (equal); Supervision (equal); Writing – original draft (equal).

The data that support the findings of this study are available within the article and its supplementary material.