Increased Curie temperature and enhanced perpendicular magneto anisotropy of Cr₂Ge₂Te₆/NiO heterostructures

Recently, magnetism in layered van der Waals (vdW) materials has attracted great attention because of their unique magnetic properties stemming from their two-dimensional (2D) nature.^1,2 Such layered vdW materials provide an opportunity to fabricate heterostructures free from constraints in conventional film growth and promise a unique route to explore new functionality of these materials based on the electric field and crystalline symmetry.^3,4 One of the major issues in vdW ferromagnets is that the ferromagnetic transition temperature (Curie temperature) is relatively low. CrI₃ and Cr₂Ge₂Te₆ were reported as atomically thin-form ferromagnets in 2017, where the Curie temperature ranges from 30 K (bilayer Cr₂Ge₂Te₆) to 45 K (monolayer CrI₃), being intriguingly low compared to the bulk values of 61 K (bulk CrI₃) and 66 K (bulk Cr₂Ge₂Te₆).^1,2 To enhance the Curie temperature, a variety of approaches using interface and gap engineering have been proposed including the dielectric effect, spin–orbit coupling proximity, charge transfer, and interface hybridization.⁵ However, only a few approaches have been implemented and reported so far, such as electric gating.^4,6 Therefore, it is important to search for other effective approaches to enhance the Curie temperature. Here, we study magnetic properties in heterostructures between antiferromagnet NiO and vdW ferromagnet Cr₂Ge₂Te₆. We will report magneto-optical Kerr effects (MOKEs) and detect hysteresis arising from ferromagnetism.

Cr₂Ge₂Te₆ was earlier reported by Carteaux et al.,⁷ and recently, there is significantly increasing interest because this is a layered two-dimensional magnet with mechanical cleavability down to atomically thin layers. The Cr₂Ge₂Te₆ single crystals used in this work were grown via a self-flux technique. First, 100 mg of Cr powder, 200 mg of Ge powder, and 2 g of Te were sealed in a quartz tube. The mixture was heated to 1050 °C and held for 30 h, and then, it was cooled down to 475 °C in 10 days, and finally, the Ge-Te flux was removed using a centrifuge at this temperature. The magnetic properties of the bulk crystals were characterized by magnetometry using the MPMS (magnetic property measurement system), and the Curie temperature was found to be T_C ≈ 66 K from the minimum of dM/dT curves (measured with the magnetic field of 50 mT in the c-axis). The observed magnetic properties were consistent with the previous reports.^1,4,7–13 Cr₂Ge₂Te₆ crystals were mechanically cleaved onto a silicon substrate in ambient conditions (with the SiO₂ thickness of 285 nm). The thickness of the flakes was characterized by using an atomic force microscope. Atomically thin flakes were visible with thickness-dependent color contrast due to the interference effect as shown in Fig. 1(a). Sputtering was performed using a NiO target with a base pressure of ca. 1 × 10⁻⁵ Pa, an Ar pressure of 0.2 Pa, and 200 Watt RF power. Various deposited NiO films with different thicknesses ranging from 20 nm to 100 nm are explored in this work. After the sputtering of the NiO layer, the color of both the silicon substrate (with NiO) and the Cr₂Ge₂Te₆ flakes changed as shown in Fig. 1(b) because the interference condition was modified.

Polar magneto-optical Kerr Effect (MOKE) measurements were performed in an Oxford MicrostatMO system in the Faraday configuration. A temperature stabilized laser diode (635 nm in wavelength) was used to deliver a linearly polarized laser beam that was focused onto the sample at normal incidence using a 0.6 NA 100X long working distance objective. The estimated power delivered to the sample is less than 3 μW. The laser beam was intensity modulated using a chopper, and the reflected beam was sent through a Wollaston prism oriented at 45° with respect to the initial polarization to split the beam into two. The split beams were collected by two photodiodes arranged in a differential mode whose output was sent to an SR570 current preamplifier (Stanford research systems). The output of the amplifier was then measured using a lock-in referenced to the chopper frequency. In order to complete the measurements, each photodiode was covered in turn to obtain the total intensity signal. This is used to normalize the previously measured signals that will yield a quantity proportional to the Kerr effect signal (rotation of the polarization angle). Calibration was performed to obtain the actual conversion factor between radians and the normalized signal by tilting the polarizer by a few fractions of a degree and recording the normalized signal.

Figure 1(c) shows MOKE curves (Kerr rotation angle vs magnetic field) for a sample before NiO deposition. The magnetic field was applied perpendicular to the substrate because the easy axis of Cr₂Ge₂Te₆ was reported in that direction. For the flakes with the thicknesses more than 5 nm, we observed clear hysteresis. For the flakes with a thickness of 5 nm or less, we did not observe clear hysteresis, which can be attributed to the weaker two-dimensional magnetism¹ and/or to the possibly not-pristine surface or film degradation (e.g., oxidization). After a 20-nm-thick NiO layer was deposited, we measured the same flakes for comparison. These flakes, for the ones that showed a clear hysteresis before the deposition, showed an increase in the coercive field and a change in the hysteresis into a rectangular shape as shown in Fig. 1(d). This clearly indicates the enhanced perpendicular anisotropy induced by depositing the NiO layer.

Importantly, Cr₂Ge₂Te₆/NiO not only enhances the perpendicular anisotropy but also increases the Curie temperature. Figure 2(a) shows the temperature dependence of the hysteresis curves in the MOKE signal for the sample shown in Fig. 1 (Cr₂Ge₂Te₆ thickness 7 nm, position 2). We characterized the Curie temperature as the midpoint between the temperature in which the coercive field is no longer seen and the one in which we still see a coercive field. For position 2, we observed that the Curie temperature increased by 30 K to ∼85 K after NiO deposition. Similarly, we observed an increased Curie temperature to 85 K at position 1 (Cr₂Ge₂Te₆ thickness 8 nm), to 70 K at position 3 (Cr₂Ge₂Te₆ thickness 6 nm), and to 70 K at position 4 (Cr₂Ge₂Te₆ thickness 5.5 nm). For comparison, we measured another sample with 50 nm thickness of NiO, while the thickness of the flake was similar to that of position 2. We observed an even larger increase in the Curie temperature up to 115 K as shown in Fig. 2(b), which is about twice of the Curie temperature on the Cr₂Ge₂Te₆ flakes without NiO. In this sample, we also observed further enhanced perpendicular anisotropy as the coercive field was even higher than that of a sample with a NiO thickness of 20 nm.

Figure 2(c) shows the hysteresis of a 202-nm-thick Cr₂Ge₂Te₆ flake with NiO of 20 nm. The hysteresis loop at T = 7 K shows small perpendicular anisotropy (coercive field of nearly 0 Oe, similar to that measured in bulk Cr₂Ge₂Te₆, see the supplementary material). At elevated temperatures (T = 55 K and above, where ferromagnetism disappears in bulk Cr₂Ge₂Te₆ with typical T_C ∼ 60 K), the hysteresis becomes more rectangular in shape and indicates the presence of some NiO-induced enhancement of perpendicular anisotropy even for such thick flakes. We also raise the possibility in heterostructures of relatively thick Cr₂Ge₂Te₆ that we could be accessing to the (different) magnetism of the Cr₂Ge₂Te₆/NiO interface [observed in the data above 55 K in Fig. 2(c)] separately from that of bulk Cr₂Ge₂Te₆ itself (observed in the data at 7 K). For the case of thin flakes, as exemplified in Figs. 2(a) and 2(b), one likely cannot separate the signal from the Cr₂Ge₂Te₆/NiO interface and the bulk Cr₂Ge₂Te₆ itself but rather should consider it as the whole Cr₂Ge₂Te₆ affected by the interface with NiO.

Figure 3 summarizes the Curie temperatures for Cr₂Ge₂Te₆ flakes with various thicknesses, both for those without and with NiO of 20 nm, 35 nm, 50 nm, and 100 nm in thickness together with the data in the literature for Cr₂Ge₂Te₆ without NiO.^1,4 The data clearly show that Cr₂Ge₂Te₆/NiO can enhance the Curie temperature. For very thin Cr₂Ge₂Te₆ flakes (thickness less than 5 nm), we were not able to see a clear enhancement, possibly being attributed to the defective layers (we note that the monolayer is ∼0.7 nm according to the crystallographic study⁷). We avoided unnecessarily exposing the samples to air (the exposure time before NiO deposition is typically within five minutes). For thin Cr₂Ge₂Te₆ flakes (thickness more than 5 nm and less than 30 nm), the increase in the Curie temperature is strong for the 50-nm-thick NiO. We note different optical constants and the interference effect on Si/SiO₂/Cr₂Ge₂Te₆/NiO with different thicknesses of NiO layers, as the color of Si/SiO₂/Cr₂Ge₂Te₆/NiO is different (in fact, the substrate color of Si/SiO₂/NiO is also clearly different upon the thicknesses of NiO layers, for example, between 50-nm-thick NiO and 20-nm-thick NiO). In addition, the opposite sign of the MOKE hysteresis curves between Figs. 1(c) and 1(d) is supposedly due to different optical constants.¹⁴ While such a difference in optical constant and interference (giving also different effective optical penetration depths) can affect the MOKE signal (sign and amplitude) due to different NiO thicknesses, it is more difficult to explain the difference in magnetic properties (Tc and anisotropy) of the materials themselves. One of such possible factors would be the difference in crystalline quality and interfaces for different thicknesses of NiO layers. As a controlled experiment to probe this question, we characterized the crystallinity of NiO with different thicknesses for Si/SiO₂/Pt/NiO of 50-nm-thick NiO and 20-nm-thick NiO by X-ray diffraction (SmartLab, RIGAKU Corporation). The (111) and (002) peaks are clearly visible (see the supplementary material for details). The full width half maximum (FWHM) of the (111) peak is 0.60 deg and 0.70 deg for 50-nm and 20-nm thick NiO, respectively. These results were concluded to be the 10–15 nm grain size depending on the details of analysis. The grain size between the two NiO layers of different thicknesses vary only by ∼15%. It is not clear presently if the difference in the Curie temperature of Cr₂Ge₂Te₆/NiO can be related to this relatively moderate difference in crystallinity. Another possible mechanism might be related to the strain, as we noticed that wrinkles appeared in Cr₂Ge₂Te₆ after the NiO deposition. This mechanism could be consistent with the middle panel of Fig. 3, where the thick Cr₂Ge₂Te₆ flakes (thickness of more than 30 nm) show little difference in Curie temperature between NiO layers of different thicknesses. For thick Cr₂Ge₂Te₆ flakes, we still see the increase in Curie temperature as we discussed in the previous paragraph. The Curie temperature in thick flakes with NiO is higher than that of the bulk Curie temperature reported in various literatures^1,4,7–13 as shown in the right panel of Fig. 3. This higher Tc in the thick flakes could mostly be contributed by the (enhanced) magnetism at the Cr₂Ge₂Te₆/NiO interface as we discussed earlier [Fig. 2(c)]. While the precise mechanism of the increase in Curie temperature and the enhancement in perpendicular anisotropy for Cr₂Ge₂Te₆/NiO interface magnetism cannot be clear yet at this stage, so far there are a few reports on the correlations between the perpendicular anisotropy and the Curie temperature in other systems. Recently, it has been reported that a two-dimensional magnet with additional perpendicular anisotropy shows an increase in the Curie temperature.^15,16 Bonding between 3d electrons in the transition metal and 2p electrons in oxygen is known to induce high perpendicular anisotropy for MgO/Fe and MgO/Co from both experimental and theoretical studies,^17,18 which may provide a microscopic mechanism for the enhanced perpendicular anisotropy and increase in the Curie temperature observed in the present studies. We also note that we did not observe any exchange bias (for example reported earlier for the NiO/Ni₈₁Fe₁₉ interface¹⁹) in our hysteresis loop. It is noted that the magnetic field was not applied when we made the interface during the NiO deposition.

In summary, we studied the magnetic properties of a two-dimensional van der Waals magnet and Cr₂Ge₂Te₆ flakes covered by NiO thin films by magneto-optical Kerr effects. We characterized the hysteresis loops of Cr₂Ge₂Te₆ flakes before and after NiO deposition. Cr₂Ge₂Te₆/NiO showed a strong increase in Curie temperature and a clear enhancement in perpendicular anisotropy as evidenced by the increase in the coercive field and the change in hysteresis into a rectangular shape. We observed the Curie temperature as high as 115 K for Cr₂Ge₂Te₆/NiO with a NiO thickness of 50 nm, more than twice the one for Cr₂Ge₂Te₆ without NiO. The Curie temperature increase for Cr₂Ge₂Te₆/NiO is observed for Cr₂Ge₂Te₆ with thicknesses ranging from 5 nm to 200 nm. Even Cr₂Ge₂Te₆/NiO with 200 nm thick Cr₂Ge₂Te₆ flakes showed the higher Curie temperature than bulk Cr₂Ge₂Te₆. These results indicate the magnetic properties of two-dimensional van der Waals materials can be controlled by employing the heterostructure and interface with other materials.

See the supplementary material for the hysteresis curve of bulk Cr₂Ge₂Te₆ and XRD data of Si/SiO₂/Pt/NiO.

We thank K. Saito for the growth of NiO thin films and the XRD measurement, and P. Upadhyaya, D. Xiao, A. Rustagi, T. Nakanishi, and A. Lu for fruitful discussions. This work was supported in part by a Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology (MEXT), JSPS KAKENHI (Grant Nos. 18H03858, 18H04473, 18H05840, 18H04471, 17-18H05326, 18H04304, 18H03883, and 18F18328), Sumitomo Foundation (Grant No. 180953), the WPI‐AIMR's fusion research program under World Premier International Research Center Initiative (WPI), MEXT, Japan, the Center for Science and Innovation in Spintronics and Inter-University Cooperative Research Program of the Institute for Materials Research, Tohoku University (Proposal No. 19G0210, Cooperative Research and Development Center for Advanced Materials), Purdue University, and the National Science Foundation (Grant Nos. ECCS1711332 and DMR1838513). Xing-Chen Pan acknowledges the support from an International Research Fellowship of Japan Society for the Promotion of Science [Postdoctoral Fellowships for Research in Japan (Standard)].