Structural and magnetic properties of Mn₃Ge films with Pt and Ru seed layers

Over the last decades, manganese-based compounds have attracted considerable attention due to their diverse magnetic characteristics^1,2. Among those compounds, Mn-Ge binary alloys can crystallize into a variety of structures with very distinct magnetic properties. Several equilibrium structures such as Mn₁₁Ge₈, Mn₅Ge₃, Mn₂Ge, Mn₅Ge₂, and Mn₃Ge are known to exist according to the phase diagram^3,4. In addition to them, other non-equilibrium structures were also reported, including MnGe, Mn₃Ge₅, and MnGe₄^5–7.

Mn₃Ge, the compound of main interest in our work, is known to have two equilibrium crystal structures of D0₁₉ hexagonal and D0₂₂ tetragonal phases, which can be selected by the annealing conditions⁸. The high temperature hexagonal (D0₁₉) phase belongs to the space group of P6₃/mmc, and shows a non-collinear antiferromagnetic spin order known as an inverse triangular spin structure below the Néel temperature of T_N ∼ 380 ^○C^9,8,10,11. This non-collinear spin structure has also been confirmed in other hexagonal Mn₃X (X=Sn, Ga) systems, and recently many extensive studies have been conducted. Despite almost non-existent magnetization, these hexagonal antiferromagnets exhibit large spontaneous transverse responses such as anomalous Hall effect^12–15, anomalous Nernst effect^16–19, and magneto-optical Kerr effect²⁰ at room temperature. They are considered to be promising materials for spintronics applications²¹, and a spin Hall effect with an anomalous sign change, called the magnetic spin Hall effect, has also been found²². The thin film fabrication of these non-collinear antiferromagnets Mn₃X is also under intense investigation, which has led to reports on the exchange bias effect^23–28 and large room temperature transverse responses²⁹. These phenomena have also been observed in the case of bulk crystals.^12–14,30

The low temperature tetragonal Mn₃Ge shows a collinear ferrimagnetic structure with the space group I4/mmm and the Curie temperature T_C has been found to be 920 ^○C^9,10,11. This material is known as a half-metallic Heusler system and has been studied energetically for a long time. The magnetic moments of Mn atoms are aligned along the c-axis and neighboring atoms couple antiferromagnetically, which leads to low saturation magnetization (M_s)^31,32. It has been predicted to possess a large perpendicular magnetic anisotropy, a low damping constant, and a high spin polarization^33,34. All these features could lead to an extremely large tunnel magnetoresistance (TMR), and thus the tetragonal Mn₃Ge is a promising candidate for a magnetic random-access memory (MRAM)³⁵. Many reports on tetragonal Mn₃Ge films, which exhibit useful magnetic properties, as mentioned above, have already been conducted, e.g. an epitaxial Mn₃Ge (001) film grown on a SrTiO₃ substrate³¹ or on a MgO (001) substrate with a Cr (001) buffer layer³⁶.

Furthermore, in the case of Mn₃Ge cubic structures, they has been confirmed to possess both ferromagnetic and antiferromagnetic properties depending on which of various crystalline forms is adopted, such as a L1₂ alloy^9,37, an α-Mn type thin film³⁸ or γ-Mn type alloys^39–42. The cubic L1₂ phase (space group $Pm3¯m$⁠, T_C = 400 K) was obtained by a belt-type high-pressure apparatus under 6.2 GPa and 1000 ^○C⁹. This compound shows ferromagnetism and metallic conductivity. The α-Mn type (space group $I4¯3m$⁠, Mn₃Ge = 334 K) film was synthesized by the molecular beam epitaxy (MBE) technique using a semi-insulating GaAs(001) substrate³⁸. This synthesis of the α-Mn type structure was made possible by additional driving forces coming from lattice mismatch strain between crystalline substrates and Mn₃Ge. Moreover, γ-Mn type alloys can be obtained in a wide compositional range including Mn₃Ge by using the melting method. Although they are known to be antiferromagnets, the proposed T_N varied from as low as 150 K to as high as 700 K according to the composition. Recently, similar γ-Mn type antiferromagnets such as γ-Mn₃Ir have been gathering attention due to their large exchange bias effect, which plays a crucial role in magnetic memory devices to pin the ferromagnetic layer^43,44. In this context, more detailed investigations of the antiferromagnet γ-Mn type cubic phase is needed. In order to clarify the properties related to the exchange bias effect, synthesis of thin film has been deemed particularly necessary.

The Mn₃Ge has diverse intriguing phases which can be synthesized by varying the synthesis conditions. Thus, we found it crucial to focus on the proposed phases with interesting magnetic properties , in order to enable the development of practical applications. In the present work, we have fabricated a Mn₃Ge thin film with an antiferromagnetic phase in a cubic structure, a combination which has never been reported before, and a ferrimagnetic phase in D0₂₂ tetragonal structure by respectively employing Pt and Ru seed layers in the DC magnetron sputtering method. We report the structural, magnetic, and transport properties of the antiferromagnetic cubic Mn₃Ge film, in comparison with the ferrimagnetic tetragonal phase.

For the thin film synthesis, we employ DC magnetron sputtering using a chamber with a base pressure of 5 × 10⁻⁷ Pa. The Pt or Ru seed layers are deposited at room temperature onto a Si/SiO₂ amorphous substrate. On top of each seed layer, we fabricate Mn₃Ge from the Mn_2.7Ge alloy target at a high temperature of 450 ^○C. The sputtering power and Ar gas pressure are optimized to be 50 W/0.5 Pa, 60 W/0.9 Pa, and 60 W/0.8 Pa for Pt, Ru, and Mn₃Ge layers, respectively. For this condition, our X-ray reflectivity measurements find the deposition rate to be 10.0 nm/s (Pt), 4.0 nm/s (Ru), and 0.2 nm/s (Mn₃Ge). The composition of the film is examined by Scanning Electron Microscopy/Energy Dispersive X-ray Spectroscopy (SEM-EDX). Mn_3.12Ge_0.88 and Mn_3.10Ge_0.9 are observed for films with the Pt and Ru seed layers, respectively. The surface morphology is investigated by the atomic force microscope (AFM) method. A typical root mean square surface roughness is ∼ 2.0 nm in Ru/Mn₃Ge and ∼ 0.3 nm in Pt/Mn₃Ge. In order to study the exchange bias effect, an additional stack was prepared by the deposition of Py on top of those two films under the same conditions, with the Py layer sputtered at room temperature. The entire sample fabrication procedure is conducted successively without breaking the vacuum.

The structural analysis is performed by X-ray diffraction (XRD) spectroscopy (CuKα with a wavelength of 1.54 Å.) The magnetization is measured with a commercial SQUID magnetometer (MPMS, Quantum Design). The diamagnetic contribution of the Si/SiO₂ amorphous substrate is separately measured and subtracted from the magnetization below to estimate the signal from the samples. The Hall resistivity (ρ_H) is measured by a commercial physical property measurement system (PPMS, Quantum Design). The Hall resistivity ρ_H is defined as V_y ⋅ d/I_x, where I_x, V_y, and d stand for the current, the transverse voltage and the thickness,respectively.

The crystal structure of the films with varying seed layer is investigated by the XRD method. Fig. 1(a) shows the XRD patterns of the Mn₃Ge films (100nm) with the Pt and Ru seed layer. We have confirmed that both films are the single phase of Mn₃Ge and no additional peaks coming from plausible impurity phases are detected within the limits of experimental accuracy. Besides the peaks from the Si/SiO₂ substrate, the Mn₃Ge with the Ru layer presents three different peaks indicating a randomly oriented polycrystalline is formed. All the peaks can be indexed by the Bravais lattice with the lattice constants a = 2.69Å and c = 3.62Å, consistent with the tetragonal I4/mmm symmetry of D0₂₂ Mn₃Ge (Fig. 1(b))⁴. The ratio of the peak intensity is almost consistent with the theoretical simulation.

As the seed layer changes from Ru to Pt, the XRD patterns exhibit two peaks different from the tetragonal phase. According to this pattern, Mn₃Ge film with the Pt layer is oriented in the (111) direction and the estimated lattice constant is a = 3.74Å. The peak in this pattern at 42.01 ^○ is particularly close to the tetragonal (112) peak, however, Pt/Mn₃Ge shows totally different magnetic properties from Ru/Mn₃Ge, as will be seen. And thus, the possibility that Pt/Mn₃Ge might also have tetragonal structure is denied. Moreover, the lattice constant of the reported α-Mn type Mn₃Ge cubic phase (a = 8.88ų⁸) is quite different from that of our film, which suggests that the Pt/Mn₃Ge sample have another crystal structure. We assume that the XRD patterns are close to the simulation of Mn₃Ge cubic phase calculated based on the γ-Mn type structure, and confirm their consistency. The γ-Mn system is known to generally have 3Q (non-coplanar) spin order in the face-centered-cubic (fcc) structure⁴⁵. The crystal and (3Q) magnetic structure in the γ-Mn case is presented in Fig. 1(c). We note that in Mn-Ga compounds, Mn₃Ga heteroepitaxy cubic thin films with disordered fcc structure (Fig. 1(d)) have recently been reported^46,47. They have also been confirmed to possess an antiferromagnetic state. It should be mentioned that reference Mn₃Ge films with no seed layer (directly deposited on the amorphous substrate) do not show any observable peaks except for the peaks from the substrate.

As for verifying the magnetic phase properties of the Mn₃Ge films on the Ru and Pt seed layer, we performed the magnetization measurement under a field of up to 5 T at room temperature. Fig. 2(a) shows the result for the film with the Ru seed layer. The magnetization M exhibits anisotropic behaviour. In the case the field is applied along the in plane direction, a hysteresis with the spontaneous component of 50 emu/cc at H = 0 was observed (red line in Fig. 2(a).) Meanwhile when the field is applied along the out of plane direction, no clear hysteresis is presented and M becomes smaller at H = 0 to be 10 emu/cc (blue line in Fig. 2(a).). The magnetization is not fully overlapped and a small hysteresis is observed even in the out of plane configuration possibly due to the polycrystalline character of the film. Fig. 2(b) exhibits the magnetization in the film with the Pt layer. It is clearly seen that the magnetization decreases compared to the tetragonal phase and any hysteresis is no longer observed. It is listed as one of the future tasks to examine the temperature dependence of magnetic susceptibility to clarify the T_N of the γ-Mn type Mn₃Ge film.

To reveal the transport properties related to their magnetic phases, we measure the Hall resistivity ρ_H at 300 K. The magnetic field is applied along the out of plane direction (perpendicular to the film surface). The Hall resistivity of the Mn₃Ge with the Ru layer (Fig. 3(a)) is saturated at the coercivity of 0.7 T, which is consistent with the results seen in the magnetization curve (Fig. 2(a)), where the hysteresis loop is not closed under the applied field of up to 5 T. The large coercivity is in good agreement with previously reported D0₂₂ tetragonal Mn₃Ge films^32,36. From the high-field data with a linear dependence, we calculate the carrier densities of n = 3.0 × 10²² cm⁻³, which is also similar to the reports¹³. The inclination of the Hall resistivity loop may come from the small ordinary Hall resistivity.

The result of the film with the Pt layer (Fig. 3(b)) again shows significantly different properties from the tetragonal phase. It does not include hysteresis, and almost no anomalous Hall resistivity is observed. This Hall measurement result, together with the obtained magnetization curve, suggests that the Pt(35)/Mn₃Ge(100) layer should have either paramagnetic or antiferromagnetic state at room temperature.

The exchange bias (EB) effect induced at the interface between a ferromagnet (FM) and an antiferromagnet (AFM) has been exploited for spintronics applications such as magnetoresistive random access memory (MRAM)^35,48 and magnetic sensors^49,50. In these devices, EB plays the role of pinning the magnetization of a reference layer along one direction so that the two distinct states in magnetoresistance effect can be easily distinguished⁵¹.

From the perspective of AFM materials, however, EB could be a strong evidence to prove the presence of AFMs itself in multilayered structure. For example in hexagonal Mn₃X (X = Ge, Ga, Sn) compounds which are known to show antiferromagnetism, EB effect has been confirmed in their thin films^23–28. EB causes the enhancement of coercivity and the shift in the center of the magnetization hysteresis loop of the FM layer. This shift, called EB field (H_ex), occurs along the applied field axis, and always in the opposite direction of the field. The shift in the hysteresis loop is thought to be related to frustration of uncompensated spins at the AFM/FM interface⁵¹. The strength of the EB depends strongly on the magnetic structure, anisotropy, crystalline order, thickness, and texture of the AFM for a given system²⁵.

In order to clarify the magnetic properties of our film with the Pt layer, we have investigated an in-plane EB effect. Fig. 4(a) shows in-plane magnetization curves of Pt(35)/Mn₃Ge(100)/Py(5) sample measured at various temperatures after field annealing at 400 K under 5 T. This graph is after correction for a vertical shift at each temperature under 350 K and normalization based on the saturated magnetization M_s of ∼ 700 emu/cc at 1 T. The M_s of Pt(35)/Mn₃Ge(100)/Py(5) sample should arise from the 5 nm ferromagnetic Py layer, which is close to the bulk value for Py (typically around 750 ∼ 800 emu/cc⁵².) The horizontal shift of the magnetization hysteresis loops under 350 K indicates the presence of the exchange coupling between the Py layer and the Mn₃Ge on the Pt layer.

It should be noted that the blocking temperature T_B at which the exchange bias vanishes is not so clear if one considers only this experimental result. Basically T_B is known to be lower than T_N of the AFM layer because of a finite size effect^53,54. Futher detailed investigation to determine T_B such as a measurement using York Protocol⁵⁵ would be interesting. Therefore, here we can only mention that the T_N and T_B of our Mn₃Ge film with the Pt layer is higher than 350 K.

The magnitude of the in-plane exchange bias field H_ex measured for both Pt(35)/Mn₃Ge(100)/Py(5) and Ru(20)/Mn₃Ge(100)/Py(5) samples at each temperature from 400 K to 200 K is plotted in Fig. 4(b). As can be seen, any sizable EB effect in Ru(20)/Mn₃Ge(100)/Py(5) film which has the tetragonal structure is not detected over the measured temperature range. We would like to attribute the EB effect to the AFM order of the γ-Mn type cubic Mn₃Ge, which causes an anisotropy in the Py film induced at the interface. As presented in Fig. 4(b), H_ex takes the largest value of 11.5 mT at 200 K. This H_ex is comparable to other γ-Mn compounds^43,44. In contrast to those other compounds, however, Mn₃Ge is inexpensive and abundant on the earth. It should also be noted that, since a sharper increase of H_ex is observed at the low temperature region for other compounds, a similar behaviour is expected and needs to be further confirmed in the case of our film. It should be noted that the coercivity of the Py layer is also enhanced compared to a single Py film, whose magnitude is usually fixed around 3 mT as shown in Fig. 4(b).

We have succeeded in fabricating thin films of two different Mn₃Ge phases by employing seed layers on top of thermally oxdized Si substrates. Among them, the Mn₃Ge film on the Pt layer may have the γ-Mn type cubic structure with antiferromagnetic properties. Although this antiferromagnetic thin film consists of Mn and Ge and does not include any rare earth metal, it demonstrates large exchange bias at room temperature. The Mn-Ge system is attractive regarding various intriguing phases and our result paves the way to investigate yet unknown phases even further.

We thank Takashi Nishikawa, Zhu Zheng, and Danru Qu for useful discussions. We also thank Daisuke Hamane for the SEM-EDX and M. Lippmaa for the AFM measurements. This work is partially supported by CREST(JPMJCR18T3), Japan Science and Technology Agency, by Grants-in-Aids for Scientific Research on Innovative Areas (15H05882 and 15H05883) from the Ministry of Education, Culture, Sports, Science, and Technology of Japan, and by Grants-in-Aid for Scientific Research (19H00650).