Highly Reduced Saturation Magnetization in Epitaxially Grown Ferrimagnetic Heusler Thin Films.

The key of spintronic devices using the spin-transfer torque phenomenon is the effective reduction of switching current density by lowering the damping constant and the saturation magnetization while retaining strong perpendicular magnetic anisotropy. To reduce the saturation magnetization, particular conditions such as specific substitutions or buffer layers are required. Herein, we demonstrate highly reduced saturation magnetization in tetragonal D022 Mn3-x Ga thin films prepared by rf magnetron sputtering, where the epitaxial growth is examined on various substrates without any buffer layer. As the lattice mismatch between the sample and the substrate decreases from LaAlO3 and (LaAlO3)0.3(Sr2AlTaO6)0.7 to SrTiO3, the quality of Mn3-x Ga films is improved together with the magnetic and electronic properties. Especially, the Mn3-x Ga thin film epitaxially grown on the SrTiO3 substrate, fully oriented along the c axis perpendicular to the film plane, exhibits significantly reduced saturation magnetization as low as 0.06 μB, compared to previous results. By the structural and chemical analyses, we find that the predominant removal of Mn II atoms and the large population of Mn3+ ions affect the reduced saturation magnetization. Our findings provide insights into the magnetic properties of Mn3-x Ga crystals, which promise great potential for spin-related device applications.

Introduction

Spintronics has received intensive interest because of further degree of freedom by spin transport, leading to efficient applications in information storage and processing.[1,2] To realize the elaborate spintronic devices, low damping constant, small saturation magnetization, large spin polarization, and strong perpendicular magnetic anisotropy (PMA) are requisitely demanded.[3−6] Recent progressive achievements have highlighted to generate the spin-polarized current, particularly in spin-transfer-torque (STT) memory devices.[7,8] The reduction of switching current density based on the STT phenomenon has garnered attention in magnetic-tunnel-junction (MTJ) technologies.[9] Mn2-based Heusler compounds are exploited as promising magnetic components in MTJ devices, which possess low saturation magnetization, strong PMA, and high Curie temperature.[10,11] Mn3Ga exhibits a variety of magnetic properties, which strongly depend on the crystal structure.[12−16] The hexagonal D019 phase in equilibrium is antiferromagnetic with noncollinear triangular magnetic structure, demonstrating remarkable exchange bias effect due to frustrated magnetic structure.[12,13,17] The metastable cubic L21 phase is discussed as an antiferromagnetic half-metal,[3] which is experimentally observed with zero net magnetization.[14,18] When the cubic structure is distorted along the c direction, one obtains the tetragonal D022 phase, which is ferrimagnetic with two inequivalent positions and magnetic moments of Mn atoms.[15,16,18−22] As shown in Figure a, one Mn I and two Mn II sublattices occupy the 2b (0, 0, 1/2) and 4d (0, 1/2, 1/4) positions, respectively, and the Ga atom is at the 2a (0, 0, 0) position.[19,23] From neutron experiments, the Mn I atoms at 2b with a magnetic moment of −3.07 μB are ferromagnetically aligned along the c axis, and antiferromagnetically coupled with Mn II atoms at 4d with 2.08 μB, leading to a ferrimagnetic state with a net moment of 1.09 μB.[18] Thus, the magnetization is strictly governed by the number of Mn I and Mn II atoms as well as the magnetic coupling between Mn atoms.

Figure 1

(a) Crystal and magnetic structure of Mn3Ga and (b) XRD patterns of Mn3–Ga films deposited on STO, LSAT, LAO substrates, where the substrate peaks are indicated by circle, triangle, and diamond symbols, respectively. The y-scale is plotted in logarithmic scale.

In experiment, this material is usually present in a lack of Mn form, that is Mn3–Ga, where the saturation magnetization, MS is enhanced with the Mn vacancy, x. Here, the mostly accepted picture is that Mn I deficiency increases the magnetization, whereas Mn II deficiency decreases the magnetization. Thus, the increased MS is believed to be because of the removal of Mn I because the removal of Mn I weakens the hybridization between Mn I and Ga along the c axis and elongates the c-axis lattice parameter.[23−25] Such increased MS can be an impediment to reduce the switching current density in spintronic device applications using the STT phenomenon. Recently, the substitution of transition metals (Tr = Fe, Co, Ni, and Cu) in Mn3–TrGa has been extensively studied as one way to effectively reduce the magnetization,[26−30] but often leading to incomplete PMA properties.[30,31] In an ordered manner, the transition metal element is considered to occupy the Mn II site at the 4d position. This result is compared to the easy removal of Mn I in Mn3–Ga. On the other hand, some theoretical studies have proposed that the total magnetic moment can decrease with the lack of Mn because of possible removal of Mn II instead of Mn I.[23,30,31] However, there has been no experimental result on the decrease of MS by the removal of Mn II in Mn3–Ga.

In order to overcome the lattice mismatch at the interface and obtain the epitaxial thin films of Mn3–Ga, various substrates and numerous buffer layers have been used.[4,32−35] Most films are grown with the easy magnetization pointing along the tetragonal c axis, which is perpendicular to the film plane. There have been some reports asserting the successful growth of epitaxial Mn3–Ga thin films.[16,34,35] Nevertheless, there has been observed a small kink structure around zero field in the magnetic hysteresis, which is ascribed to a small fraction of c-axis-oriented magnetic components in the film plane.[33] Furthermore, experiments have provided a variety of MS values ranging from 0.3 to 1.6 μB,[32,36] which depend on the substrates and buffer layers. Here is a challenging issue as to how the magnetic properties are evolved by improving the quality of epitaxial films.

Holding this kind of issues on which the Mn site is easily removed and substituted in different ways, herein, we have chosen a fairly Mn-deficient Mn3Ga with the D022 tetragonal structure for obtaining epitaxial thin films. We report a highly reduced MS in Mn3–Ga thin films, directly grown on various substrates of SrTiO3 (STO), (LaAlO3)0.3(Sr2AlTaO3)0.7 (LSAT), and LaAlO3 (LAO) without any buffer layer. Although all the samples exhibit low MS, compared to previously reported values, the Mn3–Ga film that epitaxially grows on the STO substrate shows the lowest MS, where the lattice mismatch between the sample and the substrate is the smallest among the growth substrates. Reducing the lattice mismatch from LAO and LSAT to STO improves the film quality, which is evident in the X-ray diffraction (XRD) and transmission electron microscopy (TEM) results. The magnetic and electronic properties are also dependent on the growth substrates. Especially, in the Mn3–Ga film epitaxially grown on the STO substrate, MS reaches about 10 emu/cm3 (=0.06 μB), which is one order lower than the previous results.[32,36] From the analyses of crystal structure and chemical state of Mn3–Ga thin films, the most possible origins for the low MS in Mn3–Ga are discussed with the predominant removal of Mn II atoms and the large population of Mn3+ ions. This result has a good advantage for spin-related applications such as reducing the critical current density of the STT-based devices.

Results and Discussion

The chemical composition has been analyzed by the scanning electron microscopy–energy-dispersive spectroscopy (SEM–EDS) measurements, giving rise to the composition of roughly Mn/Ga = 2.2:1 for all the films deposited on different substrates of STO, LSAT, and LAO, although we used a single target of Mn3Ga. This demonstrates clear Mn deficiencies from the stoichiometric Mn3Ga phase, that is denoted as Mn3–Ga. The structural changes of Mn3–Ga films depending on the substrates of STO, LSAT, and LAO have been investigated by XRD analyses. In Figure b, the sample on STO displays the tetragonal D022 structure (space group I4/mmm) labeled with the c-axis indices of (002) and (004), which are perpendicular to the film plane. The c-axis lattice parameter is estimated to be 7.17 Å, which is the same value reported by other literatures on the tetragonal Mn3Ga phase.[16,18,23] For the LSAT substrate, the (004) diffraction peak exhibits a broad shoulder around 2θ = 51°, reflecting the presence of a small misalignment from the c axis. The evaluated c-axis lattice parameter from the main (004) peak is slightly larger, c = 7.33 Å, as indicated in Table . The lattice parameter expansion along the c axis may be attributed to the reduction in the a-axis lattice parameter of the substrate to keep the mass density constant. For films grown on the LAO substrate, only peaks from the substrate are observed, although the existence of Mn and Ga elements was confirmed in the SEM–EDS results.

Table 1

Lattice Constant (asub) of STO, LSAT, and LAO Substrates, Lattice Mismatch between Mn3–Ga and Substrate, a- and c-Lattice Parameters of Mn3–Ga Obtained from XRD and TEM Measurements, Saturation Magnetization (MS), Coercive Field (HC), Magnetic Anisotropy Constant (Ku), and RRR of Mn3–Ga Films on STO, LSAT, LAO Substrates

	asub substrate (Å)	lattice mismatch (%)	c XRD (Å)	a TEM (Å)	c TEM (Å)	MS (emu/cm3)	HC (kOe)	Ku (Merg/cm3)	RRR
STO	3.905	0.896	7.17	3.901	7.188	9.49	16.0	0.902	1.48
LSAT	3.868	1.86	7.33	3.877	7.378	19.1	22.0	1.81	1.08
LAO	3.790	3.96				21.3	19.7		1.03

More specific crystal structure of the Mn3–Ga samples can be clarified through high-resolution TEM (HRTEM) measurements. Figure shows the representative cross-sectional HRTEM images and the fast Fourier transform (FFT) patterns of Mn3–Ga films deposited on STO, LSAT, and LAO substrates. All the samples display sharp interfaces within 1 nm. In Figure a, the crystal lattice of the Mn3–Ga film on the STO substrate is characterized to be indexed to the D022 structure, being consistent with the XRD result. The FFT plot for the STO substrate shows only the diffraction peaks corresponding to (200) and (004) orientations, parallel and perpendicular to the film plane, respectively. The well-oriented lattice fringe in the HRTEM image and the well-defined spot configuration in the FFT plot imply the epitaxial nature of the Mn3–Ga film deposited on the STO substrate. From these results, we obtain the lattice parameters of a = 3.901 Å and c = 7.188 Å, which also agree well with the XRD results. The a-axis lattice parameter is almost identical to that (=3.905 Å) of the STO substrate, so that the lattice mismatch between the Mn3–Ga sample and the STO substrate is quite small (=0.89%). It may provide the environment of the epitaxial growth. For the LSAT substrate in Figure b, we also find the well-oriented lattice fringe in the HRTEM image and the well-defined spot configuration in the FFT plot, similar to the Mn3–Ga film on STO. Moreover, we observe moiré interference patterns as shown in the inset of Figure b, which suggests the presence of small misalignment or misorientation in lattice. The FFT pattern of the whole area shows the weakened intensity of the (004) orientation, also reflecting a small misalignment in the c-axis-oriented matrix. This is consistent with the XRD result showing the broad (004) peak. The lattice parameter of c = 7.378 Å, close to the value obtained from the XRD measurement, is larger than that for the Mn3–Ga film on STO, whereas the a-axis lattice parameter of 3.877 Å is smaller. Because the lattice parameter of the LSAT substrate is smaller than that of the STO substrate, it is reasonable to obtain the reduced a-axis lattice parameter in the Mn3–Ga film on the LSAT substrate, followed by the expanded c-axis lattice parameter, in a viewpoint that the mass density keeps constant. The lattice mismatch obtained for the LSAT substrate is slightly larger (=1.86%), which may be a source for the misalignment in lattice. However, for the LAO substrate, small grains are observed in the entire area as shown in Figure c, leading to the scattered FFT patterns by the superposed small grains. This represents a considerable amount to deviate from the well oriented single-crystal phase, which is responsible for no observable XRD peaks in the case of the LAO substrate. These structural changes depending on the substrates can have influence on the electronic and magnetic properties.

Figure 2

TEM images at the interface between the Mn3–Ga sample and the substrates of (a) STO, (b) LSAT, and (c) LAO, and the FFT patterns obtained from each sample and the substrate. The inset in (b) shows the moiré interference pattern for Mn3–Ga film deposited on the LSAT substrate.

The electrical resistivity ρ(T) curves of Mn3–Ga films on different substrates are shown in Figure , where the values are normalized to the value at 4 K for better comparison. All the Mn3–Ga films exhibit metallic behavior with a positive slope of ρ(T) over the entire temperature range. Note that at low temperatures below 50 K, the ρ(T) curves for LSAT and LAO substrates show small upturns, whereas ρ(T) for the STO substrate is almost constant. Such resistivity minimum has been frequently observed in metallic systems with high chemical disorder, structural disorder, or grain boundary scattering.[37−39] Especially, the disorder-related phenomena, for example, the orbital two-channel Kondo effect has been exploited in MnGa and MnAl, which are strongly dependent on the structural disorder.[40,41] In the ρ(T) data for Mn3–Ga films on LSAT and LAO substrates, we have obtained the resistivity upturn with a clear transition from the ln T scaling to the T1/2 scaling, similar to the orbital two-channel Kondo behavior with enhanced disorder.[40,41] Thus, the upturn features of ρ(T) in this study can be related to the structural disorder of the films on LSAT and LAO substrates, based on the results obtained from the structural analyses. From the temperature dependence, we determine the residual resistivity ratio (RRR) given by RRR = ρ(300 K)/ρ(4 K), which decreases from 1.48 for STO substrate to 1.10 for LSAT and 1.04 for LAO. This indicates the higher quality of Mn3–Ga film on STO substrate than that on other substrates, which is in good agreement with the results analyzed from the XRD and TEM measurements.

Figure 3

Electrical resistivity ρ(T) of Mn3–Ga films deposited on STO, LSAT, and LAO substrates, which are normalized to the value at 4 K for comparison. The insets show the low-temperature plots for the films on LSAT, and LAO substrates.

Figure a–c show the in-plane and out-of-plane magnetization M(H) curves measured at 300 K for Mn3–Ga films deposited on STO, LSAT, and LAO substrates, respectively. The magnetic signal of the sample is obtained by directly subtracting the background signal measured with only the substrate. The physical parameters obtained from the M(H) curves are summarized in Table , where the structural parameters are also listed. With increasing the lattice mismatch from STO and LSAT to LAO, the saturation magnetization MS of out-of-plane component increases, and the in-plane magnetic component is enhanced, while keeping the coercivity around 20 kOe. The M(H) curve for the sample on the STO substrate seems to be well c-axis oriented because there is no in-plane magnetic component around zero field. The most prominent result in the magnetic properties is the low MS value (=10 emu/cm3), especially for the STO substrate, which is much smaller than those (50–270 emu/cm3) reported previously.[32,36] Because of the low MS, the magnetic anisotropy constant Ku is also found to be small, compared to the previous reports.[4,16] For the LSAT substrate, the MS value of out-of-plane component is still small about 20 emu/cm3, and both in-plane and out-of-plane magnetization curves exhibit small kinks at low magnetic fields. Such kink structures can be understood with the small misorientation, which has been already discussed in the XRD and TEM results. The lattice misorientation is maximized in Mn3–Ga film on the LAO substrate, which shows no specific preferred growth orientation. As already observed in the FFT pattern of TEM, there is a considerable amount deviating from the oriented single-crystal phase, giving rise to the polycrystalline nature with no magnetic anisotropy.

Figure 4

In-plane and out-of-plane magnetization M(H) measured at room temperature for Mn3–Ga films deposited on (a) STO, (b) LSAT, and (c) LAO substrates.

As discussed in the introduction part, there are two possible structural situations for Mn removal.[23,24] Taking into consideration the ferrimagnetic state of Mn3Ga, the removal of Mn I (Mn II) increases (decreases) the total magnetic moment. The removal of Mn I atoms from the 2b position is based on the experimental observation that the magnetization increases with increasing the deficiency of Mn, x in Mn3–Ga.[24,25,32,36] However, the experimentally observed increment of magnetization with the Mn deficiencies is too small to be explained with only the removal of Mn I, so that the simultaneous removal of Mn II atoms has been proposed for another scenario,[23,30] but it is still more probable that more Mn I atoms are removed than Mn II atoms. In this study, the MS value obtained in the epitaxial Mn3–Ga film on the STO substrate is roughly 10 emu/cm3, which is much smaller than those reported in previous studies.[32,36] Although not so small, relatively low MS values of about 50 emu/cm3 have been reported using the buffer layers such as GaAs and Cr/Pt, where the crystalline quality is improved by the reduction of lattice mismatch.[32,36] On the other hand, our Mn3–Ga films are epitaxially grown on the STO substrate without any buffer layer and exhibit surprisingly low MS values while keeping the PMA. The absence of buffer layer may provide an environment for making a metastable state, in which Mn II atoms are predominantly removed.

According to earlier reports,[23] the c-axis lattice parameter has been used as a measure to determine which Mn site is vacant in Mn3–Ga. The c-axis lattice parameter increases with increasing the vacancy of Mn I by weakening the hybridization between Mn I and Ga along the c axis, whereas the vacancy of Mn II has little effect on the c-axis lattice parameter. In the present work, the c-axis lattice parameter of Mn3–Ga grown on the STO substrate is almost same as the value for Mn3Ga (where Mn I and Mn II sites are fully occupied) and slightly smaller than the value for Mn2Ga (where only Mn I site is removed), implying that the Mn I site is almost occupied and there is a considerable amount of vacant Mn II sites. This result can also explain the low MS value. When Mn I is removed, as described above, the MS value should increase. However, when Mn II atoms are solely removed, net magnetization should go to zero because of the fully compensated magnetic state. Thus, in our Mn3–Ga thin films, the low MS is attributed to the predominant removal of Mn II atoms.

The predominant removal of Mn II in our system can be related to the strain between the sample and the substrate because MS of the sample changes depending on the substrate. A recent report has demonstrated that the strained or relaxed Mn3Ga films grown on different buffer layers exhibit distinguishable magnetic properties: more relaxed Mn3Ga film deposited on Mo buffer layer shows lower magnetization.[42] Fairly small strain effect for STO substrate can induce the low MS as a result of the removal of Mn II. For the LSAT substrate, the strain is supposed to be large, and hence the total magnetization increases and the in-plane ferromagnetic component appears at low fields. This soft ferromagnetic phase seems to arise from the strained regions at the interface, as proposed in the strained Mn3Ga film.[26] The strain effect is more pronounced on the LAO substrate with the largest lattice mismatch (∼3.96%). The polycrystalline nature with no magnetic anisotropy gives rise to the sufficiently large magnetization in both in-plane and out-of-plane configurations.

In addition to the structural factor of predominant removal of Mn II atoms as one possible origin to reduce MS, we have examined the influence of chemical states of Mn ions by using the X-ray photoelectron spectroscopy (XPS). Figure depicts the Mn 2p3/2 spectrum of Mn3–Ga thin film grown on the STO substrate and the deconvoluted curves for metallic Mn, Mn2+, and Mn3+ ionic states, fitted by the Gaussian–Lorentzian asymmetric function. The peaks at 640.8 and 641.9 eV correspond to Mn2+ and Mn3+ states, respectively. The ratio of Mn3+/Mn2+ valence states is estimated to be 9.93. Because the effective magnetic moments of the Mn3+ ionic state is smaller than that of Mn2+, the larger population of Mn3+ ions can be another possible origin of the reduced MS. Further experiments or theoretical calculations on strain-considered crystal and band structures, however, are needed to study the correlation between the Mn removal and the valence states of Mn atoms.

Figure 5

XPS of Mn3–Ga film deposited on STO substrate. The open square symbols represent the measured data, and the solid line represents the best fitted curve with the metallic Mn (dash-dot line) and two ionic states of Mn2+ (dotted line) and Mn3+ (dashed line).

Conclusions

Tetragonal Mn3–Ga films are grown on STO, LSAT, and LAO substrates without any buffer layer by rf magnetron sputtering. The structural, electronic, and magnetic properties are strongly dependent on the substrate. As the lattice mismatch between the sample and the substrate decreases from LAO and LSAT to STO, the film quality is improved together with the magnetic and electronic properties. In particular, Mn3–Ga grown on the STO substrate is an epitaxial film, which leads to the magnetic moment fully oriented along the c axis perpendicular to the film plane. The predominant Mn II deficiencies at 4d position and the less strained interface are verified by the elaborate XRD and TEM analyses, resulting in the enormously reduced MS as low as 0.06 μB (=10 emu/cm3). Although the Mn valence state is not characterized at a specific Mn site, large populated Mn3+ ions play a role in reducing the MS in Mn3–Ga thin films. Because the low MS is essential for reducing the critical switching current density, the epitaxially grown Mn3–Ga thin films can provide beneficial platform for developing effective STT-based devices.

Experimental Details

Growth of Mn3Ga Thin Films on Different Substrates

Mn3–Ga thin films were deposited directly on three different substrates of STO (001), LSAT (001), and LAO (001) without any buffer layers by rf magnetron sputtering. A single target of Mn3Ga was used in a vacuum chamber with the base pressure of 10–7 Torr. Varying the growth conditions, such as substrate temperature from 300 to 450 °C, argon gas pressure from 2 to 7 mTorr, and rf gun power from 30 to 55 W, was examined. Then, we found the optimal growth conditions showing the clear PMA characteristics at each substrate; 400 °C/5 mTorr/35 W for STO, 400 °C/5 mTorr/45 W for LSAT, and 350 °C/5 mTorr/45 W for LAO. More detailed description including the growth process and the corresponding magnetic properties can be found in Supporting Information. The thickness was approximately 100 nm for all the Mn3–Ga thin films. After cooling to room temperature, the Mn3–Ga thin films were capped by a 5 nm thick SiO2 layer to prevent oxidation.

Characterization and Measurements

The structural analyses were performed by XRD using a Rigaku SmartLab diffractometer with Cu Kα radiation and TEM using a Tecnai G2 F30. The chemical compositions were determined using EDS attached to a standard SEM. To examine the chemical states of the samples, XPS experiments were carried out with PHI 5000 VersaProbe (Ulvac-PHI) operated with a 15 kV accelerating voltage in a vacuum of 2 × 10–7 Pa. The magnetic and transport properties were measured by using a superconducting quantum interface device-vibrating sample magnetometer and Gifford-McMahon refrigerator. The van der Pauw method was used to measure the in-plane electrical resistivity. All the measurements were performed at room temperature because the Curie temperature of Mn3–Ga is above 700 K.