Perpendicular magnetic anisotropy of (001)-textured poly-crystalline MnAlGe films

Cu 2 Sb-type intermetallic compound, MnAlGe, is known to exhibit uniaxial magnetocrystalline anisotropy and relatively small saturation magnetization, which is suitable for spintronic application, e.g. spin-transfer phenomena requiring small critical current density. Ge-concentration dependence of the crystal structures, saturation magnetization, M s , and perpendicular magnetic anisotropy, K u was investigated for MnAlGe films prepared onto silicon substrates with a thermally oxidized amorphous layer. For the stoichiometric and Ge-rich samples, the films exhibited (001)-texture with perpendicular magnetization. The maximum values of M s and K u were 270 emu/cm 3 and 4.8 × 10 6 erg/cm 3 , respectively, which were comparable values with those reported for bulk and epitaxially grown films in literatures.


INTRODUCTION
Cu 2 Sb-type intermetallics, (XX')Z, form a crystal structure shown in Figure 1 which is a tetragonal C38 phase of the space group P4/nmm, No. 129. For typical compounds, the X atom is a 3d-transition metal, the X' atom is a 3d-transition metal or a group 13 element in the periodic table, and Z atom is a group 15 element. The atomic positions are the Wyckoff 2a (000), ( 1 2 1 2 0), 2c (0 1 2 uc), ( 1 2 0ūc), and 2c (0 1 2 vc), ( 1 2 0vc), for X, X', and Z atoms, respectively. Magnetism of Mn-based intermetallics, such as Mn 2 Sb, (Mn-Cr)Sb, and CuMnAs showing the Cu 2 Sb-type crystal structure varies with the composition, [1][2][3][4][5] and the itinerant nature of electrons has been of interest. 6,7 Among the materials, equiatomic MnAlGe is known to exhibit relatively small saturation magnetization, Ms, and uniaxial magnetocrystalline anisotropy, [8][9][10] which are attractive for spintronics applications utilizing spin-transfer phenomena. In addition to the magnetic properties, the MnAlGe and the related compounds, such as (Mn-Cr)AlGe offer another merit of relatively stable (001)-oriented film growth, [11][12][13] which fits with MgO-based magnetic tunnel junctions with the (001)-orientation for large tunnel magnetoresistance. 14,15 The crystal structure and magnetic properties of MnAlGe in film form have been investigated intensively for epitaxially grown samples, and the dependence on film compositions was reported. 16,17 On the other hand, correlations between magnetic properties and the chemical order of the compounds have not been investigated systematically. In addition, for the poly-crystalline MnAlGe films, although the literature reported perpendicular magnetization, 13 the details for crystal structures and quantitative magnetic properties has been unclear. In this study, Ge concentration dependence is investigated in poly-crystalline Mn-Al-Ge films.

EXPERIMENTAL PROCEDURES
Film samples were deposited using ultra-high-vacuum magnetron sputtering system with base pressure less than 3 × 10 −7 Pa.  The films were deposited onto thermally oxidized silicon substrates. The stacking structure was as follows: Substrate | Mn-Al-Ge 100 nm | MgO 2 nm | Ta 3 nm. Co-sputtering technique was used for the deposition of Mn-Al-Ge layer with an Mn-Al alloy target and a Ge target, and four series of samples were prepared with different Ge concentration including no Ge sample, Mn-Al, as summarized in Table I. All layers were deposited at room temperature, and after depositing capping layers consisting of MgO | Ta, post-annealing was carried out using a vacuum furnace. The annealing temperatures were in the range of 200 -500 ○ C and changed in 100 ○ C increments. The crystal structures were characterized using x-ray diffraction, XRD, for the out-of-plane and in-pane configurations. In addition, XRD measurements for the grazing incidence angle were also carried out around the 011 and 112 diffractions. The magnetization curves were measured by using vibrating sample magnetometer, VSM. The maximum applied fields were 25 kOe and 90 kOe for the measurements with the perpendicular-to-plane and in-plane directions, respectively. All measurements were carried out at room temperature.

RESULTS AND DISCUSSION
The out-of-plane and in-plane XRD profiles are shown in Figure 2. Diffractions from the Cu 2 Sb-type structure are marked by ▼, and the diffraction indexes are shown on the top of the graphs. The diffractions from the silicon substrates and an unknown phase are marked by * and ▽, respectively. The annealing temperatures are 400 ○ C for the no-Ge and Ge-poor samples ( Fig. 2(a-d)), and 500 ○ C for the stoichiometric and Ge-rich samples ( Fig. 2(e-h)), which are optimum annealing temperatures showing maximum Ms as it is discussed in the following part except for Ge-poor samples showing no hysteresis loop for all the annealing temperatures. In the XRD profiles, the no-Ge sample exhibits a 111 diffraction which is clearly seen for the out-of-plane geometry shown in Fig. 2(a). Although the intensity is very weak, a tiny diffraction is also seen for the in-plane geometry ( Fig. 2(b)). The 111 diffraction is possibly from the face-centered cubic, fcc, or face-centered tetragonal, fct, structure. With a small Ge concentration, the Ge-poor sample exhibits several weak diffraction peaks from the Cu 2 Sb-type structure and possibly from the fcc or fct structure. On the other hand, the stoichiometric and Ge-rich samples clearly exhibit diffraction peaks from the Cu 2 Sb-type structure, in addition, a series of 00l (l = 1, 2, 3 and 4) diffractions are observed in the out-ofplane XRD spectra, while those diffractions are invisible for the inplane configuration suggesting that the films were grown with the ARTICLE scitation.org/journal/adv (001)-texture. Note that an unknown diffraction peak is also observed in the out-of-plane XRD profile of the Ge-rich sample. The annealing temperature dependence exhibited negligibly small difference in the XRD patterns in the temperature range of 300 -500 ○ C, however, no diffraction peak was observed in all samples annealed at 200 ○ C for which no hysteresis loop was observed in magnetization curves measurements. Lattice parameters for c-axis (out-of-plane), a-axis (in-plane), and c/a ratio are summarized as a function of the annealing temperature for stoichiometric and Ge-rich samples in Figure 3. The lattice parameters were not evaluated for no-Ge or Ge-poor samples, because diffraction peaks were very weak and insufficient to determine both values for the cand a-axes quantitatively. Concerning the stoichiometric samples, the lattice parameter of c-axis slightly increase with the annealing temperature, while a-axis exhibits nearly no change, and as a result, c/a ratio slightly increases. The trend is similar for the Ge-rich samples with that of the stoichiometric samples, however, the change of c-axis is relatively large. The annealing temperature dependence of lattice parameters is possibly caused by relaxation of strain which was, e.g., induced during sputtering process, and an increase of volume fraction of the Cu 2 Sb phase. Compared with values in a bulk sample 8 and calculated values, 11 the value is smaller (larger) for c-(a-)axis, which results in the relatively small c/a ratio for the present film samples, and the deviation from the literature values is larger in the Ge-rich samples than that in the stoichiometric samples.
Relative integrated diffraction peak intensities of I 001 /I 002 and I 011 /I 112 are summarized as a function of the annealing temperature in Figure 4. Simulation values for a fully ordered MnAlGe with the Cu 2 Sb-type structure are also indicated using broken lines in the figure, for which the calculated energetically stable lattice parameters in Ref. 11 were used. Here, the ratio of I 001 /I 002 depends on the layer-by-layer chemical order of the Mn-and the (Al-Ge)-layers in the Cu 2 Sb-type structure. On the other hand, the ratio of I 011 /I 112 depends on the chemical order between the Al sites and the Ge sites in the plane. Note that the integrated peak intensities also depend on the values of uc and vc, however, possible changes of uc and vc are neglected in the following discussion because of no experimental values due to the limited XRD peaks for the present film samples. Although all hkl diffractions are allowed for the present P-lattice in the Bravais family even in case of the disordered situation, those specified diffraction intensity ratios become weak because of different atomic scattering factors. For the stoichiometric samples, the values of I 001 /I 002 exhibit nearly no change between the annealing temperatures of 300 and 400 ○ C, and it slightly increases at 500 ○ C. Regarding I 011 /I 112 for the stoichiometric samples, the values increase with the annealing, for which the difference is relatively large between the data points at 300 and 400 ○ C. These results suggest that the chemical order for the (001)-planes were slightly increased by the annealing at 500 ○ C, while the chemical order insides the Al-Ge planes were drastically promoted by the annealing at 400 ○ C. On the other hand for the Ge-rich samples, both values of I 001 /I 002 and I 011 /I 112 increase with the annealing temperature, suggesting that both chemical order for (001)-planes and Al-Ge sites were improved by the annealing over 300 ○ C.
Magnetization curves are shown in Figure 5. The annealing temperatures were 400 ○ C for Fig. 5(a, b), and 500 ○ C for Fig. 5(c, d). Magnetization curves for all samples including other annealing temperatures are shown in Figure S1 (a, b), and 500 ○ C for (c, d). and ∥ represent the applied magnetic field directions of perpendicular-to-plane and in-plane, respectively. and the squareness, each other. The trend is similar for other annealing temperatures of no-Ge samples, however, Ms was less than 100 emu/cm 3 (Fig. S1(e)). For the Ge-poor sample, no-hysteresis is observed for both perpendicular and in-plane directions, which were nearly the same for other annealing temperatures (Figs. S1(b, f, j, n)). The reason is unclear for no hysteresis in the Ge-poor samples, however, the reduction of Curie temperature and/or the cancelation of spin moments might happen considering the very week XRD patterns suggesting a small volume fraction of the Cu 2 Sb-type structure as well as a possible mixture of the fcc/fct phase. On the other hand, although finite coercivity is found for the in-plane magnetization curves, both stoichiometric and Ge-rich samples clearly exhibit perpendicular magnetization. From the magnetization curves, Ms and the effective perpendicular magnetic anisotropy energy, K eff u were evaluated. In this study, K eff u was calculated from the area enclosed by the perpendicular-to-plane and in-pane magnetization curves. Using the K eff u , the perpendicular magnetic anisotropy energy, Ku, of films is defined as Ku = K eff u + 2πM 2 s . The values of Ms and Ku are summarized as a function of the annealing temperature in Figure 6 for the no-Ge (only Ms values are shown), stoichiometric and the Ge-rich samples. For the no-Ge samples, Ms values exhibit maximum at the annealing temperature of 400 ○ C and decreases at 500 ○ C. The drop of Ms at the high-temperature annealing is possibly caused by an appearance of another phase than the ferromagnetic τ-phase in Mn-Al which is a meta-stable phase. For the stoichiometric and Ge-rich samples, both Ms and Ku increase with the annealing temperature, and the stoichiometric samples exhibit larger values for all the temperature range. The maximum values of Ms and Ku are 270 emu/cm 3 and 4.8 × 10 6 erg/cm 3 , respectively, for the stoichiometric sample annealed at 500 ○ C. These maximum values are comparable with those for a single-crystalline bulk and an epitaxially grown MnAlGe film reported in literatures. 10,18 Regarding correlations between the magnetic properties and the integrated diffraction intensity ratios shown in Fig. 4, firstly in the Ge-rich samples, I 001 /I 002 closely correlates with both Ms and Ku. In addition, for the stoichiometric samples, enhancements of Ms and Ku from the annealing temperatures of 400 to 500 ○ C possibly correspond to the small increase of I 001 /I 002 . On the other hand, regarding the ratios of I 011 /I 112 , even though the values increase with the annealing temperature, no clear correlation is seen in the magnetic properties: Relatively large Ms and Ku were also achieved in the samples annealed at 300 ○ C especially in the stoichiometric sample, for which the Al-Ge sites are considered to be nearly random. These experimental results suggest that the order of the (001)-planes is crucial for the magnitude of the Mn-moment and anisotropy, which is reasonable because it was reported that the Mn atoms present strong itinerant nature when those locate at the Wyckoff 2a positions, while the nature of the electron changed when those locate at the 2c positions. 7

SUMMARY
Ge-concentration dependence of crystal structures, Ms, and Ku were investigated for poly-crystalline MnAlGe films onto thermally oxidized silicon substrates. The films clearly exhibited the Cu 2 Sb-type structure with (001)-orientation for the stoichiometric and Ge-rich samples, while other samples with no-Ge and Ge-poor concentration exhibited weak XRD peaks with no particular orientation. Hysteresis loops showing the perpendicular ARTICLE scitation.org/journal/adv magnetization were achieved in the stoichiometric and the Ge-rich samples. The maximum values of Ms and Ku of 270 emu/cm 3 and 4.8 × 10 6 erg/cm 3 , respectively, were achieved in the stoichiometricsample, which were comparable with those reported in singlecrystalline bulk and the epitaxially grown film samples in the literatures.

SUPPLEMENTARY MATERIAL
See a supplementary material for magnetization curves for all samples in this study.