# Fileset [20241126_RDE_manuscript_MVA_CPP-GMR.pdf](https://mdr.nims.go.jp/filesets/115db015-ab7e-4b38-8239-f538f5f91acb/download) ## Creator [Hirofumi Suto](https://orcid.org/0000-0003-4387-5862), [Vineet Barwal](https://orcid.org/0000-0001-9445-5900), [Keisuke Masuda](https://orcid.org/0000-0002-6884-6390), [Kodchakorn Simalaotao](https://orcid.org/0000-0002-6098-4422), [Taisuke Sasaki](https://orcid.org/0000-0002-5952-7638), [Yoshio Miura](https://orcid.org/0000-0002-5605-5452), Hiroo Tajiri, [Loku Singgappulige Rosantha Kumara](https://orcid.org/0000-0001-9160-6590), [Tomoyuki Koganezawa](https://orcid.org/0000-0002-9302-5025), [Yuya Sakuraba](https://orcid.org/0000-0003-4618-9550) ## Rights [In Copyright](http://rightsstatements.org/vocab/InC/1.0/) ## Other metadata [Negative spin polarization and effect of composition on the atomic order and electronic structure of Mn 2 VAl Heusler alloy thin films](https://mdr.nims.go.jp/datasets/166ce7b7-1206-4056-9203-6c08c5e10d67) ## Fulltext 1 Negative spin polarization and effect of composition on the atomic order and electronic 1 structure of Mn2VAl Heusler alloy thin films 2 Hirofumi Suto1*, Vineet Barwal1, Keisuke Masuda1, Kodchakorn Simalaotao1, Taisuke Sasaki1, 3 Yoshio Miura1, Hiroo Tajiri2, Loku Singgappulige Rosantha Kumara2, Tomoyuki Koganezawa2, 4 and Yuya Sakuraba1 5 1Research Center for Magnetic and Spintronic Materials, National Institute for Materials Science 6 (NIMS), Tsukuba, 305-0047, Japan 7 2Japan Synchrotron Radiation Research Institute (JASRI), Kouto, 679-5198, Japan 8 *SUTO.Hirofumi@nims.go.jp 9 10 Abstract 11 We explored Mn2VAl (MVA) Heusler alloy thin films to investigate their potential as a material 12 with negative spin polarization. Density-of-state calculations showed that MVA has a gap in the 13 majority-spin state at the Fermi energy, leading to negative spin polarization. Negative spin 14 polarization was found to be higher in the L21-ordered state than that in the B2-ordered state. VMn 15 antisites (V atoms occupying Mn sites) introduced an in-gap state that significantly reduced 16 negative spin polarization, while MnV and AlV antisites preserved the energy gap and thus weakly 17 affected negative spin polarization. High-quality MVA films were fabricated via magnetron 18 sputtering at elevated temperatures, achieving the B2 and L21 order parameters of 0.88 and 0.5 at 19 600°C in the stoichiometric composition, respectively. The comparison of stoichiometric and 20 various off-stoichiometric samples revealed that the Mn-rich and Al-rich compositions showed 21 an improved ordering and a smaller number of VMn antisites. The benefit of using off-22 stoichiometric compositions was further highlighted by the obtained negative magnetoresistance, 23 which was measured via epitaxial MVA/Ag spacer/CoFe current-perpendicular-to-plane giant 24 magnetoresistance devices. Devices with the MVA composition of Mn2.2V0.6Al1.2 showed a very 25 large negative magnetoresistance of −4.4%, indicating high negative spin polarization in MVA. 26 Furthermore, highly efficient spin-transfer torque generation, with a torque direction opposite to 27 that seen in positive spin polarization materials, was demonstrated via spin injection from MVA. 28 mailto:SUTO.Hirofumi@nims.go.jp2 I. Introduction 29 Co-based Heusler alloys have been studied extensively for the use in various spintronic 30 applications [1–7] due to their predicted high spin polarization (P) and high Curie temperature. 31 Furthermore, large magnetoresistance (MR) and high spin-transfer torque (STT) efficiency have 32 been experimentally demonstrated in these materials [8–15]. Contrary to Co-based alloys, Mn-33 based Heusler alloys have not been as extensively studied despite the theoretical predictions of 34 high P and large perpendicular magnetic anisotropy in them [16,17,26–32,18–25]. Therefore, this 35 study focuses on a Mn-based Heusler alloy Mn2VAl (MVA) [16,17,22–25], which has a Curie 36 temperature of 768 K and saturation magnetization (Ms) smaller than that of most Co-based 37 Heusler alloys due to its ferrimagnetic ordering. The small Ms of ferrimagnetic Heusler alloys is 38 beneficial for reducing the critical current density for magnetization switching by STT, which is 39 ideal for low-power operation. Moreover, MVA is expected to possess high negative spin 40 polarization that originates from the electronic structure with a gap in the majority-spin state at 41 the Fermi energy (EF) [18,21]. Negative P denotes that the spin momentum direction of spin-42 polarized conduction electrons is opposite to the net magnetization direction. This behavior is 43 opposite to that exhibited by Co-based Heusler alloys with high positive P, whose electronic 44 structure has a gap in the minority-spin state at EF. Negative P materials exert an STT with torque 45 direction opposite to that seen in positive P materials, which can increase the design freedom of 46 STT-based devices beyond the limits of conventional positive P materials. Device structures that 47 use negative P materials have been proposed for energy-assisted writing in hard disk drive (HDD) 48 applications [33–38]. Furthermore, the small Ms of MVA results in a small disturbance in the 49 writing field of the write head of an HDD, making MVA a very suitable candidate for this 50 application. 51 52 Until now, negative P values have been studied primarily by measuring negative MR in MR 53 devices that combine a positive and a negative P material, where the measured resistance becomes 54 3 high (low) for the parallel (antiparallel) magnetization configuration. This is the opposite to that 55 occurs in a standard MR device comprising two positive P materials. In giant MR (GMR) studies, 56 CoFeGd, FeCr, FeV, NiCr, and Fe4N have been reported to exhibit negative MR [39–44]. 57 However, the amplitude of the negative MR ratio was less than 1% at room temperature in the 58 case of pseudo-spin-valve current-perpendicular-to-plane GMR (CPP-GMR) devices, indicating 59 the low negative P of these devices. Recently, we demonstrated a relatively large negative MR of 60 −1.8% in Mn2VGa, which has the same crystal structure and magnetic ordering as MVA [27]. 61 The large negative MR of Mn2VGa indicates high potential of Mn-based Heusler alloys. 62 Moreover, negative P materials have been studied for fabricating tunnel MR (TMR) devices, and 63 they show a larger negative MR ratio than those used in GMR devices [26,45,46]. The 64 controllability of the STT direction by negative P materials makes these types of materials more 65 attractive as a source of STT rather than a source of MR. In this respect, these materials are more 66 suitable in GMR devices as such devices induce a stronger STT than TMR devices owing to their 67 higher current tolerance. However, only a few studies have investigated the STT of negative P 68 materials [41,44]. Therefore, implementing high negative P in GMR devices and evaluating the 69 STT effect is critical to the development of spintronic devices utilizing negative P. 70 71 In this study, we investigated the potential of MVA as a negative P material by examining the 72 effects of thermal treatment and composition on atomic ordering and by evaluating MR and STT 73 properties using CPP-GMR devices. In addition, first-principles calculations were conducted for 74 obtaining the density of states (DOSes), antisite formation energy, and transport properties. High-75 quality MVA films were fabricated via magnetron sputtering at deposition temperatures of 500°C 76 and 600°C. Off-stoichiometric MVA was found to show improved ordering and reduced number 77 of VMn antisites, which are crucial in obtaining high negative P according to the first-principles 78 calculation. The amplitude of negative MR obtained using CPP-GMR devices increased in off-79 stoichiometric compositions, and devices with the structure of Mn2.2V0.6Al1.2/Ag spacer/CoFe 80 4 exhibited a very large negative MR of −4.4%, indicating high negative P in MVA. Additionally, 81 STT generation via spin injection from MVA was demonstrated. The evaluated STT efficiency 82 was higher than that of FeCr, one of the widely studied negative P materials, demonstrating the 83 advantage of MVA. 84 85 86 II. Experimental and computational methods 87 The DOSs of B2- and L21-ordered MVA were calculated using the density functional theory 88 (DFT) and Korringa–Kohn–Rostoker method offered in the Akai-KKR software package [47–89 49]. Generalized gradient approximation (GGA) was used for estimating the exchange-correlation 90 energy [50], and disordered states were treated within coherent potential approximation. For all 91 these calculations, we used a lattice constant of a = 5.875 Å. Brillouin-zone integrations for the 92 self-consistent field and DOS calculations were performed using 16 × 16 × 16 and 25 × 25 × 25 93 k-points. The imaginary part of energy for the DOS calculation was set to 0.00005 Ry. Formation 94 energies were calculated by means of the DFT implemented in the Vienna ab initio simulation 95 program [51]. We calculated the formation energy Ef(X) of each defect X using the equation 96 𝐸𝑓(𝑋) = 𝐸tot(𝑋) − 𝐸tot(Mn2VAl) + ∑ 𝑛𝑖𝜇𝑖𝑖 [52,53], where 𝐸tot(𝑋) is the total energy of the 97 supercell, including defect X, and 𝐸tot(Mn2VAl) is the total energy of antisite-free MVA in an 98 equivalent supercell. A MVA supercell contained 32 atoms. The difference between this number 99 of atoms and that of the stoichiometric composition is considered to be ni = +1 (−1) for an excess 100 (deficiency) of element i, and 𝜇𝑖 is its chemical potential. In this study, we used the values of 𝜇Mn, 101 𝜇V, and 𝜇Al derived from the energies of α-Mn, body-centered cubic V, and face-centered cubic 102 Al, respectively. We adopted the GGA for yielding the exchange-correlation energy and used 103 projected augmented wave pseudopotential [54,55] to properly investigate the effect of core 104 electrons. The Brillouin-zone integration was performed with 8 × 8 × 8 k-points for each supercell. 105 5 First-principles calculations for ballistic transmittance based on the Landauer formula [56] were 106 performed using the QUANTUM ESPRESSO code [57,58]. For the exchange-correlation energy, 107 we adopted the ultrasoft pseudopotential method and GGA [50]. A set of 10 × 10 × 1 grids of k-108 points was used for Brillouin-zone integrations. Cutoff energies for the wave function and charge 109 density were set to 40 (Ryd) and 400 (Ryd), respectively. Stacking models of MVA/spacer/MVA 110 (001), where spacer materials were Ag, V, and Cr, were constructed using tetragonal supercells, 111 with the in-plane lattice parameter of the supercell set to 4.061 Å. The interface of the 112 MVA/spacers (001) featured two types of termination: Mn and VAl terminations. Additionally, 113 Al terminations were considered specifically for the Ag spacer. Stacking models comprised seven 114 atomic layers that contained all spacers, with the 17, 15, and 15 atomic layers of MVA being 115 employed for the Mn, VAl, and Al terminations, respectively. 116 117 Three types of MVA samples were prepared on MgO (001) substrates via magnetron sputtering. 118 The order of layers in each sample was from bottom to top, and numbers shown in parentheses 119 indicate their thicknesses in nanometers. Single-layer MVA films (type-A sample) consisted of 120 MVA (35)/Ru (2), where the Ru layer serves as a passivation layer. CPP-GMR stacks for MR 121 measurements (type-B samples) consisted of 122 Cr(5)/Ag(100)/Cr(5)/W(5)/MVA(10)/Ag(5)/Co50Fe50(7)/Ru(8). CPP-GMR stacks for STT 123 measurements (type-C samples) consisted of 124 Cr(5)/Ag(100)/Cr(5)/W(5)/MVA(15)/Ag(5)/Fe20Ni80(5)/Ru(8). MVA layers were grown by the 125 co-sputtering of Mn, V, and Al targets. The MVA composition was controlled by changing the 126 sputtering power and measured via X-ray fluorescence (XRF) analysis calibrated with standard 127 samples whose composition was analyzed using inductively coupled plasma mass spectrometry. 128 The MVA layer was deposited either at an elevated substrate temperature (Ts) or at room 129 temperature and then subsequently annealed at a post-annealing temperature (Tp). The layers 130 above the MVA layer were deposited at room temperature after the sample had cooled down. In 131 6 the type-A samples, the following eight MVA compositions were prepared: Mn2VAl, Mn2.2V0.8Al, 132 Mn1.8V1.2Al, Mn2V1.2Al0.8, Mn2V0.8Al1.2, Mn2.2VAl0.8, Mn1.8VAl1.2, and Mn2.2V0.6Al1.2. In the type-133 B and type-C samples, selected compositions were used. Table 1 shows the XRF compositions of 134 the type-A samples deposited at Ts = 600°C, showing that the deviation between the nominal and 135 actual compositions is less than 1.3%. The compositions used in the type-A, B, and C samples are 136 shown in Table 1. 137 138 Table 1. Compositions of MVA samples measured via XRF analysis. 139 Nominal composition XRF composition of type-A samples with Ts = 600°C Type-A samples Type-B samples Type-C samples Mn (at%) V (at%) Al (at%) 1 Mn2VAl 49.26 25.97 24.77 ✓ ✓ 2 Mn2.2V0.8Al 54.54 20.90 24.57 ✓ ✓ 3 Mn1.8V1.2Al 43.98 30.72 25.30 ✓ 4 Mn2V1.2Al0.8 48.76 31.13 20.11 ✓ 5 Mn2V0.8Al1.2 49.73 20.24 30.03 ✓ ✓ 6 Mn2.2VAl0.8 54.66 25.67 19.68 ✓ 7 Mn1.8VAl1.2 45.11 25.25 29.63 ✓ 8 Mn2.2V0.6Al1.2 54.25 15.70 30.05 ✓ ✓ ✓ 140 The crystal structure and atomic ordering of the prepared films were investigated via laboratory 141 X-ray diffraction (lab-XRD) with a Cu–Kα radiation source. The degrees of the B2 and L21 orders 142 ( 𝑆𝐵2 and 𝑆𝐿21 ) were calculated as SB2 = √I002exp/I004expI002sim/I004sim and SL21 = √I111exp/I444expI111sim/I444sim , where Ihklexpand 𝐼hklsim 143 represent the experimental and simulated hkl peak intensities, respectively [59,60]. In the XRD 144 simulation model of the off-stoichiometric composition, abundant elements simply occupy the 145 site of deficient elements. In addition, anomalous X-ray diffraction (AXRD) measurements that 146 can distinguish Mn and V with close atomic number were conducted at BL13XU in SPring-8 to 147 quantitatively analyze site occupation [61,62]. The 004 and 111 diffraction peaks were measured 148 at energy around the Mn K-absorption edge (6.539 keV) and corrected with respect to the Lorentz 149 factor, self-absorption, and area. The experimental integrated intensities were then compared with 150 simulated |𝐹ℎ𝑘𝑙|2. Here, 𝐹ℎ𝑘𝑙 is the structural factor of the hkl diffraction peak and expressed as 151 7 follows: 𝐹004 = 2𝑓𝑋 + 𝑓𝑌 + 𝑓𝑍 and 𝐹111 = 𝑓𝑌 − 𝑓𝑍 , where 𝑓𝑋 , 𝑓𝑌 , and 𝑓𝑍 are the energy-152 dependent atomic scattering factors of the X (Mn), Y (V), and Z (Al) sites, respectively. The 153 magnetic properties of MVA were evaluated by measuring the magnetization (M) versus in-plane 154 magnetic field (H) curves using a vibrating sample magnetometer at room temperature. Ms was 155 calculated from the M values estimated at sufficiently high H, and the sample thickness was 156 estimated using X-ray reflectivity. 157 158 The type-B and type-C samples were processed in pseudo-spin-valve CPP-GMR devices using 159 the following procedure. First, thin films were patterned into circular pillars with a diameter of 160 80 nm via electron-beam lithography and Ar-ion milling. After patterning, the pillars were 161 covered with a thin Ta adhesion layer and a SiO2 passivation layer. After the lift-off of the 162 covering layers, a Au top electrode was fabricated. A total of 150 devices were prepared on one 163 sample. Resistance versus in-plane magnetic field (R–H) measurements were conducted using the 164 four-probe method. All devices were measured with an auto-prober system that has an H range 165 of 0.25 T at room temperature. Few of these devices were additionally analyzed using a physical 166 property measurement system (PPMS) at lower temperatures. The MR ratio was defined as 𝑀𝑅 =167 (𝑅 − 𝑅max) 𝑅max⁄ , where 𝑅max is the maximum R in the measured H range. Note that this 168 definition differs from the conventional definition using parallel and antiparallel resistances 169 because the parallel resistance was not always measurable because MVA magnetization was not 170 saturated in the H direction in the measured H range and antiparallel resistance was not 171 measurable because the complete antiparallel configuration was not obtained owing to the gradual 172 change in MVA magnetization. 173 174 The STT measurements of type-C samples were conducted on circular devices with a diameter 175 of 80 nm using the PPMS. This method is described in great detail in Ref. [44]. First, a sufficient 176 perpendicular field (Hz) was applied to saturate the MVA and NiFe magnetic layers in the 177 8 perpendicular direction. Then, R was measured by sweeping the bias current (Ib). Positive Ib was 178 defined as the current flow from top to bottom. By applying sufficient Ib, NiFe magnetization 179 reverses against Hz due to the STT induced by spin injection from the MVA layer, and this 180 magnetization reversal was reflected as the R change through the MR effect. The R–Ib curves were 181 fitted phenomenologically using 𝑅 = 𝑓(𝐼b) +𝛥𝑅2(1 + erfc (𝐼b−𝐼c𝐼width)). Here, f is a second-order 182 polynomial function representing the R change due to the Joule heating by applying Ib, ΔR 183 represents the R change due to magnetization reversal, erfc is the error function, and critical 184 current (Ic) corresponds to the Ib value at the center of the R change. Iwidth represents the Ib width 185 of the R change, and approximately 85% of the R change occurs in the range of Ic ± Iwidth. 186 187 The cross-sectional microstructure of samples was analyzed using high-angle annular dark-field 188 scanning transmission electron microscopy (HAADF-STEM), nano-beam electron diffraction 189 (NBED), and energy-dispersive X-ray spectroscopy (EDS). 190 191 192 III. Results and discussions 193 A. First-principles calculations of DOSs and formation energies of antisites and vacancies 194 Figure 1(a) shows the energy dependence of DOSs for L21- and B2-ordered Mn2VAl together 195 with P at EF and Ms. Both results exhibit a gap structure around EF in the majority-spin state. The 196 resulting negative P is very high (−95%) for the L21-ordered state and decreases to −76% for the 197 B2-ordered state, indicating that the L21-ordered state is desirable for obtaining a higher negative 198 P. Additionally, we investigated the effect of off-stoichiometry on P by changing the Mn/V ratio 199 because these two elements primarily determine the ferrimagnetic and spintronic properties of 200 MVA. Figure 1(b) shows the energy dependence of the DOS for L21-ordered Mn2.2V0.8Al and 201 Mn1.8V1.2Al. The Mn-rich composition maintains the gap structure with a slightly smaller negative 202 9 P than that observed in the stoichiometric case. In contrast, an in-gap state appears in the V-rich 203 composition, considerably decreasing negative P. The effect of VMn antisites on negative P has 204 also been reported in the case of Mn2VGa, which has a similar DOS with a gap in the majority-205 spin state [26–28]. Additionally, calculations were performed for the MVA compositions of 206 Mn2V0.8Al1.2 and Mn2.2V0.6Al1.2 as they exhibit higher ordering in the experimental results, which 207 we discuss later. In these compositions, the energy gap is maintained with a P of approximately 208 −80%. In all compositions except for Mn1.8V1.2Al, the calculated Ms is close to the following 209 values estimated from the Slater–Pauling rule: 2 μB/f.u. (Mn2VAl and Mn2.2V0.6Al1.2), 1.6 μB/f.u. 210 (Mn2.2V0.8Al), and 2.4 μB/f.u. (Mn2V0.8Al1.2). For Mn1.8V1.2Al, the deviation in the calculated Ms 211 from the Slater–Pauling value of 2.4μB/f.u. is relatively large, which coincides with the 212 deterioration in the gap structure. 213 214 We calculated the formation energies of antisites and vacancies in MVA to gain insight into site 215 occupation in off-stoichiometric samples, as shown in Table 2. The formation energy of the Mn 216 vacancies is higher than those of the VMn and AlMn antisites, indicating that in Mn-deficient 217 compositions, V or Al atoms occupy the Mn site instead of forming Mn vacancies. Similarly, V 218 and Al vacancies are not expected to form because of their high formation energies. For 219 comparison, we refer to the case of Co2MnSi [53,63]. In this system, the formation energy of 220 CoSi antisites is very high (2.25 eV and 2.3 eV), which rules out this type of antisite formation 221 and helps to predict site occupation in off-stoichiometric compositions. Here, the first value 222 presented in parentheses is adopted from Ref. [53], and the second is from our own calculation 223 of the same system. Contrary to studies on Co2MnSi, in this study, we do not rule out any specific 224 types of antisites in MVA because all antisites similarly have relatively small formation energies. 225 The formation energies of VAl and AlV are especially small, suggesting that disorder between 226 these two atoms can easily occur in MVA. 227 228 10 Table 2. Formation energies of antisites and vacancies. 229 Formation energy (eV) MnV antisite 0.93 MnAl antisite 0.85 VMn antisite 0.73 VAl antisite 0.37 AlMn antisite 0.58 AlV antisite 0.29 Mn vacancy 2.1 V vacancy 3.0 Al vacancy 3.9 230 B. Growth of single-layer MVA: effect of thermal treatment and composition tuning 231 In this section, we discuss the growth of MVA as a result of different thermal treatments and 232 varying composition. Figure 2(a) shows the out-of-plane XRD profiles of the type-A 233 stoichiometric Mn2VAl samples prepared at Ts = 300ºC–600ºC and Tp = 500ºC–600ºC. All results 234 show only the diffraction peaks related to the (001) plane, indicating (001)-oriented growth. At 235 elevated deposition temperatures, the 002 superlattice peaks implying the presence of the B2 order 236 were not observed at Ts = 300ºC, have appeared very faintly at Ts = 400ºC, and were clearly 237 observable at Ts = 500°C and 600ºC. In post-annealing samples, the 002 peak appeared clearly at 238 Tp =600ºC. Figure 2(b) shows the corresponding XRD profiles along the [111] direction at χ = 239 54.7°. The 111 superlattice peaks indicate the presence of the L21 order, and they appeared only 240 for samples where Ts = 500ºC–600ºC or Tp = 600ºC. In all the samples, no peaks other than those 241 associated with the full-Heusler crystal phase were observed, showing that single-phase MVA 242 films were obtained at all temperatures. Figure 2(c) shows the Ts and Tp dependence of 𝑆𝐵2 and 243 𝑆𝐿21. The 𝑆𝐵2 value was 0.7 at 500ºC and increased to 0.88 at 600ºC. Meanwhile, the 𝑆𝐿21 value 244 was approximately 0.5 at 600ºC. In post-annealed samples, 𝑆𝐵2 and 𝑆𝐿21 were smaller than those 245 observed for elevated temperature deposition samples, indicating that the latter approach achieves 246 the better ordering of MVA. The small 𝑆𝐿21 was consistent with the calculated formation energies 247 of antisites, implying that the mixing of V and Al can occur easily. Figure 2(d) shows the Ts and 248 11 Tp dependence of Ms. Ms showed a similar trend to that seen for order parameters. The 249 experimental values were smaller than the Slater–Pauling value of 2 μB/f.u., reflecting disorder in 250 our films. 251 252 In addition to stoichiometric samples, we also prepared off-stoichiometric samples. Based on 253 the results of the thermal treatment study, Ts = 500ºC and 600ºC were used to obtain high ordering. 254 For each sample, two among Mn, V, and Al atoms were selected, and their balance changed by 255 ±0.2, resulting in samples 2–7, as listed in Table 1. Additionally, a Mn2.2V0.6Al1.2 sample was 256 prepared. We confirmed the (001)-oriented single-phase growth of MVA for all compositions 257 using XRD (selected data are later shown in Fig. 4). Figure 3(a) shows a bar graph for the Ms of 258 the type-A samples with different compositions. The Ms values show clear composition 259 dependence, but they do not change significantly for each composition when the temperature is 260 varied between Ts = 500ºC and 600ºC. In Fig. 3, we also introduced a parameter, Ms ratio, which 261 is defined as the experimental Ms normalized by the Ms calculated based on the Slater–Pauling 262 rule. The Ms ratio is expected to be close to unity when the gap in the electronic structure is 263 preserved and there is no disorder in MVA [20]. Here, the temperature effect on Ms was neglected 264 considering the high Curie temperature of MVA. The right-hand side vertical axis of Fig. 3(a) 265 shows the Ms ratio for samples with Ts = 600ºC using a circle symbol. Ms ratios higher than that 266 of the stoichiometric composition were obtained for the following compositions in the ascending 267 order: Mn-rich V-deficient composition (Mn2.2V0.8Al), Al-rich V-deficient composition 268 (Mn2V0.8Al1.2), and the combination of these two (Mn2.2V0.6Al1.2). These results suggest that these 269 compositions can promote better ordering than that is achievable with the stoichiometric 270 composition. Note that the calculated DOS demonstrates that these off-stoichiometric 271 compositions possess relatively high negative P. Figure 3(b) shows the in-plane M–H curves for 272 type-A samples (Ts = 600ºC) of Mn2VAl, Mn2.2V0.8Al, Mn2V0.8Al1.2, and Mn2.2V0.6Al1.2. The M–273 H curve for Mn2VAl displays a gradual change over a wide H range of ±0.5 T with a coercive 274 12 field (Hc) of ~65 mT. The M–H curve for Mn2.2V0.8Al is similar to that for Mn2VAl. For Al-rich 275 compositions (Mn2V0.8Al1.2 and Mn2.2V0.6Al1.2), Ms increases and Hc decreases to ~40 mT, with 276 magnetization curves showing improved squareness. 277 278 Figure 4(a) shows the out-of-plane XRD profiles for type-A samples (Ts = 600ºC) of Mn2VAl, 279 Mn2.2V0.8Al, Mn2V0.8Al1.2, and Mn2.2V0.6Al1.2. The 004 and 002 reflections were observed in all 280 four samples, indicating the single-phase (001)-oriented growth and presence of B2 ordering. 281 Figure 4(b) shows the intensity ratio of the 002 peak over the 004 peak. The intensity ratios for 282 the Mn2V0.8Al1.2 and Mn2.2V0.6Al1.2 samples were found to increase in comparison with the 283 stoichiometric and Mn2.2V0.8Al compositions. This increase in the intensity ratio for the Al-rich 284 compositions is explained by the fact that the intensity of the 002 reflection primarily depends on 285 the amount of Al in the V–Al plane as the atomic scattering factors of Mn (Z = 25) and V (Z = 286 23) are similar and that of Al (Z = 13) is smaller. As seen in the right-hand axis of Fig. 4(b), we 287 calculated the SB2 ordering parameter considering the off-stoichiometry. SB2 increases for the Al-288 rich compositions. Because all four samples have a stoichiometric or a slightly Mn-rich content, 289 the higher B2 ordering indicates that the X site is more dominantly occupied by Mn atoms, with 290 the smaller numbers of V and Al atoms occupying this site. Considering that VMn antisites 291 significantly influence the negative P of this Heusler alloy, MVA films with their X sites occupied 292 primarily by Mn atoms are expected to have high negative P. 293 294 To further improve our understanding of the type of disorder and reveal site occupation in our 295 samples, we conducted AXRD measurements on type-A samples (Ts = 600ºC) of Mn2VAl, 296 Mn2.2V0.8Al, and Mn2V0.8Al1.2. Figures 5(a) and 5(b) show the X-ray energy dependence of the 297 004 and 111 reflections obtained around the Mn K-absorption edge, respectively. The energy 298 profile of the 004 peak shows a dip at the Mn K-absorption edge for all compositions in a similar 299 manner. The energy profile of the 111 peak is almost flat for Mn2VAl and Mn2V0.8Al1.2 but shows 300 13 a dip for Mn2.2V0.8Al. The experimental results were analyzed by fitting them using simulations. 301 We introduced parameters A and B, which represent the fractions of disorder between Mn atoms 302 in the X site and V atoms in the Y site as well as Mn atoms in the X and Al atoms in the Z site, 303 respectively. Initially, disorder between the Y and Z sites is first neglected. Then, the atomic 304 scattering factors of the X, Y, and Z sites of the Mn2VAl sample are expressed as follows: FX =305 (49.2650− 𝐴 − 𝐵) 𝑓Mn + (0.7450+ 𝐴)𝑓V + 𝐵𝑓Al , FY = 𝐴𝑓Mn + (1 − A)𝑓V , and FZ = 𝐵𝑓Mn +306 0.2325𝑓V + (24.7725− B)𝑓Al, where fMn, fV, and fAl are the atomic scattering factors of each element. 307 Three sets of values were used for A and B to find the best fit, (0, 0.07), (0.06, 0.06), and (0.15, 308 0.04), which all reproduce the SB2 value of 0.88 observed in lab-XRD. The range of B is smaller 309 than that of A because the 002 reflection is more sensitive to B due to the larger difference in the 310 atomic scattering factor between Mn and Al than that between Mn and V. Figure 5(c) shows the 311 simulated curve for the 004 reflection of the stoichiometric Mn2VAl sample. In the case of the 312 004 peak, simulated curves are the same regardless of the values of A and B because the 004 peak 313 reflects the summation of the scattering factors of each site. The simulated curves reproduce the 314 experimental energy profile, demonstrating the accuracy of the simulation. Figure 5(d) shows the 315 simulated 111 reflection intensity based on the XRF composition of the Mn2VAl and Mn2.2V0.8Al 316 samples. Depending on A and B values, simulated curves for the stoichiometric sample show 317 either a convex, flat, or concave energy profile. Because the 111 peak intensity was determined 318 based on the difference in the scattering factor between the Y and Z sites, a peak (dip) appears 319 when the A (B) parameter is higher than the other one while the profile becomes flat for A = B. 320 For the stoichiometric Mn2VAl sample, the experimental 111 reflection profile is almost flat, 321 which is consistent with the simulated curve for A = 0.06 and B = 0.06. These results indicate that 322 there is no clear preference in the formation of Mn–V and Mn–Ga disorders in MVA, as suggested 323 earlier when discussing the formation energy of different disorder types in MVA. In the case of 324 Mn-rich Mn2.2V0.8Al, the 111 profile shows a dip, indicating that the excess Mn atoms 325 14 preferentially occupy the empty V sites. The simulated curve for the Mn2.2V0.8Al sample where A 326 = 0.1 and B = 0.06 (calculated from SB2 = 0.85 observed via XRD) reproduces its experimental 327 profile. Similar to the stoichiometric sample, the flat profile obtained for the Mn2V0.8Al1.2 sample 328 indicates that the amount of Mn–V and Mn–Ga disorders present in this sample is similar. 329 330 We now consider disorder between the Y and Z sites. Note that this type of disorder changes the 331 111 peak intensities but does not affect the shape of the energy profile. By applying Y–Z disorder 332 to the AXRD analysis to satisfy the conditions needed to obtain 𝑆𝐿21 observed via XRD, site 333 occupation was determined as follows: [Mn1.85V0.09Al0.06][Mn0.06V0.71Al0.23][Mn0.06V0.24Al0.7] for 334 Mn2VAl, [Mn1.84V0.1Al0.06][Mn0.22V0.55Al0.23][Mn0.12V0.19Al0.69] for Mn2.2V0.8Al, and 335 [Mn1.91V0.04Al0.05][Mn0.04V0.71Al0.25][Mn0.04V0.06Al0.9] for Mn2V0.8Al1.2. Notably, the Al-rich 336 composition has the smallest number of VMn antisites. 337 338 The anisotropic magnetoresistance (AMR) was measured for type-A samples (Ts = 600ºC) of 339 Mn2VAl, Mn2.2V0.8Al, Mn2V0.8Al1.2, and Mn2.2V0.6Al1.2 (data are shown in the Supplemental 340 Material [64]). All the MVA samples exhibit positive AMR ratios for the current along the MVA 341 [110] direction, and negative ratios for the current direction along the MVA [100] direction. Such 342 sign reversal by the current direction was reported in the case of Mn2VGa and analyzed from the 343 viewpoint of DOS, and the analysis result supported the negative P of Mn2VGa [28]. The similar 344 AMR behavior observed in Mn2VGa and MVA supports that these materials have similar 345 electronic band structure with negative P. The amplitude of the AMR ratio changed with the 346 composition, and the change was especially large for Mn2.2V0.6Al1.2, implying the modification in 347 the electronic band structure by the composition tuning. 348 349 C. Negative MR in CPP-GMR devices 350 15 In this section, we discuss the fabrication of CPP-GMR devices and investigate their MR 351 properties. Figure 6(a) shows the stacking structure of the type-B samples used to fabricate CPP-352 GMR devices. The MVA layer compositions were Mn2VAl, Mn2.2V0.8Al, Mn2V0.8Al1.2, and 353 Mn2.2V0.6Al1.2. The deposition temperature was Ts = 500ºC as this temperature achieved a flat 354 surface with an average roughness (Ra) of approximately 0.3 nm. For Ts = 600ºC, the roughness 355 increased to Ra = 1.2 nm. Each MVA layer was grown on a Cr/W buffer layer. In our previous 356 study of CPP-GMR using Mn2VGa, we found that these buffer layers were effective in 357 suppressing the diffusion of Mn2VGa [27]. Figure 6(b) shows an example of R–H curves for 358 Mn2VAl. The behavior of the R–H curve in the case of the downward H sweep can be understood 359 based on the schematics of the magnetization configuration shown above the figure. As seen in 360 the M–H curves in Figure 3(b), MVA magnetization changes gradually while the CoFe layer 361 switches at around the zero field. The abrupt change in R seen in Figure 6(b) corresponds to the 362 magnetization switching of the CoFe layer, and afterward, the trend of the R change alters from 363 negative to positive. This result implies that R is higher (lower) when the magnetization 364 configuration of the CoFe and MVA layers is more parallel (antiparallel), and the maximum MR 365 ratio reaches −1.5%. A negative MR indicates negative P in the MVA layer because the bulk spin 366 asymmetry of CoFe and interfacial spin scattering asymmetry of CoFe/Ag are positive [65]. 367 Owing to the incomplete antiparallel configuration of devices resulting from the gradual change 368 in MVA magnetization, the full MR ratio could not be evaluated. 369 370 Figures 6(c)–6(e) show an example of the R–H curve for the Mn2.2V0.8Al, Mn2V0.8Al1.2, and 371 Mn2.2V0.6Al1.2 samples, respectively. Negative MR is observed in all three compositions. Figure 372 6(f) shows a summary of the MR ratios obtained from all devices for each sample. Defective 373 devices that showed abnormal R were excluded. The shapes of R–H curves exhibit device-to-374 device variations within each sample because of the incomplete antiparallel configuration of 375 devices. Namely, devices with a more antiparallel magnetization configuration near the zero field 376 16 exhibit a larger negative MR. Another possible reason for this variation in the magnitude of MR 377 is the spatialy nonuniform ordering within the thin film of each sample. The negative MR ratio is 378 statistically enhanced using off-stoichiometric compositions, reaching a maximum of −4.4 % for 379 the Mn2.2V0.6Al1.2 sample. This enhancement in the negative MR is consistent with the high Ms 380 ratio observed for these compositions, demonstrating that improved ordering via composition 381 tuning increases negative P in MVA. The resistance area (RA) product is calculated to be 24 382 mΩ·μm2. Figure 6(g) shows the temperature dependence of the MR ratio for Mn2.2V0.6Al1.2. The 383 negative MR ratio increases as the temperature is reduced, reaching a maximum of −7.9%. 384 385 D. STT in CPP-GMR devices 386 We next investigated STT induced via spin injection from the MVA layer. Figure 7(a) shows 387 the stacking structure of the type-C samples used for the STT measurement. Compared with the 388 type-B samples, the upper magnetic layer in the type-C sample was changed from CoFe to NiFe 389 to reduce Ms. This is necessary to induce the magnetization reversal of this layer using STT within 390 the Ib tolerance of the device. Furthermore, the MVA layer thickness was increased to stabilize 391 MVA magnetization against STT. Based on the results presented in the previous section, the 392 MVA layer composition was adjusted to Mn2.2V0.6Al1.2 and a deposition temperature of Ts = 500ºC 393 was used. Figure 7(b) shows an example of the R–H curve. Similar to the results presented in the 394 previous section, negative MR was observed. Compared with the type-B samples, the amplitude 395 of the negative MR ratio was smaller, decreasing to −1.8% due to the different upper magnetic 396 layer. Figure 7(c) shows R–Ib curves measured at several Hz values. The Hz values were large 397 enough to align MVA and NiFe magnetizations along the field direction at zero bias. R–Ib curves 398 exhibit an overall parabolic increase due to Joule heating. Additionally, R–Ib curves show a 399 decrease in the positive Ib region when sufficient Ib was introduced. This decrease in R is attributed 400 to magnetization reversal detected through the negative MR. 401 402 17 In the following, we deduce that magnetization reversal occurs in the NiFe layer and not in the 403 MVA layer. The layer that exhibits magnetization reversal was not determined solely by the Ib 404 direction because the injection of transmitted spin from the MVA layer with negative P and 405 injection of reflected spin from the NiFe layer with positive P enable STT to reverse the 406 magnetization of the opposite layer in the positive Ib region. The STT required for magnetization 407 reversal is proportional to the magnetic volume, defined as the product of Ms and thickness, and 408 Gilbert damping constant (α). Although the MVA layer was three times as thicker as the NiFe 409 layer, which was intended to stabilize MVA magnetization against spin injection, the magnetic 410 volume of the MVA layer was calculated to be 4.8 nm T and was similar to that of the NiFe layer 411 (4.5 nm·T) because of the smaller Ms of MVA. However, the α value of Mn2.2V0.6Al1.2 was 0.025, 412 which was larger than that of NiFe whose α was 0.011. This indicates that MVA magnetization 413 was more stable against STT than NiFe magnetization, suggesting that magnetization reversal 414 occurred in the NiFe layer. The α value was estimated from the ferromagnetic resonance 415 measurement of the type-A sample of Mn2.2V0.6Al1.2 (data are shown in the Supplemental 416 Material [64]), and high α observed in MVA has been reported previously [24]. To be more 417 precise, the STT efficiency of each layer needs to be considered, which we discuss later. 418 419 To compare STT induced by materials with negative and positive P, we show results for an 420 identical set of measurements conducted for a CPP-GMR device in which the lower magnetic 421 layer was replaced with Co2FeGa0.5Ge0.5 (CFGG), with high positive P. The results are from 422 Ref. [66] and shown in the inset in Fig. 7(c). The layer structure of the CPP-GMR device was 423 Cr(5)/Ag(100)/CFGG(15)/Ag(7)/Fe20Ni80(7.5)/Ru(8), and the post-annealing of the sample was 424 performed at 500°C after the deposition of the CFGG layer. The increase in R due to 425 magnetization reversal appeared in the negative Ib region. Because NiFe and CFGG have positive 426 P, the current direction determined that magnetization reversal occurred in the NiFe layer. The 427 18 magnetization reversal of the CFGG layer was not detected in the positive Ib region in the 428 measured range because of its large magnetic volume. This comparison clearly shows that Ib 429 polarity for the magnetization reversal of the Heusler alloy was opposite for MVA and CFGG, 430 reflecting the opposite sign of P in these two materials. 431 432 In addition, the dependence of STT on the relative angle between two magnetic layers could be 433 seen in this comparison. The Iwidth values estimated by fitting were approximately 0.52 mA for 434 the CFGG device and 2.2 mA for the MVA device at μ0Hz = 2.2 T, which indicated that 435 magnetization reversal occurred more abruptly with Ib for CFGG than that for MVA. When both 436 the magnetic layers have positive P, the angle dependence of STT efficiency was lower (higher) 437 near the parallel (antiparallel) configuration [67]. Therefore, after commencing magnetization 438 reversal from the parallel configuration (zero angle), the STT efficiency increased with the 439 increase of the magnetization angle, which resulted in abrupt magnetization reversal, 440 corresponding to the small Iwidth of the device using CFGG. Meanwhile, when the two magnetic 441 layers have positive and negative P values, the STT efficiency exhibited the opposite angular 442 dependence [67], resulting in gradual magnetization reversal that corresponds to the large Iwidth 443 of the MVA device. 444 445 According to theory, when the magnetization is near the equator in the middle of magnetization 446 reversal at a certain current density, the damping and STT of the system balance out. The critical 447 current density 𝐽𝑐 that satisfies this balance is expressed as 448 𝐽c = 𝜇02|𝑒|ℏ𝜂𝛼𝑀sNiFe𝑑𝐻eff. (1) 449 450 19 Here, 𝑑 = 5 nm is the thickness, 𝑀sNiFe is the Ms of the NiFe layer, and 𝛼 = 0.011 is the damping 451 constant of the NiFe layer. ℏ is Planck’s constant, and 𝑒 is the elementary charge. The effective 452 field Heff includes Hz, the demagnetizing field of the NiFe layer, and the dipolar field from the 453 MVA layer, all of which have only a z-direction component on average because of the circular 454 pillar shape of the device. This equation means that Jc depends linearly on Hz and that STT 455 efficiency 𝜂 can be estimated from the slope of this linear relationship. We assumed that NiFe 456 magnetization was on the equator at Ic, which was estimated from fitting the R–Ib curves. Figure 457 7(d) shows the dependence of Ic on Hz. In addition to the data shown in Fig. 7(c), data from other 458 devices fabricated on the same sample were measured and are shown on the same graph. A clear 459 linear dependence was observed in multiple devices, which is consistent with the theoretical 460 model. From the slope, 𝜂 was estimated to be −0.72. This value is much higher than −0.15 461 reported for FeCr [44], demonstrating that MVA with high negative P could induce larger STT. 462 Furthermore, the 𝜂 value in the MVA layer was higher than that in the NiFe layer, indicating that 463 STT induced by the MVA layer, which acts upon the NiFe layer, was stronger than that of NiFe 464 acting upon MVA. The higher STT efficiency of the MVA layer, together with its higher stability, 465 verified that magnetization reversal occurred only in the NiFe layer. 466 467 E. Cross-sectional TEM of GMR stacks 468 We conducted cross-sectional TEM to confirm that the designed GMR stacks were properly 469 fabricated. Figures 8(a) and 8(b) show the HAADF-STEM image and corresponding EDS 470 mapping of the type-B sample with a Mn2.2V0.6Al1.2 layer, respectively. In the EDS mapping, only 471 the Fe signal from the CoFe layer and the Mn signal from the MVA layer are shown for simplicity. 472 The STEM and EDS results confirm sharp and flat interfaces between neighboring layers with no 473 interdiffusion. The layer thicknesses are consistent with the designed values. The inset in Fig. 8(a) 474 shows the NBED patterns obtained for the MVA layer. The 002 and 111 spots are observed in 475 20 these patterns, confirming the existence of the B2 and L21 ordering in the thin film, which is 476 consistent with the XRD results shown earlier. Figures 8(c) and 8(d) show the HAADF-STEM 477 images obtained for the MVA/Ag/CoFe and W/MVA/Ag structures, respectively. Atomically 478 sharp interfaces were observed at the CoFe/Ag, Ag/MVA, and MVA/W interfaces, with atomic 479 dislocations being denoted using ⊥ symbol in these figures. Figure 8(e) shows atomic-resolution 480 EDS elemental maps of Ag, Mn, V, and Al in the MVA/Ag interface. An Al-rich termination 481 layer is formed at the interface. The second layer from the interface consists of Mn, and the third 482 layer consists of alternately arranged V and Al. Below the second layer, the structure is uniform 483 and comprises the alternate stacking of a Mn-rich layer and the V- and Al-rich layers, indicating 484 the presence of the L21 structure in the MVA film. 485 486 F. Transmittance calculations 487 We calculated the transmittance of the MVA/spacer/MVA (001) structure to examine the Fermi 488 surface matching of MVA with a spacer and to gain insight into the suitability of different spacer 489 materials with MVA to best utilize the negative P of this material. To find an ideal spacer material, 490 in addition to Ag used previously, we explored transmittance in the V and Cr spacer layers. 491 Figures 9(a)–9(c) show the in-plane wave vector dependence of minority-spin transmittance in 492 the parallel magnetization configuration for the Ag, V, and Cr spacers, respectively. The minority 493 spin was considered here because of the negative P of MVA. The Mn and VAl terminations were 494 considered. For the Ag spacer, the Al termination observed via TEM was also considered, where 495 V atoms in the VAl termination were replaced by Al atoms. Table 3 shows the interfacial RA 496 calculated from transmittance. The V spacer exhibits the lowest interfacial RA, regardless of the 497 termination. The Ag spacer shows a slightly higher interfacial RA than V, and the Cr spacer 498 displays a significantly higher interfacial RA. This result indicates that V is a more suitable spacer 499 material than Ag, which we used in our earlier experiments; furthermore, Cr is not a suitable 500 21 spacer material for any MVA devices. For all spacer materials used here, the Mn termination 501 demonstrates a smaller interfacial RA than the VAl termination. In the case of the Al termination 502 for the Ag spacer, the RA value is almost the same as that of the VAl termination. 503 504 The results obtained for MVA are now compared with those for a CFGG/Ag/CFGG(001) stack, 505 for which a high MR ratio has been reported experimentally. The interfacial RA of the majority-506 spin electron in this system is calculated to be 1.77, 1.69, and 2.32 mΩ·μm2 for the FeGa, FeGe 507 and, Co terminations, respectively. The majority spin was considered because of the positive P of 508 Co2FeGa0.5Ge0.5 The interfacial RA is much higher in the MVA/Ag/MVA(001) structure than in 509 the CFGG/Ag/CFGG(001) structure, which partly explains the smaller magnitude of the negative 510 MR ratio in the MVA system than that observed in the system using a Co-based Heusler alloy 511 and a Ag spacer. The high interfacial RA indicates weak hybridization between Ag and Mn (V) 512 orbitals. Although the V spacer shows some promising initial results, further research is necessary 513 to find an optimal spacer material to increase the MR ratio so that it is comparable to the very 514 large MR ratio observed in Co-based Heusler/Ag systems. 515 516 Table 3. Interfacial RA of the MVA/spacer/MVA (001) structure 517 Interfacial RA (mΩ·μm2) Mn termination VAl termination Al termination Ag spacer 6.02 6.142 6.130 V spacer 4.288 5.480 - Cr spacer 9.282 9.975 - 518 VI. Conclusion 519 Mn2VAl Heusler alloy thin films were investigated for spintronic applications utilizing the 520 negative spin polarization of the material. The DOS calculations showed that MVA has a negative 521 P due to a gap in the majority-spin state, which is higher in MVA with the L21 order than in MVA 522 with the B2 order. VMn antisites considerably reduced the negative P of the material, while MnV 523 22 and AlV antisites in the Mn-rich and Al-rich compositions weakly influenced P. These results 524 indicate that improving the ordering and suppressing VMn antisites are crucial for achieving high 525 negative P in MVA. The thermal treatment of MVA was explored to improve the atomic ordering 526 of the material. The B2 and L21 orders were observed in stoichiometric Mn2VAl samples grown 527 at Ts = 500°C and MVA samples post-annealed at Tp = 600°C. MVA grown at elevated 528 temperatures showed higher order than samples grown via post-annealing. Samples with off-529 stoichiometric compositions (Mn2.2V0.8Al, Mn2V0.8Al1.2, and Mn2.2V0.6Al1.2) exhibited better 530 ordering than the stoichiometric sample. Furthermore, the Al-rich composition was found to 531 reduce VMn antisites. We fabricated GMR stacks comprising MVA/Ag/CoFe, for which TEM 532 analysis revealed sharp and flat interfaces between each layer, without any notable interdiffusion. 533 A negative MR ratio was observed in the fabricated CPP-GMR devices, which provided evidence 534 for negative P in MVA, and the amplitude of negative MR increased when employing samples 535 with an off-stoichiometric composition for fabricating devices. A maximum negative MR ratio of 536 −4.4% at room temperature was observed for the Mn2.2V0.6Al1.2 sample, which is the largest 537 negative MR ever reported for a pseudo-spin-valve CPP-GMR, indicating the high negative P of 538 MVA. In addition, STT induction via spin injection from the MVA layer was demonstrated to be 539 more efficient than that obtained in FeCr. These results demonstrate that MVA is a potential 540 candidate for spintronic applications due to its high negative spin polarization. In particular, the 541 demonstration of efficient STT generation in MVA-based devices is important from an 542 application point of view as negative P materials are of interest as an STT source due to their 543 ability to control the direction of torque. 544 545 546 Acknowledgments 547 Synchrotron XRD measurement were performed with the approval of Japan Synchrotron 548 Radiation Research Institute (JASRI) with the proposal No. 2021A1300 and 2023A1563. The 549 23 authors thank Dr. T. Kubota at Tohoku University for fruitful discussions on Mn-based Heusler 550 alloys, Dr. P.D. Bentley at National Institutes for Quantum Science and Technology (QST) for 551 reviewing the English language of this manuscript, and M. Inoue at NIMS for technical support. 552 553 Funding 554 This work was partially supported by the Advanced Storage Research Consortium (ASRC), JSPS 555 KAKENHI (grant numbers 21K20434 and 23K03934], and MEXT Initiative to Establish Next-556 generation Novel Integrated Circuits Centers (X-NICS) (grant number JPJ011438). 557 558 Data availability 559 The data that support the findings of this study are available from the corresponding author upon 560 reasonable request. 561 562 Supplementary Material 563 The supplemental material presents FMR measurements and anisotropic MR measurements of 564 the single-layer MVA samples. 565 566 24 References 567 [1] J. C. Slonczewski, Current-Driven Excitation of Magnetic Multilayers, J. Magn. Magn. 568 Mater. 159, L1 (1996). 569 [2] L. Berger, Emission of Spin Waves by a Magnetic Multilayer Traversed by a Current, 570 Phys. Rev. B 54, 9353 (1996). 571 [3] S. I. Kiselev, J. C. Sankey, I. N. Krivorotov, N. C. Emley, R. J. Schoelkopf, R. A. 572 Buhrman, and D. C. Ralph, Microwave Oscillations of a Nanomagnet Driven by a Spin-573 Polarized Current, Nature 425, 380 (2003). 574 [4] M. Hosomi et al., A Novel Nonvolatile Memory with Spin Torque Transfer 575 Magnetization Switching: Spin-RAM, Tech. Dig. - Int. Electron Devices Meet. IEDM 576 2005, 459 (2005). 577 [5] F. Macià, A. D. Kent, and F. C. Hoppensteadt, Spin-Wave Interference Patterns Created 578 by Spin-Torque Nano-Oscillators for Memory And, Nanotechnology 22, 095301 (2011). 579 [6] J. Torrejon et al., Neuromorphic Computing with Nanoscale Spintronic Oscillators, Nat. 580 2017 5477664 547, 428 (2017). 581 [7] T. Kanao, H. Suto, K. Mizushima, H. Goto, T. Tanamoto, and T. Nagasawa, Reservoir 582 Computing on Spin-Torque Oscillator Array, Phys. Rev. Appl. 12, 024052 (2019). 583 [8] Y. Sakuraba, M. Hattori, M. Oogane, Y. Ando, H. Kato, A. Sakuma, T. Miyazaki, and 584 H. Kubota, Giant Tunneling Magnetoresistance in Co2MnSi∕Al–O∕Co2MnSi Magnetic 585 Tunnel Junctions, Appl. Phys. Lett. 88, 192508 (2006). 586 [9] H. Liu, Y. Honda, T. Taira, K. Matsuda, M. Arita, T. Uemura, and M. Yamamoto, Giant 587 Tunneling Magnetoresistance in Epitaxial Co2MnSi/MgO/Co2MnSi Magnetic Tunnel 588 Junctions by Half-Metallicity of Co2MnSi and Coherent Tunneling, Appl. Phys. Lett. 589 101, 132418 (2012). 590 25 [10] S. Li, Y. K. Takahashi, T. Furubayashi, and K. Hono, Enhancement of Giant 591 Magnetoresistance by L21 Ordering in Co2Fe(Ge0.5Ga0.5) Heusler Alloy Current-592 Perpendicular-to-Plane Pseudo Spin Valves, Appl. Phys. Lett. 103, 042405 (2013). 593 [11] M. Jourdan et al., Direct Observation of Half-Metallicity in the Heusler Compound 594 Co2MnSi, Nat. Commun. 5, 3974 (2014). 595 [12] J. W. Jung, Y. Sakuraba, T. T. Sasaki, Y. Miura, and K. Hono, Enhancement of 596 Magnetoresistance by Inserting Thin NiAl Layers at the Interfaces in 597 Co2FeGa0.5Ge0.5/Ag/Co2FeGa0.5Ge0.5 Current-Perpendicular-to-Plane Pseudo Spin 598 Valves, Appl. Phys. Lett. 108, 102408 (2016). 599 [13] S. Andrieu, A. Neggache, T. Hauet, T. Devolder, A. Hallal, M. Chshiev, A. M. Bataille, 600 P. Le Fèvre, and F. Bertran, Direct Evidence for Minority Spin Gap in the 601 C o 2 MnSi 603 Heusler Compound, Phys. Rev. B 93, 094417 (2016). 604 [14] K. Elphick, W. Frost, M. Samiepour, T. Kubota, K. Takanashi, H. Sukegawa, S. Mitani, 605 and A. Hirohata, Heusler Alloys for Spintronic Devices: Review on Recent Development 606 and Future Perspectives, Sci. Technol. Adv. Mater. 22, 235 (2021). 607 [15] K. Hamaya and M. Yamada, Semiconductor Spintronics with Co2-Heusler Compounds, 608 MRS Bull. 47, 584 (2022). 609 [16] Y. Yoshida, T. Nakamichi, and M. Kawakami, Magnetic Properties of a Ternary Alloy 610 Mn0.5v0.5-YAly, J. Phys. Soc. Japan 50, 2203 (1981). 611 [17] T. Nakamichi and C. V. Stager, Phenomenological Formula of NMR Satellite of Heusler 612 Alloys and Magnetic Structure of Mn2VAl, J. Magn. Magn. Mater. 31–34, 85 (1983). 613 [18] R. Weht and W. Pickett, Half-Metallic Ferrimagnetism in (Formula Presented), Phys. 614 Rev. B - Condens. Matter Mater. Phys. 60, 13006 (1999). 615 26 [19] C. Jiang, M. Venkatesan, and J. M. D. Coey, Transport and Magnetic Properties of 616 Mn2VAl: Search for Half-Metallicity, Solid State Commun. 118, 513 (2001). 617 [20] I. Galanakis, P. H. Dederichs, and N. Papanikolaou, Slater-Pauling Behavior and Origin 618 of the Half-Metallicity of the Full-Heusler Alloys, Phys. Rev. B 66, 174429 (2002). 619 [21] K. Ozdogan, I. Galanakis, E. Şaşioglu, and B. Aktaş, Search for Half-Metallic 620 Ferrimagnetism in V-Based Heusler Alloys Mn 2VZ (Z ≤ Al, Ga, In, Si, Ge, Sn), J. Phys. 621 Condens. Matter 18, 2905 (2006). 622 [22] T. Kubota, K. Kodama, T. Nakamura, Y. Sakuraba, M. Oogane, K. Takanashi, and Y. 623 Ando, Ferrimagnetism in Epitaxially Grown Mn2VAl Heusler Alloy Investigated by 624 Means of Soft X-Ray Magnetic Circular Dichroism, Appl. Phys. Lett. 95, 222503 (2009). 625 [23] R. Y. Umetsu and T. Kanomata, Spin Stiffness Constant of Half-Metallic Ferrimagnet in 626 Mn-Based Heusler Alloys, Phys. Procedia 75, 890 (2015). 627 [24] K. Fukuda, M. Oogane, and Y. Ando, Structural and Magnetic Properties in Mn2VAl 628 Full-Heusler Epitaxial Thin Films, IEEE Trans. Magn. 53, 2600304 (2017). 629 [25] S. Yamada, K. Kudo, R. Sadakari, and K. Hamaya, Epitaxial Mn2VAl Films with L21 -630 Ordered Structure for All-Heusler Stacks, J. Magn. Magn. Mater. 561, 169644 (2022). 631 [26] C. Klewe, M. Meinert, J. Schmalhorst, and G. Reiss, Negative Spin Polarization of 632 Mn2VGa Probed by Tunnel Magnetoresistance, J. Phys. Condens. Matter 25, 76001 633 (2013). 634 [27] H. Suto, V. Barwal, K. Simalaotao, Z. Li, K. Masuda, T. Sasaki, Y. Miura, and Y. 635 Sakuraba, Negative Spin Polarization of Mn2VGa Heusler Alloy Thin Films Studied in 636 Current-Perpendicular-to-Plane Giant Magnetoresistance Devices, J. Appl. Phys. 135, 637 203901 (2024). 638 [28] Z. H. Li et al., Enhancing Atomic Ordering, Magnetic and Transport Properties of 639 Mn2VGa Heusler Alloy Thin Films toward Negatively Spin-Polarized Charge Injection, 640 Acta Mater. 276, 120110 (2024). 641 27 [29] H. Kurt, K. Rode, P. Stamenov, M. Venkatesan, Y.-C. Lau, E. Fonda, and J. M. D. Coey, 642 Cubic Mn 643 2 Ga Thin 644 Films: Crossing the Spin Gap with Ruthenium, Phys. Rev. Lett. 112, 027201 (2014). 645 [30] T. Kubota et al., Magnetoelastic Anisotropy in Heusler-Type Mn 2 − 647 δ CoGa 648 1 + δ 649 Film, Phys. Rev. Mater. 6, 044405 (2022). 650 [31] C. Banerjee, N. Teichert, K. E. Siewierska, Z. Gercsi, G. Y. P. Atcheson, P. Stamenov, 651 K. Rode, J. M. D. Coey, and J. Besbas, Single Pulse All-Optical Toggle Switching of 652 Magnetization without Gadolinium in the Ferrimagnet Mn2RuxGa, Nat. Commun. 11, 653 4444 (2020). 654 [32] P. D. Bentley et al., Cubic-Type Heusler Compound 655 Mn 2 FeGa Thin 656 Film with Strain-Induced Large Perpendicular Magnetic Anisotropy, Phys. Rev. Mater. 657 7, 064404 (2023). 658 [33] J.-G. Zhu, X. Zhu, and Y. Tang, Microwave Assisted Magnetic Recording, IEEE Trans. 659 Magn. 44, 125 (2008). 660 [34] S. Bosu, H. Sepehri-Amin, Y. Sakuraba, S. Kasai, M. Hayashi, and K. Hono, High 661 Frequency Out-of-Plane Oscillation with Large Cone Angle in Mag-Flip Spin Torque 662 Oscillators for Microwave Assisted Magnetic Recording, Appl. Phys. Lett. 110, 142403 663 (2017). 664 [35] W. Zhou, H. Sepehri-Amin, T. Taniguchi, S. Tamaru, Y. Sakuraba, S. Kasai, H. Kubota, 665 and K. Hono, Inducing Out-of-Plane Precession of Magnetization for Microwave-666 28 Assisted Magnetic Recording with an Oscillating Polarizer in a Spin-Torque Oscillator, 667 Appl. Phys. Lett. 114, 172403 (2019). 668 [36] N. Asam, H. Suto, S. Tamaru, H. Sepehri-Amin, A. Bolyachkin, T. Nakatani, W. Zhou, 669 H. Kubota, and Y. Sakuraba, Analysis Method of a Spin-Torque Oscillator Using Dc 670 Resistance Change during Injection Locking to an External Microwave Magnetic Field, 671 Appl. Phys. Lett. 119, 142405 (2021). 672 [37] M. Takagishi, N. Narita, H. Iwasaki, H. Suto, T. Maeda, and A. Takeo, Design Concept 673 of MAS Effect Dominant MAMR Head and Numerical Study, IEEE Trans. Magn. 57, 674 3300106 (2021). 675 [38] Y. Nakagawa, M. Takagishi, N. Narita, A. Takeo, and T. Maeda, Multiple Spin Injection 676 Into Coupled Field Generation Layers for Low Current Operation of MAMR Heads, 677 IEEE Trans. Magn. 58, 3201005 (2022). 678 [39] C. Bellouard, H. Rapp, B. George, S. Mangin, G. Marchal, and J. Ousset, Negative Spin-679 Valve Effect i/A Trilayers, Phys. Rev. B - Condens. Matter Mater. Phys. 53, 5082 (1996). 680 [40] C. Vouille, A. Barthélémy, F. E. Mpondo, A. Fert, P. A. Schroeder, S. Y. Hsu, A. Reilly, 681 and R. Loloee, Microscopic Mechanisms of Giant Magnetoresistance, Phys. Rev. B - 682 Condens. Matter Mater. Phys. 60, 6710 (1999). 683 [41] M. AlHajDarwish, H. Kurt, S. Urazhdin, A. Fert, R. Loloee, W. P. Pratt, and J. Bass, 684 Controlled Normal and Inverse Current-Induced Magnetization Switching and 685 Magnetoresistance in Magnetic Nanopillars, Phys. Rev. Lett. 93, 157203 (2004). 686 [42] X. Li, H. Li, M. Jamali, and J.-P. Wang, Damping Constant Measurement and Inverse 687 Giant Magnetoresistance in Spintronic Devices with Fe4N, AIP Adv. 7, 125303 (2017). 688 [43] H. Suto, T. Nakatani, Y. Kota, A. Nagarjuna, H. Iwasaki, K. Amemiya, T. Mitsui, S. 689 Sakai, S. Li, and Y. Sakuraba, Study on FeCr Thin Film for a Spintronic Material with 690 Negative Spin Polarization, J. Magn. Magn. Mater. 557, 169474 (2022). 691 29 [44] H. Suto, T. Nakatani, N. Asam, H. Iwasaki, and Y. Sakuraba, Evaluation of Spin-692 Transfer-Torque Efficiency Using Magnetization Reversal against a Magnetic Field: 693 Comparison of FeCr with Negative Spin Polarization and NiFe, Appl. Phys. Express 16, 694 013003 (2023). 695 [45] K. Sunaga, M. Tsunoda, K. Komagaki, Y. Uehara, and M. Takahashi, Inverse Tunnel 696 Magnetoresistance in Magnetic Tunnel Junctions with an Fe4N Electrode, J. Appl. Phys. 697 102, 013917 (2007). 698 [46] S. Yasui, S. Honda, J. Okabayashi, T. Yanase, T. Shimada, and T. Nagahama, Large 699 Inverse Tunnel Magnetoresistance in Magnetic Tunnel Junctions with an Fe 3 701 O 702 4 Ele, Phys. Rev. Appl. 15, 034042 (2021). 703 [47] J. Korringa, On the Calculation of the Energy of a Bloch Wave in a Metal, Physica 13, 704 392 (1947). 705 [48] W. Kohn and N. Rostoker, Solution of the Schrödinger Equation in Periodic Lattices 706 with an Application to Metallic Lithium, Phys. Rev. 94, 1111 (1954). 707 [49] Akai-KKR (Machikaneyama) Http://Kkr.Issp.u-Tokyo.Ac.Jp. 708 [50] J. P. Perdew, K. Burke, and M. Ernzerhof, Generalized Gradient Approximation Made 709 Simple, Phys. Rev. Lett. 77, 3865 (1996). 710 [51] G. Kresse and J. Furthmüller, Efficient Iterative Schemes for Ab Initio Total-Energy 711 Calculations Using a Plane-Wave Basis Set, Phys. Rev. B 54, 11169 (1996). 712 [52] C. G. Van De Walle and J. Neugebauer, First-Principles Calculations for Defects and 713 Impurities: Applications to III-Nitrides, J. Appl. Phys. 95, 3851 (2004). 714 [53] B. Hülsen, M. Scheffler, and P. Kratzer, Thermodynamics of the Heusler Alloy Co 716 2 − x 717 30 Mn 1 + 718 x 740 2 MnSi Investigated through 741 Saturation Magnetization and Tunneling Magnetoresistance Ratio, Phys. Rev. B 89, 742 014428 (2014). 743 31 [64] See Supplemental Material at [URL Will Be Inserted by Publisher] for FMR 744 Measurements and Anisotropic MR Measurements of the Single-Layer MVA Samples., 745 (n.d.). 746 [65] J. W. Jung, Z. Jin, Y. Shiokawa, and M. Sahashi, Investigations of Interface Spin 747 Asymmetry and Interfacial Resistance in Fe x Co100 − x /Ag Interface, J. Appl. Phys. 748 117, 17A323 (2015). 749 [66] V. Barwal, H. Suto, and Y. Sakuraba, Reducing Critical Current for Spin-Transfer-750 Torque-Induced Magnetization Reversal in CPP-GMR Devices: The Effect of Low 751 Damping and Interfacial Spin Scattering Asymmetry of Co2FeGa0.5Ge0.5 Heusler 752 Alloy, Prep. (n.d.). 753 [67] J. Barnaś, A. Fert, M. Gmitra, I. Weymann, and V. K. Dugaev, From Giant 754 Magnetoresistance to Current-Induced Switching by Spin Transfer, Phys. Rev. B - 755 Condens. Matter Mater. Phys. 72, 024426 (2005). 756 757 758 32 Fig. 1 (one column) 759 760 Fig. 1. First-principles calculations of the spin-dependent total DOSs for (a) L21- and B2-ordered 761 Mn2VAl, (b) L21-ordered Mn2.2V0.8Al and Mn1.8V1.2Al, and (c) L21-ordered Mn2V0.8Al1.2 and 762 Mn2.2V0.6Al1.2. Spin polarization (P) at EF and Ms are also shown. 763 764 33 Fig. 2 (one column) 765 766 Fig. 2. (a) Out-of-plane and (b) [111]-direction XRD profiles for type-A Mn2VAl samples (Ts = 767 300ºC–600ºC and Tp = 500ºC–600ºC). The marked temperature above each line is the same for 768 (a) and (b), and data are offset for clarity. (c) Temperature dependence of SB2 and SL21 are 769 calculated from peak intensities in (a) and (b). (d) Process temperature dependence of Ms. 770 771 34 Fig. 3 (one column) 772 773 Fig. 3. (a) Composition dependence of Ms obtained for type-A samples (Ts = 500ºC and 600ºC), 774 and black circles represent Ms ratio for Ts = 600ºC. (b) M–H curves for type-A samples (Ts = 775 600ºC) of Mn2VAl, Mn2.2V0.8Al, Mn2V0.8Al1.2, and Mn2.2V0.6Al1.2. 776 777 35 Fig. 4 (one column) 778 779 Fig. 4. (a) Out-of-plane and (b) [111]-direction XRD profiles for type-A samples (Ts = 600ºC) of 780 Mn2VAl, Mn2.2V0.8Al, Mn2V0.8Al1.2, and Mn2.2V0.6Al1.2. Data are offset for clarity. (b) 781 Composition dependence of the intensity ratio of the 002 peak over the 004 peak and 782 corresponding SB2. 783 784 36 Fig. 5 (one column) 785 786 Fig. 5. X-ray energy dependence of the experimentally yielded intensities of (a) 004 and (b) 111 787 reflections around the Mn K-absorption edge obtained for type-A samples (Ts = 600ºC) of 788 Mn2VAl, Mn2.2V0.8Al, and Mn2V0.8Al1.2. Curves were normalized by the intensity at 6.4 keV and 789 offset for clarity. Legends are common for (a) and (b). Corresponding simulations of (c) 004 and 790 (d) 111 reflections. Simulated profiles for various Mn–V (A) and Mn–Al (B) disorders are overlaid 791 on the experimental results. 792 793 37 Fig. 6 (one column) 794 795 Fig. 6. (a) Structure of type-B samples used to fabricate CPP-GMR devices. (b)–(e) Examples of 796 R–H curves measured for CPP-GMR devices comprising Mn2VAl, Mn2.2V0.8Al, Mn2V0.8Al1.2, and 797 Mn2.2V0.6Al1.2. Solid and dashed lines represent the downward and upward H sweeps, respectively. 798 (b) Schematics depicting the magnetization configurations of the MVA and CoFe layers for the 799 downward H sweep. (e) Composition dependence of the MR ratio distribution. (g) MR ratio as a 800 function of measurement temperature for the Mn2.2V0.6Al1.2 sample. 801 802 38 Fig. 7 (one column) 803 804 Fig. 7. (a) Structure of the type-C samples. Arrow indicates the positive current direction. (b) 805 Example of R–H curve. Solid and dashed lines represent the downward and upward H sweeps, 806 respectively. (c) R–Ib curves and Hz values together with fitting curves and 𝐼c. Data are offset for 807 clarity. Inset shows reference results for an identical set of measurements conducted on a CPP-808 GMR with CFGG instead of MVA, which are from [66]. (d) Ic–Hz curves from the results from 809 (c) and Ic–Hz curves from several other devices from the same sample. 810 811 39 Fig. 8 (two column) 812 813 Fig. 8. (a) HAADF-STEM image of the type-B sample with a Mn2.2V0.6Al1.2 layer. The inset shows 814 NBED patterns obtained for the MVA layer. (b) Corresponding EDS mapping. Only the Fe (Mn) 815 signal from the CoFe (MVA) layer is shown for simplicity. (c) and (d) magnified HAADF-STEM 816 image at the MVA/Ag/CoFe and W/MVA/Ag regions, respectively. Yellow ⊥ symbols indicate 817 atomic dislocations. (e) Atomic-resolution EDS mapping at the MVA/Ag interface. 818 819 40 Fig. 9 (two column) 820 821 Fig. 9. In-plane wave vector dependence of minority-spin transmittance in the parallel 822 magnetization configuration calculated for (a) Mn2VAl/Ag/Mn2VAl (001) with Mn, VAl, and Al 823 terminations, (b) Mn2VAl/V/Mn2VAl(001) with Mn and VAl terminations, and (c) 824 Mn2VAl/Cr/Mn2VAl(001) with Mn and VAl terminations 825 826 41 Supplemental Material 827 Negative spin polarization and effect of composition on the atomic order and electronic 828 structure of Mn2VAl Heusler alloy thin films 829 Hirofumi Suto1, Vineet Barwal1, Keisuke Masuda1, Kodchakorn Simalaotao1, Taisuke Sasaki1, 830 Yoshio Miura1, Hiroo. Tajiri2, Loku Singgappulige Rosantha Kumara2, Tomoyuki. 831 Koganezawa2,and Yuya Sakuraba1 832 1Research Center for Magnetic and Spintronic Materials, National Institute for Materials Science 833 (NIMS), Tsukuba, 305-0047, Japan 834 2Japan Synchrotron Radiation Research Institute (JASRI), Kouto, 679-5198, Japan 835 836 Ferromagnetic resonance (FMR) measurement of the single layer MVA sample for 837 Mn2.2V0.6Al1.2 838 Figures S1(a) and S1(b) show the frequency dependence of the FMR resonance field and half-839 width of the FMR peaks measured for the type-A sample (Ts = 600°C) for Mn2.2V0.6Al1.2, 840 respectively. The FMR measurements were conducted by applying a in-plane magnetic field (H). 841 From the slope of the halfwidth, the damping parameter of Mn2.2V0.6Al1.2 was estimated to be 842 0.025. 843 844 FIG. S1. (a) Resonance field of FMR versus frequency for the type-A sample (Ts = 600°C) for 845 Mn2.2V0.6Al1.2. (b) Corresponding half-width of the FMR peaks versus frequency. 846 847 Anisotropic magnetoresistance measurements of the single layer MVA samples 848 Figures S.2(a)-(d) shows the anisotropic magnetoresistance (AMR) ratio as a function of 𝜙 for 849 type-A samples (Ts = 600ºC) of Mn2VAl, Mn2.2V0.8Al, Mn2V0.8Al1.2, and Mn2.2V0.6Al1.2. The 850 current was applied along MVA [110] and MVA [100] and the measurement temperature was set 851 from 10 to 300 K. All the MVA samples exhibit positive AMR ratios for the current along the 852 MVA [110] direction, and negative ratios as the current direction aligned with the MVA [100] 853 direction. Such sign reversal of the AMR ratio by the current direction was reported in Mn2VGa 854 and explained from the viewpoint of DOS [1]. The similarity in the AMR measurement between 855 MVA and Mn2VGa is consistent with the fact that these materials have similar DOS. The AMR 856 results changed with the compositions as follows. The results for Mn2Val and Mn2.2V0.8Al are 857 similar, showing small temperature dependence. In Mn2V0.8Al1.2, the AMR results for the current 858 along [110] shows gradual decrease with lowering the temperature. In Mn2.2V0.6Al1.2, the AMR 859 results for both [110] and [100] current directions decrease with lowering the temperature. The 860 amplitude of negative AMR for the [100] current direction was enhanced in comparison with the 861 other compositions. These changes imply the modification in the electronic band structure. 862 42 Further analysis is necessary to understand the origin of the composition dependence, which is 863 beyond the scope of this study. 864 865 FIG. S2. Angle dependence of the AMR ratio of or the type-A sample (Ts = 600°C) for (a) 866 Mn2VAl, (b) Mn2.2V0.8Al, (c) Mn2V0.8Al1.2, and (d) Mn2.2V0.6Al1.2 measured from 10 to 300 K. 867 The current was applied along the MVA [110] and MVA [100] directions. The legends are 868 common in (a)-(d). 869 870 Reference 871 [1] Z.H. Li, H. Suto, V. Barwal, K. Masuda, T.T. Sasaki, Z.X. Chen, H. Tajiri, L.S.R. Kumara, 872 T. Koganezawa, K. Amemiya, S. Kokado, K. Hono, and Y. Sakuraba, “Enhancing atomic 873 ordering, magnetic and transport properties of Mn2VGa Heusler alloy thin films toward 874 negatively spin-polarized charge injection,” Acta Mater. 276, 120110 (2024). 875 876 −90 0 90 180 270−0.5−0.4−0.3−0.2−0.10.00.10.20.3AMR ratio (%)φ (deg.)MVA[110] MVA[100] 300K 300K 200K 200K 100K 100K 50K 50K 10K 10K−90 0 90 180 270−0.5−0.4−0.3−0.2−0.10.00.10.20.3AMR ratio (%)φ (deg.)−90 0 90 180 270−0.5−0.4−0.3−0.2−0.10.00.10.20.3AMR ratio (%)φ (deg.)−90 0 90 180 270−0.5−0.4−0.3−0.2−0.10.00.10.20.3AMR ratio (%)φ (deg.)(a) Mn2VAl (b) Mn2.2V0.8Al(c) Mn2V0.8Al1.2 (d) Mn2.2V0.6Al1.2