# Fileset

[20250121_MDR_MS_Large MR and high STT achived in CMFG.pdf](https://mdr.nims.go.jp/filesets/31de319b-3cd2-4827-a107-f8578197bba4/download)

## Creator

Vineet Barwal, [Hirofumi Suto](https://orcid.org/0000-0003-4387-5862), [Ryo Toyama](https://orcid.org/0000-0002-7398-5803), [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), [Yuya Sakuraba](https://orcid.org/0000-0003-4618-9550)

## Rights

[Creative Commons BY Attribution 4.0 International](https://creativecommons.org/licenses/by/4.0/)

## Other metadata

[Large magnetoresistance and high spin-transfer torque efficiency of Co2Mn<i>x</i>Fe1−<i>x</i>Ge (0 ≤ <i>x</i> ≤ 1) Heusler alloy thin films obtained by high-throughput compositional optimization using combinatorially sputtered composition-gradient film](https://mdr.nims.go.jp/datasets/37da7acd-9d24-43f2-a9d5-d8b4fc65d333)

## Fulltext

1  Large Magnetoresistance and High Spin-Transfer Torque Efficiency of Co2MnxFe1-xGe (0 ≤ x 1 ≤ 1) Heusler Alloy Thin Films Obtained by High-Throughput Compositional Optimization 2 Using Combinatorially Sputtered Composition-Gradient Film 3 Vineet Barwal*, Hirofumi Suto*, Ryo Toyama, Kodchakorn Simalaotao, Taisuke Sasaki, Yoshio 4 Miura and Yuya Sakuraba 5 Research Center for Magnetic and Spintronic Materials, 6  National Institute for Materials Science (NIMS), Tsukuba, 305-0047, Japan 7 *BARWAL.Vineet@nims.go.jp 8 *SUTO.Hirofumi@nims.go.jp 9 ABSTRACT 10 Half-metallic ferromagnetic Heusler alloys having high spin polarization are the promising 11 candidates to realize large magnetoresistance (MR) ratio and high spin-transfer torque (STT) 12 efficiency in next-generation spintronic devices. Since Heusler alloy properties are sensitive to 13 composition, optimizing the composition is crucial for enhancing device performance. Here, we 14 report the fabrication of high-performance current-perpendicular-to-plane giant magnetoresistance 15 (CPP-GMR) devices using Co2MnxFe1-xGe (0 ≤ x ≤ 1) Heusler alloy, employing high-throughput and 16 detailed composition optimization method. The method combined composition-gradient films and 17 local measurements to enable the composition variation from Co2FeGe to Co2MnGe to be efficiently 18 studied on a single library sample with a small composition interval. The CPP-GMR devices 19 fabricated from stacks annealed at 250°C showed a clear composition dependence of MR with the 20 maximum of MR ratio ~ 38% in the Mn-rich region of x = 0.85. By increasing the annealing 21 temperature to 350°C, the MR ratio increased to ~ 45% along with high STT efficiency ~ 0.6 in the 22 broad composition range of 0.2 ≤ x ≤ 0.7. The optimal composition for the highest MR changed with 23 annealing temperature because of the stability of the GMR stack being higher in the lower x range. 24 The record high MR for the all-metal CPP-GMR devices, at low annealing temperature of 250°C was 25 achieved by the detailed composition optimization. These results present the high potential of 26 Co2MnxFe1-xGe and provide a comprehensive guidance on the composition optimization for achieving 27 large MR ratio and high STT efficiency in the CPP-GMR devices.  28 mailto:*BARWAL.Vineet@nims.go.jpmailto:SUTO.Hirofumi@nims.go.jp 2  I. INTRODUCTION 29   Half-metallic ferromagnetic (HMF) Heusler alloys are an important class of materials for spintronic 30 applications due to 100% spin polarization (P) as predicted by their electronic band structure 1,2. Co-31 based HMF Heusler alloy are the promising Heusler candidates owing to high P ~ 100% and high 32 Curie temperature (Tc) far above room temperature. For example, employing HMF Heusler alloys 33 such as Co2MnSi (CMS)3, Co2Fe0.4Mn0.6Si (CFMS)4–8, Co2FeAl0.5Si0.5 (CFAS)9, Co2FeGa0.5Ge0.5 34 (CFGG)10,11 thin films as ferromagnetic (FM) electrodes in current-perpendicular-to-plane giant 35 magnetoresistance (CPP-GMR) demonstrated large magnetoresistance (MR) ratio ~ 40% with typical 36 resistance area (RA) ~0.05 𝛺μm2. These all metallic CPP-GMR device with large MR and low RA 37 show potential for read sensors for next-generation hard disk drives (HDDs) with data recording 38 density beyond 4 Tbit/inch2.7,12–17 The increase in areal density of HDDs require scaling down the 39 size of read-head sensors with sufficient signal-to-noise (SNR) ratio and a high bit resolution. At 40 reduced sensor dimension, the lower impedance of CPP-GMR structure provides an advantage over 41 tunnel magnetoresistance (TMR) based FM/insulator/FM magnetic tunnel junction (MTJ) structure 42 for read-head sensors. The large MR ratio and low impedance in CPP-GMR devices results in 43 obtaining high SNR ratio which is significant for high density recording in HDDs. Highly spin 44 polarized Heusler alloys can also achieve high spin-transfer-torque (STT) efficiency (η) in addition 45 to high MR.2,18,19 Both, MR and η depends on the spin polarization of conduction electrons in FM 46 layers of CPP-GMR and MTJs. The required current density for STT-induced magnetization 47 switching in STT magnetoresistive random access memory (MRAM) and STT-induced persistent 48 oscillations of magnetization in spin-torque oscillators (STOs) is inversely proportional to the η 49 parameter. Higher STT efficiency results in lower operational current for STT MRAMs making them 50 more energy efficient and enhancing the endurance20. Also, high η in STOs results in large-amplitude 51 magnetization oscillation of large magnetic volume which is necessary to attain higher magnitude of 52  3  generated microwave field 21–24. Thus, high STT efficiency is significant for the efficient operation of 53 STT-MRAMs and STOs. 54   Studies have demonstrated that the structural and magnetic properties such as P, Tc, and Gilbert 55 damping of Heusler alloys are very sensitive to and can be controlled by composition25–27. 56 Compositional tuning of Heusler thin films as FM electrodes in CPP-GMR devices is carried out to 57 enhance the device performance5,28,29. This is generally done for selected compositions of Heusler, 58 through stack deposition and device fabrication for each composition, which is time- and resource-59 consuming. The combinatorial sputtering can break through the limitation by enabling efficient and 60 systematic investigation of a wide range of material compositions on a single library sample.30,31 We 61 recently studied the MR and STT efficiency in CPP-GMR devices containing composition-gradient 62 CoxFe1-x (0 ≤ x ≤ 1) system at fine x interval.32 The approach significantly enhanced the throughput 63 of the material synthesis and characterization but has never been applied for the CPP-GMR with the 64 HMF Heusler electrodes. 65   In addition to the composition tuning, the HMF Heusler alloys for spintronic applications often 66 require higher degree of atomic order because the half-metallic energy gap is collapsed by the atomic 67 disorder and the formation of detrimental antisites.17,33,34 For several potential applications such as 68 the read head of HDDs, high atomic order is required to be obtained at lower process temperature 69 around 300°C. Among half-metallic Co-based Heusler alloys, Co2MnGe (CMG) and Co2FeGe (CFG) 70 are considered the most suitable for achieving high atomic order through low-temperature annealing. 71 This is because they are intermetallic compounds with high order-disorder transition temperatures, 72 indicating high thermal stability of the L21-ordered structure. Such alloys are expected to show very 73 high driving force for the chemical ordering from the disordered state in thin films. Therefore, 74 Co2MnxFe1-xGe (CMFG), the mixture of CMG and CFG should have a high chemical ordering 75 behavior at low-temperature annealing35. There are several studies on CMFG Heusler based CPP-76  4  GMR devices using Co2Mn0.6Fe0.4Ge (x = 0.6) 36–39. Other than this composition, Page et al. 77 investigated the properties of polycrystalline Co2Mn1-xFexGe (x = 0, 0.1, 0.2, 0.4) Heusler alloy FM 78 layer within the CPP-GMR spin valves and found high MR ratio of ~15% (change in resistance area 79 product {𝛥RA} ~ 6 m𝛺 𝜇m2) independent of Fe content40. Nakatani et al. reported that the MR ratio 80 did not change significantly by varying Mn and Fe concentrations within the observed range for the 81 polycrystalline CMFG. They found that the Co:Mn(Fe) ratio has a wide window for obtaining large 82 MR ratio, while the Co:Ge ratio needs to be tuned to 2:1 in order to maximize the CPP-GMR output35. 83 However, no studies for the epitaxial CMFG films within the CPP-GMR stacks have been done 84 varying the composition, especially Mn:Fe ratio, in fine increments. In the previous studies involving 85 tuning of Fe/Mn ratios (x) in Co2FexMn1-xSi/Ag/ Co2FexMn1-xSi CPP-GMR devices with various x, it 86 was shown that the addition of Fe in CMS improved the interfacial exchange stiffness of Co at the 87 Ag spacer, resulting in an enhancement of interfacial spin scattering asymmetry parameter, thereby 88 increasing the MR ratio5. This indicates that the tuning of Mn/Fe ratio in CMFG is crucial to obtain 89 high MR and STT in CPP-GMR devices. 90 Here, we report the large MR ratio and high STT efficiency in CPP-GMR devices containing 91 epitaxial Co2MnxFe1-xGe (0 ≤ x ≤ 1) Heusler alloy thin films achieved through the high throughput 92 and detailed composition optimization method. Combinatorial sputtering method was employed to 93 achieve a composition variation from CFG to CMG on a single library sample. The samples were 94 analyzed locally at the various composition points by X ray diffraction (XRD), MR measurement, 95 and scanning transmission electron microscopy (STEM), thereby enhancing the throughput and 96 resolution in characterization. We observed a clear composition dependence of MR for the CPP-GMR 97 stacks annealed at 250°C, with the maximum MR ratio ~ 38% in the Mn-rich region of x = 0.85. By 98 increasing the annealing temperature to 350°C, the MR ratio was increased to 45% along with high 99 STT efficiency ~ 0.6 over a broad composition range (0.2 ≤ x ≤ 0.7). At 350°C annealing, the optimum 100  5  composition shifted to lower x range as the stability of the GMR stack was higher in this x range. The 101 results reveal efficient composition tuning and provides comprehensive guidance for the selection of 102 CMFG composition to obtain large MR ratio and high STT efficiency in the CPP-GMR devices. 103 II. EXPERIMENTAL AND COMPUTATIONAL DETAILS 104 A. Sample growth and device fabrication 105   Epitaxial thin film stacks including CMFG composition-gradient layers were deposited on 2 × 2 cm2 106 MgO (001) substrate using combinatorial sputtering system (CMS-A6250X2, Comet, Inc.). The base 107 pressure was lower than 6 × 10-6 Pa. Prior to deposition, the MgO substrates were cleaned by 108 ultrasonication with acetone, ethanol, and de-ionized water for 5 minutes each, followed by in-situ 109 flash annealing at 600°C for 30 minutes in the deposition chamber. Figure 1(a) and 1(b) illustrate two 110 different stack configurations named Type-I, and Type-II, respectively. The Type-I stack were 111 comprised of CPP-GMR stack with both magnetic layers being Co2MnxFe1-xGe, which was used to 112 evaluate the MR output of the devices. The Type-II stack was comprised of CPP-GMR stack with 113 magnetic layers being Co2MnxFe1-xGe and Ni80Fe20, which was used for the STT measurement. The 114 Type-II stack was designed to have a thicker CMFG bottom electrode (~15 nm) as compared to the 115 top NiFe (~7 nm) to make the CMFG layer resistant to the STT from the NiFe layer. The bottom 116 CMFG acting as a spin injection layer (SIL) induces STT and reverses the magnetization of the top 117 NiFe layer acting as a free layer (FL). Three Type-I sample stacks with different thermal treatments 118 were prepared: as-deposited sample and samples subjected to in-situ post-annealing at 250°C and 119 350°C for 30 min, respectively. For the Type-II stack, sample with 350°C post annealing was prepared. 120 The red arrows in the figure 1(a) and 1(b) indicate the layer after which in-situ post annealing was 121 done. The top 8 nm Ru cap layer was deposited after cooling down the sample to room temperature. 122 Note that there was a difference between the set temperature at the controller and the actual 123 temperature at the substrate. For example, the set temperature of 200°C (300°C) at the controller 124  6  attains maximum actual temperature ~ 250°C (350°C) at the substrate because of the overshooting in 125 heating process and reaches to ~ 220°C (320°C) in about 30 minutes (annealing time in the present 126 experiment). Here, we used the maximum actual temperature at the substrate (250°C or 350°C) for 127 simplicity. 128   The CFG(CMG) film was deposited by co-sputtering of Co67Fe33(Co60Mn40) and Ge targets using 129 DC and RF power sources. The actual compositions for the Co2MnGe and Co2FeGe nominal 130 compositions films were calibrated separately using X-ray fluorescence (XRF) spectroscopy, yielding 131 values of Co54.49Mn23.65Ge21.86 and Co47.11Fe30.88Ge22.01, respectively. Figure 1(c) illustrates the 132 combinatorial sputtering process. The CMFG composition-gradient film was obtained by the 133 alternating deposition of CFG and CMG wedge-shaped layers using a linear moving mask and 134 substrate rotation in the manner described below. We have used the similar combinatorial sputtering 135 process in our previous studies to obtain well-controlled composition gradient film25,32,41,42.  136 1. CFG deposition: wedge-shaped CFG layer was deposited using a linear moving mask, 137 defined to move along the X direction from -6 to +6 mm with respect to the center of the 138 substrate. This corresponds to the composition-gradient width of 12 mm with a thickness 139 gradient of 0 to 0.5 nm from -6 mm to +6 mm and the uniform CFG region from -9.5 to -6 140 mm. The speed of the mask was determined by the deposition rate of CFG (0.16 Å/s). 141 ensuring a maximum thickness for one-unit layer of 0.5 nm. This thickness was chosen to be 142 close to the lattice parameter of the Heusler alloys. 143 2. Substrate rotation: The substrate was rotated by 180°. 144 3. CMG deposition: Wedge-shaped CMG layer was then deposited using the mask moving 145 along the X direction from -6 to +6 mm with respect to the center of the substrate, at a speed 146 determined by the CMG deposition rate (0.30 Å/s), which corresponds to the composition-147 gradient width of 12 mm with a thickness gradient of 0 to 0.5 nm and the uniform CMG 148  7  region from -9.5 to -6 mm. 149 4. Cyclic process: Steps 1 to 3 producing flat 0.5-nm-thick one-unit layer of CMFG was 150 repeated 2n times to obtain the desired thickness of n nm. 151   The X axis in Fig 1(a), and 1(b), corresponds to the composition-gradient direction and the dotted 152 lines on the stacks indicate the separation between the composition-gradient region at the center and 153 uniform composition region for the CFG and CMG on the left and right side of the sample. The 154 software was equipped with a recipe option so that the deposition can be completed in an automated 155 mode. Figure 1(d) shows the variation of estimated composition for the CMFG film along the X axis, 156 calculated using the compositions of CFG and CMG determined by XRF. Upper abscissa scale shows 157 the variation of nominal Mn content (x) from 0 (CFG side) to 1 (CMG side). The actual composition 158 can be deduced from the nominal x content by using the formula: 159 𝐶𝑀𝐹𝐺 = 𝑥 × (𝐶𝑜𝐶𝑀𝐺 + 𝑀𝑛𝐶𝑀𝐺 + 𝐺𝑒𝐶𝑀𝐺) + (1 − 𝑥) × (𝐶𝑜𝐶𝐹𝐺 + 𝐹𝑒𝐶𝐹𝐺 + 𝐺𝑒𝐶𝐹𝐺), (1) 162   where 𝑁CMG/CFG  represents the composition of 𝑁(Co/Mn/Ge)  atom in CMG/CFG, determined 160 from XRF measurement. 161   For the CPP-GMR device fabrication, the Type-I and Type-II sample were patterned into circular 163 and elliptical pillars with designed dimensions (80 × 80 nm2, 140 × 70 nm2, 100 × 100 nm2, and 200 164 × 100 nm2) using a combination of electron-beam lithography, photolithography and Ar-ion milling 165 techniques. After patterning, the pillars were passivated by a SiO2 layer, and an Au top electrode was 166 deposited. The pillars with different size were distributed across 20 rows and 64 columns with a fixed 167 pillar size in a row and 5 rows for each device size. The composition-gradient region of 12 mm was 168 approximately divided into 48 columns by fabricating pillars at an interval of 250 μm. A total of 20 169 devices at one composition were fabricated.  170  8   FIG. 1. Sample configuration for the (a) Type-I and (b) Type-II sample structures with red arrows indicating layer after which in-situ post annealing was done. (c) Schematic showing the combinatorial sputtering deposition process. (d) Variation of atomic concentration in Co2MnxFe1-xGe composition-gradient film along the X axis. The shaded region highlights the 12 mm composition-gradient width across the 19 mm deposition length scale. B. Characterization 171   Structural characterization was done using XRD equipped with a Cu Kα radiation source (λ = 1.5418 172 Å) for the Type-I sample stacks prior to the device fabrication. The X-rays were collimated using a 173 0.5 mm incident slit and irradiated at various positions in steps of 1 mm along the X axis to span the 174 gradient as well as the uniform composition regions. The XRD scans were recorded using a two-175 dimensional X-ray detector. 176   The resistance versus magnetic field (R-H) measurements were carried out for the Type-I and Type-177 II samples on the CPP-GMR devices with all four pillar sizes using an auto prober system via four-178 probe method applying in-plane magnetic field in the range of ± 30 mT. For the MR analysis, the data 179 points were filtered based on the observed RP (resistance in the parallel configuration), ensuring they 180 fell within ± 30% of the device resistance calculated using the resistivity values for each layer and 181  9  the designed pillar size. The parasitic resistance (Rparasitic) was then determined by plotting 𝛥R = ǀRP – 182 RAPǀ (where RAP is the resistance in the anti-parallel configuration) as a function of RP and then linearly 183 fitting the data points to extract Rparasitic, which corresponds to the RP intercept at 𝛥R=0. The Rparasitic 184 was typically around 0.5 𝛺. Intrinsic MR ratio was calculated using the formula (ǀRP – RAPǀ)/(RP – 185 Rparasitic) × 100%. For each composition, the average MR ratio was calculated and devices showing a 186 deviation of more than ± 20% from the average were considered defective and their data excluded. 187 Observed MR ratio was calculated using the formula (ǀRP – RAPǀ)/RP × 100%. The 𝛥RA was calculated 188 using the designed device area. STT induced magnetization reversal measurements were carried out 189 for the Type-II sample with circular pillars having designed diameter of 80 nm. The circular pillar’s 190 cross-sectional area was estimated to be approximately 10.58 × 10−3 μm2 using scanning electron 191 microscopy. We used the recently proposed method of STT induced magnetization reversal against 192 the magnetic field to evaluate the STT efficiency43. The method provides straightforward approach to 193 analyze STT efficiency in the fabricated CPP-GMR devices. The STT efficiency can also be evaluated 194 through alternative techniques that involve STT-induced magnetization dynamics. These methods 195 include inducing magnetization switching of a magnetic layer with magnetic anisotropy or conducting 196 spin torque ferromagnetic resonance (FMR) measurements. The former method is affected by 197 stochastic switching, influenced by thermal fluctuations in magnetization, while the latter necessitates 198 high-frequency measurements at FMR frequencies (typically in the GHz range) and requires precise 199 calibration of bias current, as signals at these frequencies are prone to attenuation. In contrast, the 200 magnetization switching in the present method is achieved solely by the balance between STT and 201 damping, allowing us to neglect thermal effects, which simplifies the analysis. 202   High-angle annular dark-field scanning transmission electron microscope (HAADF-STEM) 203 observations and energy-dispersive X-ray spectroscopy (EDS) were performed for the Type-I sample 204 stacks using a FEI Titan G2 80-200 TEM with a probe aberration corrector operating at 200 kV. Thin 205  10  foil specimens for STEM observations were prepared by the focused ion beam lift-out technique using 206 a FEI Helios G4 UX. 207 C. Computational Methods 208 A model of the L21 Heusler structure was employed, consisting of four atoms arranged on 209 interpenetrating face-centered cubic (fcc) sublattices. Specifically, the basis included two Co atoms 210 at (0, 0, 0) and (0.5, 0.5, 0.5), one Ge atom at (0.75, 0.75, 0.75), and disordered Mn and Fe atoms at 211 (0.25, 0.25, 0.25), with a composition ratio of MnxFe1-x (0 ≤ x ≤ 1). The lattice parameter (a) for the 212 L21 ordered Co2MnxFe1-x Ge, was determined by applying Vegard’s law, interpolating between the 213 identical lattice parameters of the terminal alloys, Co2FeGe 44 and Co2MnGe45. The determined lattice 214 parameter, a = 5.743 Å was used for all the calculations. Based on density functional theory (DFT), 215 the generalized gradient approximation (GGA) proposed by Perdew, Burke, and Ernzerhof (PBE) 46 216 was adopted for the exchange and correlation energy. The calculations were performed using the 217 Vienna ab initio simulation package (VASP)47,48 code with the projector-augmented wave (PAW) 218 49,50method. The kinetic energy cutoff for the plane-wave basis set was set as 400 eV. Atomic disorder 219 is described by the virtual crystal approximation (VCA)51. Numerical Brillouin zone (BZ) integrations 220 are performed by using Monkhorst-Pack52 special k-point meshes of 10 × 10 × 10 for self-consistent 221 field calculations and 20 × 20 × 20 for density of states (DOS) calculations.  222   223  11  III. RESULTS AND DISCUSSION 224 A. Effect of Mn/Fe ratio in CMFG on the electronic states of CMFG 225 In order to investigate the effect of Mn:Fe composition on the spin polarization of Co2MnxFe1-xGe, 226 we calculated the spin resolved DOS. Figure 2(a) and 2(b) shows the calculated energy dependence 227 of the spd- and sp-orbital DOSs for L21-ordered Co2MnxFe1-xGe (0 ≤ x ≤ 1) at an x interval of 0.1, 228 respectively. The calculations clearly demonstrate a shift in EF position with respect to composition, 229 while the overall DOS shape is preserved. The spin polarization was calculated using the standard 230 definition P[%] = 100 × (𝐷↑ − 𝐷↓)/(𝐷↑ + 𝐷↓), where 𝐷↑ and 𝐷↓ are the majority- and minority-spin 231 DOSs at the Fermi level (EF). From the previous studies, it is known that the sp electrons having small 232 effective mass contributes majorly to the electrical conduction process and thus sp spin polarization 233 (Psp) is a good representative parameter for the spin-dependent transport in the CPP-GMR devices. 234 Figure 2(c) shows the calculated energy dependence of Psp of L21-ordered CMFG (0 ≤ x ≤ 1), and Fig. 235 2(d) shows x dependence of the higher and lower energy edge of the gap estimated at Psp = 95 %. The 236 EF was observed to be inside the gap for 0.3 ≤ x ≤1, indicating the almost half-metallic character, as 237 confirmed in the Psp at EF (Fig.2(c) inset). Because high P and EF position close to the gap center are 238 considered beneficial for obtaining large MR ratio and high STT efficiency, these results indicate that 239 there is an optimal range of x. 240  241  242  243  12   FIG. 2 First-principles calculations of the (a) spd-total DOSs, (b) sp-orbital DOSs and (c) sp spin polarization (Psp) for the L21-ordered Co2MnxFe1-xGe (0 ≤ x ≤ 1) at an x interval of 0.1. Inset in (c) shows the change in Psp with Mn content, x. (d) Higher and lower energy edge of the gap estimated at Psp = 95 % for different Mn content ( 0 ≤ x ≤ 1) at an x interval of 0.1. The legends are common for (a), (b) and (c).    244  13  B. XRD analysis 245   Figure 3a illustrates the orientation relationship for the CMFG film epitaxially grown on MgO (001) 246 substrates with MgO (001) ‖ Cr(001) ‖ Ag (001) ‖ CMFG (001) and MgO[100] ‖ Cr[110] ‖ Ag [100] ‖ 247 CMFG [110]. Figure 3b, 3c and 3d shows the out of plane 𝜃-2𝜃 XRD scans at χ = 0° and 54.7° of the 248 Type-I sample stack at the selected Mn compositions (x =1, 0.92, 0.75, 0.58, 0.42, 0.25, 0.08, and 0) 249 for the as-deposited, 250°C and 350°C post-annealed (PA) samples, respectively. The lattice 250 parameter evaluated from the 004 diffraction peak shows negligible change with composition, being 251 0.575 ± 0.001 nm from the CFG to the CMG side. The lattice mismatch between CMFG and Ag is 252 estimated to be < 0.5% from the lattice parameter (a = 0.574 nm) of the bulk CMFG and √2a = 0.577 253 nm for Ag. This value matches well with the reported value 40. The 111 superlattice peaks were 254 detected by tilting the sample at χ = 54.7° with respect to the film-surface normal direction. The 255 observation of the low intensity 002 superlattice diffraction peak in the as-deposited sample shows 256 large A2-type disorder with the presence of partial B2 ordering (atomic order between Co and (Fe, 257 Mn, Ge) sites. The enhancement in the intensity of the 002 diffraction peak along with the appearance 258 of low intensity 111 diffraction in the 250°C PA sample shows the enhanced B2 ordering and presence 259 of L21 ordered phase (atomic order between (Fe, Mn) and Ge sites) in the sample. Further 260 enhancement of the 002 and 111 superlattice diffraction peak intensity for the 350°C PA sample shows 261 the enhancement of the B2 and L21 ordering with annealing temperature, respectively. 262   For the 350°C PA sample, low-intensity peak at 2𝜃 = 75.5° was observed at the χ = 0° scan for the 263 Mn-rich compositions (x = 1 and 0.92), marked by the symbol ♣. This peak can be attributed to 264 secondary phase at higher annealing temperature and possibly due to diffraction from either one or a 265 combination of hcp Co1.75Ge 004, or fcc Co 220 or hcp Co 110 planes53. 266   14   FIG. 3. (a) Schematic showing fcc MgO 001, bcc Cr 001, fcc Ag 001, Co2MnxFe1-xGe (CMFG) Heusler 001 plane with epitaxial relationship between MgO and Cr; Cr and Ag; and Ag and CMFG. 𝜃-2𝜃 XRD scans for the Type-I structures along the 002 and 111 plane at χ = 0°and χ = 54.7° respectively, for the (b) as-deposited sample, (c) 250°C and (d) 350°C post-annealed sample. The * symbol marks the peak position for the Ru 102 plane and ♣ marks the peak position for the hcp Co1.75Ge 004, or fcc Co 220, or hcp Co 110 planes.   267  15  C. Magnetotransport properties 268   We studied the MR output for the CPP-GMR devices fabricated from the Type-I samples. Figure 4 269 shows the intrinsic MR ratio variation with Mn content for the as-deposited, 250°C PA and 350°C PA 270 samples. The change in observed MR ratio and 𝛥RA with respect to Mn content is shown in figure 271 S1(a) and S1(b) in the supplementary material, respectively. In-plane R-H curves for the as-deposited, 272 250 °C and 350 °C PA Type-I samples at three different compositions viz. Co2FeGe, Co2Mn0.5Fe0.5Ge 273 and Co2Mn0.9Ge0.1.are shown in the supplementary figure S2. In the as-deposited sample, the MR 274 ratio exhibited gradual change with Mn content. The CFG side showed lower MR ratio ~ 2.5% and 275 the CMG side showed slightly higher MR ratio ~ 5%. In the 250°C PA sample, the MR was greatly 276 enhanced and exhibited the following clear composition dependence. The maximum MR ratio was ~ 277 25% at x = 0, increased gradually with increasing x, and showed a maximum MR ratio of ~38% at 278  FIG. 4. Change in intrinsic magnetoresistance (MR) ratio with Mn content for the Type-I CPP-GMR sample stack with different annealing temperatures.  16  around x= 0.85. Then it decreased to the CMG side. In the 350°C PA sample, MR was further 279 enhanced with maximum MR ratio of ~ 45% in the broad x range. The trend of MR ratio versus x 280 changed from the case of 250°C PA sample. First, it increased in the x range of 0-0.2, stayed almost 281 constant in broad x range of 0.2-0.7, and then abruptly decreased near the CMG side. In addition, the 282 distribution in MR increases significantly with increased annealing temperature. The MR analysis 283 was also performed for the 350°C PA Type-II sample, see figure S3 and S4 in the supplementary 284 material. A drop in MR was observed around the pure CMG side (x = 1), as observed in the 350°C 285 PA Type-I sample.  286   An overall improvement in MR ratio with increasing annealing temperature can be attributed to the 287 enhanced atomic order with increasing annealing temperature, as discussed in the following. Atomic 288 ordering in Heusler is very crucial to obtain good magnetic properties. In-general, for Co2YZ (Y = 289 Cr, Mn and Z = Si, Ge, Al) HMF Heusler compounds, the perfectly ordered structure has the highest 290 P. While the B2 type disorder (swapping between Y and Z sites) does not reduces P significantly, the 291 A2 type disorder (swapping between Co and Y/Z sites) reduce P considerably3334. The MR ratio in 292 CPP-GMR devices being proportional to P of FM electrodes, is enhanced with increasing annealing 293 temperature. The as deposited sample mainly showed A2 disordered phase and exhibited low MR 294 ratio. On increasing PA temperature to 250°C we observed significant enhancement in MR ratio due 295 to enhanced B2 ordering in the sample. The MR ratio increased further as the L21 ordering sets in for 296 the 350°C PA sample. The maximum MR ratio for the 250°C PA Type-I sample reached to a high 297 value of 38% around x = 0.85. In contrast, the 350°C PA Type-I sample showed the drop in MR ratio 298 at x = 0.85, which was optimum at 250°C, deteriorated at 350°C. The drop in MR ratio can partially 299 be attributed to the existence of secondary phase in the CMG side, as observed in the XRD. 300 Interdiffusion of CMFG with Ag buffer and spacer is also expected to affect the MR property, which 301 we will discuss later from the STEM observation. As a result, the optimal composition range shifted 302  17  to the lower x for the 350°C PA Type-I sample, and maximum 𝛥RA values in the range ~ 4 – 8 mΩ 303 μm² was observed (see figure S1(b) in the supplemental material). The high MR ~ 38 % for x = 0.85 304 and MR ~ 45 % for x range of 0.2 ≤ x ≤ 0.7 in CMFG was achieved at relatively low annealing 305 temperature of 250°C and 350°C, respectively. These MR values are the highest among all-metal 306 CPP-GMR devices at lower annealing temperatures (Ref. table S1 in the supplementary material {see 307 also references 54–57 therein}). 308 D. Cross sectional TEM observation 309  FIG. 5. Cross sectional HAADF-STEM images along with EDS elemental maps of Ag, Co, Ru, Pt and Au and the corresponding EDS line profile along the direction indicated by white arrow for the (a) as-deposited, (b) 250°C and (c) 350°C PA Type-I sample at x = 0.5 in CMFG.   We performed cross-sectional TEM analysis to elucidate the effect of annealing temperature and 310  18  composition on the interface between the spacer and the CMFG layers. Figures 5(a), (b) and (c) show 311 the cross-sectional HAADF-STEM images for the as-deposited, 250°C and 350°C PA Type-I samples 312 at the composition of x = 0.5 along with EDS elemental maps of Ag, Co, Ru, Pt, and Au and the 313 corresponding EDS line compositional profile analyzed along the direction indicated by the white 314 arrow. We see clear and well separated interfaces of the CMFG layers with Ag buffer and AgSn spacer 315 layer for the as-deposited, 250°C and 350°C PA Type-I samples. The surface roughness/irregularity 316 in the Ag buffer was observed to propagate to the successive CMFG, AgSn and Ru layers in the as-317 deposited sample. Interestingly, the film flatness improves with increasing annealing temperature for 318 the 250°C and 350°C PA samples. The composition of the CMFG layer from the EDS line profiles 319 matches well with the composition values obtained from the XRF results. 320  FIG. 6. Cross sectional HAADF-STEM images along with EDS elemental maps of Ag, Co, Ru, Pt  19  and Au and the corresponding EDS line profile along the direction indicated by white arrow for the (a) x = 0 [CFG], (b) x = 0.5 [Co2Mn0.5Fe0.5Ge] and (c) x = 1 [CMG] composition for the 350°C PA Type-I sample.   Figure 6 shows the cross-sectional HAADF-STEM images of the 350°C PA Type-I sample at three 321 different compositions, viz. x = 0, 0.5 and 1. The EDS elemental maps of Ag, Co, Ru, Pt, and Au and 322 the corresponding EDS line compositional profile are analyzed along the direction indicated by white 323 arrow. The interface between the spacer and CMFG layers is sharp without interdiffusion of the 324 constituent elements for x = 0 and x = 0.5. In contrast, significant interdiffusion of the constituent 325 elements was observed in the spacer and CMFG layers in x = 1, resulting in a gradual change in the 326 composition profile. In addition, the spacer became discontinuous as the Co-rich phase was formed 327 through the spacer. The drop in MR ratio for the x = 1 CMFG composition can be attributed to the 328 change in the chemical composition of the Heusler alloy electrodes and the formation of the 329 discontinuous spacer region by interdiffusion. Our result is consistent with the report that Co2FeZ 330 type Heusler alloy shows high thermal tolerance for a Ag spacer in comparison to the Co2MnZ type 331 Heusler10. The fact that Fe has lower solubility in Ag than Mn increases the robustness of the interface 332 between the Fe rich composition in CMFG and Ag in the multilayered structure against high 333 temperature annealing.  334  20  E. Evaluation of STT efficiency 335  FIG. 7. R-Ib curves at several constant 𝜇0Hz values for the 350°C PA Type-II sample at three different compositions (a) x = 0, (b) x = 0.5, and (c) x = 0.9. (d) Variation of critical current (Ic) with 𝜇0Hz at several CMFG compositions (x = 0, 0.2, 0.4, 0.5, 0.6, 0.7, 0.9) and (e) Variation of STT efficiency (𝜂) with Mn content.   We performed the STT induced magnetization reversal measurements to evaluate the composition 336 dependent STT efficiency for the 350°C PA Type-II sample using devices with circular pillar 337 geometry having designed diameter of 80 nm. Figures 7(a), 7(b) and 7(c) shows several resistance 338 versus bias current (R-Ib) curves measured at constant 𝜇0Hz ( magnetic field applied perpendicular to 339 the sample surface) values varying from 1.5 T to 3 T in steps of 0.1 T, for three compositions: x = 0 340 (Co2FeGe), x = 0.5 (Co2Mn0.5Fe0.5Ge), and x = 0.9 (Co2Mn0.9Fe0.1Ge), respectively. Sufficiently large 341 𝜇0 Hz (> 1.5 T) was applied to saturate the magnetization of both the FL and SIL parallel to the 342  21  magnetic field at zero bias current. For both signs of bias, the R-Ib curves exhibited an overall 343 parabolic increase in R which can be ascribed to the Joule heating. The additional step in R appears 344 on applying sufficiently large negative bias, which corresponds to the reversal of FL (NiFe) 345 magnetization due to the STT from SIL (CMFG). The sign of the bias required for the reversal is as 346 expected for the spin-transfer induced FL reversal. For the applied biased current range, the 347 magnetization reversal of the SIL at positive bias was not observed, due to the large magnetic volume 348 of SIL and low STT efficiency of the NiFe. As observed in the R-Ib curves, the Ib amplitude required 349 to induce the magnetization reversal increases with 𝜇0Hz. The value of the required current density(~ 350 4×107 A/cm2) for magnetization switching is comparable to those reported previously in the CPP-351 GMR structures6,58,59. 352   The R-Ib curves were fitted phenomenologically using the following equation:  353 R = 𝑓(𝐼b) +𝛥R2(1 + erfc (𝐼b − 𝐼c𝐼width)) , (2) 364 where f is a second-order polynomial function representing the R change due to the temperature 354 change, 𝛥R represents the amount of the R change due to the magnetization reversal, erfc is the error 355 function, 𝐼c  corresponds to the 𝐼b  value at the center of the R change, and 𝐼width  represents the 𝐼b 356 width of the R change covering approximately 85% of the R change. Figure 7(d) shows the change in 357 𝐼c with 𝜇0Hz for several CMFG compositions. We applied spin transfer torque model, considering 358 that the SIL magnetization is in the +z direction owing to the 𝜇0Hz, the FL magnetization rotates near 359 the equator owing to the balance of STT and damping term. The critical current density 𝐽𝑐  that 360 satisfies balance between the damping and STT terms is proportional to 𝐻eff and is expressed as:  361 𝐽c = 𝜇02|𝑒|ℏ𝜂𝛼𝑀sFL𝑑𝐻eff. (3)362 Here 𝑑 (7 nm) is the thickness of the NiFe FL. 𝑀sFL (0.9 T) is the saturation magnetization and 𝛼 363  22  (0.011) is the damping constant of the NiFe determined for the 365 MgO//Cr(5nm)/Ag(5nm)/Ni81Fe19(7nm)/Ru(8nm) stack using vibrating sample magnetometer and 366 ferromagnetic resonance measurement. The STT efficiency, 𝜂 is estimated from the slope of the linear 367 relation between 𝐽c and 𝐻eff excluding the effect of the dipolar field from the SIL and demagnetizing 368 field of the FL, which are included in 𝐻eff. As observed from the figure 7(d), 𝐼c changes linearly with 369 𝐻𝑧 which is consistent with the model. The STT efficiency is a dimensionless quantity, dependent on 370 the spin polarization of the conduction electron in the FM layers and the relative angle between 371 magnetization directions of the FL and SIL. Slonczewski gave a formula for the 𝜂, suitable for the 372 CPP-GMR junctions by considering a free electron model as: 373 𝜂 = [−4 + (1 + 𝑃)3(3 + 𝐩 ∙ 𝐦FL) 4𝑃3 2⁄⁄ ]−1. (4) 381 𝐩 is a unit vector in the direction of spin polarization of SIL and 𝐦FL is the unit vector in the direction 374 of FL magnetization. Figure 7(e) shows the 𝜂 at several CMFG compositions for the Type-III sample. 375 The 𝜂 exhibits high value ~ 0.6 across the composition range 0 ≤ x ≤ 0.8, decline slightly with value 376 reaching close to 0.3 towards the Mn rich (x = 0.9) composition. The drop in 𝜂  at higher Mn 377 composition could be related to the existence of secondary phases as discussed previously in terms 378 of drop in MR ratio. The high value of 𝜂 indicates high spin polarization in epitaxially grown CMFG 379 Heusler thin films over a broad composition range.  380  23  F. Temperature dependence of MR and STT 382  FIG. 8. Temperature dependence of (a) MR ratio and (b) STT efficiency (𝜂) at several Co2MnxFe1-xGe compositions (x = 0, 0.2, 0.4, 0.5, 0.6, 0.7, 0.9) for the 350°C PA Type-II sample.   We performed the temperature dependent MR and STT measurement for the 350°C PA Type-II 383 sample. Figure 8(a) shows the change in observed MR ratio with measurement temperature at several 384 Co2MnxFe1-xGe compositions (x=0, 0.2, 0.4, 0.5, 0.6, 0.7, 0.9), calculated from the out-of-plane R-H 385 measurements for the CPP-GMR devices. The measurement temperature was varied from 50 K to 386 300 K in steps of 50 K. Figure 8 (b) shows the variation of STT efficiency with measurement 387 temperature for the same devices as used for the R-H measurement. The temperature dependent out-388  24  of-plane MR curves, recorded at different temperature for three different compositions (x = 0, x = 0.5, 389 and x = 0.9) and R-Ib curves at 50K for the same compositions are shown in figure S5 and S6 in the 390 supplementary material, respectively. The increase in MR ratio at low temperature in Heusler alloy 391 based CPP-GMR is understood to be due to increased spin-dependent scattering effect. The 392 temperature dependence of MR ratio (𝛥R50K/𝛥R300K) is higher for Fe rich composition (1.93) than Mn 393 rich composition (1.74), indicating that Fe rich composition has a higher spin asymmetry of scattering 394 of the conduction electrons than Mn rich composition at low temperatures40. MR ratios show large 395 temperature dependence in several studies where the MR increases by a factor of two or three from 396 room temperature to low temperature3,5,13. The STT efficiency also improves with decreasing 397 temperature, changes more gradually at lower temperatures (200 K-50 K) as compared to the change 398 in MR% with temperature. 399 G. Discussion and comparison with other CPP-GMR structures 400   The MR and STT are the characteristic magnetotransport phenomenon closely related to each other 401 through the spin polarization of charge carriers. In a FM/non-magnet/FM structure, the MR is related 402 to the component in the interlayer which is parallel to the magnetization of the second layer and the 403 STT is related to the absorption of transverse component of the spin polarized current (generated by 404 spin polarization of SIL layer) by the free FM layer. We recently explained the qualitative agreement 405 between the MR and the STT response in CPP GMR devices fabricated from NiFe(10nm)/CoxFe1-406 x(5nm)/Cu(3nm)/NiFe(5nm) GMR stacks containing composition-gradient CoxFe1-x (0 ≤ x ≤ 1) layer 407 in the SIL. MR and STT were found to be correlated to the bulk spin polarization of the SIL32. There 408 we observed maximum MR ratio ~2.8% and STT efficiency ~ 0.4 for the composition range 0.3 ≤ x 409 ≤ 0.65 in CoxFe1-x. For the present study comprising CMFG Heusler films based CPP-GMR devices, 410 we observed approximately 2-fold increment in MR ratio along with 75% enhancement in the STT 411 efficiency as compared to CoxFe1-x (0 ≤ x ≤ 1) film studied previously, highlighting the higher spin 412  25  polarization in the CMFG Heusler thin films. We obtained large MR ratio ~ 38 % at relatively low 413 annealing temperature of 250°C for the Mn-rich CMFG composition (x = 0.85). The MR ratio exhibits 414 1.6 fold increase when x is varied from 0 to 0.85 in CMFG for the 250°C PA sample. Although the 415 Co/Ge ratio changes ~16% from CFG to CMG side in the present study, Ref. [35] reports that a 15% 416 change in Co/Ge ratio in CMFG led to a ~ 1.1-fold change in the MR ratio. This suggests that the 417 Mn/Fe ratio plays a more significant role in enhancing the MR ratio. The larger MR ratio > 45% along 418 with high STT efficiency ~0.6 was obtained for a broad composition range of 0.2 ≤ x ≤ 0.7 at 350°C 419 annealing temperature. The optimal composition range for the high MR changed with annealing 420 temperature. Mn being more prone to diffusion, Fe rich FM electrode composition was found to be 421 more stable against diffusion at the interface between FM electrode and spacer at higher annealing 422 temperature. The large MR ratio obtained over broad composition range (0.2 ≤ x ≤ 0.7) for the 350 °C 423 annealed type-I sample having L21 ordered CMFG electrodes is in qualitative agreement with the 424 large values of calculated Psp ( > 95 %) over the broad composition range (0.3 ≤ x ≤ 1). Although 425 experimental and theoretical results are consistent in that the MR ratio and Psp show a decrease in 426 CFG side, decrease in the MR ratio is rather gentle in contrast to the loss of half-metallic character in 427 the calculated Psp, which can be attributed to the following reason. The GGA method is insufficient 428 for systems with well-localized d-electron orbitals because the GGA method tends to delocalize these 429 electrons, leading to an inaccurate representation of the electron correlation. Consequently, 430 incorporating an on-site Coulomb interaction (U), as in the GGA + U approach60,61 is recommended. 431 Previous studies argued the requirement of U to obtain more accurate DOS in Co-based Heusler 432 alloy62. Addressing this limitation in future work could improve the alignment between theory and 433 experiment. 434  26   FIG. 9. Intrinsic MR ratio v/s annealing temperature for all-metal CPP-GMR devices containing different Heusler thin film as FM electrodes3,7,55,8–11,13,36–38. The symbol color represents the types of Heusler materials, and the circular and square symbols represent the data for polycrystalline (poly.) and epitaxial (epi.) CPP-GMR devices. The data from different studies on the same Heusler material are shown by the same color with different shade. SV (PSV) in the legend stands for spin valve (pseudo spin valve) CPP-GMR devices. The star symbols show the present data for the CPP-GMR devices containing composition-gradient CMFG Heusler thin films.   Figure 9 shows the intrinsic MR ratio v/s annealing temperature for the present CPP-GMR stacks 435 containing composition-gradient CMFG films, shown by star symbols, in comparison to all-metal 436 spin valves and pseudo spin valves devices containing polycrystalline CMG, CFG, Co2Mn0.6Fe0.4Ge, 437 CFGG and CFAS Heusler thin films shown by circular symbols and epitaxial CMS, CFAS, CMFS 438 and CFGG Heusler thin films shown by square symbols, as FM electrodes3,7,55,8–11,13,36–38. Compared 439 to the literature values, we achieved highest MR ratio ~ 38% and 53% at x = 0.85 and x = 0.13, 440 respectively in the present CMFG Heusler based CPP-GMR devices at low annealing temperature of 441 250°C and 350°C, respectively. The Heusler alloy based CPP-GMR devices showing large MR ratio 442 (>30%), often require high temperature annealing more than 450°C to achieve good magneto 443 transport properties. For practical applications, for example in the read head sensor’s fabrication 444  27  process, the maximum applicable annealing temperature could be 350°C12. The large MR ratio 445 obtained at lower annealing temperature in the present study highlights the advantage of using CMFG 446 Heusler thin films in the CPP-GMR devices for the future read head sensor applications. 447 IV. CONCLUSION 448   We fabricated high performance CPP-GMR devices from the GMR stacks containing composition-449 gradient Co2MnxFe1-xGe (0 ≤ x ≤ 1) film by high throughput and detailed composition optimization 450 method. The combinatorially sputtered composition-gradient film and local measurements enabled 451 efficient compositional optimization at fine composition interval. The XRD measurement conducted 452 at several positions on the GMR stacks with different compositions revealed the epitaxial and largely 453 single-phase structure throughout the composition range and improvement in atomic ordering with 454 annealing temperature. The CPP-GMR devices fabricated from the stacks annealed at 250°C exhibit 455 clear composition dependence of MR with a maximum of change in 𝛥RA ~ 5 m𝛺 𝜇m2 and MR ratio 456 ~ 38 % in the Mn-rich region of x = 0.85. The MR ratio was further enhanced with increasing 457 annealing temperature to 350°C and we observed maximum MR ratio close to ~ 45% (𝛥RA ~ 8 m𝛺 458 𝜇m2) along with high STT efficiency ~ 0.6 over a broad composition range 0.2 ≤ x ≤ 0.7. The optimal 459 composition range shifted to lower x with increasing annealing temperature as the stability of the 460 interface between FM electrode and spacer was changed and higher for Fe rich composition. We 461 achieved record high MR for the all-metal CPP-GMR devices at low annealing temperature of 250°C 462 by the present composition optimization method. The results provide comprehensive guidance on the 463 composition optimization to obtain large MR ratio and high STT efficiency in the CPP-GMR devices 464 using Co2MnxFe1-xGe at lower process temperature. 465  466 SUPPLEMENTARY MATERIAL 467  28  See the supplementary material for details on the change in observed MR ratio and 𝛥RA with Mn 468 content in CMFG for the as-deposited, 250°C and 350°C PA Type-I sample; in-plane R-H curves for 469 the as-deposited, 250 °C and 350 °C PA Type-I samples at three different compositions viz. Co2FeGe, 470 Co2Mn0.5Fe0.5Ge and Co2MnGe; in-plane R-H curves for the 350 °C PA Type-II sample at three 471 different compositions viz. Co2FeGe, Co2Mn0.5Fe0.5Ge and Co2Mn0.9Ge0.1; change in observed MR 472 ratio and 𝛥RA with Mn content in CMFG for the 350°C PA Type-II sample; out-of-plane MR curves 473 in the temperature range of 300 K to 50 K for the 350°C PA Type-II sample at three different 474 compositions viz. Co2FeGe, Co2Mn0.5Fe0.5Ge and Co2Mn0.9Fe0.1Ge; R-Ib curves at 50 K for several 475 constant μ0Hz values for the 350°C PA Type-II sample at three different compositions viz. Co2FeGe, 476 Co2Mn0.5Fe0.5Ge and Co2Mn0.9Fe0.1Ge; and the table showing the summary of MR properties studied 477 at room temperature for the CPP-GMR devices containing Co2MnxFe1-xGe (0 ≤ x ≤ 1) Heusler as 478 ferromagnetic layer electrode with metallic spacers, which includes Refs. [54-57]. 479  480 ACKNOWLEDGEMENTS 481 This work was supported by the Advanced Storage Research Consortium (ASRC), JST CREST Grant 482 No. JPMJCR21O1, MEXT Program: Data Creation and Utilization-Type Material Research and 483 Development Project Grant Number JPMXP1122715503, and MEXT Initiative to Establish Next-484 generation Novel Integrated Circuits Centers (X-NICS) grant number JPJ011438. 485  486  487  488 AUTHOR DECLARATION 489  29  Conflict of Interest 490 The authors have no conflicts to disclose. 491 Author Contributions 492 Vineet Barwal: Data curation (equal); Formal analysis (equal); Investigation (equal); Visualization 493 (equal); Writing - original draft (equal); Writing - review and editing (equal). Hirofumi Suto: 494 Conceptualization (equal); Formal analysis (equal); Methodology (equal); Project Administration 495 (equal); Resources (equal); Supervision (equal); Writing-review and editing (equal). Ryo Toyama: 496 Formal analysis (supporting); Investigation (supporting); Writing-review and editing (supporting). 497 Kodchakorn Simalaotao: Formal analysis (supporting); Investigation (supporting); Writing-review 498 and editing (supporting). Taisuke Sasaki: Investigation (equal); Writing-review and editing 499 (supporting). Yoshio Miura: Formal analysis (supporting); Investigation (supporting); Writing-500 review and editing (supporting). Yuya Sakuraba: Conceptualization (equal); Project Administration 501 (equal); Resources (equal); Supervision (equal); Writing-review and editing (equal). 502 DATA AVAILABILITY 503 The data that support the findings of this study are available within the article and its supplementary 504 material. 505   506  30  REFERENCES 507 1 K. Elphick, W. Frost, M. Samiepour, T. Kubota, K. Takanashi, H. Sukegawa, S. Mitani, and A. 508 Hirohata, Sci. Technol. Adv. Mater. 22, 235 (2021). 509 2 A. Hirohata and D.C. Lloyd, MRS Bull. 47, 593 (2022). 510 3 Y. Sakuraba, K. Izumi, T. Iwase, S. Bosu, K. Saito, K. Takanashi, Y. Miura, K. Futatsukawa, K. 511 Abe, and M. Shirai, Phys. Rev. B - Condens. Matter Mater. Phys. 82, 094444 (2010). 512 4 J. Sato, M. Oogane, H. Naganuma, and Y. Ando, Appl. Phys. Express 4, 113005 (2011). 513 5 Y. Sakuraba, M. Ueda, Y. Miura, K. Sato, S. Bosu, K. Saito, M. Shirai, T.J. Konno, and K. Takanashi, 514 Appl. Phys. Lett. 101, 252408 (2012). 515 6 H. Narisawa, T. Kubota, and K. Takanashi, Appl. Phys. Express 8, 063008 (2015). 516 7 T. Kubota, Y. Ina, Z. Wen, H. Narisawa, and K. Takanashi, Phys. Rev. Mater. 1, 044402 (2017). 517 8 Z. Wen, T. Kubota, Y. Ina, and K. Takanashi, Appl. Phys. Lett. 110, 102401 (2017). 518 9 T.M. Nakatani, T. Furubayashi, S. Kasai, H. Sukegawa, Y.K. Takahashi, S. Mitani, and K. Hono, 519 Appl. Phys. Lett. 96, 212501 (2010). 520 10 S. Li, Y.K. Takahashi, T. Furubayashi, and K. Hono, Appl. Phys. Lett. 103, 042405 (2013). 521 11 Y. Du, T. Furubayashi, T.T. Sasaki, Y. Sakuraba, Y.K. Takahashi, and K. Hono, Appl. Phys. Lett. 522 107, 112405 (2015). 523 12 T. Nakatani, Z. Gao, and K. Hono, MRS Bull. 43, 106 (2018). 524 13 J.W. Jung, Y. Sakuraba, T.T. Sasaki, Y. Miura, and K. Hono, Appl. Phys. Lett. 108, 102408 (2016). 525 14 Y.K. Takahashi, A. Srinivasan, B. Varaprasad, A. Rajanikanth, N. Hase, T.M. Nakatani, S. Kasai, 526 T. Furubayashi, and K. Hono, Appl. Phys. Lett. 98, 152501 (2011). 527 15 B. Büker, J. Jung, T. Sasaki, Y. Sakuraba, Y. Miura, T. Nakatani, A. Hütten, and K. Hono, Phys. 528 Rev. B 103, L140405 (2021). 529 16 T. Kubota, Z. Wen, and K. Takanashi, J. Magn. Magn. Mater. 492, 165667 (2019). 530 17 Y. Chikaso, M. Inoue, T. Tanimoto, K. Kikuchi, M. Yamanouchi, T. Uemura, K. Inubushi, K. 531 Nakada, H. Shinya, and M. Shirai, J. Phys. D. Appl. Phys. 55, 345003 (2022). 532  31  18 Z. Wen, H. Sukegawa, S. Kasai, K. Inomata, and S. Mitani, Phys. Rev. Appl. 2, 024009 (2014). 533 19 J. Zhang, T. Phung, A. Pushp, Y. Ferrante, J. Jeong, C. Rettner, B.P. Hughes, S.H. Yang, Y. Jiang, 534 and S.S.P. Parkin, Appl. Phys. Lett. 110, 172403 (2017). 535 20 A. V. Khvalkovskiy, D. Apalkov, S. Watts, R. Chepulskii, R.S. Beach, A. Ong, X. Tang, A. 536 Driskill-Smith, W.H. Butler, P.B. Visscher, D. Lottis, E. Chen, V. Nikitin, and M. Krounbi, J. Phys. 537 D. Appl. Phys. 46, 139601 (2013). 538 21 S. Okamoto, N. Kikuchi, M. Furuta, O. Kitakami, and T. Shimatsu, J. Phys. D. Appl. Phys. 48, 539 353001 (2015). 540 22 W. Zhou, H. Sepehri-Amin, T. Taniguchi, S. Tamaru, Y. Sakuraba, S. Kasai, H. Kubota, and K. 541 Hono, Appl. Phys. Lett. 114, 172403 (2019). 542 23 M. Takagishi, N. Narita, H. Iwasaki, H. Suto, T. Maeda, and A. Takeo, IEEE Trans. Magn. 57, 543 3300106 (2021). 544 24 H. Suto, M. Takagishi, N. Narita, H. Iwasaki, T. Nagasawa, G. Koizumi, A. Takeo, and T. Maeda, 545 J. Appl. Phys. 129, 103901 (2021). 546 25 R. Toyama, V.K. Kushwaha, T.T. Sasaki, Y. Iwasaki, T. Nakatani, and Y. Sakuraba, APL Mater. 547 11, 101127 (2023). 548 26 Y. Sakuraba, S. Kokado, Y. Hirayama, T. Furubayashi, H. Sukegawa, S. Li, Y.K. Takahashi, and 549 K. Hono, Appl. Phys. Lett. 104, 172407 (2014). 550 27 C. Guillemard, W. Zhang, G. Malinowski, C. de Melo, J. Gorchon, S. Petit‐Watelot, J. Ghanbaja, 551 S. Mangin, P. Le Fèvre, F. Bertran, and S. Andrieu, Adv. Mater. 32, 1908357 (2020). 552 28 M. Inoue, B. Hu, K. Moges, K. Inubushi, K. Nakada, M. Yamamoto, and T. Uemura, Appl. Phys. 553 Lett. 111, 082403 (2017). 554 29 S. Li, Y.K. Takahashi, Y. Sakuraba, N. Tsuji, H. Tajiri, Y. Miura, J. Chen, T. Furubayashi, and K. 555 Hono, Appl. Phys. Lett. 108, 122404 (2016). 556 30 A. Frisk, M. Ahlberg, G. Muscas, S. George, R. Johansson, W. Klysubun, P.E. Jönsson, and G. 557 Andersson, Phys. Rev. Mater. 3, 074403 (2019). 558 31 H. Masuda, R. Modak, T. Seki, K. Uchida, Y.-C. Lau, Y. Sakuraba, R. Iguchi, and K. Takanashi, 559 Commun. Mater. 1, 75 (2020). 560  32  32 V. Barwal, H. Suto, T. Taniguchi, and Y. Sakuraba, Sci. Technol. Adv. Mater. Methods 3(1), 561 2286944 (2023). 562 33 S. Picozzi, A. Continenza, and A.J. Freeman, Phys. Rev. B - Condens. Matter Mater. Phys. 69, 563 094423 (2004). 564 34 Y. Miura, K. Nagao, and M. Shirai, Phys. Rev. B - Condens. Matter Phys. 69, 144413 (2004). 565 35 T. Nakatani, S. Li, Y. Sakuraba, T. Furubayashi, and K. Hono, IEEE Trans. Magn. 54, 3300211 566 (2018). 567 36 T. Nakatani, G. Mihajlović, J.C. Read, Y.S. Choi, and J.R. Childress, Appl. Phys. Express 8, 568 093003 (2015). 569 37 Y.S. Choi, T. Nakatani, J.C. Read, M.J. Carey, D.A. Stewart, and J.R. Childress, Appl. Phys. 570 Express 10, 013006 (2017). 571 38 S. Li, T. Nakatani, K. Masuda, Y. Sakuraba, X.D. Xu, T.T. Sasaki, H. Tajiri, Y. Miura, T. 572 Furubayashi, and K. Hono, Acta Mater. 142, 49 (2018). 573 39 T. Nakatani, S.K. Narayananellore, L.S.R. Kumara, H. Tajiri, Y. Sakuraba, and K. Hono, Scr. Mater. 574 189, 63 (2020). 575 40 M.R. Page, T.M. Nakatani, D.A. Stewart, B.R. York, J.C. Read, Y.-S. Choi, and J.R. Childress, J. 576 Appl. Phys. 119, 153903 (2016). 577 41 R. Toyama, S. Kokado, K. Masuda, Z. Li, V.K. Kushwaha, T.T. Sasaki, L.S.R. Kumara, T. 578 Koganezawa, H. Tajiri, T. Yamazaki, M. Kotsugi, Y. Iwasaki, and Y. Sakuraba, Phys. Rev. Mater. 7, 579 084401 (2023). 580 42 R. Toyama, W. Zhou, and Y. Sakuraba, Phys. Rev. B 109(1), 054415 (2024). 581 43 H. Suto, T. Nakatani, N. Asam, H. Iwasaki, and Y. Sakuraba, Appl. Phys. Express 16, 013003 582 (2023). 583 44 B. Balke, S. Wurmehl, G.H. Fecher, C. Felser, M.C.M. Alves, F. Bernardi, and J. Morais, Appl. 584 Phys. Lett. 90, 172501(2007). 585 45 P.J. Webster and K.R.A. Ziebeck, J. Phys. Chem. Solids 34, 1647 (1973). 586 46 J.P. Perdew, K. Burke, and M. Ernzerhof, Phys. Rev. Lett. 77, 3865 (1996). 587  33  47 G. Kresse and J. Hafner, Phys. Rev. B 47, 558 (1993). 588 48 G. Kresse and J. Furthmüller, Comput. Mater. Sci. 6, 15 (1996). 589 49 P.E. Blöchl, Phys. Rev. B 50, 17953 (1994). 590 50 G. Kresse and D. Joubert, Phys. Rev. B 59, 1758 (1999). 591 51 L. Bellaiche and D. Vanderbilt, Phys. Rev. B 61, 7877 (2000). 592 52 J.D. Pack and H.J. Monkhorst, Phys. Rev. B 16, 1748 (1977). 593 53 Z. Chen, Y. Sakuraba, Y. Miura, Z. Li, T. Sasaki, H. Suto, V.K. Kushwaha, T. Nakatani, S. Mitani, 594 and K. Hono, J. Appl. Phys. 132, 183902 (2022). 595 54 K. Nikolaev, P. Kolbo, T. Pokhil, X. Peng, Y. Chen, T. Ambrose, and O. Mryasov, Appl. Phys. 596 Lett. 94, 222501 (2009). 597 55 M.J. Carey, S. Maat, S. Chandrashekariaih, J.A. Katine, W. Chen, B. York, and J.R. Childress, J. 598 Appl. Phys. 109, 093912 (2011). 599 56 T.P. Ginley, J.A. Borchers, B.J. Kirby, C.L. Dennis, M.J. Carey, and J.R. Childress, J. Appl. Phys. 600 109,07B110 (2011). 601 57 H.S. Goripati, T. Furubayashi, Y.K. Takahashi, and K. Hono, J. Appl. Phys. 113, 043901 (2013). 602 58 J.A. Katine, F.J. Albert, R.A. Buhrman, E.B. Myers, and D.C. Ralph, Phys. Rev. Lett. 84, 3149 603 (2000). 604 59 G. Mihajlović, T.M. Nakatani, N. Smith, J.C. Read, Y.S. Choi, H.W. Tseng, and J.R. Childress, 605 IEEE Magn. Lett. 6, 3001104 (2015). 606 60 B. Balke, G.H. Fecher, H.C. Kandpal, C. Felser, K. Kobayashi, E. Ikenaga, J.-J. Kim, and S. Ueda, 607 Phys. Rev. B 74, 104405 (2006). 608 61 K. Nawa and Y. Miura, RSC Adv. 9, 30462 (2019). 609 62 K. Sumida, Y. Fujita, W. Zhou, K. Masuda, I. Kawasaki, S. Fujimori, A. Kimura, and Y. Sakuraba, 610 Phys. Rev. B 108, L241101 (2023). 611  612  613  34  Supplementary Material 614 MR analysis for the Type-I sample 615  FIG. S1. Change in (a) observed MR ratio and (b) 𝛥RA with Mn content in CMFG for the as-deposited, 250°C and 350°C PA Type-I sample. Figures S1(a) and S1(b) show the change in observed magnetoresistance (MR) ratio and change in 616 resistance area product (𝛥RA) with Mn content for the as-deposited, 250°C and 350°C PA Type-I sample. 617 The change in observed MR ratio and 𝛥RA with Mn content is consistent with the change in intrinsic MR 618 ratio with Mn content, as shown in the main manuscript. For the 250°C PA sample, we can see an 619 enhancement in MR% around x = 0.85. The observed MR ratio reaches to a maximum value of around 620 35% for 350°C PA in a broad composition range (0.2 ≤ x ≤ 0.7). 621   622  35  R-H curves for Type-I sample 623  FIG. S2. In-plane R-H curves for the as-deposited, 250 °C and 350 °C PA Type-I samples at three different compositions viz. Co2FeGe, Co2Mn0.5Fe0.5Ge and Co2MnGe. Figure S2 shows the in-plane R-H curves for the as-deposited, 250 °C and 350 °C PA Type-I samples at 624 three different compositions viz. Co2FeGe, Co2Mn0.5Fe0.5Ge and Co2Mn0.9Ge0.1. 625   626  36  R-H curves for Type-II sample 627  FIG. S3. In-plane R-H curves for the 350 °C PA Type-II sample at three different compositions viz. Co2FeGe, Co2Mn0.5Fe0.5Ge and Co2Mn0.9Ge0.1. Figure S3 shows the in-plane R-H curves for the 350 °C PA Type-II sample at three different compositions 628 viz. Co2FeGe, Co2Mn0.5Fe0.5Ge and Co2Mn0.9Ge0.1. 629 MR analysis for Type-II sample 630  FIG. S4. Change in (a) observed MR ratio and (b) 𝛥RA with Mn content in CMFG for the 350°C PA Type-II sample.   Figures S4(a) and S4(b) show the change in observed MR ratio and 𝛥RA with Mn content for the 350°C 631 PA Type-II sample. The MR ratio and 𝛥RA do not change much with Mn content, except for the drop 632 around higher Mn concentration, similar to the 350 °C PA Type-I sample. The MR ratio and 𝛥RA values 633 reduce by a factor of 0.25 for the Type-II sample as compared to Type-I sample. Change in 𝛥RA was 634  37  calculated from the MR measurements using the designed device cross sectional area. The large 635 distribution in 𝛥RA for the 350 °C PA samples, is due to increased diffusion at the interface at higher 636 annealing temperature. 637 MR curves for Type-II sample at different measurement temperatures 638  FIG. S5. Out-of-plane MR curves in the temperature range of 300 K to 50 K for the 350°C PA Type-II sample at three different compositions (a) Co2FeGe, (b) Co2Mn0.5Fe0.5Ge and (c) Co2Mn0.9Fe0.1Ge.   Figures S5 (a), (b) and (c) show the out-of-plane MR curves at different temperatures (50 K, 100 K, 150 639 K, 200 K, 250 K and 300 K) for the 350°C PA Type-II sample at three different compositions (Co2FeGe, 640 Co2Mn0.5Fe0.5Ge and Co2Mn0.9Fe0.1Ge). The MR ratio increased more than 2.5 times upon decreasing the 641 temperature from 300 K to 50 K. 642   643  38  Magnetization reversal curves at 50K for Type-II sample  644  FIG. S6. R-Ib curves at 50 K for several constant 𝜇0Hz values for the 350°C PA Type-II sample at three different compositions (a) Co2FeGe, (b) Co2Mn0.5Fe0.5Ge and (c) Co2Mn0.9Fe0.1Ge.   Figures S6 (a), (b) and (c) show the R-Ib curves at 50 K for several constant 𝜇0Hz values (1.5 T, 1.8 T, 645 2.1 T, 2.4 T, 2.7 T and 3 T) for the 350°C PA Type-II sample for three different compositions viz. Co2FeGe, 646 Co2Mn0.5Fe0.5Ge and Co2Mn0.9Fe0.1Ge. STT efficiency increased by factor of ~ 1.4 upon decreasing the 647 temperature from 300 K to 50 K. 648 Summary of MR study on Co2MnxFe1-xGe (0 ≤ x ≤ 1) [CMFG] Heusler based CPP-GMR devices 649 Table S1. Summary of MR properties studied at room temperature for CPP-GMR devices containing 650 Co2MnxFe1-xGe (0 ≤ x ≤ 1) [CMFG] Heusler as ferromagnetic layer electrode with metallic spacers. The 651 epitaxially grown stacks are indicated by acronym “epi.” and polycrystalline stacks are indicated by 652 acronym “poly.”. Numbers in the parentheses indicates layer thickness in nm. PSVs are the pseudo spin 653 valves without any pinned layer and SVs means spin valves with pinned layer. RT denotes room 654 temperature meaning no annealing was done for the CPP-GMR stacks. 655   656  39  CPP-GMR stacking structure MR ratio  (%) ΔRA (mΩ μm²) Annealing Temperature (°C) Ref. Co2MnGe (CMG) Seed layer/IrMn/CoFe/Ru/CMG(3.6)/Rh2CuSn(2.2)/CMG(3.6)/capping layer poly. SV 6.7 4 250 1 Seed layer/IrMn(7)/Co65Fe35(4)/Ru(0.8)/Co90Fe10(0.6)/CMG(4)/Co90Fe10(0.6)/Cu(8)/Co90Fe10(0.6)/CMG(4)/ Co90Fe10(0.6)/Cu(8)/Co90Fe10(0.6)/CMG(4)/Co90Fe10(0.6)/capping layer poly. SV 9 6 245 2 Seed layer/CMG(8)/Cu(5)/Co50Fe50(0.5)/CMG(4)/Co50Fe50(1)/IrMn(6)/Ru(12)/capping layer poly. SV - 2.6 245 3 Seed layer/Co50Fe50(0.5)/CMG(8)/Co50Fe50(0.5)/Cu(5)/Co50Fe50(0.5)/CMG(4)/Co50Fe50(1)/IrMn(6)/Ru(12)/capping layer poly. SV - 3.2 Co2FeGe (CFG) Seed layer/CFG(10)/Ag(5)/CFG(10)/capping layer epi. PSV - 4 300 4 Seed layer/Ru(2)/IrMn(6)/Co50Fe50(1)/CFG(4)/Co50Fe50(0.5)/ Ag(3.5)/Co50Fe50(0.5)/CFG(4)/Co50Fe50(1)/capping layer poly. SV 10 - 270/280 5 Co2MnxFe1-xGe (CMFG) Seed layer/IrMn(6)/Co50Fe50(2.7)/Ru(0.8)/Co50Fe50(0.7)/ CMFG(6.5)/Co50Fe50(0.5)/Ag90Sn10(2.5)/Co50Fe50(0.5) /CMFG(6.5)/capping layer x = 0.6 poly. SV 15 - 270/280 5 Seed layer/IrMn(5)/Co50Fe50(2.6)/Ru(0.8)/Co50Fe50(0.7)/ CoFeBTa(0.6)/CMFG(2.5)/Co50Fe50(0.2)/Ag90Sn10(4)/ Co50Fe50(0.2)/CMFG(4)/Co50Fe50(1)/Ru(0.8)/ Co50Fe50(1.2)/capping layer x = 1, 0.9, 0.8, 0.6 poly. SV 13 6 270 6 Seed layer/IrMn(5)/Co(0.4)/Co50Fe50(0.8)/ (Co40Fe40B20)0.93Ta0.07(0.9)/CMFG(2.5)/Co50Fe50(0.2)/ Ag90Sn10(3.5)/Co50Fe50(0.2)/CMFG(4)/capping layer  x = 0.6 poly. SV 18 8 270 7 Seed layer/CMFG(10)/Ag(5)/CMFG(10)/capping layer x = 0.6 epi. PSV - 9.5 400 8 Seed layer/CMFG(5)/Ag90Sn10(4)/CMFG(5)/capping layer - 7.5 300  40  x = 0.6 poly. PSV Seed layer/IrMn(6)/Co50Fe50(1)/(Co40Fe40B20)0.93Ta0.07(0.8)/ CMFG(5)/Co50Fe50(0.4)/Ag90Sn10(3.5)/Co50Fe50(0.4)/ CMFG(5)/Co50Fe50(1)/capping layer x = 0.6 poly. SV 16 - 300 Seed layer/IrMn(6)/Co50Fe50(3)/Ru(0.8)/Co50Fe50(0.6)/ (Co40Fe40B20)0.93Ta0.07(0.8)/CMFG(3)/Co50Fe50(0.4)/Ag90Sn10(3.5)/Co50Fe50(0.4)/CMFG(4)/Co50Fe50(1)/capping layer x = 0.6 poly. SV 14 4 280 Seed layer/(Co40Fe40B20)0.93Ta0.07(1.2)/CMFG(5)/Co50Fe50(0.4)/Ag90Sn10(4)/Co50Fe50(0.4)/CMFG(5)/Co50Fe50(1)/ capping layer  x = 0.6 poly. PSV 25 7.5 300 9 Seed layer/(Co40Fe40B20)0.93Ta0.07(0.6)/CMFG(4)/Co50Fe50(0.4)/Ag90Sn10(3.5)/Co50Fe50(0.4)/CMFG(5)/Co50Fe50(1)/ capping layer  x = 0.6 poly. PSV 17.8 6 300 10 Seed layer/CMFG(7)/Ag90Sn10(5)/CMFG(7)/capping layer 0.2 ≤ x ≤ 0.7 epi.PSV 35-45 4-8 350 Present case Seed layer/CMFG(7)/Ag90Sn10(5)/CMFG(7)/capping layer x = 0.85 epi. PSV 35 5 250 Seed layer/CMFG(7)/Ag90Sn10(5)/CMFG(7)/capping layer 0.8 ≤ x ≤ 1 epi. PSV 3-6 0.6-1.5 RT  657 References 658 1 K. Nikolaev, P. Kolbo, T. Pokhil, X. Peng, Y. Chen, T. Ambrose, and O. Mryasov, Appl. Phys. Lett. 94, 659 222501 (2009). 660 2 M.J. Carey, S. Maat, S. Chandrashekariaih, J.A. Katine, W. Chen, B. York, and J.R. Childress, J. Appl. 661 Phys. 109, 093912 (2011). 662 3 T.P. Ginley, J.A. Borchers, B.J. Kirby, C.L. Dennis, M.J. Carey, and J.R. Childress, J. Appl. Phys. 109, 663 (2011). 664  41  4 H.S. Goripati, T. Furubayashi, Y.K. Takahashi, and K. Hono, J. Appl. Phys. 113, 043901 (2013). 665 5 T. Nakatani, G. Mihajlović, J.C. Read, Y.S. Choi, and J.R. Childress, Appl. Phys. Express 8, 093003 (2015). 666 6 M.R. Page, T.M. Nakatani, D.A. Stewart, B.R. York, J.C. Read, Y.-S. Choi, and J.R. Childress, J. Appl. 667 Phys. 119, 153903 (2016). 668 7 Y.S. Choi, T. Nakatani, J.C. Read, M.J. Carey, D.A. Stewart, and J.R. Childress, Appl. Phys. Express 10, 669 013006 (2017). 670 8 T. Nakatani, S. Li, Y. Sakuraba, T. Furubayashi, and K. Hono, IEEE Trans. Magn. 54, 3300211 (2018). 671 9 S. Li, T. Nakatani, K. Masuda, Y. Sakuraba, X.D. Xu, T.T. Sasaki, H. Tajiri, Y. Miura, T. Furubayashi, and 672 K. Hono, Acta Mater. 142, 49 (2018). 673 10 T. Nakatani, S.K. Narayananellore, L.S.R. Kumara, H. Tajiri, Y. Sakuraba, and K. Hono, Scr. Mater. 189, 674 63 (2020). 675