# Fileset [20260713_MDR_Evaluation of spin mixing conductance in Co2FeGa0.5Ge0.5_Pt bilayer and the effect of ultrathin Cu Ni Ru Ta or Cr insertion layers.pdf](https://mdr.nims.go.jp/filesets/70f4f524-7ef4-4f07-b5c3-2651c67b5a6d/download) ## Creator Madhav M. Bhat, [H. Suto](https://orcid.org/0000-0003-4387-5862), [T. T. Sasaki](https://orcid.org/0000-0002-5952-7638), A. Perumal, A. Srinivasan, [Y. Sakuraba](https://orcid.org/0000-0003-4618-9550) ## Rights ©2026 American Physical Society[In Copyright](http://rightsstatements.org/vocab/InC/1.0/) ## Other metadata [Evaluation of spin mixing conductance in Co 2 Fe Ga 0.5 Ge 0.5 / Pt bilayer and the effect of ultrathin Cu, Ni, Ru, Ta, or Cr insertion layers](https://mdr.nims.go.jp/datasets/e9e61312-f145-46d4-816b-f1de9aed09b4) ## Fulltext 1 Evaluation of spin mixing conductance in Co2FeGa0.5Ge0.5/Pt bilayer and the effect of 1 ultrathin Cu, Ni, Ru, Ta, or Cr insertion layers 2 Madhav M. Bhat1,2, H. Suto2,*, T. T. Sasaki2, A. Perumal1, A. Srinivasan1,*, Y. Sakuraba1,2,3 3 1Department of Physics, Indian Institute of Technology Guwahati, Guwahati-781039, India. 4 2 Research Center for Magnetic and Spintronic Materials, NIMS, Tsukuba, 305-0047, Japan. 5 3Graduate School of Pure and Applied Sciences, University of Tsukuba, Tsukuba, 305-8571, 6 Japan. 7 Email: *SUTO.Hirofumi@nims.go.jp, *asrini@iitg.ac.in 8 9 ABSTRACT 10 Improving the spin mixing conductance at the ferromagnet/heavy metal (FM/HM) interface with 11 a low Gilbert damping constant in the FM layer is a key requirement for developing efficient 12 spintronic devices. To evaluate the potential of Co2FeGa0.5Ge0.5 as the FM layers in such structures, 13 epitaxial Co2FeGa0.5Ge0.5 single-layer and Co2FeGa0.5Ge0.5/Pt bilayer films were deposited on 14 MgO substrate using magnetron sputtering, and their Gilbert damping constants were evaluated 15 from ferromagnetic resonance spectra. While the single-layer samples exhibit a low damping 16 constant of ~ 1.07×10-3, an enhancement in the damping constant was observed in the bilayer 17 samples due to spin pumping across the Co2FeGa0.5Ge0.5/Pt interface, from which relatively large 18 spin mixing conductance (𝑔eff↑↓ ) of (2.78±0.09)×1019 m-2 was evaluated. Further, to investigate the 19 effect of interface property on 𝑔eff↑↓ , ultrathin (~ 0.2 nm) Cu, Ni, Ru, Ta, or Cr insertion layers were 20 introduced between the Co2FeGa0.5Ge0.5 and Pt. Inserting a non-magnetic (NM) Cu layer results 21 in a pronounced decrease of 𝑔eff↑↓ , whereas the FM Ni layer leaves 𝑔eff↑↓ nearly unchanged. In the 22 case of Ru and Ta insertion layers, a modest reduction in 𝑔eff↑↓ is observed despite their NM nature. 23 The antiferromagnetic Cr insertion layer also causes a small decrease in 𝑔eff↑↓ . The observed change 24 in 𝑔eff↑↓ depends on the insertion material, indicating that their physical properties, such as magnetic, 25 non-magnetic, antiferromagnetic, light or heavy metal, reflect on 𝑔eff↑↓ . 26 mailto:*asrini@iitg.ac.in2 1. INTRODUCTION 27 The development of high-performance non-volatile magnetoresistive random access memory 28 requires a magnetic tunnel junction with energy-efficient switching of magnetization in the active 29 ferromagnetic (FM) layer [1]. This favours the adoption of the spin-orbit torque (SOT) mechanism, 30 where spin current injection occurs from a nonmagnetic (NM) layer to the active FM layer [2,3]. 31 In this switching scheme, spin mixing conductance (𝑔eff↑↓ ) is a key parameter to quantify the spin 32 current from the NM to the FM layer. So, an FM/NM interface with high 𝑔eff↑↓ (~ 1019 to 1020 m-2) 33 is essential for improving the SOT efficiency [4–6]. 34 The value of 𝑔eff↑↓ is influenced by several key properties of the FM/NM interface. First-35 principles studies by Barati et al. [7] show that the Gilbert damping constant (α) is largely affected 36 by the interfacial electronic structure, spin-orbit coupling strength of the NM layer, and the 37 thicknesses of the FM and NM layers, which together determine 𝑔eff↑↓ . This dependence is 38 experimentally illustrated by Azzawi et al. [8] in Co/Pt, Co/Au, and Ni81Fe19/Pt bilayers. Zhang et 39 al. [9] theoretically showed that an increase in interface roughness enhanced 𝑔eff↑↓ in the 40 permalloy/Pt bilayer sample. Tokaç et al. [10] experimentally observed the effect of FM and NM 41 crystal structures at the interface in Ta/Cu/Co/Cu (or Ir)/Ta multilayers and found that 𝑔eff↑↓ is 42 enhanced when the FM layer and NM overlayers have the same crystal structure. Other studies 43 have also explored the impact of insertion layers at the FM/NM interface on 𝑔eff↑↓ . For example, 44 Kumar et al. [5] studied the effect of Cu insertion layer (0 – 10 nm) on 𝑔eff↑↓ in Co2FeAl/β-Ta bilayer, 45 and Swindells et al. [11,12] examined the effect of Au insertion layer (0 – 3 nm) in Co25Fe75/Pt 46 and Ni80Fe20/Pt bilayers. These studies have provided valuable insights into the roles of 47 proximity-induced magnetization and interface spin current transparency in modifying 𝑔eff↑↓ . 48 Considering these studies, a need for a detailed structural analysis of the NM overlayer at the 49 FM/NM interface is desired to separate the effects of insertion layers from the changes in the NM 50 layer structure that the insertion layers may induce. 51 Another crucial parameter for energy-efficient switching is a low α of the active FM layer [13]. 52 Since α is proportional to the total density of states (DOS) at the Fermi level (EF), half-metallic 53 materials can exhibit low α due to the presence of only one type of electronic spin in DOS at 54 EF [14–16]. Additionally, suppression of spin-flip scattering in half-metals further decreases α [17]. 55 Half-metallicity is also beneficial for achieving large magnetoresistance output [18–21]. In these 56 3 respects, Co2Fe-based Co2FeGa0.5Ge0.5 (CFGG) Heusler alloy has gathered considerable attention 57 with the illustration of enhanced magnetoresistance ratio in current-perpendicular-to-plane (CPP) 58 giant magnetoresistance (GMR) device [18,22,23] with a α of ~ 0.008 [24]. Regarding the NM 59 layer, heavy metals (HMs) such as Pt, Ta, W, Ir, or Hf have been reported as an effective 60 spin-current source. Such materials have a high spin Hall angle (θH) arising from large spin-orbit 61 coupling (SOC), since SOC scales with the atomic number (Z) as a function of Z4 [25,26]. 62 Based on the above material considerations, this study focuses on the CFGG/Pt system and 63 investigates its 𝑔eff↑↓ . The CFGG single-layer and CFGG/Pt bilayer samples were grown epitaxially 64 on MgO substrates, and their α was evaluated using the ferromagnetic resonance (FMR) technique. 65 The enhanced damping in the bilayer CFGG/Pt sample signifies spin pumping across the CFGG/Pt 66 interface, from which relatively high 𝑔eff↑↓ value of (2.78±0.09)×1019 m-2 was estimated. The effect 67 of ultrathin insertion layers on 𝑔eff↑↓ was further examined, highlighting the material-dependent 68 variations in 𝑔eff↑↓ . Notably, structural analysis using X-ray diffraction (XRD) and scanning 69 transmission electron microscope (STEM) analysis confirmed that the Pt layer structure was 70 unaffected by the insertion layers, which excluded the possible contributions to the change in 𝑔eff↑↓ 71 from structural modifications in the Pt layer. The change in 𝑔eff↑↓ was found to depend on the 72 insertion material, indicating that their physical properties, such as magnetic, nonmagnetic, 73 antiferromagnetic, light or heavy metal, are reflected in 𝑔eff↑↓ . 74 75 2. EXPERIMENTAL DETAILS 76 Figures 1(a)–1(c) illustrate the schematic structure of the samples used in this study, viz., CFGG 77 (7.5, 8.8, 10.5, 12.5, 14, 16.3, and 18.5 nm) single-layer samples, CFGG (7.5, 8.8, 10.5, 12.5, 14, 78 16.3, and 18.5 nm)/Pt (10 nm) and CFGG (14 nm)/Pt (0.5, 1, 2, 3, 5, and 10 nm) bilayer samples, 79 and CFGG (7.5 nm)/X (~ 0.2 nm)/Pt (10 nm) bilayer samples with insertion of X (Cu, Ni, Ru, Ta, 80 or Cr) layers. All the samples were deposited on MgO (001) substrate at ambient temperature using 81 an ultra-high vacuum magnetron sputtering system with a base pressure of ~ 10-7 Pa. Prior to the 82 deposition, substrates were cleaned with acetone, propanol, and deionized water, followed by in 83 situ argon ion milling. The CFGG layers were deposited from an alloy target of 84 Co1.95Fe0.89Ga0.48Ge0.68 and subsequently annealed in situ at 600 ℃ for 30 minutes to enhance the 85 atomic ordering. The composition of the CFGG layer was determined to be 86 4 Co2.06±0.01Fe0.99±0.01Ga0.53±0.01Ge0.42±0.01 by X-ray fluorescence spectroscopy. For the single-layer 87 samples, a 2 nm Al capping layer was deposited to avoid oxidation (not shown in Fig. 1(a)) of the 88 CFGG layer. In the bilayer samples, a Pt layer was deposited on the CFGG layer after cooling 89 down to room temperature, which served as both a spin sink and a capping layer. For the bilayer 90 samples with an insertion layer, ~ 0.2 nm thick layer of Cu, Ni, Ru, Ta, or Cr was deposited on the 91 CFGG layer at room temperature before the deposition of the Pt layer. The detailed deposition 92 parameters and purity of the sputtering sources are given in Table 1. 93 Table 1: Deposition parameters and purity of sputtering sources. 94 Material Target purity and supplier Power source Power (W) Ar flow rate (sccm) Pressure (mTorr) Deposition rate (Å/s) CFGG 99.9% Toshima Manufacturing Co., Ltd. RF 90 50 1.5 0.20 Pt 99.99% Tanaka Precious Metal Technologies Co., Ltd. DC 50 100 4.2 0.54 Cu 99.99% Kojundo Chemical Laboratory Co.,Ltd. RF 20 10 8 0.18 Ni 99.99% Toshima Manufacturing Co., Ltd. DC 30 30 5.5 0.08 Ru 99.9% Toshima Manufacturing Co., Ltd. RF 30 15 4.5 0.08 Ta 99.99% Toshima Manufacturing Co., Ltd. RF 30 10 5 0.12 Cr 99.99% Toshima Manufacturing Co., Ltd. RF 50 50 7.5 0.04 The film thickness and interfacial roughness were determined by X-ray reflectivity measurements. 95 A model consisting of CFGG and Pt layers was used to fit the XRR data of CFGG/Pt and 96 5 CFGG/X/Pt (X = Cu, Ni, Ru, Ta, or Cr) samples. The evaluated interfacial roughness values are 97 listed in Table 2. 98 Table 2: Interfacial roughness at the CFGG and Pt interface in CFGG/Pt and CFGG/X/Pt (X = Cu, 99 Ni, Ru, Ta, or Cr) samples. 100 CFGG/Pt CFGG/Cu/Pt CFGG/Ni/Pt CFGG/Ru/Pt CFGG/Ta/Pt CFGG/Cr/Pt Interface roughness (nm) ~ 0.3 ~ 0.5 ~ 0.5 ~ 0.5 ~ 0.4 ~ 0.5 The structure of the samples was analyzed by XRD with Cu-Kα radiation (λ = 1.5406 Å) and cross-101 sectional (S)TEM analysis (FEI Titan G2 80-200). The specimens for the STEM analysis were 102 fabricated by a standard lift-out method using focused ion-beam/scanning electron microscope 103 (FIB/SEM), FEI Helios G4. 104 105 Fig. 1. Schematic representation of the thin film stack of the (a) CFGG (7.5, 8.8, 10.5, 12.5, 14, 106 16.3, and 18.5 nm) single-layer, (b) CFGG (7.5, 8.8, 10.5, 12.5, 14, 16.3, and 18.5 nm)/Pt (10 nm) 107 and CFGG (14 nm)/Pt (0.5, 1, 2, 3, 5, and 10 nm) bilayer, and (c) CFGG (7.5 nm)/X (~ 0.2 nm)/Pt 108 (10 nm) bilayer with insertion layer, deposited on the MgO (001) substrate. 109 To evaluate 𝑔eff↑↓ , α of each sample was determined using in-plane broadband FMR spectra 110 recorded in the frequency range of 16 GHz to 36 GHz using an FMR spectrometer (Phase FMR – 111 40) unit attached to the physical property measurement system (PPMS, Quantum Design, USA). 112 During the measurement, lock-in detection technique was employed to improve the signal-to-noise 113 ratio, which requires a reference a.c. signal to lock on. This signal was generated by a set of 114 Helmholtz coils powered by an alternating current source, producing a small (~ 1 Oe) modulation 115 field. The obtained FMR spectra were fitted with the equation [5], 116 6 𝑑𝐼𝑑𝐻= 𝐾1[(∆𝐻 2⁄ )2 − (𝐻 − 𝐻r)2][(∆𝐻 2⁄ )2 + (𝐻 − 𝐻r)2]2 + 𝐾2 2(𝐻 − 𝐻r)(∆𝐻 2⁄ )[(∆𝐻 2⁄ )2 + (𝐻 − 𝐻r)2]2. (1) Here, dI/dH is the derivative of the microwave absorption signal with respect to the applied 117 external magnetic field (H). K1 and K2 are coefficients that correspond to the symmetric and 118 antisymmetric components of the FMR spectrum, respectively. This fitting provides the resonance 119 field (μ0Hr) and the linewidth (μ0∆H) of the FMR spectrum, from which α is calculated. The total 120 α (αtotal) obtained from the experiments contains both intrinsic (αint) and extrinsic (αext) components, 121 and αext consists of contributions from spin pumping (αsp), two-magnon scattering, radiative 122 damping, and eddy current [27]. Among these contributions, the dominant one from the additional 123 Pt layer increases αsp by spin pumping. Therefore, the enhancement in αtotal of the bilayer samples 124 from the single-layer samples (∆α) can be expressed as follows [4]: 125 ∆𝛼 = 𝛼Bilayer − 𝛼Single−layer ≅ 𝛼sp. (2) Then, 𝑔eff↑↓ is evaluated using the relation [4], 126 ∆𝛼 = 𝑔𝜇B4π𝑀s 1𝑡CFGG 𝑔eff↑↓ , (3) where, 𝑔, μB, Ms, and tCFGG refer to Lande’s 𝑔 factor, Bohr magneton, saturation magnetization, 127 and thickness of the CFGG layer, respectively. Note that, 𝑔eff↑↓ varies with the Pt layer thickness 128 because of the contribution due to spin back flow from the Pt to the CFGG layer. 129 130 3. RESULTS AND DISCUSSION 131 3.1. Structural analysis 132 Figures 2(a) and 2(b) show the 2θ-ω XRD patterns of CFGG single-layer samples with a thickness 133 (tCFGG) of 7.5, 12.5, and 18.5 nm, along the out-of-plane direction for χ = 0° and the <111> 134 direction for χ = 54.7°, respectively. The corresponding results of CFGG (tCFGG)/Pt (10 nm) bilayer 135 samples are shown in Figs. 2(c) and 2(d), respectively. In all the samples, the observed peaks of 136 CFGG 004 and Pt 002 for χ = 0° confirm epitaxial growth along the [001] direction. Additionally, 137 the presence of CFGG 002 and 111 superlattice peaks confirm the L21-type ordered crystal 138 structure. Due to the similar atomic scattering factors of Co and Fe, standard XRD techniques are 139 limited in their ability to detect Co-Fe disorder. Nonetheless, such disorder is expected to be 140 suppressed as a result of the high annealing temperature of 600 ℃ used, which was confirmed 141 through the anomalous XRD technique using synchrotron-radiated X-ray in an earlier study [28]. 142 7 143 Fig. 2. XRD patterns of 2θ-ω scans for CFGG (7.5, 12.5, and 18.5 nm) single-layer samples for 144 (a) χ = 0° and (b) χ = 54.7°. (c, d) Corresponding results for CFGG (7.5, 12.5, and 18.5 nm)/Pt (10 145 nm) bilayers. 146 147 3.2. Evaluation of spin mixing conductance across the CFGG/Pt interface 148 To assess 𝑔eff↑↓ and examine its dependence on the CFGG/Pt interface, αtotal of all samples were 149 evaluated using FMR measurements. Figure 3(a) shows a typical FMR spectrum along with the 150 fitted curve based on Eq. (1), from which μ0Hr and μ0∆H were extracted. Figure 3(b) shows the 151 variation of the evaluated μ0Hr as a function of the resonance frequency (fr), which is fitted with 152 the Kittel’s equation [6] to evaluate the gyromagnetic ratio (γ). This γ value of 1.81×1011 rad/s/T 153 is then used to calculate αtotal by fitting μ0∆H versus fr shown in Fig. 3(c), using the equation [6], 154 μ0Δ𝐻 = 4π𝛼total𝛾𝑓r + μ0∆𝐻0. (4) 20 30 40 50 60 70 80¨Log intensity (arb. unit)2q (degree) 18.5 nm 12.5 nm 7.5 nmCFGG 004CFGG 002c = 0 °¨(a)20 30 40 50 60 70 80Intensity (arb. unit)2q (degree) 18.5 nm 12.5 nm 7.5 nm(b)¨ CFGG 111CFGG 222 ¨c = 54.7 °CFGGCFGGCFGG/PtCFGG/Pt20 30 40 50 60 70 80Pt 002Log intensity (arb. unit)2q (degree) 18.5 nm 12.5 nm 7.5 nm¨ CFGG 002 ¨ CFGG 004¨c = 0 °(c)20 30 40 50 60 70 80Intensity (arb. unit)2q (degree) 18.5 nm 12.5 nm 7.5 nm(d)¨ CFGG 111¨ CFGG 222c = 54.7 °8 Here, μ0∆H0 corresponds to the linewidth broadening due to inhomogeneities in the CFGG layer, 155 and αtotal is the Gilbert damping constant obtained as a fitting parameter. 156 157 Fig. 3. (a) FMR spectrum of a 14 nm thick CFGG single-layer sample measured at 24 GHz fitted 158 with Eq. (1), (b) μ0Hr versus fr data fitted to the Kittel’s equation, and (c) μ0∆H versus fr plot along 159 with fit to Eq. (4). 160 Figure 4(a) depicts the variation of αtotal with tCFGG for both single-layer and bilayer samples. 161 For the single-layer samples, αtotal remains nearly independent of tCFGG in the thickness range of 162 18.5–8.8 nm and slightly increases for tCFGG < 8.8 nm. In bilayer samples, αtotal is significantly 163 increased compared to single-layer samples due to the transfer of spin current from the CFGG 164 layer to the Pt layer. Figure 4(b) shows the tCFGG dependence of the increase in αtotal (∆α), which 165 shows a monotonic decrease with increasing tCFGG. 166 167 Fig. 4. Variations of (a) αtotal with a tCFGG in single and bilayer samples, and (b) ∆α with tCFGG. 168 By fitting the data in Fig. 4(b) with Eq. (3), 𝑔eff↑↓ was evaluated to be (2.78±0.09)×1019 m-2, which 169 is relatively high in comparison with the previously reported values for conventional FM/Pt bilayer 170 systems such as Co/Pt ((1.42±0.01)×1019 m-2), CoFe/Pt ((2.65±0.08)×1019 m-2), and Py/Pt 171 3800 4000 15 20 25 30 35 400.20.40.60.815 20 25 30 35 406.87.27.68.08.4dI/dHm0H (Oe) Experimental Fit to Eq. (1)m0Hrm0DH(a)m0Hr (T)fr (GHz)(b)m0DH (mT)fr (GHz) Experimental Fit to Eq. (4)(c) Experimental Kittel's Fit6 8 10 12 14 16 18 2002468106 8 10 12 14 16 18 200246810 αSingle-layer αBilayeratotal ´ 10-3tCFGG (nm)(a) Δα = αBilayer - αSingle-layer Fit to Eqn. (3)Da ´ 10-3tCFGG (nm)(b)9 ((2.53±0.02)×1019 m-2) [29]. In this analysis, the effect of spin back flow is neglected because the 172 Pt layer thickness is sufficiently larger than the reported value of the Pt spin diffusion length [30]. 173 The effect of the Pt layer thickness is discussed below. 174 Figure 5 represents the variation of ∆α with the thickness of the Pt layer (tPt) for the sample 175 consisting of CFGG (14 nm)/Pt (0.5, 1, 2, 3, 5, and 10 nm) layers. One can notice the decrease in 176 ∆α value in the lower tPt range (≤ 4 nm). This reduction is attributed to spin current reflection at 177 the Pt/air interface. When the Pt layer thickness is comparable to its spin diffusion length (λPt), the 178 reflected spin current flows back into the CFGG layer, thereby reducing ∆α. 179 180 Fig. 5. Variation of ∆α with Pt layer thickness for the bilayer films with a 14 nm CFGG layer. 181 As a result, Eq. (3) gets modified to [31], 182 ∆𝛼 =𝑔μB4π𝑀s 1𝑡CFGG 𝑔eff↑↓ (1 − 𝑒−2𝑡Pt𝜆Pt⁄). (5) The terms in the bracket on the right side represent spin current reflection at the Pt/air interface, 183 for the case of perfect reflection. Thus, the factor of 2 in the exponential term represents the total 184 distance travelled by the spin current as it returns to the CFGG layer. λPt is estimated to be 185 2.4±0.5 nm from a fit to the data shown in Fig. 5 to Eq. (5). The obtained λPt is consistent with the 186 previously reported value of 2.4±0.3 nm [30]. This validates the assumption made in the earlier 187 analysis of 𝑔eff↑↓ that λPt is much smaller than the tPt of 10 nm. Thus, the estimated 𝑔eff↑↓ is unaffected 188 by the spin current reflection. 189 190 3.3. Effect of various insertion layers on spin mixing conductance 191 To explore the impact of ultrathin insertion layers on 𝑔eff↑↓ , we introduced ~ 0.2 nm thick Cu, Ni, 192 Ru, Ta, or Cr between the CFGG and Pt layers. In these experiments, tCFGG and tPt were set to 7.5 193 nm and 10 nm, respectively, because the α of a thinner CFGG layer is more sensitive to changes 194 0 2 4 6 8 1001234 Da Fit to Eq. (5)Da ´ 10-3 tPt (nm)lPt = 2.4±0.5 nm10 in spin-pumping. First, to examine the possible modifications in the Pt structure due to different 195 insertion layers, XRD patterns were recorded. Figures 6(a) and 6(b) show the 2θ-ω scans of CFGG 196 (7.5 nm)/X (~ 0.2 nm)/Pt (10 nm) samples with various insertion layers (X = Cu, Ni, Ru, Ta, or 197 Cr) along the <001> (χ = 0°) and <110> (χ = 45°) directions, respectively. It is to be noted that, in 198 Fig. 6(a), the CFGG 002 peak is not visible. However, for the same 7.5 nm CFGG layer thickness, 199 the CFGG 002 peak is clearly visible in the Figs. 2(a) and 2(c) for the CFGG (7.5 nm) and CFGG 200 (7.5 nm)/Pt (10 nm) films, respectively, when measured using a 0D-mode XRD detector. The 201 reduced intensity of the CFGG 002 peak in Fig. 6(a) is due to the use of a 2D-mode XRD detector, 202 which is employed for rocking curve measurements, as discussed later. The XRD patterns for the 203 samples with an insertion layer are almost identical to that of the bilayer sample. This indicates 204 that the [001]-direction epitaxial growth of CFGG and Pt layers was maintained irrespective of the 205 type of insertion layer, with negligible change in the crystal structure of the CFGG and Pt layers. 206 207 Fig. 6. XRD patterns of CFGG (7.5 nm)/X (~ 0.2 nm)/Pt (10 nm) samples recorded at (a) χ = 0° 208 and (b) χ = 45° for X = Cu, Ni, Ru, Ta, or Cr. (c) The corresponding rocking curves recorded about 209 the Pt 220 peak. 210 To study the structure of the Pt layer in more detail, rocking curve measurement was carried out 211 for the Pt 220 peak, as shown in Fig. 6(c). Pt 220 peak was chosen because of its larger separation 212 from the MgO peak than that in the <001> direction scan. The Pt 220 rocking curves show identical 213 shape and intensity. The full-width at half-maximum (FWHM) values of these peaks for all the 214 20 30 40 50 60 70 80¨Log intensity (arb. unit)2q (degree) CFGG/Cr/Pt CFGG/Ta/Pt CFGG/Ru/Pt CFGG/Ni/Pt CFGG/Cu/Pt CFGG/Pt¨ Pt 002CFGG 004c = 0°(a)20 30 40 50 60 70 80Log intensity (arb. unit)2q (degree) CFGG/Cr/Pt CFGG/Ta/Pt CFGG/Ru/Pt CFGG/Ni/Pt CFGG/Cu/Pt CFGG/Pt¨Pt 220c = 45°(b)175 180 1853.23°Intensity (arb. unit)b (degree) CFGG/Cr/Pt CFGG/Ta/Pt CFGG/Ru/Pt CFGG/Ni/Pt CFGG/Cu/Pt CFGG/Pt(c)3.28°3.30°3.40°3.28°3.34°FWHM =11 samples with different insertion layers are mentioned in Fig. 6(c). The nearly identical FWHM 215 values confirm that the Pt epitaxial layer is not largely influenced by the insertion layers. 216 Cross-sectional TEM analysis was performed on the CFGG (7.5 nm)/Cu (~ 0.2 nm)/Pt (10 nm) 217 structure. This sample was chosen as a representative sample to confirm the presence of the 218 ultra-thin insertion layer, given that the Cu insertion most significantly modifies 𝑔eff↑↓ , as discussed 219 later. Figures 7(a) and 7(b) show the cross-sectional high-angle annular dark-field (HAADF)-220 STEM images, energy dispersive spectroscopy (EDS) elemental maps, and line compositional 221 profiles of Mg, O, Co, Fe, Ga, Ge, Cu, and Pt at low magnification for the CFGG (7.5 nm)/Pt (10 222 nm) and CFGG (7.5 nm)/Cu (~ 0.2 nm)/Pt (10 nm) samples, respectively. 223 224 Fig. 7. Cross-sectional HAADF-STEM image and EDS elemental maps of bilayers (a) without Cu 225 insertion layer, and (b) with ~ 0.2 nm Cu insertion layer. (a) and (b) also includes the line 226 compositional profile across the interfaces. 227 The HAADF-STEM image and EDS elemental map confirm the presence of the Cu insertion 228 layer (indicated by an arrow) at the CFGG/Pt interface, further supported by the line composition 229 profile as depicted in Fig. 7(b). The line compositional profiles in Figs. 7(a) and 7(b) show the 230 oxygen distribution (~ 10-15%) throughout CFGG and Pt layers. The oxygen in these regions is 231 20 nm20 nm MgOCFGGPtMgOCFGGPt20 nm20 nm MgOCFGGPtCuMgOCFGGPtCu(a)(b)01020300.60.40.20.0Atomic fractionDistance (nm)MgOCoFeGeCuGaPtAu0102030Distance (nm)GeGa Fe CoPtAuMgO12 an artifact, resulting from oxygen adhering to the sample surface and contact with the sample being 232 analyzed. To analyze the Cu distribution in detail, a separate EDS elemental map of Cu, and the 233 corresponding multiple EDS line profiles from 5-nm-wide regions, denoted as (i)–(x), are shown 234 in Figs. 8(a) and 8(b), respectively. These results suggest the non-uniformity in the Cu insertion 235 layer. An averaged EDS line profile from the entire region indicated by the red dashed box in Fig. 236 8(a), is shown in red in Fig. 8(b). A weak Cu signal is additionally detected in the Pt region, which 237 can be attributed to the spectral overlap effects, as the Cu-K and Pt-Lλ peaks are close in energy, 238 at approximately 8.0 keV and 8.26 keV, respectively. This overlap-induced artifact is supported 239 by the fact that the Cu signal inside the Pt layer does not decay with the distance from the interface. 240 241 Fig. 8. (a) Cross-sectional EDS elemental map of the Cu insertion layer in 242 CFGG (7.5 nm)/Cu (~ 0.2 nm)/Pt (10 nm) sample, and (b) the corresponding compositional line 243 profiles across the interface at different positions denoted as (i)–(x) in (a). 244 The high-magnification HAADF-STEM images are shown in Figs. 9(a) and 9(b). The Pt layer 245 grows epitaxially on the CFGG layer regardless of the Cu layer insertion. The orientation 246 relationship between CFGG and Pt layers is described as (001)CFGG/(001)Pt, [100]CFGG/[110]Pt. 247 Given that the ultrathin insertion layer is not continuous, some portion of CFGG and Pt remain in 248 01020300102030Atomic fraction (arb. unit)Distance (nm)(i) (ii) (iii)AverageDistance (nm)(a)(b)Distance (nm)(iv) (v) (vi) (vii) (viii) (ix) (x)Cu(Average)5 nm(ii) (iii) (iv) (v) (vi) (vii) (viii) (ix) (x)(i)13 direct contact, which can partly explain why the Pt structure is largely unaffected by the deposition 249 of insertion layers. 250 251 Fig. 9. High-magnification HAADF-STEM images obtained from the samples (a) without Cu 252 insertion layer, and (b) with ~ 0.2 nm Cu insertion layer. The arrow in the figure indicates the 253 position of Cu insertion. 254 Figures 10(a) and 10(b) show the comparison of αtotal and 𝑔eff↑↓ for the samples without and with 255 insertion layer. The corresponding ratio of change in αtotal and 𝑔eff↑↓ due to different insertion layers 256 are shown in Figs. 10(c) and 10(d), respectively. The change in 𝑔eff↑↓ cannot be explained solely by 257 the changes in interface roughness observed for different insertion layers as listed in Table 2. This 258 indicates that intrinsic physical properties of the insertion layers play a dominant role in changes 259 observed in 𝑔eff↑↓ . The sample without insertion layer exhibits the highest 𝑔eff↑↓ because the direct 260 contact of CFGG with Pt facilitates efficient spin current transfer across the CFGG/Pt interface. 261 The most pronounced decrease in αtotal of ~ 22% (refer to Fig. 10(c)) is noted in the case of the Cu 262 insertion layer. Such behavior of Cu insertion has been reported in the previous studies [5], and 263 has been ascribed to a reduction of proximity-induced magnetization in the Pt layer, resulting from 264 the nonmagnetic property of the Cu insertion layer. This leads to the reduction in 𝑔eff↑↓ by ~ 32%. 265 On the other hand, no noticeable change in αtotal, and consequently in 𝑔eff↑↓ , is observed in the sample 266 with Ni insertion layer, presumably owing to its similarity in magnetic property to that of the 267 CFGG layer. A direct contact between FM and the Pt layer enables the formation of proximity-268 induced magnetization in the Pt layer [11,32,33]. In the case of nonmagnetic Ru and Ta insertion 269 layers, a relatively smaller reduction in 𝑔eff↑↓ (~ 22%) is observed in comparison to the Cu insertion 270 layer. This difference in 𝑔eff↑↓ might originate from the interfacial enhancement of spin-orbit 271 coupling resulting from direct contact of CFGG layer with a material possessing relatively high 272 spin-orbit coupling, such as Ru and Ta [7,8]. This explain the large θH in the films with Ru and 273 2 nm2 nm(a) (b)CuCFGG CFGGPt Pt14 Ta insertion as compared to the one with Cu insertion layer [34]. In the case of the Cr insertion 274 layer, 𝑔eff↑↓ decreased only slightly (~ 12%) despite its low atomic number. Previous studies on 275 Y3Fe5O12 (YIG)/NM bilayers by Du et al. [26] and Wang et al. [34] reported that Cr (~ 0.051) has 276 a smaller θH than Ta (~ 0.071), although Cr shows a relatively high θH for its small atomic number 277 due to its d-electron configuration. Therefore, 𝑔eff↑↓ is expected to be lower for the Cr insertion layer 278 compared to Ta when only θH of these materials is considered, which is contrary to the obtained 279 result shown in Fig. 10(b). This discrepancy suggests that the antiferromagnetic property of the Cr 280 insertion layer could increase the spin current transfer from CFGG to the Pt layer [35]. Based on 281 the above discussion, it is apparent that the interfacial modification by ultrathin (~ 0.2 nm) 282 insertion layers leads to a noticeable change in 𝑔eff↑↓ while not affecting the structure of the Pt layer, 283 highlighting the role of material characteristics at the interface on 𝑔eff↑↓ . 284 285 Fig. 10. (a) αtotal and (b) 𝑔eff↑↓ of bilayers without and with Cu, Ni, Ru, Ta, or Cr insertion layers. 286 The percentage variation of the (c) αtotal and (d) 𝑔eff↑↓ for different insertion layers. 287 288 4. CONCLUSIONS 289 A high 𝑔eff↑↓ at the FM/HM interface, along with low α in the FM, is essential for enabling faster 290 and more energy-efficient spintronic devices. In this study, we evaluated 𝑔eff↑↓ in CFGG/Pt bilayer 291 thin films and the effects of different ultrathin (~ 0.2 nm) insertion layers using FMR 292 measurements. The single-layer CFGG sample showed a low Gilbert damping of ~ 1.07×10-3, 293 while a notable increase, observed in CFGG/Pt bilayers, indicates efficient spin pumping across 294 0246810atotal ´ 10-3CFGG/PtCFGG/Cr/PtCFGG/Cu/PtCFGG/Ni/PtCFGG/Ru/PtCFGG/Ta/Pt(a)0.01.02.03.0g­¯eff (´1019) m-2 CFGG/PtCFGG/Cr/PtCFGG/Cu/PtCFGG/Ni/PtCFGG/Ru/PtCFGG/Ta/Pt(b)-25-20-15-10-50Change in atotal (%)(c)-40-30-20-100Change in g­¯eff (%)(d)15 the interface, from which a high 𝑔eff↑↓ of (2.78±0.09)×1019 m-2 was estimated. While investigating 295 the effect of an insertion layer on 𝑔eff↑↓ , XRD and STEM measurements confirmed that the Pt 296 structure remained largely unaffected by the insertion layers. This confirmation eliminates the 297 possibility of the insertion-layer-induced changes in the Pt layer. The insertion of the NM Cu layer 298 significantly reduces 𝑔eff↑↓ , while the FM Ni layer retains almost same 𝑔eff↑↓ . In the case of Ru and 299 Ta insertion layers, despite being nonmagnetic, a slight reduction in 𝑔eff↑↓ is observed, while the 300 antiferromagnetic Cr insertion layer also causes a small decrease. The ultra-low Gilbert damping 301 of CFGG, combined with high 𝑔eff↑↓ in CFGG/Pt bilayers, highlights their potential for energy-302 efficient SOT induced magnetization switching in the active FM layer of magnetic tunnel junctions. 303 Moreover, the effect of insertion layers provides insights into interface engineering for tuning 𝑔eff↑↓ . 304 Future scope of the study includes elucidation of the microscopic mechanism underlying the 305 influence of the insertion layer at the interface by employing theoretical analysis on αtotal and 𝑔eff↑↓ 306 to obtain a deeper insight of the underlying physics. 307 308 Acknowledgement 309 This work is supported by the Advanced Storage Research Consortium (ASRC), JST CREST 310 (Grant No. JPMJCR21O1), the MEXT Initiative to Establish Next-generation Novel Integrated 311 Circuits Centers (X-NICS) (Grant No. JPJ011438), JST ERATO “Magnetic Thermal Management 312 Materials Project” (Grant No. JPMJER2201). The authors thank Y. Miura and I. Kurniawan at 313 NIMS for fruitful discussions, and S. Kuramochi and N. Kojima at NIMS for the preparation and 314 characterization of thin films. 315 316 Data availability statement 317 The data that support the findings of this study are openly available in Materials Data Repository 318 (MDR) operated by National Institute for Materials Science at https://doi.org/10.48505/nims.6269. 319 320 References 321 [1] S. Ikegawa, F. B. Mancoff, J. Janesky, and S. Aggarwal, Magnetoresistive Random Access 322 Memory: Present and Future, IEEE Trans. Electron Devices 67, 1407 (2020). 323 16 [2] K. Dolui, U. Bajpai, and B. K. Nikolić, Effective spin-mixing conductance of topological-324 insulator/ferromagnet and heavy-metal/ferromagnet spin-orbit-coupled interfaces: A first-325 principles Floquet-nonequilibrium Green function approach, Phys. Rev. Mater. 4, 326 121201(R) (2020). 327 [3] Q. Lu, Y. Li, B. Peng, H. Tang, Y. Zhang, Z. He, L. Wang, C. Li, W. Su, Q. Yang, Z. Zhou, 328 and M. Liu, Enhancement of the spin-mixing conductance in Co-Fe-B/W bilayers by 329 interface engineering, Phys. Rev. Appl. 12, 064035 (2019). 330 [4] S. Akansel, A. Kumar, N. Behera, S. Husain, R. Brucas, S. Chaudhary, and P. Svedlindh, 331 Thickness-dependent enhancement of damping in Co2FeAl/β-Ta thin films, Phys. Rev. B. 332 97, 134421(2018). 333 [5] A. Kumar, R. Gupta, S. Husain, N. Behera, S. Hait, S. Chaudhary, R. Brucas, and P. 334 Svedlindh, Spin pumping and spin torques in interfacially tailored Co2FeAl/β-Ta layers, 335 Phys. Rev. B. 100, 214433 (2019). 336 [6] B. B. Singh, K. Roy, P. Gupta, T. Seki, K. Takanashi, and S. Bedanta, High spin mixing 337 conductance and spin interface transparency at the interface of a Co2Fe0.4Mn0.6Si Heusler 338 alloy and Pt, NPG Asia Mater. 13(9), (2021). 339 [7] E. Barati, M. Cinal, D. M. Edwards, and A. Umerski, Gilbert damping in magnetic layered 340 systems, Phys Rev B 90, 014420 (2014). 341 [8] S. Azzawi, A. Ganguly, M. Tokaç, R. M. Rowan-Robinson, J. Sinha, A. T. Hindmarch, A. 342 Barman, and D. Atkinson, Evolution of damping in ferromagnetic/nonmagnetic thin film 343 bilayers as a function of nonmagnetic layer thickness, Phys Rev B 93, 054402 (2016). 344 [9] Q. Zhang, S. I. Hikino, and S. Yunoki, First-principles study of the spin-mixing conductance 345 in Pt/Ni81Fe19 junctions, Appl. Phys. Lett. 99, 172105 (2011). 346 [10] M. Tokaç, S. A. Bunyaev, G. N. Kakazei, D. S. Schmool, D. Atkinson, and A. T. Hindmarch, 347 Interfacial structure dependent spin mixing conductance in cobalt thin films, Phys. Rev. Lett. 348 115, 056601 (2015). 349 17 [11] C. Swindells, H. Głowiński, Y. Choi, D. Haskel, P. P. Michałowski, T. Hase, P. Kuświk, and 350 D. Atkinson, Proximity-induced magnetism and the enhancement of damping in 351 ferromagnetic/heavy metal systems, Appl. Phys. Lett. 119, 152401 (2021). 352 [12] C. Swindells, H. Glowinski, Y. Choi, D. Haskel, P. P. Michalowski, T. Hase, F. Stobiecki, 353 P. Kuswik, and D. Atkinson, Magnetic damping in ferromagnetic/heavy-metal systems: The 354 role of interfaces and the relation to proximity-induced magnetism, Phys. Rev. B 105, 355 094433 (2022). 356 [13] Y. Fan, J. Wang, A. Chen, K. Yu, M. Zhu, Y. Han, S. Zhang, X. Lin, H. Zhou, X. Zhang and 357 Q. Lin, Thickness-dependent Gilbert damping and soft magnetism in metal/Co-Fe-B/Metal 358 Sandwich Structure, Nanomaterials 14, 596 (2024). 359 [14] S. Mizukami, D. Watanabe, M. Oogane, Y. Ando, Y. Miura, M. Shirai, and T. Miyazaki, 360 Low damping constant for Co2FeAl Heusler alloy films and its correlation with density of 361 states, J. Appl. Phys. 105, 07D306 (2009). 362 [15] T. Kubota, S. Tsunegi, M. Oogane, S. Mizukami, T. Miyazaki, H. Naganuma, and Y. Ando, 363 Half-metallicity and Gilbert damping constant in Co2FexMn1-xSi Heusler alloys depending 364 on the film composition, Appl. Phys. Lett. 94, 122504 (2009). 365 [16] M. M. Bhat, K. Simalaotao, H. Suto, A. Perumal, A. Srinivasan, and Y. Sakuraba, Atomic 366 order induced reduction of Gilbert damping constant and enhancement of half-metallicity in 367 off-stoichiometric Co2FeGa0.5Ge0.5 Heusler alloy thin films, J. Appl. Phys. 138, 043904 368 (2025). 369 [17] J. M. Shaw, E. K. Delczeg-Czirjak, E. R. J. Edwards, Y. Kvashnin, D. Thonig, Martin A. W. 370 Schoen, M. Pufall, M. L. Schneider, T. J. Silva, O. Karis, K. P. Rice, O. Eriksson, and H. T. 371 Nembach, Magnetic damping in sputter-deposited Co2MnGe Heusler compounds with A2, 372 B2, and L21 orders: Experiment and theory, Phys. Rev. B 97, 094420 (2018). 373 [18] Y. K. Takahashi, A. Srinivasan, B. Varaprasad, A. Rajanikanth, N. Hase, T. M. Nakatani, S. 374 Kasai, T. Furubayashi, and K. Hono, Large magnetoresistance in current-perpendicular-to-375 plane pseudospin valve using a Co2Fe(Ge0.5Ga0.5) Heusler alloy, Appl. Phys. Lett. 98, 376 152501 (2011). 377 18 [19] Y. Sakuraba, M. Ueda, Y. Miura, K. Sato, S. Bosu, K. Saito, M. Shirai, T. J. Konno, and K. 378 Takanashi, Extensive study of giant magnetoresistance properties in half-metallic 379 Co2(Fe,Mn)Si-based devices, Appl. Phys. Lett. 101, 252408 (2012). 380 [20] D. Taparia, T. Sasaki, T. Nakatani, H. Suto, Y. Miura, Z. Li, V. K. Kushwaha, K. Inubushi, 381 S. Ichikawa, K. Nakada, T. Sasaki, S. Mitani and Y. Sakuraba, Improvement in CPP-GMR 382 read head sensor performance using [001]-oriented polycrystalline half-metallic Heusler 383 alloy Co2FeGa0.5Ge0.5 and CoFe bilayer electrode, Sci. Technol. Adv. Mater. 25(1), 2388503 384 (2024). 385 [21] V. Barwal, H. Suto, R. Toyama, K. Simalaotao, T. Sasaki, Y. Miura, and Y. Sakuraba, Large 386 magnetoresistance and high spin-transfer torque efficiency of Co2MnxFe1−xGe (0 ≤ x ≤ 1) 387 Heusler alloy thin films obtained by high-throughput compositional optimization using 388 combinatorially sputtered composition-gradient film, APL Mater. 12, 111114 (2024). 389 [22] S. Li, Y. K. Takahashi, T. Furubayashi, and K. Hono, Enhancement of giant 390 magnetoresistance by L21 ordering in Co2Fe(Ge0.5Ga0.5) Heusler alloy current-391 perpendicular-to-plane pseudo spin valves, Appl. Phys. Lett. 103, 042405 (2013). 392 [23] J. W. Jung, Y. Sakuraba, T. T. Sasaki, Y. Miura, and K. Hono, Enhancement of 393 magnetoresistance by inserting thin NiAl layers at the interfaces in 394 Co2FeGa0.5Ge0.5/Ag/Co2FeGa0.5Ge0.5 current-perpendicular-to-plane pseudo spin valves, 395 Appl. Phys. Lett. 108, (2016). 396 [24] B. S. D. C. S. Varaprasad, A. Srinivasan, Y. K. Takahashi, M. Hayashi, A. Rajanikanth, and 397 K. Hono, Spin polarization and Gilbert damping of Co2Fe(GaxGe1-x) Heusler alloys, Acta 398 Mater. 60, 6257 (2012). 399 [25] Y. Tserkovnyak, A. Brataas, and G. E. W. Bauer, Enhanced Gilbert damping in thin 400 ferromagnetic films, Phys. Rev. Lett. 88, 4 117601 (2002). 401 [26] C. Du, H. Wang, F. Yang, and P. C. Hammel, Systematic variation of spin-orbit coupling 402 with d-orbital filling: Large inverse spin Hall effect in 3d transition metals, Phys. Rev. B 90, 403 140407(R) (2014). 404 19 [27] S. Hait, S. Husain, V. Barwal, N. K. Gupta, L. Pandey, P. Svedlindh, and S. Chaudhary, 405 Comparison of high temperature growth versus post-deposition in situ annealing in attaining 406 very low Gilbert damping in sputtered Co2FeAl Heusler alloy films, J. Magn. Magn. Mater. 407 519, 167509 (2021). 408 [28] K. Goto, L. S. R. Kumara, Y. Sakuraba, Y. Miura, I. Kurniawan, A. Yasui, H. Tajiri, Y. 409 Fujita, Z. Chen, and K. Hono, Effects of the atomic order on the half-metallic electronic 410 structure in the Co2Fe(Ga0.5Ge0.5) Heusler alloy thin film, Phys. Rev. Mater. 4, 114406 411 (2020). 412 [29] X. Tao, Q. Liu, B. Miao, R. Yu, Z. Feng, L. Sun, B. You, J. Du, K. Chen, S. Zhang, L. Zhang, 413 Z. Yuan, D. Wu, H. Ding, Self-Consistent determination of spin Hall angle and spin diffusion 414 length in Pt and Pd: The Role of the Interface Spin Loss, Sci. Adv. 4, 1670 (2018). 415 [30] T. Fache, J. C. Rojas-Sanchez, L. Badie, S. Mangin, and S. Petit-Watelot, Determination of 416 spin Hall angle, spin mixing conductance, and spin diffusion length in CoFeB/Ir for spin-417 orbitronic devices, Phys. Rev. B. 102, 064425 (2020). 418 [31] J. M. Shaw, H. T. Nembach, and T. J. Silva, Determination of spin pumping as a source of 419 linewidth in sputtered Co90Fe10/Pd multilayers by use of broadband ferromagnetic resonance 420 spectroscopy, Phys. Rev. B 85, 054412 (2012). 421 [32] F. Wilhelm, P. Poulopoulos, G. Ceballos, H. Wende, K. Baberschke, P. Srivastava, D. Benea, 422 H. Ebert, M. Angelakeris, N. K. Flevaris, D. Niarchos, A. Rogalev, and N. B. Brookes, 423 Layer-resolved magnetic moments in Ni Pt multilayers, Phys. Rev. Lett., 85, 413 (2000). 424 [33] M. Caminale, A. Ghosh, S. Auffret, U. Ebels, K. Ollefs, F. Wilhelm, A. Rogalev, and W. E. 425 Bailey, Spin pumping damping and magnetic proximity effect in Pd and Pt spin-sink layers, 426 Phys Rev B 94, 014414 (2016). 427 [34] H. L. Wang, C. H. Du, Y. Pu, R. Adur, P. C. Hammel, and F. Y. Yang, Scaling of spin Hall 428 angle in 3d, 4d, and 5d metals from Y3Fe5O12/metal spin pumping, Phys. Rev. Lett. 112, 429 197201 (2014). 430 [35] H. Wang, C. Du, P. C. Hammel, and F. Yang, Antiferromagnonic spin transport from 431 Y3Fe5O12 into NiO, Phys. Rev. Lett. 113, 097202 (2014). 432 20 433