Spin pumping and inverse spin Hall effect in CoFeB/IrMn heterostructures
Koustuv Roy, Abhisek Mishra, Pushpendra Gupta, Shaktiranjan Mohanty, Braj Bhusan Singh, Subhankar Bedanta
SSpin pumping and inverse spin Hall effect in CoFeB/IrMn heterostructures
Koustuv Roy, Abhisek Mishra, Pushpendra Gupta, ShaktiranjanMohanty, Braj Bhusan Singh, and Subhankar Bedanta
1, 2, ∗ Laboratory for Nanomagnetism and Magnetic Materials (LNMM), School of Physical Sciences,National Institute of Science Education and Research (NISER), HBNI, Jatni-752050, India Center for Interdisciplinary Sciences (CIS), National Institute ofScience Education and Research (NISER), HBNI, Jatni, 752050 India
The spin current based electronics, known as spintronics, is the promising technology to replacethe charge current based technology. Spintronics based devices allow miniaturization of the deviceand faster response. The study of spin current efficiency in different materials is one of the emergentresearch topics in modern days. Spin pumping and inverse spin Hall effect (ISHE) are popular toolsto study the spin to charge current conversion efficiency in the materials. In last one decade, ISHEand spin pumping are heavily investigated in ferromagnet (FM)/ heavy metal (HM) heterostructures.Recently the antiferromagnetic (AFM) materials are found to be a good replacement of heavy metals(HM) for these kind of studies. Faster dynamics, exhibiting very high spin-orbit coupling, absenceof stray field, efficient propagation of spin current make AFM to be a good replacement of HMs.Understanding the role of different FM layers in a FM/AFM heterostructure for ISHE or spinpumping study is very important for spin to charge conversion physics. In this context we haveperformed the ISHE in CoFeB/ IrMn heterostructures. Spin pumping study is carried out for Co F e B (12 nm ) /Cu (3 nm ) /Ir Mn ( tnm ) /AlO x (3 nm ) samples where t value varies from 0to 10 nm. The Cu(3 nm) buffer layer is used which is less than the spin diffusion length( ∼ g ↑↓ r ) =0.704 ± × m − is found to be maximum in IrMn(2 nm) sample from the ISHE analysis.The large Spin Hall angle was also found to be ∼ I. INTRODUCTION
Spin transport phenomena is one of the timely top-ics to develop the future spintronic devices[1, 2]. Thegeneration and manipulation of the pure spin currentis important for future technological devices and funda-mental research as well. Production of pure spin currentvia spin pumping [3–7] is an efficient method which doesnot require nanofabrication in ferromagnetic (FM)/heavymetal (HM) systems. In this process the magnetizationof FM is excited by a rf magnetic field through ferromag-netic resonance (FMR) and it generates the spin current( J s ) at the FM/HM interface, which is given by [2]: (cid:126)J s = (cid:126) π g ↑↓ r ˆ m × d ˆ mdt (1)where g ↑↓ r is the real part of spin mixing conductanceand ˆ m is the unit vector of magnetization. The proces-sion under time varying field can generate resultant dccurrent which diffuses towards FM/HM interfaces. It canbe converted into transverse charge current and hencetransverse voltage can be measured by inverse spin Halleffect (ISHE). ISHE voltage is given by [8]: V ISHE ∝ θ SH | (cid:126)J s × (cid:126)σ | (2) ∗ [email protected] where θ SH is the spin Hall angle which defines spincurrent to charge current conversion efficiency, and (cid:126)σ isthe spin matrix governed by spin polarization direction.In order to get ISHE signal, θ SH needs to be large inthe heterostructure, which depends on the spin orbit in-teraction (SOI) of the HM. The spin to charge currentconversion efficiency mostly depends on SOI of HM andits conductivity [2, 9].Efficient and high spin current generation at low powerwith cost-effective materials at room temperature is stilla challenge. Therefore, low damping FM materials (e.g.Py, CoFeB etc.) are very important to generate largespin pumping signal (ISHE voltage) by relatively sup-pressing the other spurious effects. Interface effect, im-purity, magnetic proximity effects (MPE) etc., are thespurious effects which also enhance the value of dampingconstant. Further the selection of HM is very impor-tant for obtaining high ISHE. However there are only afew HMs in the periodic table and usually they are quiteexpensive. On the other hand recent works show anti-ferromagnets(AFM) exhibit high spin-orbit coupling andefficient transfer of spin angular momentum via antifer-omagnetic spin waves [10–12]. Thus AFMs offer a goodreplacement for HMs as they also exhibit high SOI andthere are many AFMs yet to be explored. ImportantlyFM/AFM bilayers have been heavily used in spintronicsdevice application due to the exchange bias property [13].In addition to that the ISHE study on this FM/AFM bi-layers can give rise to new scope in spintronics for futuredevice applications. a r X i v : . [ c ond - m a t . m e s - h a ll ] F e b The collinear AFM (i.e.
M n Au , IrMn etc.) is re-portedly found to be exhibiting high θ SH [10, 12, 14].The collinear AFMs of CuAu-I type (e.g., IrMn, PtMn,FeMn, PdMn etc.) took great interest in this field fortheir simple crystal structure. Further they are easy togrow in the crystalline phase which is very much essentialin spintronic device application purpose. IrMn exhibitshigh θ SH within the CuAu-I AFMs. Most of the ISHEworks on IrMn have been performed with crystalline FMmaterials like Py [15–17]. There are very few reports ofspin pumping with CoFeB/IrMn bilayers [18–20]. Thepreivous works lag systematic ISHE analysis. In thiswork, we have systematically studied the ISHE on theCoFeB/Cu/IrMn heterostructures. The crystallinity ofthe FM also plays a role in spin transport phenomenaas it directly affects the interface of the FM/AFM. Wehave introduced a spacer Cu(3 nm) layer to improve thegrowth of IrMn further. The lower thickness of Cu layerin comparison to its spin diffusion length ( ∼ II. EXPERIMENTAL DETAILS
Table I shows the sample structure of our study. Allthe samples were prepared on Si(100) substrates usingmagnetron sputtering in a vacuum system with base pres-sure ∼ × − mbar. R1 and R2 are called the ref-erence samples. The thickness of the films were evalu-ated using x-ray reflectivity (XRR) (data not shown) andtransmission electron microscope(TEM) imaging. Crys-talline quality of the thin films is also characterized by X-ray diffraction (XRD) and TEM imaging. The magneticmeasurement was performed by superconducting quan-tum interference device (SQUID) based magnetometer.Ferromagnetic resonance (FMR) measurements havebeen performed in the frequency range 5-16 GHz on acoplanar wave guide in the flip-chip manner [22, 23].ISHE measurements have been performed by connect-ing a nanovoltmeter over two ends of the sample (samplesize: 3 mm × III. RESULTS AND DISCUSSIONCrystalline quality
Grazing incidence X-ray diffraction(GIXRD) was per-formed on the S5 sample with X-ray of wave-length 0.15nm. Fig. 1(a) shows the GIXRD pattern of the sampleS6. We could clearly see the XRD peaks at 43.1 ◦ and74 ◦ which correspond to IrMn(111) and Cu(220), respec-tively. The CoFeB has been grown in the amorphousform, therefore it does not show any XRD peaks.The fig.1(b) shows the TEM image of S6 sample. Itclearly shows the amorphous nature of the CoFeB and TABLE I. Samples studied for the present work (The numbersin the bracket are in the units of nm).R1 CoFeB(12)R2 CoFeB(12)/Cu(3)S1 CoFeB(12)/Cu(3)/
AlO x (3)S2 CoFeB(12)/Cu(3)/IrMn(0.5)/ AlO x (3)S3 CoFeB(12)/Cu(3)/IrMn(0.75)/ AlO x (3)S4 CoFeB(12)/Cu(3)/IrMn(2)/ AlO x (3)S5 CoFeB(12)/Cu(3)/IrMn(5)/ AlO x (3)S6 CoFeB(12)/Cu(3)/IrMn(10)/ AlO x (3) existence of crystallinity in IrMn. The Cu layer is notclearly visible in the image due to its low thickness valuein our samples. The image also indicates the interdiffu-sion of Cu layer with its neighbouring layers. Further, wehave performed line scan energy dispersive X-Ray anal-ysis (EDX) of the sample as shown in fig.1 (c,d). TheEDX result confirms the existence of all the elements inthe sample structure in table I. Magnetic Damping
In order to study the damping properties of the sam-ples, the FMR measurement was carried out in the 5-16GHz rf frequency range. Fig. 2(a) and (b) show theplots of resonance frequency ( f ) versus H r and ∆ H ver-sus f , respectively. H r and ∆ H values were evaluatedusing FMR spectra. In order to evaluate the effectivedemagnetization (4 πM eff ) and gyromagnetic ratio ( γ ),2(a) was fitted to Kittel’s equation [29] which is given as: f = γ π (cid:113) ( H K + H r )( H K + H r + π M eff ) (3)where 4 πM eff = 4 πM s + 2 K S M s t F M (4)and K s , H K , M s t F M , are perpendicular surface mag-netic anisotropy constant, anisotropy field, saturationmagnetization and thickness of FM layer, respectively.The Gilbert damping( α ) was evaluated by fitting data ofFig. 2(b) using the following expression [30]:∆ H = ∆ H + 4 πα f γ (5)where, ∆ H is the inhomogeneous broadening oflinewidth. α of a FM material depends upon the crystalstructure as well as its band structure. In presence ofany NM layer with HM, the interface plays a big role tomodify the α . In such cases, the total α of the systemcan be written as: FIG. 1. (a) GIXRD pattern, (b) high resolution cross-sectional TEM image, (c) line scan EDX (inset image shows the selectedarea),(d) corresponding EDX scan profile of sample S6. (e) Elemental mapping using EFTEM for S6 at different energy edges. α = α int + α impurity + α MP E + α sp (6)where α int is the intrinsic damping, and α impurity , α MP E , and α sp are the contribution from impurity, mag-netic proximity effect (MPE) and spin pumping to the α ,respectively [31]. FIG. 2. (a) f vs H r and (b) ∆ H vs f plots for the samplesS1 and S6. The solid lines are the best fits according to theequations 3 and 5. The ∆ H value is larger for S6(15 Oe) in comparisonto S1 (1 Oe) which indicates a good homogeneity in thesample without the IrMn layer (S1). The almost linearbehaviour of ∆ H vs f rules out possibility of two magnonmechanism in our sample. The values of α for the S1-S6 are 0.0085, 0.0102, 0.0101, 0.0100, 0.0103, 0.0099 andR1-R2 are 0.0087, 0.0081 with the error bar ± α values are almost similar in com-parison to the sample S1. Thus, we may conclude thatthe damping enhancement in our samples are actuallycoming because of IrMn layer.From fig 4, it is observed that the value of α for thesample S2 (IrMn=0.5 nm) is maximum. Beyond thisIrMn thickness limit the α value is almost saturating.This behaviour can be explained with the spin diffusionlength of IrMn ( ∼ FIG. 3. α variation as a function of IrMn thickness (S1-S6). spin pumping behaviour over the other interface effects,the α value saturates at the NM thickness equals to it’sspin diffusion length. However, we can not rule out thepossibility of other interfacial effects like MPE on the to-tal α value. In order to investigate the magnetic deadlayer formation or MPE at the interface, we performedmagnetization measurement on all the samples. The sat-uration magnetization ( M s ) of all the samples are foundto be 840.13 emu/cc (S1), 827.03 emu/cc (S2), 884.28emu/cc (S3), 882.36 emu/cc (S4), 840.11 emu/cc (S5)and 849.97 emu/cc (S6). No significant change in the M s value was observed. Thus we conclude that the dom-inant spin pumping behaviour in our samples. Inverse spin Hall effect measurement
Damping analysis indicates the presence of spin pump-ing mechanism in the samples. In order to confirm thespin pumping and qualitative analysis of the spin currentgeneration in our system, we performed ISHE measure-ments on all the samples. The measurements are carriedout at 7 GHz frequency and +15 dBm rf magnetic fieldpower.The measured electrical voltage in ISHE measurementhas different contributions other than the spin pumping, TABLE II. Fitted parameters from φ dependent voltage measurements for the samples with different IrMn thickness(S2-S6)Sample V sp (V) × − V AHE (V) × − V ⊥ AMR (V) × − V || AMR (V) × − S2 1.14 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± V meas ) measured across the sample withapplied magnetic field along with FMR signal for sample S6at the φ values of (a) 0 ◦ , (b) 180 ◦ . Experimental data isrepresented as open symbol. Solid lines are the fit to theexperimental data using Eq. (7). Short dotted and dashlines are the symmetric ( V sym ) and anti-symmetric ( V asym )components of the voltage. e.g. anisotropic magnetoresistance (AMR) and anoma-lous Hall effect (AHE). Angle dependent voltage mea-surement have been performed to rectify these spuriouseffects. Fig. 4 shows the measured voltage ( V meas ) (openblue symbol) versus H along with FMR signal (closeblack symbol) for sample S6 at the angles φ = 0 ◦ (a) and180 ◦ (b). The φ denotes the angle between the perpen-dicular direction of applied DC magnetic field ( H ) andmeasured voltage direction. It should be noted that thepumped spin current is invariant upon the sample rota-tion in the measurement. Upon the change in polarityof the voltmeter electrodes, the measured voltage shouldchange sign in case of spin pumping in the samples. Wefound that the measured voltage ( V meas ) is changing signwith the 180 ◦ rotation of the sample which is a clear in-dication of the presence of spin pumping in the sample.If the sign does not reverse with the 180 ◦ roation of thesample, the measured voltage solely comes from the spu-rious effects other than the spin pumping. It should alsobe noted that the V meas goes to zero value when the φ ∼ ◦ . According to the basic mechanism of electron spinscattering in ISHE, it happens because the spin accumu-lation is almost zero when the measured voltage directionis parallel to applied magnetic field. The maximum spinaccumulation happens in the perpendicular direction tothe applied magnetic field. In order to, quantify spinpumping contribution from the V meas , the V meas ver-sus H plots for the samples S6 (Fig. 4) was fitted withLorentzian equation[32], which is given by: FIG. 5. V sym and V asym contribution with the variation of rf magnetic field power for the sample S6. V meas = V sym (∆ H ) ( H − H r ) + (∆ H ) + V asym H ( H − H r )( H − H r ) + (∆ H ) (7)where V sym and V asym are the symmetric and anti-symmetric Lorenzian components of the measured signal.Solid red lines are fits to the experimental data. The V sym consists of major contribution from spin pumping,while minor contributions are from AMR, and AHE ef-fects. Whereas the AMR and AHE are the major contri-butions in the V asym component. Fig. 4 also representsthe plot of V sym (dotted line) and V asym (dashed line)separately for the samples S6. FIG. 6. Angle dependent (a)( φ ) V sym and (b) V asym mea-surements for samples S6. In order to quantify the spin pumping and other rec-tification contributions, In-plane angle dependent ISHEmeasurements of V meas were performed at the interval of5 ◦ for all the samples (fig.6 shows for S6). This methodis an established one to decouple the individual compo-nents from the measured voltage [31, 33, 34]. Harder et.al. [35] has considered the rectification effects i.e., theAHE contribution due to the FM layer, perpendicularAMR ( V AMR ⊥ asym/sym ) and parallel AMR ( V AMR || asym/sym ) to theapplied rf field. The relation between these rectificationeffects and V meas are as follows [33]: V asym = V AHE sin ( Φ ) cos ( φ + φ )+ V AMR ⊥ asym cos φ + φ ) cos ( φ + φ )+ V AMR || asym sin φ + φ ) cos ( φ + φ ) (8) V sym = V sp cos ( φ + φ ) + V AHE cos ( Φ ) cos ( φ + φ )+ V AMR ⊥ sym cos φ + φ ) cos ( φ + φ )+ V AMR || sym sin φ + φ ) cos ( φ + φ )(9) V sp and V AHE correspond to the spin pumping andAHE contributions, respectively. Φ is the angle between rf field and applied DC magnetic field which is 90 ◦ forour set up. Thus, the AHE does not contribute to the V sym term. The extra factor φ takes care of the mis-alignment of sample positioning in defining the φ valueduring the measurement. Thus the equation 8 and 9 canbe simplified as follows: V asym = V AHE cos ( φ + φ )+ V AMR ⊥ asym cos φ + φ ) cos ( φ + φ )+ V AMR || asym sin φ + φ ) cos ( φ + φ ) (10) V sym = V sp cos ( φ + φ )++ V AMR ⊥ sym cos φ + φ ) cos ( φ + φ )+ V AMR || sym sin φ + φ ) cos ( φ + φ ) (11)Further the total AMR contribution also can be quan-tified by the following formula [33] : V AMR = (cid:113) ( V AMR ⊥ , || asym ) + ( V AMR ⊥ , || sym ) (12)The V AMR ⊥ , || asym and V AMR ⊥ , || sym are evaluated from thein-plane angle dependent ISHE measurements by fitting V meas values by equations 10 and 11, respectively. Simi-lar angle dependent ISHE measurement was carried outfor all the samples. The extracted various rectificationcomponents for all the samples are listed in the Table II.It should be noted that the samples without IrMn layer,S1,R1 and R2 do not show any ISHE signal (data notshown). It is expected as Cu does not have any high SOC. This fact confirms that the spin pumping mecha-nism is occurring in the samples S2-S6 solely because ofIrMn layer.Table II clearly shows that the V sp is dominating overother unwanted rectification effects in all the samples.However, the AHE contribution is not negligible in thesamples. The AHE phenomena is controlled by the berrycurvature[36] of the material. It is an intrinsic propertyof the FM layer. The Co based FM materials are alwaysa potential candidate for the AHE contribution [37, 38].The anisotropic nature of the samples can be understoodfrom the finite contribution of AMR values. The posi-tive values of V sp indicates the positive spin Hall anglein IrMn, which is consistent with literature[10]. Fig. 5shows the rf magnetic field power dependent V sym and V asym contribution for the sample S6. With the increasein power, the spin angular momentum transfer from FMto NM increases linearly. The linear behaviour in Fig. 5is another signature of the spin pumping mechanism insample S6.The V sp is found to be maximum for S3 which indicatesthe smooth interface between FM and NM. The spinpumping happens more at the thickness of NM whichis equal to its spin diffusion length ( λ ). The λ IrMn ispreviously reported to be 0.7 nm[10]. The high V sp forS3 and S4 is consistent with this argument. The Eq. 17shows that the the V sp depends upon the conductivity ofthe IrMn. At higher thickness of IrMn, the conductivityof IrMn increases. This leads to decrease in V sp value forS5 and S6. In addition to this, the back spin pumpingalso dominates beyond the spin diffusion length of HMdue to the spin flip mechanism at this thickness limit.Thus the decrease in V sp value for S5 and S6 is possiblya cumulative contribution of conductivity as well as thespin back flow.Effective spin mixing conductance ( g ↑↓ eff ) is another pa-rameter which defines the efficiency of spin current prop-agation from FM to NM. Figure 7 shows the graph be-tween g ↑↓ eff and IrMn thickness. g ↑↓ eff was calculated bythe following expression using damping constant[2]: g ↑↓ eff = ∆ α πM s t CoF eB gµ B (13)where ∆ α , t CoF eB , µ B , g are the change in the α due tospin pumping, the thickness of CoFeB layer, Bohr mag-neton, Lande g- factor, respectively. The real part of g ↑↓ eff mainly contributes to the spin transport mechanism.Thus, it is very crucial to quantify the real part of spinmixing conductance g ↑↓ r . The g ↑↓ r value can be evaluatedby the following expression[39, 40]: g ↑↓ r = g ↑↓ eff [1 + g ↑↓ eff ρ IrMn λ IrMn e π (cid:126) tanh[ t IrMn λ IrMn ] ] − (14)Where ρ IrMn , λ IrMn are the resistivity and spin dif-fusion length of IrMn respectively. In order to calculate g ↑↓ r , the λ IrMn is taken to be 0.7 nm [10]. The g ↑↓ r valuesfor S2-S6 found to be 0.606, 0.690, 0.704, 0.696, 0.687 nm − . It is also to be noted that our g ↑↓ r values are oneorder less than the previously reported values with IrMn[10, 17, 20, 41]. The relatively lower value of g ↑↓ r confirmsthat the spin current flow efficiency is not so good in ourheterostructures. It is probably because Cu is not a goodspacer layer for such heterostructures. This result sup-ports our TEM image where Cu layer has been seen to beinterdiffused with the neighbouring layers. Spin interfacetransparency ( T ) is the parameter which defines the spincurrent propagation efficiency at the FM/NM interface. T value is calculated from the following expressions[42]: T = g ↑↓ r tanh ( t IrMn λ IrMn ) g ↑↓ r coth ( t IrMn λ IrMn ) + hσ IrMn e λ IrMn (15)where σ IrMn is the conductivity of IrMn layer. Themaximum g ↑↓ r is found to be in sample S4. For S4( t IrMn = 2 nm), T is evaluated to be 0.14 ± θ SH is also an-other parameter which is very much important in spin-tronic device applications. We evaluated the θ SH for theIrMn using the following expression [39, 40, 43]: | (cid:126)J s | ≈ ( g ↑↓ r (cid:126) π )( µ h rf γα ) × [ µ M s γ + (cid:112) ( µ M s γ ) + 16( πf ) ( µ M s γ ) + 16( πf ) ]( 2 e (cid:126) ) (16) V sp = ( w y σ F M t F M + σ NM t NM ) × θ SH λ NM tanh ( t NM λ NM ) | (cid:126)J s | (17)The four probe technique was used to measure theresistivity ( ρ ) of the samples. The ρ IrMn and ρ CoF eB are found to be 5.5 × − Ω.m and 8.5 × − Ω.m respec-tively. σ corresponds to the conductivity of the individuallayers. FIG. 7. (a) g ↑↓ eff , g ↑↓ r and (b) θ SH with the variation of IrMnthickness. Short dotted line in (a) is the guide to the eye. The rf field ( µ h rf ) and CPW transmission line width( w y ) value for our set up are 0.05 mT (at + 15 dBm rfpower) and 200 µ m, respectively. Fig 7 shows the varia-tion of θ SH for the sample S2-S6. The highest θ SH =0.30value is obtained for S3. This is higher than any otherreported values for IrMn and also comparable with thePt[10, 17]. IV. CONCLUSION
We presented the spin pumping and inverse spin Halleffect study in a series of CoFeB/ IrMn heterostructures.The IrMn has been grown in AFM phase. Angle de-pendent study was performed to disentangle all the spinrectification effects. Strong spin pumping contribution isfound in the samples. The spin pumping voltage satu-rates at the spin diffusion length of the IrMn thickness. V sp value decreases for higher value of IrMn thickness dueto increase in conductivity of IrMn . The maximum g ↑↓ r is evaluated to be 0.704 × m − for S4. Spin trans-parency also found to be 0.14 which indicates that Cu isnot a good spacer layer for IrMn and CoFeB. But, thehigh spin Hall angle of 0.30 in S3 predicts IrMn as a goodreplacement of HM in spintronic device applications. ACKNOWLEDGEMENT
The authors acknowledge DAE and DST, Govt. ofIndia, for the financial support for the experimental fa-cilities. KR and PG thank CSIR and UGC for their JRFfellowships, respectively. BBS acknowledges DST for IN-SPIRE faculty fellowship. The authors thank Dr. T.Ghosh for help in TEM measurement.
REFERENCES [1] S. Bader and S. Parkin, Annu. Rev. Condens. MatterPhys. , 71 (2010). [2] Y. Tserkovnyak, A. Brataas, G. E. Bauer, and B. I.Halperin, Reviews of Modern Physics , 1375 (2005). [3] J. R. S´anchez, L. Vila, G. Desfonds, S. Gambarelli, J. At-tan´e, J. De Teresa, C. Mag´en, and A. Fert, Nature com-munications , 1 (2013).[4] X. Tao, Q. Liu, B. Miao, R. Yu, Z. Feng, L. Sun, B. You,J. Du, K. Chen, S. Zhang, et al., Science advances ,eaat1670 (2018).[5] B. Heinrich, C. Burrowes, E. Montoya, B. Kardasz,E. Girt, Y.-Y. Song, Y. Sun, and M. Wu, Physical reviewletters , 066604 (2011).[6] M. Obstbaum, M. Decker, A. Greitner, M. Haertinger,T. Meier, M. Kronseder, K. Chadova, S. Wimmer,D. K¨odderitzsch, H. Ebert, et al., Physical review letters , 167204 (2016).[7] K. Chen and S. Zhang, Physical review letters ,126602 (2015).[8] E. Saitoh, M. Ueda, H. Miyajima, and G. Tatara, Ap-plied physics letters , 182509 (2006).[9] Y. Tserkovnyak, A. Brataas, and G. E. Bauer, Physicalreview letters , 117601 (2002).[10] W. Zhang, M. B. Jungfleisch, W. Jiang, J. E. Pearson,A. Hoffmann, F. Freimuth, and Y. Mokrousov, Physicalreview letters , 196602 (2014).[11] B. B. Singh, K. Roy, J. A. Chelvane, and S. Bedanta,Physical Review B , 174444 (2020).[12] B. B. Singh and S. Bedanta, Physical Review Applied , 044020 (2020).[13] J. Nogu´es and I. K. Schuller, Journal of Magnetism andMagnetic Materials , 203 (1999).[14] J. Zhou, X. Wang, Y. Liu, J. Yu, H. Fu, L. Liu, S. Chen,J. Deng, W. Lin, X. Shu, et al., Science advances ,eaau6696 (2019).[15] R. Hao, R. Zang, T. Zhou, S. Kang, S. Yan, G. Liu,G. Han, S. Yu, and L. Mei, Chinese Physics B , 037202(2019).[16] H. Wu, C. Wan, Z. Yuan, X. Zhang, J. Jiang, Q. Zhang,Z. Wen, and X. Han, Physical Review B , 054404(2015).[17] L. Frangou, S. Oyarzun, S. Auffret, L. Vila, S. Gam-barelli, and V. Baltz, Physical review letters , 077203(2016).[18] D. Jhajhria, D. K. Pandya, and S. Chaudhary, BeilsteinJournal of Nanotechnology , 2198 (2018).[19] S. Tu, J. Hu, G. Yu, H. Yu, C. Liu, F. Heimbach,X. Wang, J. Zhang, Y. Zhang, A. Hamzi´c, et al., Ap-plied Physics Letters , 222401 (2017).[20] X. Ma, G. Yu, S. A. Razavi, L. Chang, L. Deng, Z. Chu,C. He, K. L. Wang, and X. Li, Physical Review B ,104428 (2018).[21] F. Jedema, M. Nijboer, A. Filip, and B. Van Wees, Phys-ical Review B , 085319 (2003).[22] “Nanosc ab,” . [23] B. B. Singh, S. K. Jena, and S. Bedanta, Journal ofPhysics D: Applied Physics , 345001 (2017).[24] B. B. Singh, S. K. Jena, M. Samanta, K. Biswas, andS. Bedanta, ACS Applied Materials & Interfaces ,53409 (2020).[25] B. B. Singh, K. Roy, P. gupta, Pushpendra, T. Seki,K. Takanashi, and B. Bedanta, NPG Asia Mater ,11 (2021).[26] P. Gupta, B. Singh, K. Roy, A. Sarkar, M. Waschk,T. Brueckel, and S. Bedanta, Nanoscale (2021).[27] B. Sahoo, K. Roy, P. Gupta, B. Satpati, B. B. Singh,and S. Bedanta, arXiv e-prints , arXiv (2020).[28] B. B. Singh, S. K. Jena, M. Samanta, K. Biswas, B. Sat-pati, and S. Bedanta, physica status solidi (RRL)–RapidResearch Letters , 1800492 (2019).[29] C. Kittel, Physical review , 155 (1948).[30] B. Heinrich, J. Cochran, and R. Hasegawa, Journal ofApplied Physics , 3690 (1985).[31] A. Conca, S. Keller, L. Mihalceanu, T. Kehagias, G. Dim-itrakopulos, B. Hillebrands, and E. T. Papaioannou,Physical Review B , 134405 (2016).[32] R. Iguchi and E. Saitoh, Journal of the Physical Societyof Japan , 011003 (2017).[33] A. Conca, B. Heinz, M. Schweizer, S. Keller, E. T. Pa-paioannou, and B. Hillebrands, Physical Review B ,174426 (2017).[34] M. Harder, Y. Gui, and C.-M. Hu, Physics Reports ,1 (2016).[35] M. Harder, Z. Cao, Y. Gui, X. Fan, and C.-M. Hu,Physical Review B , 054423 (2011).[36] N. Nagaosa, J. Sinova, S. Onoda, A. H. MacDonald, andN. P. Ong, Reviews of modern physics , 1539 (2010).[37] W. Zhang, V. Vlaminck, J. E. Pearson, R. Divan, S. D.Bader, and A. Hoffmann, Applied physics letters ,242414 (2013).[38] G. Zahnd, L. Vila, V. Pham, M. Cosset-Cheneau,W. Lim, A. Brenac, P. Laczkowski, A. Marty, and J. At-tan´e, Physical Review B , 174414 (2018).[39] K.-R. Jeon, C. Ciccarelli, H. Kurebayashi, J. Wunderlich,L. F. Cohen, S. Komori, J. W. Robinson, and M. G.Blamire, Physical Review Applied , 014029 (2018).[40] K. Rogdakis, A. Sud, M. Amado, C. Lee, L. McKenzie-Sell, K. Jeon, M. Cubukcu, M. Blamire, J. Robinson,L. Cohen, et al., Physical Review Materials , 014406(2019).[41] J. Holanda, H. Saglam, V. Karakas, Z. Zang, Y. Li, R. Di-van, Y. Liu, O. Ozatay, V. Novosad, J. E. Pearson, et al.,Physical review letters , 087204 (2020).[42] W. Zhang, W. Han, X. Jiang, S.-H. Yang, and S. S.Parkin, Nature Physics , 496 (2015).[43] K. Ando, S. Takahashi, J. Ieda, Y. Kajiwara,H. Nakayama, T. Yoshino, K. Harii, Y. Fujikawa,M. Matsuo, S. Maekawa, et al., Journal of applied physics109