A novel method to evaluate spin diffusion length of Pt Y. Q. Zhang,1 N. Y. Sun,1 W. R. Che,1 R. Shan,1,a) and Z. G. Zhu2,3,b) 1)Shanghai Key Laboratory of Special Artificial Microstructure and Pohl Institute of Solid State Physics and School of Physics Science and Engineering, Tongji University, Shanghai 200092, China 2)School of Electronic, Electrical and Communication Engineering, University of Chinese Academy of Sciences, Beijing 100049, China. 3)Theoretical Condensed Matter Physics and Computational Materials Physics Laboratory, College of Physical Sciences, University of Chinese Academy of Sciences, Beijing 100049, China. 6 1 (Dated: 8 January 2016) 0 Spin diffusion length ofPt is evaluated via proximityeffect of spin orbitcoupling (SOC) andanomalousHall 2 effect (AHE) in Pt/Co FeAl bilayers. By varying the thicknesses of Pt and Co FeAl layer, the thickness 2 2 n dependences ofAHE parameterscanbe obtained, whichare theoreticallypredictedto be proportionalto the a square of the SOC strength. According to the physical image of the SOC proximity effect, the spin diffusion J length of Pt can easily be identified from these thickness dependences. This work provides a novel method 7 to evaluate spin diffusion length in a material with a small value. ] i PACS numbers: 71.70.Ej;73.50.Jt; 75.47.Np; 75.50.Bb c s - l r As a promptly growingresearcharea,spintronicsaims mt at using and manipulating not only the charge, but also thespininanelectronicdevice1,2. Spindependenttrans- . t portpropertyisthekeyfortheapplicationofspintronics a device,andthusithasattractedgreatattentionoverthe m past few decades3–7. Herein, spin diffusion length (λ) - d is a fundamental parameter in the study of spin depen- n dent transport, which has an inverse relationship with o the intensity of spin dependent scattering. There are c many methods such as lateral spin valve, spin pumping, [ spin-torque ferromagneticresonance,Hall crossandspin 1 absorption to gain the spin diffusion length in a normal v metal8–15. Different from above methods, a novel ap- FIG. 1. (Color online) Physical image of AHE in FM/NM 1 proach to evaluate spin diffusion length is introduced in bilayer. Gray arrow (M) indicates the magnetic moment di- 7 rection of FM layer. Yellow line indicates possible motion thiswork,throughproximityeffectofspinorbitcoupling 3 (SOC) and anomalous Hall effect (AHE)16–18. It is well pathoftheconductionelectrons. tNM andλisthicknessand 1 spin diffusion length of NM layer, respectively. known that spin orbit coupling plays a fundamental role 0 . inspindependenttransportproperties,includinganoma- 1 lous Halleffect, spinHalleffect, spintransfertorque and 0 not contribute to the SOC proximity effect but continue Dzyaloshinskii-Moriya interaction etc.17,19–22. SOC can 6 shuntingthe measurementcurrent,whichleadstoasud- 1 stronglyaffecttheintensityofspindependentscattering, denly weaken performance of spin dependent scattering. : and hence has a direct relationship with spin diffusion v Catchingthe turning point, the spindiffusion lengthcan length. If a low dimensionallayeredstructure consistsof i consequently be evaluated. X a nonmagnetic (NM) layer with strong SOC and a fer- In orderto analyze spin dependent scattering, anoma- r romagnetic (FM) layer, the conduction electrons will be a repeatedly scattered in z direction by the interfaces of lous Hall effect is employed in this work. It is now firmly established that there are two categories com- heterolayers,substrates and vacuum when the current is peting mechanisms contributing to the AHE: intrinsic flowing, as shown in Fig. 1. The conduction electrons mechanism, which originates from the anomalous veloc- will be polarized by the magnetic layer and affected by ity of the Bloch electrons induced by the SOC23–25; ex- the strong SOC layer meantime, causing enhanced spin ternal mechanism, which includes skew scattering and dependent scattering. This phenomenon is called SOC sidejump26,27. Bothintrinsicandsidejumpcontribution proximity effect. When the thickness of the NM layer is over its spin diffusion length, the added NM layer will obey the square relationshipρAH ∝ρ2xx, where ρAH and ρxx correspondtotheanomalousHallresistivityandlon- gitudinalresistivity,respectively23–26. Differently,dueto asymmetric scattering from impurities caused by SOC, a)Electronicmail: [email protected] theskewscatteringcontributionshowslineardependence b)Electronicmail: [email protected] onρxx27. Therefore,themeasuredρAH usuallyobeysthe 2 empirical Eq. (1): ρAH =aρxx+bρ2xx, (1) (a) 3 Pt (2.5 nm)/CFA (tCFA nm) whereaandbareparameters. Thisrelationshipisknown 0.6 nm as the traditional scaling. In 2001, Cr´epieux and Bruno 2 0.73 nm presentedatheoryoftheAHE,wheretheydeemthatno 0.9 nm matter skew scattering or side jump contributions, AHE 1.1 nm parameters (a and b) always obey square relationships 2.2 nm 1 withstrengthofSOC28. Afterward,Tianet al. proposed m 5 nm an expanded scaling29, where the impurity and phonon c 10 nm are assumed to have different contributions to the skew 60 70 80 90 scattering. The scaling can be rewritten as Eq. (2): H A ρAH =a′ρxx0+a′′ρxxT +bρ2xx (2) 3 Pt (tPt nm)/CFA (0.9 nm) Here, ρxx0 is the residual resistivity, ρxxT comes from ′ ′′ the scattering of excited phonons, a and a are due to 2 theskewscattering,andbisdominatedbythe sidejump 1.5 nm andintrinsiccontributions. Inthis equation,the phonon 2.5 nm 1 contribution is introduced by considering the inelastic 5 nm scattering at finite temperatures, which was formulated 10 nm (b) in terms of a multiband tight-binding model by Shitade 0 20 nm and Nagaosa30. 20 40 60 80 TwoseriesofPt(2.5nm)/Co2FeAl(tCFA nm) andPt xx cm o(tnPtthnem1)×/C1o2cFme2Aplo(l0is.9hendmM)gbOila(y1e0r0)filsmubsswtrearteesdebpyomsitaegd- FIG. 2. (Color online) ρAH/ρxx versus ρxx curves for (a) Pt (2.5 nm)/Co2FeAl (tCFA nm) and (b) Pt (tPt nm)/Co2FeAl netron sputtering. All those samples were prepared un- (0.9 nm) bilayers. All scattered symbols represent experi- ◦ der Hall bar mask and annealed at 320 C in situ. The mental data. All red lines were fitted results by ρAH = basepressureofsputteringchamberisbelow3×10−6Pa. a′ρxx0+a′′ρxxT +bρ2xx. The sputtering Ar gas with 99.999% purity was intro- ducedwithaconstantpressureof0.3Pa. Filmthickness was measured by X-ray reflectivity (XRR) using a D8 If t < λ (λ is the spin diffusion length of Pt), Discover X-ray diffractometer. The transport property Pt Pt Pt increasing the thickness of Pt layer will enhance the ef- wasmeasuredby physicalpropertymeasurementsystem fectiveSOC(whichhasthe meaningofsqureofζ here- (PPMS) from 20 to 300 K. eff after)monotonicallyobeyingEq. (3)showninFig.3(b). Onthe contrary,increasingthe thickness ofthe Co FeAl Figure 2 shows the measured results of the AHE. 2 layer will not affect the spin diffusion in the Pt layer ρAH/ρxx versus ρxx curves of the samples are given in but reduce the effective SOC since the atomic SOC in Fig. 2(a) and (b). The red lines are fitting curves. Ob- the Co FeAl layer is much weaker than that in Pt layer, viously, ρAH/ρxx versus ρxx is not linear relationship 2 which is shown in Fig. 3(a). for almost all samples. Just when Co FeAl (CFA) layer 2 is thicker than 5 nm, the relationship looks like linear. If tPt > λPt, the spin states will be lost in partial Hence,thetraditionalscalingdoesnotworkwellforvery of Pt layer where the distance to the interface of the thin bilayers. On the contrary, Eq. (2) can fit the data FM and NM layer is over the λPt already. The reason perfectly. These results are consistent with our reported is that the injected spin (along the M direction) from data31,32. ItrevealsthatthedependenceofρAH/ρxx ver- the FM layer into Pt layer will be relaxed due to the susρxx showsastraighterlineatspecialratiosamonga′, SOC of Pt. Therefore, this part of Pt only plays a role a′′ and bρxx0 only, such as the cases in a bulk material of conducting a leakage current and a leakage current and a thick film. factor has to be introduced, which is simply proposedas According to the CB theory and the physical image L(tPt) = (λPt+tCFA)/(tPt+tCFA) here. Only those Pt of the SOC proximity effect, the effective SOC strength atomic layers in the scope of λPt can make contribution of Pt/Co FeAl can be treated using tight-binding sense, to the measuredsignalof spin dependent scattering, e.g. 2 and the expression can be read as: the AHE. The generalbehaviorof factorL(tPt) is shown in Fig. 3(c). Taking into account the L(t ) factor, the ζ t +ζ t Pt ζ = CFA CFA Pt Pt, (3) effective SOC of the bilayersystem shouldtake behavior eff t +t CFA Pt shown in Fig. 3(d), in which the turning point occurs whereζ (=53.8meV)representsthe SOCstrengthof exactly at t = λ . Since the fitted parameters of the CFA Pt Pt Co FeAl,ζ (=554meV)istheSOCstrengthofPt,and AHE are theoretically proportional to the effective SOC 2 Pt t is the thickness of Co FeAl(Pt) layer33. of the bilayer system, identifying the turning points in CFA(Pt) 2 3 80 2 (/)effCFA 40 Pt (2.5 nm)/CFA (tCFA nm) Without fPatc t(otPrt nLm)/CFA (0.9 nm) 80 2 (/)effCFA -2 a’ (10) 1200 Pt (2.5 nm)/CFA (tCFA( nam)) Pt (tPt nm)/CFA (0.9( dnm)) 10 (a) (b) 40 0 0 0 5 10 0 5 10 15 20 ) 30 tCFA (nm) tPt (nm) -2 0 40 L(t) Pt01..50 (c) PPPPtttt====5812 With factor L PPPPtttt====1285 4800 2 (/)L(t)effCFAPt -1a" (1m)2000 (b) (e) 012000 (d) 0.0 0 c-1 0 5 10 15 20 0 5 10 15 20 -2 tPt (nm) tPt (nm) m-2 (c) (f) ( FIG. 3. (Color online) Thickness dependence of (ζeff/ζCFA)2 b -3 -4 given by Eq. (3) for (a) Pt (2.5 nm)/Co2FeAl (tCFA nm) 0 5 10 0 5 10 15 20 and (b) Pt (tPt nm)/Co2FeAl (0.9 nm) bilayers. (c) Thick- nessdependenceofL(tPt)forPt(tPt nm)/Co2FeAl(0.9nm) tCFA (nm) tPt (nm) bilayer. L(tPt) is a factor resulting from the leakage cur- FIG. 4. (Color online) Thickness dependences of a′, a′′ and rent for a thicker Pt layer. L(t ) = 1 when t ≤ λ ; Pt Pt Pt b for Pt (2.5 nm)/Co FeAl (t nm) in (a)-(c) and Pt (t simply L(tPt) = (λPt+tCFA)/(tPt +tCFA) when tPt > λPt. nm)/Co FeAl (0.9 nm2) bilayeCrFsAin (d)-(f). Here, a′ and aP′t′ (d) Thickness dependence of (ζ /ζ )2L(t ) for Pt (t 2 eff CFA Pt Pt indicate the contributions of skew scattering from impurities nm)/Co FeAl (0.9 nm) bilayer, corresponding to the values 2 and phonons;b represents theintrinsic andside jump contri- with the leakage current factor revision. λ is Pt diffusion Pt butions. length, which is arbitrarily taken as 1, 2, 5, 8 nm in (c) and (d). Grant Nos. 51331004, 11374228 and 11205235, the Na- tionalBasicResearchProgramofChinaunderGrantNo. the fitted parameters of the AHE provides a powerful 2015CB921501,andtheInnovationProgramofShanghai tool to fix the spin diffusion length in the Pt layer. Municipal Education Commission No. 14ZZ038. Z. G. Fig. 4 shows the thickness dependences of fitted pa- Zhu is supported by Hundred Talents Program of The rameters from Fig. 2(a) and (b). For the samples with Chinese Academy of Sciences. t = 2.5 nm, the t dependences are consistent with Pt CFA those in Fig. 3(a), indicating t <λ . For the samples 1S.A.Wolf,D.D.Awschalom,R.A.Buhrman,J.M.Daughton, Pt Pt with varied thicknesses of Pt layers but fixed thickness S. von Molnr, M. L. Roukes, A. Y. Chtchelkanova, and D. M. ′ ′′ Treger,Science 294,1488(2001). of Co2FeAl layers, the overall behavior of a, a and b 2I. Zˇuti´c, J. Fabian, and S. D. Sarma, Rev. Mod. Phys. 76, 323 are found to be similar to those of theoretical prediction (2004). shown in Fig. 3(d) including the leakage current factor. 3G.A.Prinz,Science 282,1660(1998). Thusλ isevaluatedaround5nmwhichisapproximate 4H.Nakayama,M.Althammer,Y.-T.Chen,K.Uchida,Y.Kaji- Pt wara,D.Kikuchi,T.Ohtani,S.Gepr¨ags,M.Opel,S.Takahashi, to that in Ref. [34]. R.Gross,G.E.W.Bauer,S.T.B.Goennenwein,andE.Saitoh, Phys. Rev. Lett.110,206601(2013). In conclusion, this work demonstrates a method for 5T.Lin,C.Tang,H.M.Alyahayaei, andJ.Shi,Phys. Rev. Lett. determiningthe spindiffusionlengthviaanomalousHall 113,037203(2014). effectandproximityeffectofspinorbitcoupling. Weob- 6M. Althammer, S. Meyer, H. Nakayama, M. Schreier, S. Alt- mannshofer,M.Weiler,H.Huebl,S.Gepr¨ags,M.Opel,R.Gross, tain the spin diffusion length of Pt is around5 nm. This D.Meier,C.Klewe,T.Kuschel,J.-M.Schmalhorst,G.Reiss,L. method can be used to evaluate nonmagnetic materials Shen,A.Gupta,Y.-T.Chen,G.E.W.Bauer,E.Saitoh,andS. with short spin diffusion length. However, due to the T.B.Goennenwein, Phys. Rev. B 87,224401 (2013). limit of proximity effect, it does not work for thick films 7Y.-T. Chen, S. Takahashi, H. Nakayama, M. Althammer, S. T. B. Goennenwein, E. Saitoh, and G. E. W. Bauer, Phys. Rev. B probably. 87,144411 (2013). 8P.Laczkowski,L.Vila,V.-D.Nguyen,A.Marty,J.-P.Attan´e,H. Jaffr`es,J.-M.George,andA.Fert,Phys. Rev. B 85,220404(R) (2012). ACKNOWLEDGMENTS 9V.Vlaminck,J.E.Pearson,S.D.Bader,andA.Hoffmann,Phys. Rev. B 88,064414(2013). We thank the beamline 08U1 at the Shanghai Syn- 10W. Zhang, V. Vlaminck, J. E. Pearson, R. Divan, S. D. Bader, andA.Hoffmann,Appl. Phys. Lett.103,242414(2013). chrotron Radiation Facilities (SSRF) for the sample 11A.Azevedo,L.H.Vilela-Le˜ao,R.L.Rodr´ıguez-Su´arez,A.F.L. preparation and measurement. This work was sup- Santos,andS.M.Rezende,Phys. Rev. B 83,144402 (2011). ported by the National Science Foundation of China 12J.-C.Rojas-S´anchez,N.Reyren,P.Laczkowski,W.Savero,J.-P. 4 Attan´e, C. Deranlot, M. Jamet, J. -M. George, L. Vila, and H. 24T. Jungwirth, Q. Niu, and A. H. MacDonald, Phys. Rev. Lett. Jaffr`es,Phys. Rev. Lett.112,106602 (2014). 88,207208 (2002). 13L.Liu,R.A.Buhrman,andD.C.Ralph,arXiv: 1111.3702. 25D. Xiao, M. C. Chang, and Q. Niu, Rev. Mod. Phys. 82, 1959 14G.Mihajlovi´c,J.E.Pearson,M.A.Garcia,S.D.Bader,andA. (2010). Hoffmann,Phys. Rev. Lett.103,166601(2009). 26J.Smit,Physica (Amsterdam) 24,39(1958). 15M. Isasa, E. Villamor, L. E. Hueso, M. Gradhand, and F. 27L.Berger,Phys. Rev. B 2,4559(1970). Casanova, Phys. Rev. B 91,024402(2015). 28A.Cr´epieux,andP.Bruno,Phys. Rev. B 64,014416(2001). 16Y. Q. Zhang, N. Y. Sun, W. R. Che, X. L. Li, J. W. Zhang, 29Y. Tian, L. Ye, and X. F. Jin, Phys. Rev. Lett. 103, 087206 R. Shan, Z. G. Zhu, and G. Su, Appl. Phys. Lett. 107, 082404 (2009). (2015). 30A.ShitadeandN.Nagaosa,J.Phys.Soc.Jpn.81,083704(2012). 17E.H.Hall,Philos. Mag.12,157(1881). 31W. R. Che, X. F. Xiao, N. Y. Sun, Y. Q. Zhang, R. Shan, and 18N. Nagaosa, J. Sinova, S. Onoda, A. H. MacDonald, and N. P. Z.G.Zhu,Appl. Phys. Lett.104,262404(2014). Ong,Rev. Mod. Phys.82,1539(2010). 32Y. Q. Zhang, N. Y. Sun, R. Shan, J. W. Zhang, Z. Shi, S. M. 19J.Sinova,Phys. Rev. Lett.92,126603(2004). Zhou,andG.Y.Guo,J. Appl. Phys.114,163714(2013). 20J.C.Slonczewski, J. Magn. Magn. Mater.159,L1(1996). 33Y.KotaandA.Sakuma, J. Phys. Soc. Jpn.83,034715(2014). 21I.Dzyaloshinsky,J. Phys. Chem. Solids 4,241(1958). 34S.R.Marmion,M.Ali,M.McLaren, D.A. Williams,andB.J. 22T.Moriya,Phys. Rev.120,91(1960). Hickey,Phys. Rev. B 89,220404(R) (2014). 23R.KarplusandJ.M.Luttinger, Phys. Rev.95,1154(1954).