Planar Hall driven torque in a FM/NM/FM system
Christopher Safranski, Jun-Wen Xu, Andrew D. Kent, Jonathan Z. Sun
PPlanar Hall driven torque in a FM/NM/FM system
Christopher Safranski and Jonathan Z. Sun
IBM T. J. Watson Research Center, Yorktown Heights, New York 10598, USA
Jun-Wen Xu and Andrew D. Kent
Center for Quantum Phenomena, Department of Physics,New York University, New York, NY 10003, USA
An important goal of spintronics is to covert a charge current into a spin current with a controlledspin polarization that can exert torques on an adjacent magnetic layer. Here we demonstrate suchtorques in a two ferromagnet system. A CoNi multilayer is used as a spin current source in a samplewith structure CoNi/Au/CoFeB. Spin torque ferromagnetic resonance is used to measure the torqueon the CoFeB layer. The response as a function of the applied field angle and current is consistentwith the symmetry expected for a torques produced by the planar Hall effect originating in CoNi.We find the strength of this effect to be comparable to that of the spin Hall effect in platinum,indicating that the planar Hall effect holds potential as a spin current source with a controllablepolarization direction.
The manipulation of magnetization through the useof spin torques [1] and the conversion of charge to spincurrent are intensely studied areas of condensed mat-ter physics. Investigation of these topics may pave theway towards energy efficient nanodevices for memory[2, 3] and novel computing applications [4–6]. Spin or-bit torques produced by the spin Hall and Rashba effectsin non-magnetic (NM) materials have been explored forswitching of ferromagnets [7, 8] (FM) and spin torqueoscillators [9, 10]. The polarization of the spin currentsin these systems has been dictated by sample geometryand crystal structure, limiting the form of torques thatcan be produced. Expansion of the class of materials andavailable spin polarization has been investigated throughthe means of manipulating crystal symmetry [11] andspin rotation [12, 13] at interfaces. Additionally, spin or-bit torques produced in FM materials [14–17] have alsogained interest.In FM materials, it has been proposed that the anoma-lous (AHE) and planar (PHE) Hall effects can be used toproduce spin current with a controllable spin polarizationdirection [17]. While the AHE has been experimentallyshown to inject spin current from one FM to another[18–20], torques from the PHE have not been observedin the two FM system. In this letter, we demonstratethat the PHE can be used to inject spin current fromone ferromagnet into another in a FM/NM/FM system.In previous studies of a single FM/NM system, the pla-nar Hall effect has been shown to produce spin polariza-tion with out of plane components and a unique angularsymmetry [21]. In the two FM system, we find this angu-lar symmetry can be preserved and the torque strengthis comparable to spin Hall materials. Combined withthe controllable nature of the spin current polarization,spin torques from PHE can be used to broaden the ma-terial systems and available symmetries for spintronicsresearch.The planar Hall effect produces a charge current J PHE = ∆ σ AMR ( ˆ m · E ) ˆ m flowing in the FM parallel toits magnetization [17, 22], where E ≈ E ˆ x is the appliedelectric field and ∆ σ AMR is anisotropic part of the FMconductivity. The charge current creates a spin currentwith a spin polarization aligned with the magnetization[17, 23]. From theoretical calculations [24, 25], the chargeto spin conversion efficiency in materials such as Co is ex-pected to be on the order of the spin Hall effect in Pt.To study PHE driven torques, we choose a CoNisuperlattice since it has a relatively strong PHE [22] andcan be grown with perpendicular anisotropy [26, 27].Spin current is injected from this layer into an adjacentCoFeB (CFB) layer with relatively weak PHE [28](SeeSupplementary Note 1). Using magnetron sputtering,we deposit Ta(3)/Pt(3)/CoNi/Au(3)/CFB(1.5)/Ta(3)(in nm), where the CoNi superlattice is[Co(0.65)/Ni(0.98)] /Co(0.65), onto an oxidized sil-icon substrate. FM layer thicknesses are chosen suchthat CoNi is perpendicularly magnetized with theadditional anisotroy generated by the Pt [29] layer, whilethe CFB is in-plane. The Au layer is used to break theexchange coupling between FM layers while allowingspin current carried by conduction electrons to pass.Its thickness is chosen such that the RKKY couplingacross Au is weak [30]. Vibrating sample magnetometrymeasurements in Fig. 1(a) show that a 2 kOe appliedfield can saturate both layers’ magnetization along theapplied field for in-plane and out-of-plane directions.Using E-beam lithography and ion milling techniques,we pattern 400 nm wide 3 µ m long bridges into the filmand encapsulate it with 40 nm of SiN in situ. The result-ing structure is schematically represented in Fig. 1(b). Inthis design, the leads are patterned from the same mate-rial as the bridges. The width of the leads is chosen suchthat the current density is too low to produce parasiticsignals.In order to detect spin current injection, we employspin torque ferromagnetic resonance (ST-FMR) tech- a r X i v : . [ c ond - m a t . m e s - h a ll ] O c t FIG. 1. (a) Vibrating sample magnetometry data for the film stack used showing saturation along the axes. (b) Samplegeometry and coordinate system. (c) ST-FMR voltage from self-rectification shown as a contour plot against frequency andfield with θ =205 degrees. θ = 0 deg corresponds to sample-normal. (d) ST-FMR voltage as a function of θ and applied field,with a 14 GHz applied microwave current. niques [31, 32]. An amplitude modulated microwave cur-rent is applied directly to the device with a modulationfrequency of 1117 Hz. Rectified voltages produced fromferromagnetic resonance are then measured using lock-intechniques. To produce additional spin currents, a DCcurrent is supplied to the sample through a bias tee. Anyresulting damping-like torques will then modify the FMRresonance linewidth [19, 21, 33].Figure 1(c) shows a contour plot of the measured ST-FMR signal as a function of microwave frequency andapplied field at θ =205 degrees. We observe two distinctresonances corresponding to the two ferromagnetic lay-ers. In order to determine which layer each branch isassociated with, we measure the resonance field as a func-tion of applied field angle θ in the xz plane shown as acontour plot in Fig. 1(d). A strong angular dependenceis observed when rotating from in plane to out of plane.Since the CFB was chosen to be in-plane magnetized andCoNi to be out-of-plane, the angular dependence allowsthe identification of the peaks shown in Figs. 1(c,d).We then determine the damping-like torques by a mea-surement of the resonance linewidth as a function ofDC current. ST-FMR measurements are performed at14 GHz to be in a field range where both FM are nearlycollinear with the applied magnetic field. In this con-figuration, the spin current polarization from CoNi willbe nearly collinear with the CFB layer magnetization.The resulting effect on the FMR resonance linewidth ismaximal for this relative orientation between the spincurrent and CFB magnetization direction. To the lead-ing order, the overall angular dependence for planar Halldriven torques with CFB and CoNi moments in the xz plane will then be determined by the spin current pro-jection on the CFB interface, resulting in a dependenceproportional to ( ˆ m · ˆ x )( ˆ m · ˆ z ) = cos( θ ) sin( θ ) [21].Figure 2(a) shows three ST-FMR traces taken at dif-ferent DC bias at θ =330 degrees. We observe that the linewidth of both resonance peaks is modified bythe application of current. Fitting these to the sum ofLorentzian and anti-Lorentzian functions, we see a lin-ear change in linewidth in Fig. 2(b) for both resonances.Further, we observe that when the CFB layer’s linewidthnarrows, the linewidth of the CoNi layer increases. Fromthe angular dependence of the planar Hall current, if themagnetization is rotated across the z axis to θ =205 de-grees, we would expect to see a change in the sign of thelinewidth vs bias slope. Figure 2(c) shows that the slopedoes indeed change sign.To determine the angular dependence of the torque, wemeasure the slope of linewidth vs bias d ∆ H/dI dc at mul-tiple angles in the xz plane. We exclude angles where thepeaks overlap, since fitting overlapping curves introducesadditional error and effects such as dynamic exchangecoupling that can modify resonance linewidths [34]. Fig-ure 3 shows the measured slope d ∆ H/dI dc for both layers.The dotted fit follows the expected cos( θ ) sin( θ ) angulardependence for PHE driven torques. Other sources ofspin current such as spin Hall and Rashba effects areknown to produce torques as well. However, for thegeometry used here, their spin polarization direction isalong ˆ y and would result in the d ∆ H/dI dc angular sym-metry of ( ˆ m · ˆ y ). This is inconsistent with our obser-vation. Spin currents produced by AHE have a polar-ization following ˆ m as well, however the flow directionfollows ˆ m × ˆ x . When magnetization lies in the xz planeas studied here, there is no flow of spin current in the ˆ z direction towards CFB [17].The angular dependence of the observed torques is con-sistent with the absorption of angular momentum in theCFB layer from a spin current produced in the CoNilayer. In Ref. [21], a similar torque was observed in asingle ferromagnet paired with a spin sink. When com-paring the torque on the CoNi layer in this work to Ref.[21], the sign is consist with a larger spin current transfer FIG. 2. (a) ST-FMR signal at θ =330 degrees for three different DC bias. Resonance linewidth for each layer as a function ofDC bias at (b) θ =330 degrees and (c) θ =205 degrees. from CoNi in the direction of the Au layer. While thereis a Pt layer on the other interface, its resistivity is highdue to the thin nature of the layer [35]. When consider-ing the CFB layer, the symmetry of the torque matcheswhat would be expected if the layer was receiving a trans-fer of angular momentum from the CoNi layer, in thatits linewidth decreases as the CoNi linewidth increases.While CFB is known to have AMR, the CoNi system hasa significantly higher AMR [22] than CFB [28] (See Sup-plementary Note 1). Further, the resistiviity of CFB ishigher, resulting is less charge current passing throughthis layer. As such, the CoNi layer is expected to be themain source of planar Hall driven torques.We next estimate the strength of the observed effect.Here we aim to determine the relative efficiency describedby a dimensionless coefficient η FM , similar to a spin Hallangle. We define η FM to represent the conversion effi-ciency from charge-current to the respective damping-like torque on the CoNi and CFB FM layers. Assuminga collinear geometry and to the leading order, we wouldexpect the spin current to alter the resonance linewidthlinearly for each layer with a slope: d ∆ HdI dc = (cid:126) e η F M M s t F M cos( θ ) sin( θ ) wt tot (1)where M s is the saturation magnetization, t FM is theparticular layer FM thickness, t tot is the total stackthickness, and w the bridge width. Taking M s to be800 emu/cm for both layers and using the amplitude of d ∆ H/dI dc vs θ from the fits in Fig. 3, we calculate theefficiency of the torque on the CFB layer η CFB to be0.05 and for the CoNi layer η CoNi =0.09. Comparing tothe data in Ref [21], the effective η of the spin Hall ina Pt/CoNi system is 0.07 and for the planar Hall basedtorques 0.03. This comparison shows that the spin cur- FIG. 3. Angular dependence of d ∆ H/dI dc for the CFB (top)and CoNi (bottom) layers with a fit to the expected angulardependence (dotted line). rent generation measured here is on the same order asthe spin Hall effect in Pt.In conclusion, we have observed the action of spincurrent generation in a CoNi/Au/CFB system in thelinewidths of the two FM layers. We find the angulardependence of the spin torque magnitude to be consis-tent with the planar Hall effect associated with the CoNilayer. The overall charge to spin conversion efficiency inthis structure is on par with the spin Hall effect in Pt.However, unlike the spin Hall effect, here a partially outof plane polarization is achieved. We believe PHE driventorques allow for deterministic switching of perpendicularmagnetic layers without the need for symmetry breaking.Such controllable spin polarization will also allow studiesof new spin transport configurations. ACKNOWLEDGMENT
Work done with the MRAM group at IBM T. J. Wat-son Research Center in Yorktown Heights, New York.ADK acknowledges support from the National ScienceFoundation under Grant No. DMR-1610416. Researchat NYU was supported partially by the MRSEC Programof the National Science Foundation under Award Num-ber DMR-1420073. This work was performed in part atthe Advanced Science Research Center Nano-FabricationFacility of the Graduate Center at the City University ofNew York. [1] J. Slonczewski, J. Magn. Magn. Mater. , L1 (1996).[2] S. Ikeda, K. Miura, H. Yamamoto, K. Mizunuma, H. D.Gan, M. Endo, S. Kanai, J. Hayakawa, F. Matsukura,and H. Ohno, Nat. Mater. , 721 (2010).[3] D. C. Worledge, G. Hu, D. W. Abraham, J. Z. Sun,P. L. Trouilloud, J. Nowak, S. Brown, M. C. Gaidis, E. J.O’Sullivan, and R. P. Robertazzi, Appl. Phys. Lett. ,98 (2011).[4] B. Dieny, I. L. Prejbeanu, K. Garello, P. Gam-bardella, P. Freitas, R. Lehndorff, and W. Raberg,arXiv:1908.10584 (2019).[5] J. Torrejon, M. Riou, F. A. Araujo, S. Tsunegi,G. Khalsa, D. Querlioz, P. Bortolotti, V. Cros,K. Yakushiji, A. Fukushima, H. Kubota, S. Yuasa, M. D.Stiles, and J. Grollier, Nature , 428 (2017).[6] K. Y. Camsari, B. M. Sutton, and S. Datta, Appl. Phys.Rev. , 011305 (2019).[7] L. Liu, C.-F. Pai, Y. Li, H. W. Tseng, D. C. Ralph, andR. A. Buhrman, Science , 555 (2012).[8] I. M. Miron, K. Garello, G. Gaudin, P.-J. Zermatten,M. V. Costache, S. Auffret, S. Bandiera, B. Rodmacq,A. Schuhl, and P. Gambardella, Nature , 189 (2011).[9] V. E. Demidov, S. Urazhdin, H. Ulrichs, V. Tiberkevich,A. Slavin, D. Baither, G. Schmitz, and S. O. Demokritov,Nat. Mater. , 1028 (2012).[10] Z. Duan, A. Smith, L. Yang, B. Youngblood, J. Lindner,V. E. Demidov, S. O. Demokritov, and I. N. Krivorotov,Nat. Commun. , 5616 (2014).[11] D. MacNeill, G. M. Stiehl, M. H. D. Guimaraes, R. A.Buhrman, J. Park, and D. C. Ralph, Nat. Phys. , 300(2017).[12] A. M. Humphries, T. Wang, E. R. J. Edwards, S. R.Allen, J. M. Shaw, H. T. Nembach, J. Q. Xiao, T. J.Silva, and X. Fan, Nat. Commun. , 911 (2017).[13] S.-h. C. Baek, V. P. Amin, Y.-w. Oh, G. Go, S.-J. Lee,G.-H. Lee, K.-j. Kim, M. D. Stiles, B.-g. Park, and K.-j.Lee, Nat. Mater. , 509 (2018).[14] H. Kurebayashi, J. Sinova, D. Fang, A. C. Irvine, T. D.Skinner, J. Wunderlich, V. Nov´ak, R. P. Campion, B. L.Gallagher, E. K. Vehstedt, L. P. Zˆarbo, K. V´yborn´y, A. J. Ferguson, and T. Jungwirth, Nat. Nanotechnol. , 211(2014).[15] M. Haidar, A. A. Awad, M. Dvornik, R. Khymyn,A. Houshang, and J. ˚Akerman, Nat. Commun. , 2362(2019).[16] W. Wang, T. Wang, V. P. Amin, Y. Wang, A. Radhakr-ishnan, A. Davidson, S. R. Allen, T. J. Silva, H. Ohldag,D. Balzar, B. L. Zink, P. M. Haney, J. Q. Xiao, D. G.Cahill, V. O. Lorenz, and X. Fan, Nat. Nanotechnol. ,819 (2019).[17] T. Taniguchi, J. Grollier, and M. D. Stiles, Phys. Rev.Appl. , 044001 (2015).[18] J. D. Gibbons, D. MacNeill, R. A. Buhrman, and D. C.Ralph, Phys. Rev. Appl. , 064033 (2018).[19] S. Iihama, T. Taniguchi, K. Yakushiji, A. Fukushima,Y. Shiota, S. Tsunegi, R. Hiramatsu, S. Yuasa, Y. Suzuki,and H. Kubota, Nat. Electron. , 120 (2018).[20] T. Seki, S. Iihama, T. Taniguchi, and K. Takanashi,arXiv:1906.09730 (2019).[21] C. Safranski, E. A. Montoya, and I. N. Krivorotov, Nat.Nanotechnol. , 27 (2019).[22] T. McGuire and R. Potter, IEEE Trans. Magn. , 1018(1975).[23] S. Kokado, M. Tsunoda, K. Harigaya, and A. Sakuma,J. Phys. Soc. Jpn. , 1 (2012).[24] V. P. Amin, J. Zemen, and M. D. Stiles, Phys. Rev. Lett. , 136805 (2018).[25] K. D. Belashchenko, A. A. Kovalev, and M. van Schilf-gaarde, Phys. Rev. Mater. , 011401 (2019).[26] S. Mangin, D. Ravelosona, J. A. Katine, M. J. Carey,B. D. Terris, and E. E. Fullerton, Nat. Mater. , 210(2006).[27] M. Arora, R. H¨ubner, D. Suess, B. Heinrich, and E. Girt,Phys. Rev. B , 024401 (2017).[28] K. M. Seemann, F. Freimuth, H. Zhang, S. Bl¨ugel,Y. Mokrousov, D. E. B¨urgler, and C. M. Schneider, Phys.Rev. Lett. , 20 (2011).[29] P. F. Carcia, J. Appl. Phys. , 5066 (1988).[30] S. S. P. Parkin, R. F. C. Farrow, R. F. Marks, A. Cebol-lada, G. R. Harp, and R. J. Savoy, Phys. Rev. Lett. ,3718 (1994).[31] A. A. Tulapurkar, Y. Suzuki, A. Fukushima, H. Kub-ota, H. Maehara, K. Tsunekawa, D. D. Djayaprawira,N. Watanabe, and S. Yuasa, Nature , 339 (2005).[32] J. C. Sankey, P. M. Braganca, A. G. F. Garcia, I. N.Krivorotov, R. A. Buhrman, and D. C. Ralph, Phys.Rev. Lett. , 227601 (2006).[33] S. Petit, C. Baraduc, C. Thirion, U. Ebels, Y. Liu, M. Li,P. Wang, and B. Dieny, Phys. Rev. Lett. , 077203(2007).[34] B. Heinrich, Y. Tserkovnyak, G. Woltersdorf, A. Brataas,R. Urban, and G. E. Bauer, Phys. Rev. Lett. , 4(2003).[35] M. H. Nguyen, D. C. Ralph, and R. A. Buhrman, Phys.Rev. Lett. , 1 (2016).[36] H. Nakayama, M. Althammer, Y.-T. Chen, K. Uchida,Y. Kajiwara, D. Kikuchi, T. Ohtani, S. Gepr¨ags,M. Opel, S. Takahashi, R. Gross, G. E. W. Bauer, S. T. B.Goennenwein, and E. Saitoh, Phys. Rev. Lett. ,206601 (2013).[37] J. Kim, P. Sheng, S. Takahashi, S. Mitani, andM. Hayashi, Phys. Rev. Lett. , 097201 (2016). Supplementary Material forPlanar Hall driven torque in a FM/NM/FM system
SUPPLEMENTARY NOTE 1: MAGNETORESISTANCE MEASUREMENTS
TABLE I. Summary of film level magnetoresistance determined by 4 point probe measurements. Numbers in parenthesis arelayer thickness in nm. Sample MR (%)CFB(10.0)/MgO 0.04Ta(3.0)/[Co(0.65)/Ni(0.98)] /Co(0.65)/Ta(3.0) 0.4Ta(3.0)/Au(3.0)/CFB(1.5)/Ta(3.0) 0.009Ta(3.0)/Pt(3.0)/Au(3.0)/CFB(1.5)/Ta(3.0) 0.008Ta(3.0)/Pt(3.0)/[Co(0.65)/Ni(0.98)] /Co(0.65)/Au(3.0) 0.5 The planar Hall effect is strongly related to the anisotropic magnetoresistance (AMR) in metallic ferromagnets[17,21, 28]. From literature, the AMR in the CoNi system[22] should be much larger than that found in CFB[28]. Inthe following section, we measure the strength of the magnetoresistance in the material stacks used in this study.Using four point resistance measurements on thin films, we define the strength to be
M R = 100(1 − ρ x /ρ z ), where ρ x is the resistivity with a 7 kOe magnetic field applied along the current path, and ρ z measured with field appliedperpendicular to the sample plane. The MR we report here is related to AMR, but is diluted by the resitivity of theshunting layers. We neglect the resistivity in the yy