Amplitude control of spin-triplet supercurrent in S/F/S Josephson junctions
aa r X i v : . [ c ond - m a t . s up r- c on ] O c t Amplitude control of spin-triplet supercurrent in S/F/S Josephson junctions
William Martinez, W.P. Pratt, Jr., and Norman O. Birge ∗ Department of Physics and Astronomy, Michigan State University, East Lansing, Michigan 48824, USA (Dated: October 14, 2018)Josephson junctions made with conventional s-wave superconductors and containing multiple lay-ers of ferromagnetic materials can carry spin-triplet supercurrent in the presence of certain types ofmagnetic inhomogeneity. In junctions containing three ferromagnetic layers, the triplet supercur-rent is predicted to be maximal when the magnetizations of adjacent layers are orthogonal, and zerowhen the magnetizations of any two adjacent layers are parallel. Here we demonstrate on-off controlof the spin-triplet supercurrent in such junctions, achieved by rotating the magnetization directionof one of the three layers by 90 ◦ . We obtain “on-off” ratios of 5, 7, and 19 for the supercurrent in thethree samples studied so far. These observations directly confirm one of the most salient predictionsof the theory, and pave the way for applications of spin-triplet Josephson junctions in the nascentarea of “superconducting spintronics.” PACS numbers: 74.50.+r, 74.45.+c, 75.70.Cn
When a superconducting (S) material is placed in con-tact with a non-superconducting material, the proper-ties of both materials are modified close to the interface.This “superconducting proximity effect” can extend overdistances of several hundred nanometers into the non-superconducting material at low temperatures [1]. Whenthe non-superconducting material is ferromagnetic (F),in contrast, the proximity effect decays over a very shortdistance, of order one nm in strong F materials such asFe or Co [2, 3]. This is because the electrons in a con-ventional superconductor have spin-singlet pairing sym-metry; when such a pair enters a ferromagnetic material,one electron enters the majority spin band and the otherenters the minority band. Those two bands have dif-ferent Fermi momenta, hence the pair acquires a finitemomentum, or equivalently, the pair correlation functionoscillates rapidly in space [4]. Those oscillations dephaseequally rapidly in diffusive systems, leading to a veryshort decay length of the pair correlations in F.In 2001, Bergeret et al. showed that a new type ofproximity effect can arise in S/F systems in the pres-ence of suitable forms of magnetic inhomogeneity nearthe S/F interface [5]. Specifically, the presence of non-collinear magnetizations can induce conversion of spin-singlet Cooper pairs from the conventional superconduc-tor into spin-triplet pairs with projection ± et al. [19] using S/F’/N/F/N/F”/S junctionsand by Iovan et al. [20] using asymmetric S/F’/N/F/Sjunctions. Those groups found evidence for spin-tripletgeneration occurring during the magnetization reversalprocess while sweeping an external magnetic field. Bettercontrol of the magnetic states has been achieved recentlyby several groups measuring the critical temperature T c of S/F/F trilayers, where generation of spin-triplet cor-relations results in lowering of T c [21–26]. Controlling T c , however, is less likely to be useful for future deviceapplications than controlling supercurrent.The goal of this work is to design a Josephson junctionwhere the spin-triplet supercurrent can be controllablyturned on and off by an external magnetic field and wherethese configurations persist when the field is removed.To achieve this goal, we utilized a combination of “hard”(F’) and “soft” (F”) ferromagnetic materials with vastlydifferent switching fields, with the Co/Ru/Co SAF in be-tween them. From our previous work [18, 27], we knowthat thin layers of Ni inside our Josephson junctions arequite hard, i.e. they have a large coercive field, µ H c ≈
50 mT. Ni is also very effective at producing spin-tripletsupercurrent in S/F’/SAF/F”/S junctions [18, 27]. Forthe soft layer, we chose the Permalloy alloy Ni . Fe . ,which typically has µ H c of only a few mT. The sam-ples used in this work have F’ = Ni(1.2nm) and F” =NiFe(1.0nm).Our first task was to determine the hardness of ourCo/Ru/Co SAF - in other words, at what value of theexternal magnetic field does the SAF undergo the spin-flop transition whereby the remanent magnetization di-rections of the Co layers rotate by 90 ◦ . Standard magne-tization measurements were not sufficient for this task,because the M vs H curves of our SAFs do not showany feature at the spin-flop transition; they are essen-tially linear until the SAF magnetization saturates athigh field. Instead we used the anisotropic magnetoresis-tance (AMR) to determine how the Co layers in the SAFrespond to an external magnetic field [28]. As expected,AMR measurements showed that samples with thinnerCo layers are less sensitive to the applied field, but verythin Co won’t suppress the short-range spin-singlet su-percurrent adequately relative to the long-range spin-triplet supercurrent. As a compromise, we chose to workwith d Co = 4 nm (for a total Co thickness of 8 nm) inour actual Josephson junctions. Josephson junctions con-taining such SAFs exhibit significantly suppressed spin-singlet supercurrent [16], while SAF samples with d Co =4 nm showed minimal change in AMR for applied fieldsless than 20 mT [28] – the field range relevant to theexperiments to be presented here.A second important consideration in the sample de-sign was the lateral size of the Josephson junctions. It iswell known that the critical current, I c , of a Josephsonjunction exhibits a “Fraunhofer pattern” as a functionof magnetic field applied in a direction perpendicular tothe current flow. For circular junctions with the cur-rent flowing out-of-plane and the field applied in-plane, I c follows an Airy pattern in flux with its first minimumat Φ / Φ = 1.22, where Φ = h/2e is the flux quantum.The effective magnetic flux through the junction area,Φ = µ Hw (2 λ L + d N + d F ) + µ M wd F , includes thecontribution from the external field H and from the in-ternal magnetization M , assuming the latter is uniformand collinear with H . Here w is the sample diameter, d N and d F are the thicknesses of the N and F layersin the junction, and the London penetration depth, λ L ,appears in the first term due to penetration of the ex-ternal field into the top and bottom S electrodes. Thewidth of the central lobe of the Airy pattern in magneticfield is inversely proportional to the sample diameter w ,and the central peak is displaced in the direction oppo-site to the direction of M [29, 30]. Since our experimentwill involve changing the magnetization direction of the soft NiFe layer in our samples, we anticipate shifts in theAiry pattern central peak position during the course ofthe experiment. While such shifts may be useful in theirown right [31, 32], from the point of view of this projectthey are a nuisance. To avoid this complication, we mustmake our Josephson junctions sufficiently small so thatthe widths of the Airy patterns are much larger than anydisplacements of the central peak position. For this ex-periment we fabricated samples with diameters of 0.5,0.7, and 1.0 m. Given that λ L = 85 nm for our Nb, for acircular junction of diameter w = 1.0 µ m we expect thefirst zero in the Airy pattern to occur at µ H ≈
12 mT,where we have used d N + d F = 50 nm to account for allthe ferromagnetic layers and Cu spacers in the junctions.So as long as the Airy shifts are much less than 12 mT,we can ignore them.A final consideration in the sample design is thenumber of magnetic layers in the junctions that arepatterned during the ion milling fabrication step, asopposed to being left as extended thin films. Wehave fabricated samples with various combinations ofion mill depths. In general, we found that patternedSAFs were softer than unpatterned SAFs; hence inthe samples reported here, only the top NiFe layeris patterned while both the SAF and the bottom Nilayer remain as extended films. The final Joseph-son junction samples had the following layer structure:Nb(100)/Cu(5)/Ni(1.2)/Cu(10)/Co(4)/Ru(0.75)/Co(4)/Cu(10)/NiFe(1.0)/Cu(5)/Nb(20)/Au(15)/Nb(150)/Au(20),where all thicknesses are in nm. Details of the samplefabrication procedure are provided in [33]. FIG. 1: (color online) Schematic diagram of the magnetiza-tion directions for the ferromagnetic layers in our Josephsonjunctions in the spin-triplet supercurrent “on” state (left) and“off” state (right). The sample is initialized into the on stateby a large field, µ H = 260 mT, in the longitudinal (X) direc-tion. Thereafter, only small applied fields, µ H <
20 mT, areapplied in either the X or Y directions to rotate the magneti-zation of the soft NiFe layer while leaving the magnetizationsof the hard Ni layer and Co/Ru/Co SAF unchanged.
Completed samples were mounted on a cryo-probe andmeasured at 4.2 K in a liquid helium dewar equipped witha Cryoperm magnetic shield. The probe had a pair of or-thogonal coils to provide magnetic fields at any directionin the sample plane. To initialize the magnetization ofeach layer, a large (260 mT) magnetic field is applied lon-gitudinally along the sample, i.e. along the x directionshown in the left side of Figure 1. This causes the Ni andNiFe magnetizations to align in the x-direction while thetwo Co layers in the SAF point along the ± y-directionsafter the spin-flop transition and subsequent reduction offield, creating orthogonality of the magnetizations to op-timize generation of spin-triplet supercurrent. To removetrapped flux in the Nb electrodes, the sample was brieflylifted to just above the liquid He level in the dewar, thenre-immersed. Current-voltage characteristic curves weremeasured using a 4-terminal SQUID-based self-balancingpotentiometer circuit. The current was stepped from zerowell past the critical current in each direction; the curveswere fit by the standard form for overdamped Joseph-son junctions: V = R N ( I − I c ) / , where R N is thenormal state resistance and I c is the critical current. Byrepeating this measurement while iteratively stepping H x through a field range typically -20 to 20 mT, an Airy pat-tern is measured in the longitudinal direction. This initialAiry measurement serves as a test of junction quality. FIG. 2: Evolution of the Josephson critical current, I c , of sam-ple From here, a small transverse magnetic field H y is ap-plied and removed, and I c is measured again at H = 0.This process is repeated with increasing values of H y .The results are shown in Figure 2(a) for sample I c starts at about 45 µ A, then de-creases with increasing H y until it saturates at the lowvalue of ≈ µ A for µ H y = 16 mT. This decrease isdue to rotation of the NiFe magnetization until it alignsnearly collinearly with the Co layers in the SAF, therebyturning off the spin-triplet-generation mechanism. Theprocess can be reversed by applying gradually increasingvalues of H x , as shown in Figure 2(b) – again with allmeasurements performed at H = 0. For this sample, I c returns to its initial value when µ H x reaches about 10mT. The fields required to turn the triplet supercurrenton and off vary somewhat from sample to sample; hencethe initial procedure described in Figure 2 determines themagnitude of the maximum external field that will beused for all ensuing measurements. If we apply too largea field during the “turn-off” step shown in Figure 2(a),then the supercurrent starts to increase again, indicatingthat the SAF is starting to rotate. If that happens, thesample must be re-initialized and the experiment startedover again.To further test the robustness of these observations,we carried out measurements of I c in the presence ofthe applied field, i.e. measurements of the longitudinaland transverse Airy patterns in the on and off states,respectively. Figure 3(a) shows the results for sample µ H max = 20 mT. In the initial on state, sam-ple 1 has I c ≈ µ A. We first ramp the transverse field H y from 0 to H max , and I c drops precipitously with in-creasing H y , as expected. Not only does the field ro-tate the magnetization of the NiFe and therefore turnthe spin-triplet supercurrent off, but the presence of thefield also causes I c to decrease as the first minimum inthe Airy pattern is approached. To separate the two ef-fects, a full Airy pattern is now measured as a function offield H y in the transverse direction: H max → − H max → H max . Figure 3(a) shows that I c remains low duringthis process, confirming that the spin-triplet supercur-rent remains suppressed for all fields. The whole pro-cess is now repeated while applying a longitudinal field H x . After an initial transient, Figure 3(b) shows that I c starts to rise for 2 mT < µ H x < H x approaches the location of the first minimum inthe Airy pattern. During the subsequent full longitudi-nal field sweeps, H max → − H max → H max , I c exhibits alarge-amplitude Airy pattern with maximum value closeto the initial value of I c = 76 µ A, demonstrating thatthe spin-triplet supercurrent has been turned back on.Figure 3(b) shows that there is some shift in the centralpeak position of the Airy pattern due to the changingNiFe magnetization direction, but the effect is not largeenough to obscure the results.Finally, we test the repeatability of the result by apply-ing alternately fields in the two directions: H y = H max and H x = H max , measuring I c in zero field for each itera-tion. The results are plotted in Figure 4 for all three sam-ples. After an initial transient, which is not understood, FIG. 3: (color online) Plots of critical current vs applied field,known as Fraunhofer patterns. (a) The sample starts in the“on” state at H = 0, with I c = 76 µ A. As a field is applied inthe transverse (Y) direction (black squares), I c drops rapidlydue to both the Fraunhofer physics and the turning off ofthe spin-triplet supercurrent. To verify the latter, the trans-verse field is swept from +20 to -20 mT (red circles) and thenback to +20 mT (blue triangles). The supercurrent stays lowthroughout this entire transverse Fraunhofer pattern. (b) Thesample starts in the off state at H = 0, with I c ≈ µ A. Asa longitudinal (X) field is applied, I c starts to increase dueto the spin-triplet supercurrent turning on, but immediatelydrops due to Fraunhofer physics. To verify the former, thelongitudinal field is swept from +20 to -20 mT (red circles)and then back to +20 mT (blue triangles). The Fraunhoferpattern exhibits a maximum supercurrent close to the initialvalue of 76 µ A. all samples show repeatable behavior, with I onc / I offc ≈
7, 19, and 5 for samples 1 - 3, respectively. Note thatthe value of I c in the on state varies little between thethree samples, whereas the off-state values vary consider-ably due to small uncontrolled misalignments of the layermagnetizations that produce residual spin-triplet super-current.In conclusion, we have demonstrated control of theamplitude of spin-triplet supercurrent through manipu-lation of the magnetization direction of neighboring F FIG. 4: Demonstration of “on-off” switching of the spin-triplet supercurrent. Each figure shows a different sample asfields of H max are applied alternately in the transverse andlongitudinal directions, for odd and even values of the fielditeration, respectively. The values of H max used for thesethree samples were 20 mT, 16 mT, and 10 mT for (a), (b),and (c), respectively. layers in multi-ferromagnet S/F/S Josephson junctions.In addition, we have shown that the states are stableand maintained when the magnetic field is turned off,and reproducible over multiple iterations and in multi-ple samples. This work represents a major step forwardin the control of spin-triplet correlations induced in su-perconducting/ferromagnetic heterostructures, hence inthe development of the nascent field of superconductingspintronics [34, 35].Acknowledgements: We thank B. Bi, A. Cramer, andR. Loloee for technical support, and B. Niedzielski fora critical reading of the manuscript. The work re-ported here was funded by the U.S. Department of En-ergy, Office of Basic Energy Sciences, Division of Mate-rials Sciences and Engineering under Award DEFG02-06ER46341. Sample fabrication was carried out in theKeck Microfabrication Facility at Michigan State Uni-versity. ∗ Electronic address: [email protected][1] G. Deutscher, P.G. deGennes, Proximity Effects. Su-perconductivity, edited by Parks, R.G., pp. 1005-1033(1969).[2] A.I. Buzdin, Rev. Mod. Phys.
935 (2005).[3] J.W.A. Robinson, S. Piano, G. Burnell, C. Bell, and M.G.Blamire, Phys. Rev. Lett. , 177003 (2006).[4] E.A. Demler, G.B. Arnold, and M.R. Beasley, Phys. Rev.B , 15174 (1997).[5] F.S. Bergeret, A.F. Volkov, and K.B. Efetov, Phys. Rev. Lett. , 4096 (2001).[6] A.F. Volkov, F.S. Bergeret, and K.B. Efetov, Phys. Rev.Lett. , 117006 (2003).[7] M. Eschrig, J. Kopu, J.C. Cuevas, and G. Sch¨on, Phys.Rev. Lett., 90, 137003 (2003).[8] F.S. Bergeret, A.F. Volkov, K.B. Efetov, Rev. Mod. Phys. , 1321 (2005).[9] R.S. Keizer, S.T.B. Goennenwein, T.M. Klapwijk, G.Miao, G. Xiao, and A. Gupta, Nature (London) ,825 (2006).[10] I. Sosnin, H. Cho, V.T. Petrashov, and A.F. Volkov,Phys. Rev. Lett. , 157002 (2006).[11] T.S. Khaire, M.A. Khasawneh, W.P. Pratt Jr. and N.O.Birge, Phys. Rev. Lett. , 137002 (2010).[12] J.W.A. Robinson, J.D.S. Witt and M.G. Blamire, Science , 59 (2010).[13] D. Sprungmann, K. Westerholt, H. Zabel, M. Weides andH. Kohlstedt, Phys. Rev. B , 060505 (2010).[14] M. S. Anwar, F. Czeschka, M. Hesselberth, M. Porcu,and J. Aarts, Phys. Rev. B , 100501 (2010).[15] J. Wang, M. Singh, M. Tian, N. Kumar, B.Z. Liu, C.Shi, J.K. Jain, N. Samarth, T.E. Mallouk, and M.H.W.Chan, Nature Phys. , 389 (2010).[16] M.A. Khasawneh, W.P. Pratt Jr. and N.O. Birge, Phys.Rev. B , 020506(R) (2009).[17] M. Houzet and A.I. Buzdin, Phys. Rev. B , 060504(R)(2007).[18] C. Klose, T.S. Khaire, Y. Wang, W.P. Pratt, Jr., N.O.Birge, B.J. McMorran, T.P. Ginley, J.A. Borchers, B.J.Kirby, B.B. Maranville, and J. Unguris, Phys. Rev. Lett. , 127002 (2012).[19] N. Banerjee, J.W.A. Robinson, and M.G. Blamire, Na-ture Comm. , 4771 (2014).[20] A. Iovan, T. Golod, and V.M. Krasnov, Phys. Rev. B ,134514 (2014). [21] P.V. Leksin, N.N. Garif’yanov, I.A. Garifullin, Ya.V.Fominov, J. Schumann, Y. Krupskaya, V. Kataev, O.G.Schmidt, and B. B¨uchner, Phys. Rev. Lett. , 057005(2012).[22] V.I. Zdravkov et al. , Phys. Rev. B , 144507 (2013).[23] X.L. Wang, A. Di Bernardo, N. Banerjee, A. Wells, F.S.Bergeret, M.G. Blamire, and J.W.A. Robinson, Phys.Rev. B , 140508 (2014).[24] A.A. Jara, C. Safranski, I.N. Krivorotov, C.T. Wu, A.N.Malmi-Kakkada, O.T. Valls, and K. Halterman, Phys.Rev. B , 184502 (2014).[25] M.G. Flokstra, T.C. Cunningham, J. Kim, N. Satchell,G. Burnell, P.J. Curran, S.J. Bending, C.J. Kinane,J.F.K. Cooper, S. Langridge, A. Isidori, N. Pugach, M.Eschrig, and S.L. Lee, Phys. Rev. B , 060501 (2015).[26] A. Singh, S. Voltan, K. Lahabi, and J. Aarts, Phys. Rev.X , 021019 (2015).[27] Y. Wang, W.P. Pratt, Jr., and N.O. Birge, Phys. Rev. B , 214522 (2012).[28] W. Martinez, Ph.D. thesis, Michigan State University(2015).[29] V.V. Ryazanov, Usp. Fiz. Nauk , 825 (1999).[30] T.S. Khaire, W.P. Pratt, Jr., and N.O. Birge, Phys. Rev.B , 094523 (2009).[31] R. Held, J. Xu, C.W. Schneider, J. Mannhart, and M.R.Beasley, Appl. Phys. Lett. , 163509 (2006).[32] T.I. Larkin et al., Appl. Phys. Lett. , 222601 (2012).[33] B.M. Niedzielski, E.C. Gingrich, R. Loloee, W.P. Pratt,Jr., and N.O. Birge, Supercond. Sci. Technol. , 085012(2015).[34] M. Eschrig, Physics. Today (1), 43 (2011).[35] J. Linder and J.W.A. Robinson, Nature Phys.11