Supercurrent in ferromagnetic Josephson junctions with heavy metal interlayers

Supercurrent in ferromagnetic Josephson junctions with heavy metal interlayers


The lengthscale over which supercurrent from conventional BCS, -wave, superconductors (S) can penetrate an adjacent ferromagnetic (F) layer depends on the ability to convert singlet Cooper pairs into triplet Cooper pairs. Spin aligned triplet Cooper pairs are not dephased by the ferromagnetic exchange interaction, and can thus penetrate an F layer over much longer distances than singlet Cooper pairs. These triplet Cooper pairs carry a dissipationless spin current and are the fundamental building block for the fledgling field of superspintronics. Singlet-triplet conversion by inhomogeneous magnetism is well established. Here, we describe an attempt to use spin orbit coupling as a new mechanism to mediate singlet-triplet conversion in S–F–S Josephson junctions. We report that the addition of thin Pt spin-orbit coupling layers in our Josephson junctions significantly increases supercurrent transmission, however the decay length of the supercurrent is not found to increase. We attribute the increased supercurrent transmission to Pt acting as a buffer layer to improve the growth of the Co F layer.


I Introduction

In nature there are very few examples of materials exhibiting simultaneously superconducting (S) and ferromagnetic (F) properties, due to the competition between the order parameters. Breakthroughs in materials engineering and nanolithography techniques in the last two decades have enabled the synthesis of artificial heterostructures, where two or more layers are in direct electronic contact, revealing a wealth of new physics at S–F interfaces Eschrig (2011); Linder and Robinson (2015); Eschrig (2015). Exploitation of this new physics has led to advances in the emerging field of superspintronics, which offers a new class of highly energy efficient devices, most promisingly cryogenic memory elements based on ferromagnetic Josephson junctions Bell et al. (2004); Goldobin et al. (2013); Baek et al. (2014); Niedzielski et al. (2014); Gingrich et al. (2016); Niedzielski et al. (2018); Glick et al. (2017); Dayton et al. (2018).

In the simplest case of a normal metal (N) in S–N–S Josephson junctions the critical current of the junction () will decay slowly with increasing N thickness, on a typical lengthscale 100’s nm Dubos et al. (2001); Buzdin (2005). When the N layer is replaced by a ferromagnet, S–F–S, singlet Cooper pairs entering the F layer are dephased and gain an oscillatory term with F layer thickness, driven by the same physics as the predicted Fulde-Ferrell-Larkin-Ovchinnikov (FFLO) state Fulde and Ferrell (1964); Larkin and Ovchinnikov (1965); Buzdin et al. (1982); Demler et al. (1997); Buzdin (2005). This oscillation results in a series of transitions from a zero to ground-state phase difference across the Josephson junction. Experimentally, oscillations in are observed with increasing F layer thickness Oboznov et al. (2006), typically over a lengthscale 1-5 nm for a strong ferromagnet Robinson et al. (2006); Khaire et al. (2009). With the addition of spin mixing layers on either side of the F layer it is possible to create the so-called long ranged triplet component (LRTC) Bergeret et al. (2001, 2005). Unlike singlet Cooper pairs, the LRTC is not dephased by the ferromagnet and can therefore penetrate further into the F layer than the singlet component, typically 10’s-100’s nm Keizer et al. (2006); Sosnin et al. (2006); Robinson et al. (2010); Anwar et al. (2010); Khaire et al. (2010). Experimentally, the LRTC is generated reliably through the addition of F’, F” ferromagnetic layers in S–F’–F–F”–S Josephson junctions. Where F’, F” have either intrinsic magnetic inhomogenity (for example Ho Sosnin et al. (2006); Robinson et al. (2010)) or magnetic inhomogenity which is engineered in multilayered structures Khaire et al. (2010); Anwar et al. (2012); Martinez et al. (2016). It is known theoretically, however, that this spin mixing layer need not be a ferromagnet and there are several proposals indicating that spin-orbit coupling can act as a source for the LRTC Konschelle (2014); Bergeret and Tokatly (2013, 2014); Jacobsen et al. (2015, 2016).

In this work we describe an attempt to generate the LRTC using a normal metal with strong spin-orbit coupling (). We compare the transport properties of two sets of Josephson junctions; S–F–S and SFS, where S is Nb, F is a Co/Ru/Co synthetic antiferromagnet (SAF), and is Pt (which has been shown in previous works to have strong Rashba spin-orbit coupling with Co due to broken inversion symmetry Miron et al. (2011); Haazen et al. (2013); Hrabec et al. (2014)). It is our proposal that only the set of samples containing two layers should display properties consistent with the generation of the LRTC. In -containing samples the decay of supercurrent transmitted through the F layer should take place over a longer length scale than in the non- samples, which only contain the short ranged supercurrent components (namely singlet and triplet Cooper pairs).

In a recent report of Banerjee et al., characteristics of Nb/Pt()/Co/Pt multilayers vary with  nm Banerjee et al. (2017). The changes in are attributed to spin-orbit coupling mediating singlet to triplet conversion upon applying a field either in-plane or out-of-plane (as it is argued the triplet will be generated preferentially in-plane and suppressed out-of-plane by the spin-orbit coupling). The results are presented in analogy to the triplet spin valve, where the presence of triplet correlations modifies the proximity effect, introducing measurable signatures in Fominov et al. (2010). In this present work we measure Josephson current, as opposed to , exploring a larger parameter space with a systematic approach. Josephson junction experiments are a more direct measure of LRTC generation and propagation, and are less susceptible to stray field effects than measurements of Satchell et al. (2017). We hope the data in this present work can help inform interpretation of the Banerjee et al. result, however we stress that the two works are not directly comparable.

Ii Methods

The films are deposited using DC sputtering in a vacuum system with base pressure of  Torr and partial water pressure of  Torr after liquid nitrogen cooling. The samples are grown on 0.5 mm thick Si substrates which have a native oxide layer. A uniform 200 Oe magnetic growth field is applied to the substrates. Growth is performed at an approximate Ar pressure of 2 mTorr, at a typical growth rate of 0.4 nm s for Nb and 0.1-0.2 nm s for the other materials. Growth rates are calibrated using an in situ crystal film thickness monitor and checked by fitting to Keissig fringes obtained by low angle X-ray reflectometry. All layer thicknesses (in brackets) are in nm. The bottom superconducting electrode is a multilayer [Nb(25)/Al(2.4)]/Nb(20) which grows considerably smoother than single layer Nb of comparable total thickness Wang et al. (2012); Thomas et al. (1998); Kohlstedt et al. (1996). The bottom electrode, ferromagnetic layers, any normal metal interlayers, and a capping bilayer Nb(5)/Au(15) are grown without breaking vacuum.

For electrical transport measurements, films are patterned into circular Josephson junctions of diameter; 12, 24 and 48 m using standard photolithography and ion milling methods, described in previous work Wang et al. (2012). Once the Josephson junctions are defined, the samples are returned to the DC sputtering system and the top Au(15) layer is ion milled in situ thus recovering a very clean interface before depositing the top superconducting electrode, Nb(100).

Magnetization loops of sister sheet film samples are measured using a Quantum Design SQUID VSM magnetometer at 10 K. Electrical transport is performed using a conventional four-point-probe measurement configuration at 4.2 K, employing the low noise electrical transport system described in reference Glick et al. (2017). Our system can resolve 6 pV, which is taken as a resolution limit where appropriate. For all transport measurements the field is applied parallel to the sample’s plane, and samples are measured in the as-grown magnetic state (which is set by the growth field).

Iii Magnetic Characterization

Figure 1: Magnetic hysteresis loops acquired at a temperature of 10 K. (a,b) for S–F–S type samples with the applied field oriented (a) in and (b) out of the sample plane and (c,d) for S–Pt(0.5)–F–Pt(0.5)–S type samples (c) in and (d) out of the sample plane. F is a Co(5)/Ru(0.6)/Co(5) multilayer in all cases. The diamagnetic contribution from the substrate has been subtracted and data are normalized by the saturated value of magnetization. Insets show the low field switching.

The magnetization versus field data are shown in FIG. 1, for (a,b) S–F–S and (c,d) SFS samples at 10 K, where F is a Co(5)/Ru(0.6)/Co(5) multilayer in all cases and is Pt(0.5). The choice of Pt thickness here is dictated by the transport measurements to follow. Both samples behave as synthetic antiferromagnets (SAFs). The application of an applied field in-plane (a,c) causes a spin-flop transition and then rotates the magnetizations into the direction of the applied field, saturating at about 5 kOe. Removing the field causes the Co layers in the SAF to relax antiparallel w.r.t. each other and perpendicular w.r.t. to the applied field, hence zero remanent magnetization is observed. The spin-flop transition in similar samples was confirmed in previous work by polarized neutron reflectometry Klose et al. (2012). In FIG. 1 (c), the addition of the Pt(0.5) interlayers has caused a slight reduction in the low-field susceptibility, between about -1 and 1 kOe (highlighted in the figure inset). This implies that there are magnetic phases present with different coercivities, which we can understand if the surface moments couple with the Pt layer, modifying the local anisotropy of the Co/Pt interface. Or alternatively, this may be a direct signature of the spin-flop transition in this sample.

The addition of Pt at the Co interface may induce an additional out-of-plane magnetic anisotropy Nakajima et al. (1998). The response of both samples to an out-of-plane applied field, as shown in FIG. 1 (b,d), indicates that the magnetic anisotropy of our samples lie predominantly in-plane as very little out-of-plane remanent magnetization is observed. The sample containing Pt interlayers may have a small out-of-plane component, shown in FIG. 1 (d) and inset. This is not surprising given the thickness of Co and Pt layers in this study compared to the previous work of Shepley et al. Shepley et al. (2015); where an out-of-plane magnetic anisotropy is achieved for Pt thickness of 2.5 nm and corresponding Co thicknesses in the range 0.85-1.0 nm. The reorientation transition from predominant out-of-plane to in-plane magnetic anisotropy is found at Co thickness 1.1 nm Shepley et al. (2015).

We do not expect the slight differences in magnetic switching between samples with and without Pt interlayers (FIG. 1) to affect our transport measurements, which are performed in the as-grown magnetic state.

Iv Electrical Transport

Typical IV curves and Fraunhofer () patterns for each Josephson junction size along with the collated (area times the normal state resistance) are shown in the Supplemental Materials Not (a).

iv.1 –Pt()–Co(5)/Ru(0.6)/Co(5)–Pt()–

Figure 2: Product of critical current times normal state resistance vs Pt interlayer thickness () for Josephson junctions of the form –Pt()–Co(5)/Ru(0.6)/Co(5)–Pt()–. Each data point represents one Josephson junction and the uncertainty in determining is smaller than the data points.

We first consider the transport characteristics of the set of samples; –Pt()–Co(5)/ Ru(0.6)/Co(5)–Pt()–, where  nm. A total Co thickness of 10 nm is chosen as it is known from previous works to be a thickness where LRTC and non-LRTC samples show obvious difference in (approximately an order of magnitude) Khaire et al. (2010). FIG. 2 shows the result of this study, where characteristic junction voltage (the product of the maximum measured in the Fraunhofer pattern times normal state resistance) is plotted for a series of samples with increasing . Without the Pt, our Josephson junctions have an expected small . With the addition of  nm (less than two monolayers) we see a large enhancement in and no change in , increasing the characteristic junction voltage by approximately an order of magnitude. This high remains approximately constant for  nm Pt interlayers, but for increasing Pt thicknesses beyond this there is a sharp drop in , where the critical current of our Josephson junctions appears to fall back towards the value without any Pt.

iv.2 –Pt(0.5)–Co()/Ru(0.6)/Co()–Pt(0.5)–

Figure 3: Product of critical current times normal state resistance vs total Co thickness () for Josephson junctions of the form; blue triangles –Pt(0.5)–Co()/Ru(0.6)/Co()–Pt(0.5)– and red inverted triangles –Co()/Ru(0.6)/Co()–. Each data point represents one Josephson junction and where no error bars are shown, the uncertainty in determining is smaller than the data point. The lines are fits to simple exponential decay in the range 8 nm 16 nm, with decay lengths of  nm and  nm, respectively.

With a fixed Pt thickness of 0.5 nm, guided by FIG. 2, we next vary the thickness of the F layer in –Pt(0.5)–Co()/Ru(0.6)/Co()–Pt(0.5)– and –Co()/Ru(0.6)/ Co()– samples to compare the decay length of the supercurrent with and without the Pt interlayer. For these samples as a function of total Co thickness is plotted in FIG. 3. Considering first the data for samples without Pt, where we expect to only have short ranged supercurrent components inside the layer, we observe the expected rapid decay in Khasawneh et al. (2009). Fitting to a simple exponential decay the decay length is found to be  nm. This decay length is longer than found in our previous study of a Co SAF, where the comparable samples have decay length  nm Khasawneh et al. (2009). The difference between this work and previous work is the choice of bottom electrode, where the smoother multilayer electrode in this work may have improved the growth condition of the Co, compared to the rougher single layer Nb in the previous work. Samples thicker than  nm were fabricated but show no measurable critical current.

The samples containing Pt(0.5) interlayers can be described in two regimes. In the Co thickness range 10 nm 16 nm the supercurrent is found to decay exponentially with decay length  nm. This is identical within our experimental uncertainty to the decay length of samples without Pt interlayers and suggests that short ranged supercurrent components dominate transport in our Josephson junctions. The product for the Pt interlayer samples is, however, consistently about one order of magnitude higher than the samples without the Pt interlayer. This suggests that the Pt has a role in the transmission of these short ranged supercurrent components.

Although the product of the  nm Josephson junction appears consistent with the other samples measured in this work, we believe this data point to be an anomaly. The other five Josephson junctions on this substrate showed electrical shorts when measured, indicating a failure in the lithographic processing. While it is included in FIG. 3 for completeness, the  nm sample is not used in the analysis of decay lengths. A  nm sample containing Pt(0.5) interlayers did not survive fabrication processing.

In the second regime, upon increasing the Co thickness further it is found that unlike the samples without Pt, a small residual supercurrent is observable in Josephson junctions with Co thickness  nm and  nm. The  nm sample produces a clear critical current and Fraunhofer pattern, giving us confidence that the of this sample is well above what is expected from the decay of short ranged supercurrent components, depicted by the blue line. The  nm sample shows some evidence for non-zero critical current, however the Fraunhofer pattern is not well defined, and the product is at the resolution of our instrument (approximately 6 pV), leading to a large uncertainty in this value. We have included data from the and 24 nm samples in FIG 4 of the Supplemental Materials Not (a). The decay length in this regime (where short ranged supercurrent components are vanishingly small) appears much longer, but we do not have enough data points to place a meaningful quantitative value on the decay.

iv.3 Control Samples

Figure 4: Product of critical current times normal state resistance vs selected sample for; symmetric Josephson junctions containing no interlayers, two Pt(0.5) interlayers or two Cu(0.5, 2.5) interlayers, and asymmetric Josephson junctions which contain only one Pt(0.5) interlayer. Each symbol corresponds to a set of samples grown in the same vacuum cycle, and our run-to-run variation is visible in the -F- only sample. All Josephson junctions contain a Co(5)/Ru(0.6)/Co(5) F layer. Each data point represents one Josephson junction and the uncertainty in determining is smaller than the data points.

As a first control measurement we consider the properties of a set of samples where we replace the Pt interlayer with a normal metal , Cu. Cu is not expected to contribute much to spin-orbit coupling in our system and the supercurrent carrying properties of Cu are well characterized. It is known that supercurrent decay through Cu is very slow due to to the long electron mean free path Dubos et al. (2001). Additionally, Cu is already widely implemented as a normal metal interlayer in S–F systems, where it is added into multilayer stacks as a buffer layer to improve growth conditions, and to decouple multiple F layers where independent switching of each F layer is desirable Khasawneh et al. (2009); Jara et al. (2014); Leksin et al. (2015).

FIG. 4 includes for the set of samples; –Cu()–Co(5)/Ru(0.6)/Co(5)–Cu()–, where , 0.5 and 2.5 nm (black diamonds). With no interlayer, once again a small is observed. The slightly lower of this sample set (compared to the previous sample set, blue triangles) is mostly likely run-to-run variation between samples grown in different vacuum cycles. Adding 0.5 nm of Cu does not significantly improve the transmission of supercurrent through the F layer, unlike Pt. With an increase in Cu thickness to 2.5 nm, an increase in to approximately the value of the Pt(0.5) interlayer samples is observed. This increase with thick Cu interlayers is expected from previous works and is due to improved structural properties of the Co, a mechanism outlined further in Section V.

The next control samples are asymmetric Josephson junctions containing only one Pt(0.5) interlayer, olive green stars on FIG. 4. We find that the interface on which the Pt is grown is of key importance to the observed increase in . When the Pt layer is grown only on the top interface (–Co(5)/Ru(0.6)/Co(5)–Pt(0.5)–) a small reduction in is observed. On the other hand, when the Pt layer is grown only on the bottom interface (–Pt(0.5)–Co(5)/Ru(0.6)/Co(5)–) a large increase in is observed, almost recovering the value for the symmetric Pt(0.5) samples (replotted blue triangles on FIG. 4).

V Discussion

Direct evidence for the presence of a LRTC of superconductivity in a Josephson junction is the slower decay of with increasing F layer thickness. This “smoking gun” is reliably observed in samples of the form S–F’–F–F”–S where F’, F” are aligned perpendicular to F Khaire et al. (2010); Anwar et al. (2012); Martinez et al. (2016). This increase in supercurrent decay length is often (but not necessarily) accompanied by an increase in for LRTC samples compared to non-LRTC samples. In this work for the Co thickness range 8 nm 16 nm we observe increase in , without the corresponding increase in decay length. We therefore do not believe the increased transmission of supercurrent into the Co layers for the thickness range 8 nm 16 nm can be attributed to the presence of a LRTC generated by spin-orbit coupling.

For our Josephson junctions with Co thickness  nm and  nm we measure a very small residual supercurrent. The measured of these junctions is well above what is expected from the decay of short ranged supercurrent components, the blue line in FIG. 3. The decay length of superconductivity for these Co thicknesses appears to be much longer than for thinner Co. These observations offer the best evidence in this work that spin-orbit coupling can mediate the conversion of Cooper pairs to the LRTC. This residual supercurrent is many orders of magnitude lower than comparable samples with F’, F” LRTC generating interlayers Khaire et al. (2010). This suggests at a minimum that if we attribute these observations to spin-orbit coupling generating a LRTC that the conversion efficiency of such a mechanism is poor. We must caution that the supercurrent observed in these junctions is only just above the resolution limit of our measurement apparatus.

In previous work on the transmission of singlet supercurrent through a Co/Ru/Co SAF it is found that the addition of Cu(5) interlayers changes the growth characteristic of Co, improving the mean-free-path and increasing the decay length of supercurrent from  nm to  nm, accompanied with an increase in Khasawneh et al. (2009). This improvement can be understood by the fcc Cu providing better lattice matching for the Co layer compared to growth of Co directly on bcc Nb, which introduces misfit strain at the interface.

It is found in this present study that Cu(0.5) interlayers are not thick enough to affect the growth characteristic of Co, and so no improvement in the transmission of supercurrent is observed (FIG. 4 black diamonds). However, upon growing Cu(2.5) interlayers, we recover the higher value found for the Pt(0.5) interlayers. In previous high resolution transmission electron microscopy studies it is found that Cu will grow in a nonequilibrium bcc phase on bcc Nb for Cu thicknesses up to 1.2 nm Mitchell et al. (1997); Kung et al. (1997); Geng et al. (1999). In this bcc phase the lattice parameter of Cu is  nm (close to the bulk bcc Nb lattice parameter of 0.331 nm). We would therefore not expect the thin Cu(0.5) interlayers to act as an effective seed layer for growth of the Co layers, due to it forming the nonequilibrium bcc phase. By 2.5 nm the Cu has recovered fcc growth, and the positive influence this has on the Co growth is observed in the improved .

We propose here that fcc Pt influences the growth of the Co layers by the same mechanism as fcc Cu. However, unlike Cu, Pt does not form a nonequilibrium bcc phase for very thin layers grown on bcc Nb. Hence only a thin Pt(0.5) interlayer is needed to improve the growth condition of the Co layer, and we see improved transmission of supercurrent in our Josephson junctions for these samples.

To support this interpretation of our data consider the asymmetric structures in FIG. 4. It is reasonable to assume that a seed layer preceding Co growth is most important to improve the growth condition and atomic structure of the Co. FIG. 4 shows that indeed a higher is only observed in the asymmetric sample with the bottom Pt(0.5) interlayer only. It is interesting to note, however, that to recover the highest observed in this work requires both interlayers, and that a decrease in is observed for asymmetric samples with only the top Pt(0.5) interlayer. These observations allude to some importance of having uniform strain across the Josephson junction, but this hypothesis requires further study.

We next need to understand why as the Pt layer is made thicker, the enhanced transmission of supercurrent decreases towards the value with no Pt interlayer (FIG. 2). Consider what happens to the number of Cooper pairs in the junction as the Pt interlayer is made thicker. This is described by the S–N proximity effect. For a metal such as Cu with a very long electron mean free path, the lengthscale of this proximity effect can be very long Dubos et al. (2001). Therefore in Cu interlayer samples there is little loss of Cooper pairs from the S–N proximity effect. Pt has a much shorter electron mean free path, and additionally gains a magnetic moment by proximity to Co Geissler et al. (2001); Rowan-Robinson et al. (2017). Both of these are unfavorable to Cooper pairs and will contribute to being short. Any gain in supercurrent transmission from the addition of Pt interlayers will have to compete with this short and the loss of Cooper pairs to the S–N proximity effect. Hence as the Pt interlayer is grown thicker, there are less Cooper pairs in the junction to contribute to the critical current.

Finally, we discuss our experiment in the context of theoretical predictions. Bergeret et al. consider singlet-triplet conversion in the presence of Rashba and/or Dresselhaus spin-orbit coupling Bergeret and Tokatly (2014). They provide a criteria for generating the LRTC, that the spin-orbit vector is not parallel to the exchange field (). They obtain the spin-orbit vector in equation (67) of their work, which we simplify here for a metallic system (such as ours) with finite Rashba () and zero Dresselhaus () contribution to the spin-orbit coupling as Bergeret and Tokatly (2014)


where is known in the literature as the Rashba constant and is the vector of the Pauli matrices. In other words, the spin-orbit vector has components (). If the direction of the exchange field is perpendicular to the plane, (, ), then there is no LRTC as the spin-orbit vector is parallel to the exchange field. Equally if and (in-plane magnetization), then there is no LRTC by the same logic. If and at least one of or are non-zero, then the spin-orbit vector () has a component perpendicular to the exchange field (), hence the LRTC can be created.

To address this limitation, Bergeret et al. propose performing the experiment with a magnetization in-plane F layer fabricated into a current in-plane (lateral) Josephson junction, which they show can recover LRTC generation Bergeret and Tokatly (2014). Lateral Josephson junctions containing half-metals are well established Keizer et al. (2006); Anwar et al. (2010, 2012), however substituting transition metal F layers into this geometry has proved experimentally difficult. Single crystal Co nanowires contacted by W electrodes display zero resistivity Wang et al. (2010), and recently a Josephson current has been passed laterally through Py disks when the separation between S electrodes is very small Lahabi et al. (2017). Polycrystalline Co wires show promise, however a zero resistance state is not observed Kompaniiets et al. (2014a, b). Jacobsen et al. suggest alternatively to use the current perpendicular-to-plane geometry (employed in this work) with a ferromagnetic alloy which has both in and out-of-plane magnetization components together with a source of spin-orbit coupling Jacobsen et al. (2016). This is closer to our experiment, where a small out-of-plane remanent magnetization is observed in FIG 1 (d). This out-of-plane anisotropy, however, is most likely limited to the Co/Pt interface as our samples have predominant in-plane anisotropy. This may explain why LRTC generation in our samples appears (at best) to be very poor. In future works we will replace the Co/Ru/Co SAF with a multilayer such as [Pd/Co], [Pt/Co], or [Ni/Co] where careful engineering of the layer thicknesses can promote the required canted magnetization Shepley et al. (2015); Glick et al. (2017); Gingrich et al. (2012).

Vi Conclusions

The major conclusions of this work may be summarized as follows. The growth of 0.5 nm Pt interlayers in Josephson junctions containing Co/Ru/Co ferromagnetic layers significantly enhances the transmission of supercurrent through the junction. The origin of this enhanced transmission is believed to be primarily from the fcc Pt being an effective seed layer for the growth of Co. Although most of our junctions displayed a supercurrent decay length consistent with singlet superconductivity, for the thickest Co layers a small residual supercurrent is present which may have a longer decay length. This small residual supercurrent is the best evidence in this work for spin-obit coupling mediating singlet-triplet conversion.

The data associated with this paper are openly available from the University of Leeds data repository Not (b).

We thank G. Burnell and F. S. Bergeret for many helpful discussions, B. Bi for help with fabrication using the Keck Microfabrication Facility, and R. Loloee, J. Glick and V. Aguilar for assistance with sample growth, fabrication, magnetometry and transport measurements particularly in the early stages of this project. This work was supported by the Marie Skłodowska-Curie Action “SUPERSPIN” (grant number: 743791).


  1. M. Eschrig, Phys. Today 64, 43 (2011).
  2. J. Linder and J. W. A. Robinson, Nat. Phys. 11, 307 (2015).
  3. M. Eschrig, Rep. Prog. Phys 78, 104501 (2015).
  4. C. Bell, G. Burnell, C. W. Leung, E. J. Tarte, D.-J. Kang,  and M. G. Blamire, Appl. Phys. Lett. 84, 1153 (2004).
  5. E. Goldobin, H. Sickinger, M. Weides, N. Ruppelt, H. Kohlstedt, R. Kleiner,  and D. Koelle, Appl. Phys. Lett. 102, 242602 (2013).
  6. B. Baek, W. H. Rippard, S. P. Benz, S. E. Russek,  and P. D. Dresselhaus, Nat. Commun. 5, 3888 (2014).
  7. B. M. Niedzielski, S. G. Diesch, E. C. Gingrich, Y. Wang, R. Loloee, W. P. Pratt,  and N. O. Birge, IEEE Trans. Appl. Supercond. 24, 1 (2014).
  8. E. C. Gingrich, B. M. Niedzielski, J. A. Glick, Y. Wang, D. L. Miller, R. Loloee, W. P. Pratt,  and N. O. Birge, Nat. Phys. 12, 564 (2016).
  9. B. M. Niedzielski, T. J. Bertus, J. A. Glick, R. Loloee, W. P. Pratt,  and N. O. Birge, Phys. Rev. B 97, 024517 (2018).
  10. J. A. Glick, S. Edwards, D. Korucu, V. Aguilar, B. M. Niedzielski, R. Loloee, W. P. Pratt, N. O. Birge, P. G. Kotula,  and N. Missert, Phys. Rev. B 96, 224515 (2017).
  11. I. Dayton, T. Sage, E. Gingrich, M. Loving, T. Ambrose, N. Siwak, S. Keebaugh, C. Kirby, D. Miller, A. Herr, Q. Herr,  and O. Naaman, IEEE Magn. Lett. (At Press)  (2018), 10.1109/LMAG.2018.2801820.
  12. P. Dubos, H. Courtois, B. Pannetier, F. K. Wilhelm, A. D. Zaikin,  and G. Schön, Phys. Rev. B 63, 064502 (2001).
  13. A. I. Buzdin, Rev. Mod. Phys. 77, 935 (2005).
  14. P. Fulde and R. A. Ferrell, Phys. Rev. 135, A550 (1964).
  15. A. Larkin and Y. Ovchinnikov, Sov. Phys. JETP 20 (1965).
  16. A. Buzdin, L. Bulaevskii,  and S. Panyukov, JETP Lett 35, 178 (1982).
  17. E. A. Demler, G. B. Arnold,  and M. R. Beasley, Phys. Rev. B 55, 15174 (1997).
  18. V. A. Oboznov, V. V. Bol’ginov, A. K. Feofanov, V. V. Ryazanov,  and A. I. Buzdin, Phys. Rev. Lett. 96, 197003 (2006).
  19. J. W. A. Robinson, S. Piano, G. Burnell, C. Bell,  and M. G. Blamire, Phys. Rev. Lett. 97, 177003 (2006).
  20. T. S. Khaire, W. P. Pratt,  and N. O. Birge, Phys. Rev. B 79, 094523 (2009).
  21. F. S. Bergeret, A. F. Volkov,  and K. B. Efetov, Phys. Rev. Lett. 86, 4096 (2001).
  22. F. S. Bergeret, A. F. Volkov,  and K. B. Efetov, Rev. Mod. Phys. 77, 1321 (2005).
  23. R. Keizer, S. Goennenwein, T. Klapwijk, G. Miao, G. Xiao,  and A. Gupta, Nature 439, 825 (2006).
  24. I. Sosnin, H. Cho, V. T. Petrashov,  and A. F. Volkov, Phys. Rev. Lett. 96, 157002 (2006).
  25. J. W. A. Robinson, J. D. S. Witt,  and M. G. Blamire, Science 329, 59 (2010).
  26. M. S. Anwar, F. Czeschka, M. Hesselberth, M. Porcu,  and J. Aarts, Phys. Rev. B 82, 100501 (2010).
  27. T. S. Khaire, M. A. Khasawneh, W. P. Pratt,  and N. O. Birge, Phys. Rev. Lett. 104, 137002 (2010).
  28. M. S. Anwar, M. Veldhorst, A. Brinkman,  and J. Aarts, Appl. Phys. Lett. 100, 052602 (2012).
  29. W. M. Martinez, W. P. Pratt,  and N. O. Birge, Phys. Rev. Lett. 116, 077001 (2016).
  30. F. Konschelle, Eur. Phys. J. B 87, 119 (2014).
  31. F. S. Bergeret and I. V. Tokatly, Phys. Rev. Lett. 110, 117003 (2013).
  32. F. S. Bergeret and I. V. Tokatly, Phys. Rev. B 89, 134517 (2014).
  33. S. H. Jacobsen, J. A. Ouassou,  and J. Linder, Phys. Rev. B 92, 024510 (2015).
  34. S. H. Jacobsen, I. Kulagina,  and J. Linder, Sci. Rep. 6, 23926 (2016).
  35. 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 476, 189 (2011).
  36. P. Haazen, E. Murè, J. Franken, R. Lavrijsen, H. Swagten,  and B. Koopmans, Nat. Mater. 12, 299 (2013).
  37. A. Hrabec, N. A. Porter, A. Wells, M. J. Benitez, G. Burnell, S. McVitie, D. McGrouther, T. A. Moore,  and C. H. Marrows, Phys. Rev. B 90, 020402 (2014).
  38. N. Banerjee, J. Ouassou, Y. Zhu, N. Stelmashenko, J. Linder,  and M. Blamire, arXiv preprint arXiv:1709.03504  (2017).
  39. Y. V. Fominov, A. A. Golubov, T. Y. Karminskaya, M. Y. Kupriyanov, R. Deminov,  and L. R. Tagirov, JETP letters 91, 308 (2010).
  40. N. Satchell, J. D. S. Witt, M. G. Flokstra, S. L. Lee, J. F. K. Cooper, C. J. Kinane, S. Langridge,  and G. Burnell, Phys. Rev. Applied 7, 044031 (2017).
  41. Y. Wang, W. P. Pratt,  and N. O. Birge, Phys. Rev. B 85, 214522 (2012).
  42. C. D. Thomas, M. P. Ulmer,  and J. B. Ketterson, J. Appl. Phys. 84, 364 (1998).
  43. H. Kohlstedt, F. König, P. Henne, N. Thyssen,  and P. Caputo, J. Appl. Phys. 80, 5512 (1996).
  44. C. Klose, T. S. Khaire, Y. Wang, W. P. Pratt, N. O. Birge, B. J. McMorran, T. P. Ginley, J. A. Borchers, B. J. Kirby, B. B. Maranville,  and J. Unguris, Phys. Rev. Lett. 108, 127002 (2012).
  45. N. Nakajima, T. Koide, T. Shidara, H. Miyauchi, H. Fukutani, A. Fujimori, K. Iio, T. Katayama, M. Nývlt,  and Y. Suzuki, Phys. Rev. Lett. 81, 5229 (1998).
  46. P. Shepley, A. Rushforth, M. Wang, G. Burnell,  and T. Moore, Sci. Rep. 5, 7921 (2015).
  47. (a), see Supplemental Material at [URL will be inserted by publisher] for: typical junction I–V and Fraunhofer characteristic along with collated , and data obtained on the  nm and  nm samples.
  48. M. A. Khasawneh, W. P. Pratt,  and N. O. Birge, Phys. Rev. B 80, 020506 (2009).
  49. A. A. Jara, C. Safranski, I. N. Krivorotov, C.-T. Wu, A. N. Malmi-Kakkada, O. T. Valls,  and K. Halterman, Phys. Rev. B 89, 184502 (2014).
  50. P. V. Leksin, N. N. Garif’yanov, A. A. Kamashev, Y. V. Fominov, J. Schumann, C. Hess, V. Kataev, B. Büchner,  and I. A. Garifullin, Phys. Rev. B 91, 214508 (2015).
  51. T. E. Mitchell, Y. C. Lu, A. J. Griffin, M. Nastasi,  and H. Kung, J. Am. Ceram. Soc. 80, 1673 (1997).
  52. H. Kung, Y.-C. Lu, A. J. Griffin., M. Nastasi, T. E. Mitchell,  and J. D. Embury, Appl. Phys. Lett. 71, 2103 (1997).
  53. H. Geng, J. W. Heckman, W. P. Pratt, J. Bass, F. J. Espinosa, S. D. Conradson, D. Lederman,  and M. A. Crimp, J. Appl. Phys. 86, 4166 (1999).
  54. J. Geissler, E. Goering, M. Justen, F. Weigand, G. Schütz, J. Langer, D. Schmitz, H. Maletta,  and R. Mattheis, Phys. Rev. B 65, 020405 (2001).
  55. R. Rowan-Robinson, A. Stashkevich, Y. Roussigné, M. Belmeguenai, S.-M. Chérif, A. Thiaville, T. Hase, A. Hindmarch,  and D. Atkinson, Sci. Rep. 7, 16835 (2017).
  56. J. Wang, M. Singh, M. Tian, N. Kumar, B. Liu, C. Shi, J. Jain, N. Samarth, T. Mallouk,  and M. Chan, Nature Physics 6, 389 (2010).
  57. K. Lahabi, M. Amundsen, J. A. Ouassou, E. Beukers, M. Pleijster, J. Linder, P. Alkemade,  and J. Aarts, Nat. Commun. 8, 2056 (2017).
  58. M. Kompaniiets, O. Dobrovolskiy, C. Neetzel, F. Porrati, J. Brötz, W. Ensinger,  and M. Huth, Applied Physics Letters 104, 052603 (2014a).
  59. M. Kompaniiets, O. Dobrovolskiy, C. Neetzel, E. Begun, F. Porrati, W. Ensinger,  and M. Huth, Journal of Applied Physics 116, 073906 (2014b).
  60. E. C. Gingrich, P. Quarterman, Y. Wang, R. Loloee, W. P. Pratt,  and N. O. Birge, Phys. Rev. B 86, 224506 (2012).
  61. (b),
Comments 0
Request Comment
You are adding the first comment!
How to quickly get a good reply:
  • Give credit where it’s due by listing out the positive aspects of a paper before getting into which changes should be made.
  • Be specific in your critique, and provide supporting evidence with appropriate references to substantiate general statements.
  • Your comment should inspire ideas to flow and help the author improves the paper.

The better we are at sharing our knowledge with each other, the faster we move forward.
The feedback must be of minimum 40 characters and the title a minimum of 5 characters
Add comment
Loading ...
This is a comment super asjknd jkasnjk adsnkj
The feedback must be of minumum 40 characters
The feedback must be of minumum 40 characters

You are asking your first question!
How to quickly get a good answer:
  • Keep your question short and to the point
  • Check for grammar or spelling errors.
  • Phrase it like a question
Test description