Formation and anisotropic magnetoresistance of Co/Pt nano-contacts through aluminum oxide barrier
FFormation and anisotropic magnetoresistance of Co/Pt nano-contacts throughaluminum oxide barrier
Muftah Al-Mahdawi a) and Masashi Sahashi Department of Electronic Engineering, Tohoku University, Sendai 980-8579,Japan (Dated: 14 November 2018)
We report on the observation of anisotropic magnetoresistance (AMR) in verticalasymmetric nano-contacts (NCs) made through AlO x nano-oxide layer (NOL) formedby ion-assisted oxidation method in the film stack of Co/AlO x -NOL/Pt. Analysis ofNC formation was based on in situ conductive atomic force microscopy and transmis-sion electron microscopy. Depending on the purity of NCs from Al contamination,we observed up to 29% AMR ratio at room temperature. a) Electronic mail: [email protected] a r X i v : . [ c ond - m a t . m e s - h a ll ] D ec allistic transport either through a tunneling barrier or nano-contacts (NCs), wherecontact diameter is much smaller than electron’s mean free path, offers a lot of informationabout electron’s local Density Of States (DOS) near Fermi level. In systems with Spin-OrbitCoupling (SOC), local DOS dependence on local magnetization direction leads to the ef-fects of tunneling anisotropic magnetoresistance (TAMR), for tunneling conduction, andnano-contact AMR (NC-AMR), in the case of ballistic transport through nanometer-sizedmetallic contacts. In principle, effect of DOS anisotropy should be observable up to roomtemperature but it was not, either due to randomly trapped charges at the barrier, or dueto instability of atoms in planar NCs. Vertical NCs between films provide a more stablealternative to planar NCs, but fabrication of NCs in this geometry is difficult with currentlithography techniques. Bombarding the surface of a thin Al layer with low-energy (65eV) Ar + -ions in the presence of oxygen atmosphere ( ≈ × − Pa partial pressure), theso-called Ion-Assisted Oxidation (IAO) method, can be used to form an AlO x Nano-Oxide-Layer (NOL) with multiple holes 1–2 nm in diameter. IAO method has been used to makeferromagnetic NCs between the upper and bottom layers; if short oxidation times are used(20–40 sec).
IAO method is relatively easy compared to lithography and the resultingNCs are stable even at elevated temperatures, but the formation mechanism is inherently arandom bottom-up process.By placing an interface (Co/Pt) with a strong SOC adjacent to a tunneling barrier, Ref. 2could increase TAMR ratio two-folds. In this paper, we report on the formation of Co/Ptasymmetric NCs through AlO x -NOL, and the resulting AMR behavior.To study the formation of AlO x -NOL over Co layer by IAO, we used in situ atomic force mi-croscopy in conductive mode (cAFM) with 10 − -Pa base pressure to characterize the surfacetopography and current profile of AlO x -NOL. A Si tip coated with heavily-doped diamond(CDT-CONTR) from NANOSENSORS TM was used. The film design was (thickness in nm):thermally-oxidized silicon substrate/electrode layer(Ta (5)/Cu (285)/Ta (30)/chemical-mechanical polishing)/Ta (10)/Ru (2)/Co (2)/Al (1.3)/IAO 30 sec exposure time. Afterdeposition, the sample was transferred in vacuum to cAFM chamber. Image analysis wasdone by extension scripts to Gwyddion software. Electrode layer/Ta (10)/Ru (2)/Co (2)/Al(1.3)/IAO 30 sec/Pt (7) was used for transport measurement. Co, Pt, Ru, and Ta weredeposited by dc magnetron sputtering. Ion-beam sputtering was used for Al deposition, withan assist-ion-gun used for IAO at the same chamber. All chambers are connected through2n ultra-high-vacuum (10 − -Pa base pressure) transfer tube. Current-In-Plane-Tunneling(CIPT) method was used to measure Resistance-Area (RA) product of unpatterned films.Current-Perpendicular-to-Plane (CPP) geometry pillars with circular cross-section (180–400nm in diameter) were patterned by Ar + ion-milling and lithography techniques, where RAwas also deduced from the slope of CPP pillars dc resistance vs. area inverse (1/A) plot.Cross-sectional high-resolution transmission electron micrographs (TEM) were obtained asan additional indication of the shape and material composition of NCs. Zero-bias differ-ential resistance of the NCs ( R ) dependence on magnetization direction was measured byapplying a large magnetic field and varying the angle ( θ ) between normal to film plane andthe applied field, at 300-K and 5-K temperatures. Also R dependence on temperature inthe heating direction was measured for 0-T and 0.9-T magnetic fields applied out-of-plane,after cooling from 350 K at the same magnetic field used in measurement. Angles of 0 ◦ and90 ◦ corresponds to out-of-plane and in-plane applied field directions, respectively. AMRratio is defined as ( R ( θ ) − R (0 ◦ )) /R (0 ◦ ).cAFM current and topography scan images of 200 nm ×
200 nm area at 1.8-V tip-to-samplebias are shown in Figs. 1(a),(b). Several high-current paths were observed [sample profilesare in the inset of Fig. 1(a)], which correlate with the valleys between the grains of AlO x -NOL [Fig. 1(b)]. This indicates that holes form at the grain boundaries during oxidationprocess, in agreement with previous cAFM study of AlO x -NOL over Fe Co , and thatthe top ferromagnetic layer determines MR performance in Fe x Co − x /AlO x -NOL/Fe y Co − y structures. To estimate the number of conductive paths and mean diameter, current imagewas turned into a binary image by thresholding; then number, occupancy, and mean diam-eter of paths were computed for each current threshold level [Fig. 1(c)]. A dilation-erosionfilter was used to eliminate single-pixel noise in current image beforehand. The observedconductive paths at higher threshold levels are due to metallic contacts, whereas at lowerlevels noise from tunneling through AlO x prevails. A plateau in number of contacts andoccupancy vs. threshold current at 1.6–1.8 nA, attributed to a transition in counting frommetallic contacts to tunneling noise, can be used to deduce characteristics of NCs. Numberdensity, area occupancy, and mean diameter are 300 µ m − , 0.04%, and 1.25 nm, respectively.Semiconducting tips over metal films form a Schottky barrier, with junction built-in poten-tial ( V bi ) being the difference between work functions of tip material and sample surfacemetal. The tip was calibrated by measuring the I-V curves over Al and Co single 20-nm3lms, and rectifying I-V characteristic was confirmed. Average V bi values were found todiffer by 0.79 V between Co and Al. This corresponds well with the difference betweenwork-functions of Al and Co ( ≈ x -NOL. The values of V bi showed separation into twogroups, one group had values closer to Co and the other closer to Al [Fig. 1(d)]. We implythat the conductive paths are metals with a random distribution of Co and Al. Thus, bycapping Co/AlO x -NOL with Pt, there will be a random distribution of Co/Al/Pt and Co/Ptnano-contacts. TEM images of Co/AlO x -NOL/Pt showed direct connections between Ptand Co layers through AlO x at few places [Fig. 1(e)]. The crystal structure and orientationof NC region is same as top Pt layer and different from Co under-layer. In support of expec-tations from cAFM results, Pt filled most of the contact volume with a similar NC diameterof 1.8 nm. Due to low atomic numbers of Al and O, it was not possible to distinguish Alfrom AlO x in TEM image. AlO x thickness ranged 0.75–0.9 nm which is thinner than the1.3-nm expected thickness of Al. The effective electrical thickness of AlO x was 0.8 nm,determined from BDR model fitting to I-V curves of long-time oxidized junctions (IAO180 sec). The reason for this discrepancy is due to further consideration.RA products of Co/AlO x -NOL/Pt films from measurement of CIPT and CPP-pillars resis-tance were 100–150 and 81 mΩ · µ m , respectively. Also, resistance increased with increasingtemperature and bias voltage (not shown), indicating metallic contact formation throughAlO x -NOL. This low RA, compared with FeCo-NCs, is from effect of Pt on AlO x -NOL.When Pt capping was replaced with Co, RA increased to 330 mΩ · µ m , and resistanceincreased with decreasing temperature. Pt is not inert in the presence of Co and AlO x , the effect of Pt on NCs would be to reduce the oxygen impurities and enhance metalliccontact formation.CPP pillars from same substrate showed characteristics that could be grouped into twogroups, from which the best cases are reported next. We attribute this grouping to preva-lence of either Al or Pt NCs in the area where the CPP pillar was patterned. In pillarA, R − θ curve had a low AMR ratio of 0.04% at 300 K, which increased to 0.08% at 5K under 8-T magnetic field [Fig. 2(a)]. Contrarily, pillar B had up to 28.6% AMR ratioat 300 K [Fig. 2(b)]. The AMR ratio of pillar B increased with increasing applied field[insets of Fig. 2(b)] and decreased with increasing temperature at high temperatures (T >
170 K) [Fig. 2(c)]. The qualitative proportionality of AMR ratio to
B/T can be at-4ributed to Curie-Weiss-like dependency of induced magnetization in Pt at NCs. At thecontact area between Pt and Co, the induced magnetization can show a superparamegneticbehavior. Possible mechanism for superparamagnetism is the presence of fluctuating mo-ments in Pt-NCs exchange coupled to Co layer, or diffused Co atoms, in addition toreduced “magnetic particle” size at NC. We attribute the large AMR ratio of pillar B tothe large DOS anisotropy of Co/Pt interface. In the case of NC geometry, the reduceddimensionality increases the magnitude of SOC, which would increase the DOS anisotropymore, compared to extended electrodes. We expect that the presence of the weak-SOC Alcontamination at NCs lowers the DOS anisotropy and introduces shunting paths againstpurer Co/Pt NCs, thus lowering AMR ratio in the group of pillar A [Fig. 2(d)].At 150–160 K, pillar B (and other pillars having large AMR ratio > R − θ curveat 5 K and 8 T has a widened flattening around 90 ◦ and 270 ◦ . This indicates an inducedin-plane anisotropy preventing the NC magnetic moments from aligning with applied fieldleading to an apparent additional fourfold-symmetry dependence and suppressing the AMRratio down to 0.7% at 5 K[Fig. 2(b)]. Due to small magnitude of Pt induced magnetization(0.2–0.6 µ B ), even a small induced anisotropy results in a large anisotropy field. Thetemperature of this anomaly coincides well with the blocking temperature ( T B ) of CoO inNOL systems. Formation of CoO (or Co-O bonds) beneath AlO x and around NCs duringoxidation process is within expectation [Fig. 2(d)]. We propose that below T B the exchangecoupling between Pt and CoO induces such an anisotropy. The amplitude and nature ofcoupling is difficult to determine in this report due to unknown magnetization of Pt-NCsand coupling of Pt moments to both Co and CoO, which can lead to non-trivial couplingconfigurations. The observed AMR cannot be attributed to the bulk AMR of 2-nm Co layer, as itwould amount to a 0.003% resistance change in the 81-mΩ µ m pillar RA (AssumingRA Co = 124 . µ Ω µ m , and AMRR Co = 1 . x layer with longer IAO time was used to separate Co and Pt layers completely (Co 2/Al1.3/IAO 180 sec/Al 1.3/IAO 240 sec/Pt 7, thickness in nm). CPP-RA was 10 kΩ · µ m ,and R − θ curve showed similar behavior ( R (0 ◦ ) < R (90 ◦ )) to sample B (Co/AlO x /Pt) of5ef. 2 with 0.06% tunneling AMR ratio at 3K.In conclusion, we characterized the formation of randomly distributed Al-NCs or Pt-NCswith an interface to Co through AlO x -NOL, using the bottom-up process of IAO. Because oflarge anisotropic DOS of Pt/Co system, we were able to observe up to 29% NC-AMR ratioat room temperature. Due to randomness in NC formation mechanism through AlO x -NOL,mixed results were obtained from CPP pillars made of the same film because of the presenceof Al contamination. The expected presence of CoO ( T B = 150 K in NOL structure) re-sulted in an induced in-plane anisotropy, suppressing AMR ratio at low temperature. Moreclarification is needed for the role of coupling modulation in Co-CoO/Pt NC system. Thenon-scattering transport of electrons through NCs allows for effective application of highelectric fields within the metal, the prospects of which are under investigation. The thermalstability of vertical NCs, adding the relative ease of making NCs by IAO method, promisereal-world applications, but one major hurdle is improving the control of NCs formationwith a higher yield, which is an ongoing research effort. ACKNOWLEDGMENTS
M. A. acknowledges financial support from Ministry of Education, Culture, Sports, Sci-ence and Technology (MEXT). Authors thank Dr. Y. Shiokawa and Dr. T. Nozaki for theirhelp and comments.
REFERENCES C. Gould, C. R¨uster, T. Jungwirth, E. Girgis, G. M. Schott, R. Giraud, K. Brunner,G. Schmidt, and L. W. Molenkamp, Phys. Rev. Lett. , 117203 (2004). B. G. Park, J. Wunderlich, D. A. Williams, S. J. Joo, K. Y. Jung, K. H. Shin, K. Olejn´ık,A. B. Shick, and T. Jungwirth, Phys. Rev. Lett. , 087204 (2008). K. I. Bolotin, F. Kuemmeth, A. N. Pasupathy, and D. C. Ralph, Nano Lett. , 123 (2006). A. B. Shick, F. M´aca, J. Maˇsek, and T. Jungwirth, Phys. Rev. B , 024418 (2006). K. S. Ralls, R. A. Buhrman, and R. C. Tiberio, Appl. Phys. Lett. , 2459 (1989). M. Takagishi, H. N. Fuke, S. Hashimoto, H. Iwasaki, S. Kawasaki, R. Shiozaki, andM. Sahashi, J. Appl. Phys. , 07B725 (2009).6
H. N. Fuke, S. Hashimoto, M. Takagishi, H. Iwasaki, S. Kawasaki, K. Miyake, and M. Sa-hashi, IEEE Trans. Magn. , 2848 (2007). Y. Shiokawa, M. Shiota, Y. Watanabe, T. Otsuka, M. Doi, and M. Sahashi, IEEE Trans.Magn. , 3470 (2011). D. Neˇcas and P. Klapetek, Cent. Eur. J. Phys. , 181 (2011). K. Kishi, Y. Shiokawa, H. Watanabe, Z. Zheng, and M. Sahashi , presented at IEEEInternational Magnetics Conference, Vancouver, British Columbia, Canada, 2012 (unpub-lished). W. Brinkman, R. Dynes, and J. Rowell, J. Appl. Phys. , 1915 (1970). F. Lu, M. L. Newhouse, R. Dieckmann, and J. Xue, Solid State Ionics , 187 (1995). Z. Celinski, B. Heinrich, J. F. Cochran, W. B. Muir, A. S. Arrott, and J. Kirschner, Phys.Rev. Lett. , 1156 (1990). J. Crangle and W. R. Scott, J. Appl. Phys. , 921 (1965). B. Doudin and M. Viret, J. Phys.: Condens. Matter , 083201 (2008). J. Geissler, E. Goering, M. Justen, F. Weigand, G. Sch¨utz, J. Langer, D. Schmitz,H. Maletta, and R. Mattheis, Phys. Rev. B , 020405 (2001); M. Suzuki, H. Mu-raoka, Y. Inaba, H. Miyagawa, N. Kawamura, T. Shimatsu, H. Maruyama, N. Ishimatsu,Y. Isohama, and Y. Sonobe, ibid . , 054430 (2005). M. Sahashi, K. Sawada, H. Endo, M. Doi, and N. Hasegawa, IEEE Trans. Magn. , 3668(2007). J. Slonczewski, J. Magn. Magn. Mater. , 13 (1995); H. Fukuzawa, K. Koi, H. Tomita,H. N. Fuke, H. Iwasaki, and M. Sahashi, J. Appl. Phys. , 6684 (2002).7 IG. 1. (a) cAFM current image of AlO x -NOL over Co thin-film taken. Sample line profiles showninset. (b) Superposition of current image (in light blue) over topography profile. (c) Number andoccupancy of conductive paths vs. thresholding current level. A plateau is seen at 1.6–1.8 nA.(d) Distribution of V bi measured from I-V curves of a semiconducting tip over single-film Co andAl, and Co/AlO x -NOL conductive paths. Conductive paths divided into two groups: Co-like andAl-like. (e) TEM image of Co/AlO x -NOL/Pt showed direct connections between Pt and Co layersthrough AlO x . C u rr e n t [ n A ] Scan [nm]
Line scan:
50 nm nA V b i [ V ] Co Al
Co/Al/IAO 30" Al-like V [V] I [ n A ] − Co-like nm
50 nm (a) (b)(c) (d) N u m b e r o f c on t ac t s O cc up a n cy [ × . % ] Threshold current [nA] x CoPt (e) IG. 2. (a) Pillar A (180 nm in diameter) with Al-dominant NCs had a very low AMR ratio at300 K which increased at 5 K. (b) For pillar B (200 nm in diameter) with Pt-dominant NCs, AMRratio increased substantially at 300 K but decreased at 5 K with widened flattening around 90 ◦ and 270 ◦ indicating increased in-plane anisotropy (insets are dependence of AMR ratio on appliedfield). (c) AMR ratio had an anomaly at 150–160K, which can be attributed to onset of couplingwith CoO. (d) A schematic showing the presence of both Al-NCs and Pt-NCs (the arrow refers toinduced magnetization in Pt-NCs but the direction does not represent the actual configuration),with CoO present beneath AlO x . R A [ m Ω ∙ μ m ] [ R ( . T ) - R ( T ) ] / R ( T ) [ % ] T [K] θ = 0 ∘ CPP pillar B
PtCo
AlO x Al CoO (d)
CPP pillar B RA [ m Ω · μ m ] θ [ ∘ ]
300 K
030 2 4 6 8 M R [ % ] B [T]
01 2 4 6 8 M R [ % ] B [T] (c) (b)(a)
CPP pillar A RA [ m Ω · μ m ] θ [ ∘ ]
300 K, AMRR 0.04%5 K, 0.08%8 T