Tunneling anisotropic magnetoresistance in multilayer-(Co/Pt)/AlOx/Pt structures

0.45 MB·

by  B. G. Park

Tunneling anisotropic magnetoresistance in multilayer-(Co/Pt)/AlO /Pt structures
x
B. G. Park,1 J. Wunderlich,1,2 D. A. Williams,1 S. J. Joo,3 K. Y. Jung,3
K. H. Shin∗,3 K. Olejn´ık,2 A. B. Shick,4 and T. Jungwirth∗2,5
1Hitachi Cambridge Laboratory, Cambridge CB3 0HE, United Kingdom
2Institute of Physics ASCR, v.v.i., Cukrovarnick´a 10, 162 53 Praha 6, Czech Republic
3Nano Device Research Center, Korea Institute of Science and Technology, Seoul 136-792, Korea
4Institute of Physics ASCR, v.v.i., Na Slovance 2, 182 21 Praha 8, Czech Republic
5School of Physics and Astronomy, University of Nottingham, Nottingham NG7 2RD, United Kingdom
(Dated: February 2, 2008)
8
0 We report observations of tunneling anisotropic magnetoresitance (TAMR) in vertical tunnel
0 deviceswithaferromagneticmultilayer-(Co/Pt)electrodeandanon-magneticPtcounter-electrode
2 separated by an AlOx barrier. In stacks with the ferromagnetic electrode terminated by a Co film
the TAMR magnitude saturates at 0.15% beyond which it shows only weak dependence on the
n
magneticfieldstrength,biasvoltage, andtemperature. Forferromagnetic electrodes terminatedby
a
J two monolayers of Pt we observe order(s) of magnitude enhancement of the TAMR and a strong
dependence on field, temperature and bias. Discussion of experiments is based on relativistic ab
8
initiocalculationsofmagnetizationorientationdependentdensitiesofstatesofCoandCo/Ptmodel

] systems.
i
c
PACSnumbers: 85.75.Mm,75.45.+j,75.50.Cc
s
-
l
r
t Anisotropic magnetoresistance (AMR) sensors re- nel junction with a ferromagnetic metal electrode has
m
placed in the early 1990s classical magneto-inductive recently been reported [9] in an epitaxial Fe/GaAs/Au
t. coils in hard-drive readheads launching the era of spin- stack. TheobservedTAMRinthisstructureisrelatively
a
tronics. Their utility has, however, remained limited small,bellow0.5%,consistentwiththeweakSO-coupling
m
partly because the response of these ferromagnetic resis- inFe. Inthis paperwe presenta study ofverticaltunnel
-
d tors to changes in magnetization orientation originates devices in which the ferromagnetic electrode comprises
n from generically subtle spin-orbit (SO) interaction ef- alternatingCoandPtfilms. Webuildupontheextensive
o fects [1]. Currently widely used giant magnetoresistance literature[10,11,12,13,14,15]onferromagnetic/heavy-
c [2] and tunneling magnetoresistance(TMR) [3] elements element transition metal multilayers in which large and
[
comprising (at least) two magnetically decoupled ferro- tunablemagneticanisotropiesaregeneratedattheinter-

1 magnetic layers provided a remarkably elegant way of facesbycombinedeffectsofinducedmomentsandstrong
v
tying the magnetoresistance response directly to the fer- SO-coupling. Weshowthatplacingtheseinterfacesadja-
2
romagnetic exchange splitting of the carrier bands with- cent to the tunnel barrier opens a viable route to highly
9
1 out involving SO-coupling. Large magnetoresistances in sensitive metal TAMR structures.
1 thesedevicesare,nevertheless,obtainedattheexpenseof
. a significantly increased structure complexity, necessary
1
0 to guarantee independent and different magnetization Sample A
Pt(5nm) Sample B
8 switchingcharacteristicsandspin-coherenceoftransport
0 between the ferromagnetic layers. PAt(0lO.5xn(2mn)m)
:
v Studies of AMR effects [4, 5, 6, 7] in ferromagnetic Co(1nm)/Pt(1nm)/Co(1nm)
i
X semiconductor tunneling devices showed that AMR re- Pt(10nm)/Ta(5nm)
SiO SiO
r sponse can in principle be huge and richer than TMR, 2 2
a with the magnitude and sign of the magnetoresistance 0.15
0 B
dependent on the magnetic field orientation and electric
%) 0.10
fields. Subsequenttheoreticalworkpredicted[8]thatthe R ( -5
tunnelingAMR(TAMR)effectisgenericinferromagnets AM M M 0.05
with SO-coupling, including the high Curie temperature T-10
transition metal systems. A detailed investigationof the -15 -0.1 B(T) 0.1 -0.1B(T) 0.1 0.00
TAMR is therefore motivated both by its intricate rela- 0 30 60 90 0 30 60 90
tivistic quantumtransportnatureandbyits potentialin !(deg.) !(d eg.)
FIG. 1: Upper panels: schematic layer structures of device
moreversatilealternativestocurrentTMRdeviceswhich
A (left) and device B (right). Lower panels: corresponding
will not require two independently controlled ferromag-
TAMR traces defined as (R(θ)−R(0))/R(0) at -5 mV bias
netic electrodes and spin-coherent tunneling.
and4K.Insets: SQUIDmagnetizationmeasurementsinout-
Experimental demonstration of the TAMR in a tun- of-plane magnetic fields.
2
Our tunneling devices, schematically illustrated in strongly influenced by the nature of the surface layers
Fig. 1, were grown by magnetron sputtering on a ther- of the ferromagnetic electrode [16, 17] we attribute the
mally oxidized Si wafer. The Ta/Pt seed layer was cho- large TAMR signal in sample A to the induced moment
sen to initiate growth of a textured Pt(111)/Co ferro- andstrongSO-couplinginthe two-monolayerPtfilmin-
magnetic film with a strong out-of-plane magnetocrys- serted between Co and the tunnel barrier. In sample B,
talline anisotropy. The tunnel barrier was fabricated the nearest Pt/Co interfaces to the barrier is coveredby
by plasma oxidation of 1.6 nm Al layer and the growth 1 nm (five-monolayer)Co film and therefore this tunnel-
wascompletedbysputteringanon-magneticPtcounter- ing device behaves effectively as a Co/barrier/normal-
electrode. Two types of multilayers are investigated: metal stack for which the magnetotransport anisotropy
samples of type A have the alternating sequence of Pt is expected to be comparably weak as in the previously
and Co layers beneath the AlO barrier terminated by studied Fe-based TAMR device [8, 9].
x
0.5nm(two-monolayer)PtfilmwhilesamplesBhavethis The distinct phenomenologies of samples A and B are
topPtfilmintheferromagneticelectrodeomitted. Pillar further detailed in Figs. 2 and 3. Fig. 2(a) shows field-
transport devices of 10-100 µm diameter were patterned sweep magnetoresistance measurements in sample A for
inthemultilayersusingphotolithographyandionetching θ =0and90◦atthreedifferentbiasvoltagesandthecor-
with corresponding resistances of the order ∼ 10−100 responding TAMR curves are plotted in Fig. 2(b). The
kΩ. Devices were mounted on a rotating sample holder initialonsetoftheTAMR,seenclearlyinthe10mVbias
allowing the application of magnetic fields up to 10 T measurementandassociatedwithrotationofthemagne-
at an arbitrary in-plane and out-of-plane angle. Four- tization from the out-of-plane easy-axis direction in the
point transport measurements were used to determine in-plane field sweep, is followed by a further dependence
the TAMR. oftheTAMRonthefieldstrengthathigherfieldswithno
signsofsaturationup to the maximum measuredfieldof

26 5 10T.Since appreciablemagnetoresistanceis observedin
=90 10mV both in-plane and out-of-plane field sweeps and the sign
25 0
of the magnetoresistance changes from positive to neg-
=0 (a) (b)
ative for different bias voltages we exclude the Lorentz
28
=0 0 T force origin of the TAMR in our thin tunnel-barrier de-
!)27 CoPt/AlOx/Pt A vices. Apart from the strong field and bias dependence
R (k26 -54 mKV %) -5 MR ( wweithalstoemobpseerravteurlear(gseeevairnisaettioinnoFfitgh.e2T).AMThResmeaogbnsietruvdae-
25 =90 R| (15 -10%) tions are unique to the samples of type A; samples of
B M
=0 TA 0 type B show only weak dependence of their TAMR on
| 0 30
24 T (K) 0 field strength, bias and temperature, as shown in Fig. 3.
=90 -10mV
23 -5 V (m V)
-10 -5 0 5 10-10 -5 0 5 10 -30 0 30 0.2
B (T) B (T) (a)
134.1 =90
RFI(G0).)/2R: ((0a))(Rbe)siosftasnamcepalendA(mb)eaTsAurMedRadte1fi0n,e-d5,aasn(dR-(1900◦m)−V !k)133.8 =0 0.1R (%)
bfoirasthaend-5atm4VKb.iIansseatn:dte1m0pTerfiaetludr.edependenceoftheTAMR R (133.5 MR0%].2 TAM
A[
133.2 T0.0 (b)
-10 B(T) 10
0.0
LowerpanelsofFig.1comparecharacteristicmagneti- -3 -2 -1 0 1 2 3 50 100
B (T) Temperature (K)
zation and magnetotransport anisotropy data measured
in samples A and B. Wide and square hysteresis loops FIG.3: (a)ResistanceandTAMR(inset)definedas(R(90◦)−
in perpendicular magnetic fields, shown in the insets, R(0)/R(0) of sample B at 10 mV bias and 4 K. (b) Temper-
confirm the presence of strong magnetic anisotropy with ature and bias dependenceof the TAMR at 10 T field.
out-of-plane easy axis in both devices. The anisotropic
magnetotransport in the two tunneling devices is, how- OurtheoreticaldiscussionoftheobservedTAMRphe-
ever,fundamentally different. The TAMR tracesplotted nomenologiesinthetwotypesofCo/Pttunnelingdevices
inFig.1aredefinedas(R(θ)−R(0))/R(0)andaretaken followsthe approachappliedpreviouslyto ferromagnetic
at 10 T magnetic field rotating from the perpendicular semiconductor TAMR structures [4]. The analysis is
◦
(θ =0)to the in-plane(θ =90 )direction. Inbothsam- basedoncalculatingdensityofstates(DOS)anisotropies
ples the TAMR has the expected uniaxialsymmetry but in the ferromagnetic electrode with respect to the ori-
the magnitude of the effect is enhanced by 2 orders of entation of the magnetic moment and assumes propor-
magnitude in sample A. As the transport characteristics tionality betweenthe DOS andthe differentialtunneling
of TMR or TAMR tunneling devices are expected to be conductance anisotropies.
3
The SO-coupled band structures are obtained within dependence on the energy are markedly different from
thelocalspindensityapproximationusingtherelativistic the calculations in Fig. 4.
version of the full-potential linearized augmented plane-
20
wavemethod (FP-LAPW)inwhich SOinteractionisin- 15 CoPt !=0 Co
!=90
cluded in a self-consistent second-variational procedure CoPt 10 D
V) O
[8, 18]. The magnetic force theorem [19] is used to eval- e S
1/10 0 /D
uchaatergteheanDdOsSpiannidseontrsoitpieys: csatalcrutilnatgedfrofmor stehlfe-cmonasgisnteetnict DOS ( Co -100OS (%
moment aligned along one of principal axes, the mo- 5 (a) (b) )
-20
ment is rotated and a single energy-band calculation
-100 -50 0 50 100 -50 0 50 100
is performed for the new orientation of magnetization. Energy (meV) Energy (meV)
55 15
The DOS anisotropies result from SO-coupling induced !=0
changesinthe bandeigenvalues. Inorderto increasethe M"# 50 !=90 10
accuracyinDOSevaluation,thesmoothFourierinterpo- V (1/ 45 5 G/G
lation scheme of Pickett et al. [20] is used together with dI/d 40 00 (%
the linear tetrahedron method. G= -5)
To model sample A we performed thin-slab calcula- 35 (c) (d)
-10
tions for five-monolayer Co sandwitched between two- -20 -10 0 10 20 -10 0 10 20
monolayerPtfilms. Sample B is modeled by considering V (mV) V (mV)
the thin Co film only. The calculated DOS as a func- FIG. 4: (a) Theoretical DOSs at θ = 0 and 90◦ and
tion of energy measured from the Fermi level are shown (b) the relative DOS anisotropy defined as (DOS(90◦) −
in Fig. 4(a) for the in-plane and out-of-plane magnetiza- DOS(0)/DOS(0) as a function of energy measured form the
tion orientations. For the Co/Pt model system we find Fermi level for the Pt/Co and Co model systems. (c) Corre-
a complex structure of the DOS near the Fermi energy spondingexperimentaldifferentialconductancesand(d)rela-
withthemainfeaturesshiftedby10’sofmeVwhencom- tivedifferentialconductanceanisotropiesasafunctionofbias.
paring the two magnetization orientation curves. The
Co-film,onthe otherhand,hasanearlyfeaturelessDOS Theoreticalexpectationthatlatticedistortionscansig-
near the Fermi level which depends very weakly on the nificantly alter the TAMR response is echoed in comple-
magnetizationdirection. The relative difference between mentaryexperiments,showninFig.5(b),wherewestudy
DOSs for the in-plane and out-of-plane magnetizations, the TAMRasa functionofthe in-planeangleinsamples
whichwerelatetothetunnelingconductanceanisotropy, of type A and B. A uniaxial in-plane magnetocrystalline
shows an oscillatory behavior as a function of energy for anisotropy is generated in these samples by applying in-
the Co/Pt slab with a magnitude of up to ∼ 20%. For planemagneticfieldduringthegrowth. Thisisanestab-
the Cofilmthemagnitudeandthe energydependenceof lishedproceduretocontrolmagneticanisotropiesintran-
the anisotropyis substantially weaker. These theoretical sition metal structures and in Co/Pt systems has been
resultsarequalitativelyconsistentwithmeasurementsin ascribed to anisotropic bonding effects at the Co/Pt in-
samples A and B presented above. terface[21]. Wecanthereforeexpectthattheinducedin-
To facilitate a more direct comparisonbetween theory planemagneticanisotropywillbereflectedintheTAMR
andexperimentweshowinthe lowerpanelsofFig.4the characteristicsonly for sample A in which the Pt/Co in-
measured differential conductances and the correspond- terfaceisincontactwiththetunnelbarrier. IndeedFig.4
ing conductance anisotropies as a function of the bias showsanegligibledependence insampleBofthe TAMR
voltage in sample A. Both the relative shift of the dif- on the in-plane angle ϕ towards which magnetization is
ferential conductances for the two magnetization orien- rotated from the out-of-plane direction. In sample A,
tations and the oscillatory behavior of the conductance on the other hand, the dependence on ϕ has a uniaxial
anisotropy with changing bias reflect qualitatively the character consistent with the induced in-plane magne-
anisotropic DOS behavior of the Co/Pt system seen in tocrystalline anisotropy and, notably, the magnitude of
the theory curves. the variation with ϕ is comparable to the magnitude of
In the calculations shown in Fig. 4 we assumed re- the θ-dependent TAMR shown in Fig. 1. This result il-
laxed Co-Co and Co-Pt interlayer spacings in the stack. lustrates the potential for controlling the characteristics
To estimate effects of lattice parameter variations on ofhighlysensitiveTAMRdevicesbylatticestructureen-
the TAMR we performed additional calculations (see gineering. We alsonote that the observationof the large
Fig.5(a))withtheCo-Cospacinginthefilmcorrespond- in-plane TAMR signal in our devices confirms the SO-
ing to bulk hcp Co. Qualitatively we obtained similar coupling origin of the measured tunneling anisotropy ef-
shifts in the complex DOS patterns for the two magne- fects.
tization orientations and comparable overall magnitude We conclude with a brief discussion of the observed
ofthe DOS anisotropiesbut the details ofthe oscillatory magneticfieldandtemperaturedependenceoftheTAMR
4
frequent with increasing temperature which leads to av-
(a) eraging out of the TAMR signal.
=0
V)15 =90 (b) Sample A 0 Sample B To summarize, our work demonstrates the prospect
e
DOS (1/10 AMR (%)- 100000 ...010 00.1 ftttoihhoreensretemrasayelnitszsatiielntmigostnsrhusimhgcthoeuultyraledlsss.ienancnSslditutridavoetefefigtTnhieAeesMttfuhoRnircinkfdugnetevousifsrceetahsrneedisncehcatoorrimcachenpsoooi--ff
-100 -50 0 50 100 T
Energy (meV) -100.1 0.1 sition of the films to optimize the SO-coupling and ex-
change splitting of layers adjacent to the tunnel barrier.
(c)7.0 )!162 18 0 "(deg) Improving the crystalline quality of the barrier and the
V(1/M)!6.5 R (k16146 V (mV) 48 ! B bmaargrnieirt/uedleecatnroddsetainbtielirtfyacoefstshheouTlAdMalRsoelffeaedct.to enhanced
dI/d6.0 We acknowledge support from the TND Frontier
" Project funded by KISTEP, the KIST Institutional pro-
5.5
-10 0 10 gram, from the Korea Research Council of Fundamental
V (mV)

Science &Technology(KRCF), fromthe EUGrantIST-
◦
FIG. 5: (a) Theoretical DOSs at θ = 0 and 90 for an unre-
015728, UK Grant GR/S81407/01, and Czech Repub-
laxedPt/Comodelsystems. (b)ExperimentalTAMRdefined
as (R(θ=90◦,ϕ)−R(θ=0)/R(θ=0) for samples A and B. lic Grants 202/05/0575, 202/04/1519, FON/06/E002,
∗
AV0Z1010052,and LC510. Corresponding author.
(c) Resistance measurements (inset) in two independent bias
sweeps showing a discrete step in resistance for the upper
curve. Corresponding differential conductances (main panel)
showingarelativeshift intheoscillatory biasdependencefor
thetwo measurements.
[1] T.McGuireandR.Potter,IEEETrans.Magn. 11,1018
(1975).
[2] M. N.Baibich et al., Phys. Rev.Lett. 61, 2472 (1988).
in sample A and implications our work suggests towards
[3] J.S.Moodera,L.R.Kinder,T.M.Wong,andR.Meser-
the realization of TAMR structures with optimized sen- vey,Phys. Rev.Lett. 74, 3273 (1995).
sitivity and field and thermal stability. Enhanced, com- [4] C. Gould et al., Phys. Rev.Lett. 93, 117203 (2004).
pared to bulk systems, and temperature dependent sus- [5] L. Brey, C. Tejedor, and J. Fern´andez-Rossier, Appl.
ceptibility of heavy element transition metals has been Phys. Lett. 85, 1996 (2004).
studied in thin films exchange coupled to neighboring [6] C. Ru¨ster et al., Phys.Rev.Lett. 94, 027203 (2005).
ferromagnetic layers or in dilute ferromagnetic alloys [7] J.Wunderlichetal.,Phys.Rev.Lett.97,077201(2006).
[8] A.B.Shick,F.M´aca,J.Maˇsek,andT.Jungwirth,Phys.
[10, 22]. Depending on the thickness of the film or con-
Rev. B 73, 024418 (2006).
centrationofthe ferromagneticelements inthe dilute al- [9] J. Moser et al., Phys. Rev.Lett. 99, 056601 (2007).
loys the systems can exhibit superparamagnetic behav- [10] Z. Celinski et al., Phys.Rev.Lett. 65, 1156 (1990).
ior ascribed to the presence of fluctuating moments in [11] P. Bruno, J. Magn. Soc. Jpn.15, 15 (1991).
the heavy element transition metals [23]. Experimen- [12] S. Ferrer et al., Phys. Rev. B 56, 9848 (1997).
tally, magnetization in these systems can show Curie- [13] F. Wilhelm et al., Phys. Rev.Lett. 85, 413 (2000).
Weiss-like scaling with B/T, and remain unsaturated at [14] J. Geissler et al., Phys.Rev. B 65, 020405 (2001).
[15] H. Nemoto and Y. Hosoe, J. Appl. Phys. 97, 10J109
high fields and low temperatures. We infer that similar
(2005).
physics applies to our two-monolayer Pt film coupled to
[16] E. Y. Tsymbal and D. G. Pettifor, J. Phys. Condens.
theneighboringColayerortointermixedCoatomsinthe Matter 9, L411 (1997).
Pt. TheobservedapproximatescalingoftheTAMRwith [17] A. N. Chantis, K. D. Belashchenko, E. Y. Tsymbal, and
B/T, implicitly contained in Fig. 2, can then be partly M.vanSchilfgaarde,Phys.Rev.Lett.98,046601(2007).
ascribed to the proportionality between TAMR and the [18] A. B. Shick, D. L. Novikov, and A. J. Freeman, Phys.
unsaturated polarization of the Pt layer adjacent to the Rev. B 56, R14259 (1997).
[19] A.I.Liechtenstein,M.I.Katsnelson,V.P.Antropov,and
tunnel barrier.
V. A.Gubanov, J. Magn. Magn. Mater. 67, 65 (1987).
Another effect which contributes to the suppressionof [20] W.E.Pickett,H.Krakauer,andP.B.Allen,Phys.Rev.
theTAMRinourdeviceswithtemperatureistelegraphic B 38, 2721 (1988).
noiseassociatedwithrandomlytrappedchargespresum- [21] M. Einax et al., J. Phys.: Condens. Matter 19, 086227
ably in the AlO barrier or at the interface between the (2007).
x [22] J. Crangle and W. R. Scott, J. Appl. Phys. 36, 921
barrieranthePtthinfilm[24]. Atlowtemperatures,the
(1965).
resultingjumps betweendistinctresistancestatescanbe
[23] T. Moriya, in Spin Fluctuations in Itinerant Electron
clearly identified, as shown in Fig. 5(c), together with
Magnetism (Springer,Berlin, Heidelberg, 1985).
shiftsinthecorrespondingoscillatorybiasdependenceof [24] E.R.Nowak,M.B.Weissman,andS.S.P.Parkin,Appl.
the differential conductance. These jumps become more Phys. Lett. 74, 600 (1999).