Spin injection in a single metallic nanoparticle: a step towards nanospintronics

0.17 MB·English

by  A. Bernand-Mantel

APL
Spin injection in a single metallic nanoparticle: a step towards nanospintronics
A. Bernand-Mantel, P. Seneor, N. Lidgi, M. Mun˜oz,∗ V. Cros, S.
Fusil,† K. Bouzehouane, C. Deranlot, A. Vaures, F. Petroff, and A. Fert
Unit´e Mixte de Physique CNRS/Thales and Universit´e Paris-Sud 11
91767, Palaiseau, France
6 (Dated: February 3, 2008)
0
We have fabricated nanometer sized magnetic tunnel junctions using a new nanoindentation
0
technique in order to study the transport properties of a single metallic nanoparticle. Coulomb
2
blockadeeffectsshowclearevidenceforsingleelectrontunnelingthroughasingle2.5nmAucluster.

n The observed magnetoresistance is the signature of spin conservation during the transport process
a through a non magnetic cluster.
J

9 PACSnumbers: 85.75.-d,75.47.-m,73.63.-b
1

] Spintronics debuted with the discovery of giant Co 20
ci magnetoresistance[1] effect in magnetic multilayers in Insulator
10
s which a single dimension was reduced to the nanome-
- Co 0
trl tteurresrawnigteh.twoTrheidsuficeedlddiwmaesnstiohnens liekxetnenadneodwirteos[2s]trauncd- Resist Co Y(nm)-10
m nanopillars[3, 4] or nanotubes[5]. Today, a challenge -20
t. for spintronics is the study of spin transport proper- Al2O3 Au -30
a ties in structures based on 0D elements in which the Co 0 100 200 300 400
X(nm)
m three dimensions have been reduced. In particular, we
- have in mind systems in which the reduction of the size FIG. 1: Left: Schematic cross-section of thewhole patterned
d
leads to both Coulomb blockade and spin accumulation
structure showing the top/bottom Co electrodes and the 2D
n
effects[6, 7, 8]. Transport studies on systems including assembly of nanoparticles embedded in a thin alumina layer.
o
c mesoscopic islands[9, 10, 11] or granularfilms[12, 13, 14] The circle represents the zone zoomed in the bottom draw-
[ have paved the way to understanding the effect of con- ing. Right: Cross section of an AFM tapping mode scan of
finement on charge and spin transport properties in a nanoindent before filling with the Co top electrode. The
1
metallic nano-objects. However, so far, very few tech- effectivenanocontact cross section is less than 10nm.
v
niques allowto contacta single isolatednanometer sized
9
3 object[15, 16, 17] to study the effect of confinement on
4 spin transport. to form the first tunnel barrier. Then, an ultrathin layer
1 of Au (0.2 nm nominal thickness) is deposited on top of
Inthis letterwe presenttheexperimentalachievement
0 the bilayer. The 3D growth (see [18]) of the sputtered
6 of a new technique allowingus to inject anddetect spins
goldontopofaluminaproducesaself-formednanoparti-
0 ina singleisolatednanometersizedcluster. We thenob-
cleslayer. Aplaneviewtransmissionelectronmicroscopy
/ tain information on both spin and single electron trans-
t (TEM) picture of the Au nanoparticles 2D self-assembly
a port in the nanoparticle. In this technique, a ferromag-
m netic nanocontactwith a cross sectionof ∼5−10 nm in is shown in Figure 2. The size distribution of the Au
nanoparticles is characterized by a 2 nm diameter mean
- diameteriscreatedonabilayerassociatingacobaltlayer
d value and a 0.5 nm standard deviation with a density of
andanultrathinaluminalayerinwhicha2Dassemblyof
n 1.7 1012 cm−2. Finally, the Au clusters are capped by
gold nanoparticles is embedded (see Fig 1 for a sketch).
o another Al layer (0.6nm) subsequently oxidized in pure
As we will show, this structure allows the tunneling of
c
: electrons into and out of a single Au nanoparticle. O2 with the same process used to form the first tunnel
v barrier. This creates a Co/Al2O3 bilayer with Au nan-
i The whole structure is elaborated in a sputtering sys- oclusters embedded in the thin alumina layer.
X tem (base pressure 5×10−8 mbar) with Ar gas at a dy-
To define an electrical contact on a single cluster
r namic pressure of 2.5×10−3 mbar. The deposition of
a we use a 4 steps process combining AFM and optical
a bilayer of Co(15nm)/Al(0.6nm) is followed by the oxi-
lithographic techniques (for further technical details see
dization of the Al layer in pure O2 (50 mbar for 10 min) Ref[19]). Forthefirststep,aphotoresistlayerof35nmis
spin coated over the whole Co/Al2O3/Au/Al2O3 struc-
ture. Then, contact zones are defined using a second
photoresist layer and a standard UV lithography pro-
∗Now at Laboratorio de F´isica de Sistemas Pequen˜os y Nanotec- cess. During the second step a conductive boron doped
nolog´ia, Consejo Superior de Investigaciones Cient´ificas, Serrano
diamond AFM tip is used to nanoindent the thin resist.
144,28006Madrid,Spain
†AlsoatUniversit´ed’Evry, Bat. des Sciences, ruedupereJarlan, Theconductancebetweentheconductivesampleandthe
91025Evry,France AFM tip is monitored in real time. In the late stage of
2
500
1x1015 (cid:15)(cid:16)(cid:5)(cid:3)
400
1x1015 (cid:15)(cid:2)(cid:5)(cid:3)
300
-2Density (m)468xxx111000111444 dV(nS) 120000 (cid:0)(cid:4)(cid:1)(cid:5)(cid:2)(cid:3)(cid:3)
2x1014 dI/ 0
0 -100
10 nm 0,0 0,5 1,0 1D,i5ame2t,e0r (n2m,5) 3,0 3,5 4,0 -200 (cid:6)(cid:7)(cid:8)(cid:9)(cid:10)(cid:11)(cid:12)(cid:7)(cid:13)(cid:14)(cid:0)(cid:1)(cid:2)(cid:3)
-400 -200 0 200 400
FIG. 2: Left: Binarized plane view of self-assembled Au Voltage (mV)
nanoparticles observed by TEM. See Ref. [18] for further in-
formation on the binarization process. Right: Size distribu-
FIG.3: Differentialconductancecurvesat,190K,120K,60K
tion of the same self-assembly. Fitting gaussian (bold line),
and4.2KforaCo/Al2O3/Au0.2nmnominal/Al2O3/Co/Au
parameters are 2 nm mean diameter and 0.54 nm standard
nanocontact. Monte-Carlo simulation (plain line) at 4.2 K
deviation.
and with a background charge offset Q0/e=0.4. The other
simulationparametersareC1=0.4aF,C2=1.14aF,R2=3R1.
the nanoindentation a tunneling current is established
between the tip and the sample. The exponential thick- the conductance curves. At 4.2 K, the Coulomb block-
nessdependenceallowstopreciselycontroltheendofthe ade peaks are clearly visible, with an equal spacing of
indentation process. The tip is then retracted and the Vs ≃140mV between peaks. As the highestcapacitance
sampleisleftwithananometerscaleholeonthesurface. governsthepeakspacing,weextractCmax=1.14aFfrom
After this nano-indentation process, the holes can be in- Vs through Cmax=e/Vs. This parameter is used as an
spectedbytappingmodeAFMusingultrasharptips. On inputfor the singlenanoparticleMonte-Carlosimulation
figure 1, we show the cross section of a typical hole after (MOSES) [24] which faithfully reproduces the peaks of
enlargementby a 30 s O2 plasma etch. Onthe crosssec- the 4.2 K conductance curve in Figure 3. In this sim-
tion one can see that the contact area section is in the ulation, the structure is modelized as two tunnel junc-
10 nm range. The samples presented in this letter use a tions in series. Using C2=Cmax=1.14 aF and T=4.2 K,
shorterenlargingplasmaetchof20sgivingholessections the best fitting parameters for the two tunnel junctions
below the 10 nm range. However, due to a tip nominal are C1=0.4 aF, R2=3R1 and a background charge of
radius of 5 nm, holes having a contact cross-section be- Q0/e=0.4. AV2 termis addedtothe simulatedcurveto
low 10 nm can not be properly imaged[19]. From the take into account the quadratic variation of the tunnel
nanoparticles density of 1.7 1012.cm−2 and assuming a conductanceversusvoltage[21]. The chargingenergyEC
disc shape area for the contact surface, there is an av- of the nanoparticle can then be calculated assuming the
erage of 0.3-1.2 nanoparticles per nanocontact hole for nanoparticle total capacitance is CT=C1+C2. One finds
holes diameter in the 5-10 nm range. The next step, is EC =e2/2CT ≃50 meV which is in agreementwith the
thefillingoftheholebyasputteredCo15nm/Au50nm fading of the Coulomb blockade features above 120 K.
counter electrode, just after a short O2 plasma. Finally Using electrostatic simulations and the capacitances ex-
we use standard optical lithography and ion beam etch- tracted from the fit, we can determine the nanoparticle
ing techniques to define the top electrode. This allows diametertobe ≃2.5nminexcellentagreementwiththe
us toobtainasingle clusterper nanocontactwitha high average size determined by TEM.
expectation value. In Figure 4, I(V) and dI/dV curves at 4.2 K of the
The patterned samples were measured in a variable same sample are shown after a sweep at higher voltage.
temperature cryostatfrom 4.2 K to 300 K. On Figure 3, A change in the background charge has occurred. The
onecanseethedI/dV(V)curvesofananopatternedsam- curves are almost symmetric in voltage bias which re-
ple at different temperatures. Above 190 K the conduc- flects the very low offset of the new background charge.
tancecurvepresentstunnellikefeatureswithoutanyevi- A simulation curve using the same set of parameters as
dence ofCoulombblockade. At120K,shoulderscharac- before, except for a smaller charge offset of Q0/e=0.07,
teristic of Coulomb blockade start to appear in the con- is also shownon the figure. A clearevidence that we ob-
ductance curve and get more pronounced as tempera- serve single electron tunneling through a single isolated
ture is lowered. Considering a single nanoparticle, for nanoparticleisthatthesamesetofjunctionsparameters
Coulomb blockade to be observed at 120 K, the charg- are used to obtain an excellent fit of our data with two
ingenergyEC ofthe nanoparticlehastobe greaterthan different background charges.
about kT120K which suggest that the charging energy Wenowfocusonspindependenttransportinthesam-
is at least ≃ 10 meV. Below 60 K an asymmetry in ple. In figure 5, we present a R(H) curve showing evi-
the shoulder position, typical of the presence of a non denceoftunnelmagnetoresistance(TMR).Due toshape
zeroQ0 backgroundoffsetcharge[20], canbe observedin anisotropy, the top ”cone” and the bottom plane mag-
3
15 netic electrodes have distinct coercive fields. Therefore
Data
10 Simulation itenablestheswitchingfromparalleltoanti-parallelcon-
figurationsfortheelectrodesmagnetizations. Theoccur-
5 rence of TMR is a direct proof of spin dependent trans-
A) 0 port in the nanostructure,which indicates spin injection
n fromone electrode intothe clusterandspindetectionby
I(
-5 the second electrode. As the nanoparticle is non mag-
netic,the observation of TMR means that spin informa-
-10 tion is conserved in the transport process through the
nanoparticle. ThesignoftheTMReffect,whichrulesout
-15
S imulation direct tunneling between the two magnetic electrodes,
Data can be understood in the framework of spin accumula-
150
tiononthenon-magneticnanoclusterintheco-tunneling
regime as discussed in references[22, 23].
S)
n100
V(
d
dI/ 50 In summary, we have developed an original process
to investigate the spin transport properties of a single
nanoparticleandprovidedevidenceforitssuccessfulreal-
0
-200 -100 0 100 200 isation. Ourapproachpavesthewayforamorein-depth
Voltage (mV) studyofmagneto-Coulombphenomenainnanosizedclus-
ters.
FIG.4: Top: I(V)curve(-(cid:4)-)measuredat4.2Kforthesam-
pleshowninfigure3afterachangeinthebackgroundcharge.
Thelineisasimulationusingtheparametersgiveninthefig-
ure 3 caption but with Q0=0.07 e. Bottom: dI/dV(V) curve
(-(cid:4)-) obtained from the derivative of the above I(V) curve.
The line represents thedifferentiation of thesimulation.
650
)640
WM
e (630
c
n
a620
st
si
e610
R
Acknowledgments
600
-90 -60 -30 0 30 60 90
Magnetic field (Oe)
FIG. 5: Resistance versus magnetic field obtained at 20 mV We acknowledge the support of the EU through the
and 4 K for thesample of figures 3 and 4. Nanotemplates project (NMP-CT-2004-505955).
[1] M. N. Baibich, J. M. Broto, A. Fert, F. nguyen van 401, 572 (1999).
Dau, F. Petroff, P. Etienne, G. Creuzet, A. Friedrich, [6] J.BarnasandA.Fert,Phys.Rev.Lett.80,1058(1998).
and J. Chazelas, Phys. Rev.Lett. 61, 2472 (1988). [7] H. Imamura, S. Takahashi, and S. Maekawa, Phys. Rev.
[2] L. Piraux, J. George, J. Despres, C. Leroy, E. Ferain, B 59, 6017 (1999).
R.Legras, K. Ounadjela, and A. Fert, Appl.Phys.Lett. [8] A. Brataas, Y. V. Nazarov, and J. Inoue, Phys. Rev. B
55, 2484 (1994). 59, 93 (1999).
[3] J.A.Katine,F.J.Albert,R.A.Buhrman,E.B.Myers, [9] M.ZaffalonandB.J.vanWees,Phys.Rev.B71,125401
and D. C. Ralph,Phys. Rev.Lett. 84, 3149 (2000). (2005).
[4] J.Grolier, V.Cros, A.Hamzic,J.-M.George, H.Jaffres, [10] K. Ono, H.Shimada, and Y.Ootuka, J. Phys. Soc. Jpn.
A. Fert, G. Faini, J. B. Youssef, and H. LeGall, Appl. 66, 1261 (1997).
Phys.Lett. 78, 3663 (2001). [11] C.D.Chen,W.Kuo,D.S.Chung,J.H.Shyu,andC.S.
[5] K. Tsukagoshi, B. W. Alphenaar, and H. Ago, Nature Wu, Phys.Rev.Lett. 88, 047004 (2002).
4
[12] S. Mitani, S. Takahashi, K. Takanashi, K. Yakushiji, [19] K. Bouzehouane, S. Fusil, M. Bibes, J. Carrey, T. Blon,
S.Maekawa, and H.Fujimori, Phys.Rev.Lett. 81, 2799 M. Le Du, P. Seneor, V. Cros, and L. Vila, Nanoletters
(1998). 3, 1599 (2003).
[13] K.Yakushiji,F.Ernult,H.Imamura,K.Yamane,S.Mi- [20] A. E. Hanna and M. Tinkham, Phys. Rev. B 44, 5919
tani,K.Takanashi,S.Takahashi,S.Maekawa,andH.Fu- (1991).
jimori, NatureMat. 4, 57 (2005). [21] R. D. W.F. Brinkman and J. Rowell, J. Appl.Phys. 41,
[14] L. F. Schelp, A. Fert, F. Fettar, P. Holody, S. Lee, J. L. 1915 (1970).
Maurice, F. Petroff, and A. Vaures, Phys. Rev. B 56, [22] J. Martinek, J. Barnas, S. Maekawa, H. Schoeller, and
R5747 (1997). G. Schon,Phys.Rev. B 66, 014402 (2002).
[15] D. C. Ralph, C. T. Black, and M. Tinkham, Phys. Rev. [23] I. Weymann, J. Konig, J. Martinek, J. Barnas, and
Lett.74, 3241 (1995). G. Schon,Phys.Rev. B 72, 115334 (2005).
[16] R. Desmicht, G. Faini, V. Cros, A. Fert, F. Petroff, and [24] Monte-Carlo Single-Electronics Sim-
A.Vaures, Appl.Phys. Lett.72, 387 (1998). ulator (MOSES), available at
[17] J. Von Delft and D. Ralph,Phys. Rep. 345, 61 (2001). http://hana.physics.sunysb.edu/software/MOSES.htm
[18] J.Carrey,J.-L.Maurice,F.Petroff,andA.Vaures,Phys.
Rev.Lett. 86, 4600 (2001).