2012约瑟夫森结

Home Search Collections Journals About Contact us My IOPscienceStructural and electrical characterization of ultra-thin SrTiO3 tunnel barriers grown over YBa2Cu3O7 electrodes for the development of high T c Josephson junctionsThis content has been downloaded from IOPscience. Please scroll down to see the full text.2012 Nanotechnology 23 495715(/0957-4484/23/49/495715)View the table of contents for this issue, or go to the journal homepage for moreDownload details:IP Address: 58.198.97.239This content was downloaded on 24/05/2014 at 08:23Please note that terms and conditions apply.IOP P UBLISHING N ANOTECHNOLOGY Nanotechnology23(2012)495715(6pp)doi:10.1088/0957-4484/23/49/495715Structural and electrical characterization of ultra-thin SrTiO3tunnel barriers grown over YBa2Cu3O7electrodes for the development of high T c Josephson junctionsL Avil´e s F´e lix1,M Sirena1,2,L A Ag¨uero Guzm´a n3,J Gonz´a lez Sutter3,S Pons Vargas3,L B Steren2,4,R Bernard5,J Trastoy5,J E Villegas5,J Bri´a tico5,N Bergeal6,J Lesueur6and G Faini71Centro At´o mico Bariloche,Instituto Balseiro—CNEA,Universidad Nacional de Cuyo,AvenidaBustillo9500,8400Bariloche,Rio Negro,Argentina2Consejo Nacional de Investigaciones Cient´ıficas y T´e cnicas,Avenida Rivadavia1917C1033AAJCABA,Argentina3Universidad de Concepci´o n,Victoria631,2613Concepci´o n,Chile4Centro At´o mico Constituyentes,Avenida Gral.Paz1499,San Mart´ın1650,Buenos Aires,Argentina5Unit´e Mixte de Physique CNRS/THALES,Universit´e Paris Sud11,F-91767Palaiseau Cedex,France6Laboratoire de Physique et d’Etude des Mat´e riaux,UMR8213/CNRS,ESPCI ParisTech,10rueVauquelin,Paris F-75005,France7LPN-CNRS,Route de Nozay,F-91460Marcoussis,FranceE-mail:lavilesf@Received11August2012,infinal form24October2012Published16November2012Online at /Nano/23/495715AbstractThe transport properties of ultra-thin SrTiO3(STO)layers grown over YBa2Cu3O7electrodeswere studied by conductive atomic force microscopy at the nano-scale.A very good control ofthe barrier thickness was achieved during the deposition process.A phenomenologicalapproach was used to obtain critical parameters regarding the structural and electricalproperties of the system.The STO layers present an energy barrier of0.9eV and anattenuation length of0.23nm,indicating very good insulating properties for the developmentof high-quality Josephson junctions.(Somefigures may appear in colour only in the online journal)1.IntroductionA superconductor–insulator–superconductor Josephson junc-tion(JJ)is formed by two superconductor electrodes separated by a very thin insulating barrier composed of a few atomic layers[1].Cooper pairs canflow through this barrier thanks to the coupling of the superconducting electrodes.If a constant voltage is applied on the junction,an alternating current with a frequency in the range of1–1000GHz can be generated in the system[2].These two particular phenomena have many technological applications,such as the fabrication of a voltage standard[3],SQUIDs and other magnetic sensors[4], quantum computing(qubits)[5],and the fabrication of ultra-fast superconducting microelectronics based on rapid singleflux quantum logic[6].However,the fabrication of a homogeneous JJ array is complicated,especially for highT c superconducting(HTSc)materials[7–9].A successful JJ technology requires an improvement and a better control of the JJ characteristics,in particular the JJ energy,i.e.the I c R n product,where I c is the critical current of the junction and R n its resistance in the normal state.This could be achieved with a better control of the barrier properties,reducing the thickness of the insulating barrier,in order to increase the superconducting coupling between the electrodes.In recent years,the improvement in microfabrication techniques has allowed the size of the operating devices to be reduced below the micro-scale.Moreover,recently,researchers from the Unit´e Mixte de Physique CNRS/Thales have developed a fabrication technique using AFM controlled nanoindentation for the development of nanojunctions[10].Typical sizes of these devices go from10×10nm2to50×50nm2, and with this size reduction the homogeneity of the barrier and the probability to obtain a working device should increase significantly.However,a good characterization of the insulating barrier at the nano-scale must be made in order to improve the performance of tunnel junction devices. We have recently developed a phenomenological approach using conductive atomic force microscopy(CAFM)[11,12] that allows one to obtain important electrical information to optimize the quality of JJ devices,e.g.the barrier thickness required to totally cover the superconducting electrodes and the physical properties of the barrier(its thickness distribution,the attenuation length and the energy barrier).In recent years,CAFM has become an extremely useful technique to study and characterize the insulating barriers of tunnel junction devices,such as magnetic tunnel junctions[13–15]or spinfilters.However,little work has been done using CAFM for the electrical characterization of insulating layers for the development of JJ[16].CAFM is a technique capable of performing local resistance measurements between the tip and the sample and allows one to obtain simultaneously the topography and the conductivity maps of different systems at the nano-scale[17,18].This technique also provides qualitative information about the insulating barrier homogeneity,directly detecting pinholes or hotspots,which can shortcut the electrodes reducing the quality of the JJ.The phenomenological approach used in this work is,basically,a generalization of the theoretical Simmons’model[19].Special attention was paid to analyze the influence of the surface defects in the topography and the conductivity map of these systems.We would like to stress that the use of the CAFM in the frame of this phenomenological approach allows one to obtain critical information about the structural and electrical properties of ultra-thin insulating layers,which are one of the key components not only of superconductor Josephson junctions but also of many other technological devices,such as magnetic tunnel junctions[20],multiferroic tunnel junctions and spinfilters[10],new memristors technologies[21]and tunneling electroresistance devices[22].Moreover,improved characterization methods of ultra-thin oxides are needed,as their importance continues to grow as new properties and new functionalities are discovered[23].2.Experimental detailsYBa2Cu3O7/SrTiO3(YBCO/STO)bilayers were grown on single-crystal(100)STO substrates by pulsed laser deposition with a KrF excimer laser(248nm)of780mW from stoichiometric ceramic targets.STO layers with thicknesses of0.4,0.8,1.2,1.6and2nm were deposited over YBCO electrodes with a thickness of75nm.The bilayers were grown in situ at664◦C using an oxygen pressure of0.355mbar. After deposition,the temperature of the sample was decreased to500◦C for an hour in order to assure the oxygen content of the YBCO and STO layers.Typical YBCO electrodes present a superconducting transition temperature of85K.CAFM measurements were performed at room tem-perature on a Veeco Dimension3100scanning probe microscopy AFM with a CAFM module and using a boron-doped diamond conductive tip(Bruker DDESP-10, spring constant=20–80N m−1)in contact mode.The CAFM tip was polarized with a positive voltage while the YBCO electrode was grounded.The minimum detectable current in the CAFM used is50pA,and the maximum current is480nA,under a bias voltage ranging from0.01to 12V.The CAFM is operated with a constant force(0.5V of deflection setpoint),with the laser reflecting always at the same position of the cantilever,and with a constant bias voltage to reducefluctuations in the mapping of the tunnel currents as a consequence of external parameters[16]. Different scan sizes and different scan orientations were used to rule out the presence of tip artifacts.CAFM images and I(V)curves were always taken in untouched areas in order to reduce the chemical modification of the surface due to the polarization voltage of the tip.I(V)curves for samples with different thicknesses of the insulating layer were obtained by holding the tip steadily in contact with the surface of the samples.Then the CAFM set up varies the applied voltage in the‘ramp’mode and measures the current for different values of the polarization voltage.In order to reduce thefluctuations of the I(V)measurements and the electrical contact noise, several I(V)curves(∼25measurement)were performed at different places on the samples and then averaged.3.Results and discussionFigure1shows the topographic(left)and CAFM(right) images measured simultaneously for YBCO/STO bilayers with different thicknesses of the insulating layer at a constant applied voltage.The scanning area is15×15µm2.In the electrical images,the bright areas indicate high tunneling currents.As expected,the current through the STO barrier decreases as the STO thickness increases.The samples present a very low roughness,∼0.5nm(∼1u.c.)and a low density of surface defects(∼0.07defµm−2),with heights going from 10to50nm.CAFM images show that the resistance through the grain boundaries of YBCO is higher than the resistance through the center of the grains.In general,higher values of tunneling current were detected at the center of the grain. Grains present a diameter between180and220nm with a separation between them in the range from80to120nm.Figure1.Topographic(left)and CAFM(right)15µm×15µm images of YBCO/STO bilayers grown over STO substrates with different STO thicknesses(2u.c.(a),3u.c.(b)and4u.c.(c)).The applied voltage was3V.CAFM current distributions(figure2)for the samples with STO thicknesses smaller than4u.c.present a high current peak that seems to correspond to the existence of pinholes in the barrier.However,the density of pinholes reduces rapidly for thicknesses greater than3u.c.The CAFM current distribution can be wellfitted with a log-normal distribution,as expected from the exponential decay of the tunneling current with the barrier thickness[24].The current distributionfitting for the sample with a barrier thickness of 3u.c.is not as good as thefitting for the other samples.The experimental data seem to show a double peak distribution (dotted lines infigure3),meaning thefitting with a single distribution peak(single line)is not as good as thefitting for the other samples.This is probably due to the existence of two areas with a slightly different thickness or a small change in the tip–surface resistance during the scan.However, the effect is small and was not observed for the other samples.The current distribution widths obtained from the fits corresponds to a barrier thickness distribution width of σd∼0.6nm(1.5u.c.),in agreement with the observed topographic roughness of the samples.The evolution of the current distributions as a function of the barrier thickness shows the good control and reproducibility of the insulating layers grown by pulsed laser deposition.Figure3shows characteristic I(V)curves for YBCO/STO bilayers with different thicknesses of the insulating barrier, obtained with the CAFM.The experimental data arefitted using a phenomenological approach based on the Simmons model[19]:ln(I(V,d))=A0(d)+α(d)ln V−B(d)V.(1)We have used threefitting parameters A0,αand B that depend on the barrier nature or its thickness(d).In the Simmons model,for polarization voltages higher than theFigure2.Current distributions for YBCO/STO bilayers.The curves correspond to different thicknesses of the barrier( =5u.c., =4u.c., =3u.c.and =2u.c.)at3V.The lines arefittings using the log-normal distribution(seetext).Figure3.I(V)curves for the YBCO/STO bilayers for different thicknesses of the STO barrier.The solid lines are linearfits of theexperimental data(ln I=A0+αln V−B(d)V ).The inset presents theaverage values of A0andαas a function of the barrier thickness(d) obtained from thefitting of the I(V)curves.energy of the barrier,α=2.As shown in the inset offigure3, A0andαdepend linearly on the barrier thickness.B was found to be smaller than the experimental resolution and was assumed to be zero.So,in general,for voltages higher than 1V,we can rewrite equation(1)as[11,12]:I=e a0Vα0e−dλwithλ=1|a |−α ln V,(2)whereλis the electronic attenuation length of the carriers in the barrier and a0=A0(d=0);α0=α(d=0);a =∂A0∂d<0 andα =∂α∂d.In order to reduce the currentfluctuations and to obtain accurate values of A0andα,several I(V)curves were measured,placing the conductive tip in different zones of the sample and averaging.The I(V)curves show a more abrupt change of the tunneling current as a function of the polarization voltage than the expected behavior predictedby parison between the values ofαas a function of the attenuation length for three systems modeled with the same phenomenological approach.The curves correspond to different thicknesses of the barrier( =2u.c., =4u.c., =5u.c.).Inset: attenuation length(open squares)and energy barrier(open circles) as a function of the polarization voltage for the YBCO/STO bilayers.the Simmons model[19].Indeed,for the samples studied in the present work,α0varied between4and7,larger than the theoretical value(α0=2)corresponding to the high-voltage tunneling regime(V>φ/e,whereφis the energy barrier of the junction)[19].To obtainλ,we have plotted ln(I/Vα0)= a0−d/λas a function of d for different polarization voltagesand a linearfit of the experimental data was performed.The obtained value ofλ=0.23nm is in good agreement with the values reported in the references mentioned above[11,12], for similar insulating barriers.The value of the attenuation length increases when the applied voltage increases,with a value of0.23nm for1.25V(inset offigure4).Once the attenuation length is obtained,the energy barrier can be estimated,by considering a rectangular barrier shape,for different applied voltages.Under these assumptions,in the Simmons model[18,19],the energy barrier is given by φ=(3he8π√2m∗Vλ)2/3,where h is the Planck constant,m∗is the effective mass of the charge carriers,e is the electron charge and V is the applied voltage.m∗is given by the band structure of STO,and for the present calculations it has been considered that m∗=m e(giving an upper limit of the energy barrier).The energy barrier varies between0.8and1.2eV,being around 0.90eV at1.25V(inset offigure4),higher than the values reported in the literature in similar systems[12,18,25].The variation of the energy barrier could be related to reproducible uncertainties of the measurements,since no change ofφwith the polarization voltage is usually expected.It should be noticed that the assumption of the Fowler–Nordheim regime is not always satisfied for the samples with thinner barriers.This is probably why the I(V)curves for the samples with thicker barriers(i.e.5u.c.)and higher voltages present a better linear behavior than the samples with thinner barriers.However,this effect seems to be small.As mentioned before,the experimental data show a faster increase of the tunneling current with increasingapplied voltage than the behavior predicted by the Simmonsmodel.According to this model,in the Fowler–Nordheim(FN)regime,the current follows a V2law(α0=2).This square dependence of the tunneling current with theapplied voltage arises from the assumption that,for thehigh-voltage regime,the effective thickness of the rectangularpotential barrier,d eff,decreases linearly as the applied voltageincreases and d eff=dφ/eV[19].However,it seems thatthis is not the case for this system,and a similar behaviorwas observed for Nb/Ba0.05Sr0.95TiO3(BSTO)[12]andLa0.75Sr0.25MnO3(LSMO)/BSTO bilayers[11].Moreover,experimental evidence indicates thatα0depends on theinsulating nature of the barrier.Therefore,in analogy tothe Simmons model,an effective thickness of the potentialbarrier of the form d eff=dφ/eVα0/2(generallyα0>2)was considered.This indicates a decreasing effective thickness ofthe insulating barrier as the value ofα0increases,and alsogives a qualitative idea about the deformation of the effectivebarrier for the charge carriers.For high applied voltages,andfor a given d(e.g.afixed value ofα0),the tunnel transporttakes place through an effective barrier thinner than thatexpected in the Simmons model due to d eff∝1/Vα0/2.Asimilar result is obtained when the effect of the image forceis considered[19].The image force rounds the corner ofthe barrier,reducing its effective thickness and increasing thetunneling current of the junction.It was noted by Simmonsthat the image potential is a hyperbolic function,which resultsin an elliptic integral when included in the general equationto calculate the tunneling current,and can only be solvednumerically.The phenomenological method proposed in thiswork gives a simple way to evaluate the importance of theimage forces in these systems.The increase ofα0withincreasing barrier thickness seems to indicate that the more‘insulating’(thicker insulating barriers)the system is,themore abrupt(higherα)is the change of the tunneling currentwith the applied voltage.In order to test this hypothesis,αwas plotted as function of the attenuation length,λ,for threedifferent systems,YBCO/STO,Nb/BSTO and LSMO/BSTO(figure4).Thefigure shows that for these systems,αincreasesas the attenuation length decreases,indicating that‘moreinsulating’materials with lower attenuation lengths,presenta more important change of the tunneling current with theapplied voltage.A similar effect was observed theoretically bySimmons[19],when the influence of the barrier’s dielectricconstant was considered to calculate the resistance of thetunnel junction as function of the applied voltage.The originof this phenomenon is not clear and more work is needed inorder to provide a new insight about the tunneling mechanismin these systems.4.ConclusionsWe have used a phenomenological approach based on theSimmons model to analyze the electrical transport throughultra-thin SrTiO3layers grown over YBa2Cu3O7electrodes.The proposed method helps one,in a direct way,to study andto optimize the growth of insulating barriers over high criticaltemperature superconductor electrodes in order to improve the performance and application of high T c Josephson junctions. We have found that the tunneling of the carriers seems to be the main mechanism for the electrical transport in these systems.In general,a STO barrier thickness of at least1.6nm (4u.c.)is required to obtain a good insulation of the YBCO electrode for small areas.The SrTiO3layers present an energy barrier of0.9eV,with an attenuation length of0.23nm, indicating their good insulating properties.The deposition method allows the growth of high-quality ultra-thin STO layers over high T c superconducting electrodes,with a very good control of the barrier thickness,for the fabrication of Josephson junctions.AcknowledgmentsThe authors would like to thank R Benavides for his extraordinary technical support.The authors would also like to recognize the important work,help and support from Dr J Guimpel and Dr H Pastoriza for the use of micro-and nanofabrication facilities and A Butera for the critical reading of the manuscript.This work was partially supported by the ANPCYT(PICT PRH2008-109),Universidad Nacional de Cuyo(06/C328)and the international cooperation program MINCyT-ECOS(France)A10E05.J Trastoy acknowledges the support from Fundaci´o n Barri´e(Galicia,Spain). References[1]Josephson B D1962Phys.Lett.1251[2]Barone A and Paterno G1982Physics and Applications of theJosephson Effect(New York:Wiley)pp309–24[3]Benz S P,Hamilton C A,Burroughs C J,Harvey T E andChristian L A1997Appl.Phys.Lett.711866[4]Adachi S,Tsukamoto A,Oshikubo Y,Hato T andTanabe K2011Physics C4711258[5]Nakamura Y,Nakamura Y,Terai H,Inomata K,Yamamoto T,Qui W and Wang Z2011Appl.Phys.Lett.99212502 [6]Hassel J,Helist¨o P,Sepp¨a H,Kunert J,Fritzsch L andMeyer H G2006Appl.Phys.Lett.89182514[7]Bergeal N,Lesueur J,Sirena M,Faini G,Aprili M,Contour J-P and Leridon B2007J.Appl.Phys.102083903 [8]Hamilton Clark A,Burroughs Charles J and Benz Samuel P1997IEEE Trans.Appl.Supercond.73756[9]Bernard R,Briatico J,Sirena M,Crete D G andContour J-P2007IEEE Trans.Appl.Supercond.173589 [10]Bouzehouane K,Fusil S,Bibes M,Carrey J,Blon T,Le Dˆu M,Seneor P,Cros V and Vila L2003Nanoletters31599 Bouzehouane K,Fontcuberta J,Barth´e l´e my A andFert A2007Nature Mater.6296–302[11]Sirena M2011J.Appl.Phys.110063923[12]Sirena M,Avil´e s F´e lix L,Carvacho Vera G A,Navarro Fern´a ndez H L,Steren L B,Bernard R,Bri´a tico J,Bergeal N,Lesueur J and Faini G2012Appl.Phys.Lett.100012602[13]Canizo-Cabrera A,Li Simon C,Shu M-F,Lee J-M,Garcia V V,Chen C C,Chen J C,Wu J C,Takahashi M andWu T-H2005IEEE Trans.Magn.40887[14]Luo E Z,Wong S K,Pakhomov A B,Xu J B,Wilson I H andWong C Y2001J.Appl.Phys.905202[15]Ando Y,Kameda H,Kubota H and Miyazaki T2000J.Appl.Phys.875206[16]Lang K M,Hite D A,Simmonds R W,McDermott R,Pappas P and Martinis J M2004Rev.Sci.Instrum.752726[17]Bhutta K M,Schmalhorst J and Reiss G2009J.Magn.Magn.Mater.3213384[18]Infante C,Sanchez F,Laukhin V,P´e rez del Pino A,Foncuberta J,Bouzehouane K,Fusil S andBarth´e lemy A2006Appl.Phys.Lett.89172506[19]Simmons John G1963J.Appl.Phys.341793[20]Tsoi M2008Nature Phys.417[21]Chanthbouala A et al2012Nature Mater.11860[22]Garc´ıa V,Fusil S,Bouzehouane K,Enouz-Vedrenne S,Mathur N D,Barth´e l´e my A and Bibes M2009Nature48081[23]Warusawithana M P et al2009Science324367[24]Bardou F1997Europhys.Lett.39239[25]Fix T,Da Costa V,Ulhaq-Bouillet C,Colis S,Dinia A,Bouzehouane K and Barth´e l´e my A2007Appl.Phys.Lett.91083104。