Quasi one-dimensional transport in single GaAs/AlGaAs core-shell nanowires D. Lucot, F. Jabeen, J.-C. Harmand, G. Patriarche, R. Giraud, G. Faini, and D. Mailly Laboratoire de Photonique et de Nanostructures (LPN-CNRS), route de Nozay, 91460 Marcoussis, France (Dated: January 4, 2011) WepresentanoriginalapproachtofabricatesingleGaAs/AlGaAscore-shellnanowirewithrobust and reproducible transport properties. The core-shell structure is buried in an insulating GaAs overlayerandconnectedasgrowninatwoprobeset-upusingthehighlydopedgrowthsubstrateand a top diffused contact. The measured conductance shows a non-ohmic behavior with temperature and voltage-bias dependences following power laws, as expected for a quasi-1D system. 1 Semiconductor nanowires (NWs) are promising candi- achievemodulation-dopedGaAs/AlGaAscore-shellNWs 1 dates for the realization of innovative nano-devices for embedded in a GaAs matrix. The NWs are individually 0 electronics as well as for photonics[1]. They also rep- connected directly on their growth substrate and show a 2 resent a new test bed system in semiconductor mate- smallcontactresistanceenablinglow-temperaturetrans- n rial science to explore the fundamental properties of 1D port measurements. Clear signatures of their quasi-1D a systems[2]. Usual fabrication of such nano-objects by characterareobserved,openingupnewopportunitiesfor J lithography and etching techniques is limited by sur- the investigation of quantum transport. 2 face damage and roughness which, at very small sizes, have a dominant effect on their physical properties. In ] i contrast, the metal particle-mediated vapor-liquid-solid c (VLS) growth mechanism allows to obtain NWs of high s - crystalline quality with uniform nanometer-scale diame- rl ters. Moreover, this bottom-up approach gives the pos- t sibility to achieve heterostructure material combinations m that are not possible in bulk semiconductors[3]. In par- . t ticular core-shell heterostructures formed by the growth a of crystalline overlayers around the initial NW reduces m surface states which can act as scattering or recombina- - tion centers[4]. d n The GaAs/AlGaAs material system, which presents o very large band offsets and small lattice mismatches, c [ has been studied and used extensively to fabri- cate heterostructures with complex band engineering FIG. 1: Scanning Electronic Microscope (SEM) pictures of 1 freestanding25nmdiameterGaAsNWsbefore(a)andafter along the growth direction of the crystal[5]. The v (b) 40nm AlGaAs shell growth. (c) Tilted SEM micrograph modulation-doping concept[6] was first implemented in 1 ofthetopendofGaAs/AlGaAscore-shellNWafterapartial this system and high electron mobility transistors were 2 buryinginGaAs. (d)SchematicsofaGaAs/AlGaAsmodula- 4 demonstrated[7]. Nowadays, a low-temperature elec- tion doped core-shell NW buried into a semi-insulator GaAs 0 tronmobilityofseveral106cm2/V.siscurrentlyobtained matrix. . in 2D structures [8]. To achieve 1D carrier confine- 1 0 ment it is thus of particular interest to use the well- GaAs(111)Bsubstratesarepatternedwithnano-sized 1 known GaAs/AlGaAs system and realize core-shell NW golddotsrealizedbyEBLandlift-off,atwelldefinedsur- 1 by wrapping a GaAs NW core with an AlGaAs shell face locations. The samples are introduced in the MBE : layer. VLS technique has been used to produce such v equipment and deoxidized at relatively low temperature Xi GaAs/AlGaAs core-shell NWs[9–11], but up to now (350◦C) under an atomic hydrogen flux. Then, the sam- theseheterostructureshavebeenmainlycharacterizedby pletemperatureisincreasedupto550◦CforAu-assisted r photoluminescence[11, 12] due to difficulty in achieving a VLS growth[13]. At this temperature, the gold dots are good electrical contacts. alloyed with Ga and form small droplets. The exact The standard technique to contact a NW is to me- size of the droplet is determined by the original diam- chanically detach them from the growth substrate, and eter of the EBL patterning, the thickness of the Au film after dispersion on an insulating substrate, to evaporate and the growth conditions. Ga and As fluxes are sup- 4 contact pads by lithography. This technique may suf- plied to form nominally undoped vertical GaAs NWs by fer from the random positioning of the wires, from the VLS,inepitaxialrelationshipwiththesubstrate. Atthis oxidation of the AlGaAs shell and from the presence of stage, the crystalline structure of these NWs is wurtzite- pinned charges in the insulating subtrate. In this let- like with several stacking faults[14]. The NWs present a ter,wepresentanoriginalmethodusingmolecularbeam regular hexagonal cross-section limited by (1010) facets. epitaxy (MBE) and electron-beam lithography (EBL) to Alongtheirlength,theyhaveaconstantlateralsize,close 2 tothediameterofthecatalystdrop(Fig. 1a). Thisdiam- eter can be easily tuned between 20 nm and 100 nm and the NW length, fixed by the growth time, is about 1µm. Then, an AlGaAs shell is formed on the NW facets (Fig. 1b). The sidewall growth is promoted by the presence of AlandbyswappingtheAs fluxtoAs flux,theadatom 4 2 diffusion length on the sidewall facets being strongly re- duced under these conditions[15]. The nominal Al com- position of the shell is 30%. A first radial layer of 10 nm is formed, then Si flux is supplied to give a Si δ-doping with a nominal sheet concentration of 1×1012cm−2 on the facets. The shell is completed with a 30 nm AlGaAs layer. Hence, in this structure, we expect a transfer of freecarriersintheGaAscorewherenodopingimpurities are introduced. Because it results from epitaxial nucle- ation on the NW sidewalls, the shell adopts the wurtzite structure of the core. It has to be noted that during the shell formation, similar layers deposit on the substrate surface and on top of the NWs (Fig. 1d). FIG. 2: Typical temperature dependence of the resistance R (T)normalizedtoR(300K)forasingleGaAs/AlGaAscore- Finally, the resulting core-shell NWs are buried (Fig. shell NW with a 25 nm GaAs core diameter. (inset) SEM 1c)byanundopedGaAsovergrowthperformedathigher picture of a single NW partially buried in an undoped GaAs temperature (630◦C). At this temperature, the VLS matrixandelectricallycontactedtoaNi/Ge/Auelectrodeat growth is inhibited and there is no lateral growth on the its top end. NWsidewalls. GaAsgrowsonlyonthesubstratesurface, adopting its zinc blende bulk structure and embedding progressively the NWs. We have shown that this pro- of multi walled carbon nanotube (CNT) [17]. A good cess transforms the initial wurtzite crystal phase of the contact coupling to NW is of particular interest because NWs in the zinc blende phase adopted by the burying it sets our device in a conduction regime where classi- layer[14]. The phase transformation is very favorable in calCoulombchargingeffectsduetothepresenceoflarge that it eliminates the stacking faults pre-existing in the tunnel barriers are negligible [17, 18]. Moreover, the dis- NWs. Wethusendupwithperfectcubiccore-shellNWs tinctionbetweentwo-andfour-terminalresistanceisless embedded in a cubic matrix. Since they no longer have pertinent and the two-points configuration used in the free sidewall facets, these NWs are perfectly protected presentstudyisexpectedtocapturemostoftheintrinsic against aging by surface oxidation and their electronic physics. properties are not influenced by surface states. This is Now we focus on the temperature dependence of the confirmed by the excellent reproducibility of our trans- NW resistance. As shown on figure 2, two conduction port measurements over several months. regimes can be distinguished. As the temperature is de- The planarization of the sample induced by the GaAs creased from RT to 80 K, the resistance exhibits little layerovergrowthenablestheconnectionofsingleNWsdi- variations. Asimilarbehaviorhasbeenalreadyreported rectly on their substrate (Fig. 1d). A first contact is ob- in Si/Ge core-shell NWs [4] and is ascribed to the weak tainedontheemergingpartoftheNWbyusingaconven- acousticphononcontributionin1Dsystems. Below80K, tionalNi/Ge/AufilmdepositionpatternedbyEBL(Fig. the resistance increases while lowering the temperature. 2. inset). A second one is directly taken on the backside This increase cannot be explained by a strong localiza- of the highly doped n-type substrate. After a rapid (5 tion effect. Indeed, we cannot fit our experimental data s)thermalannealingofthedeviceat430◦C,weobtained withanactivationlaw(ln(R)∝1/T)oravariablerange good ohmic current-voltage (I-V) characteristics with a hoppingbehaviour(ln(R)∝1/T1/2,fora1Dwire),con- total two-probe resistance ranging from 400Ω to 2kΩ at firming the absence of any large tunnel barrier within or room temperature (RT), for all the 12 different single at the ends of the NW. NWs of 25 nm GaAs core diameter investigated[16]. The low temperature behavior is further investigated The electrical properties of single NWs are measured by measuring the differential conductance G=dI/dV of in this two-probe configuration in a 4He VTI cryostat a single NW device as function of the bias-voltage, for using current-biased lock-in techniques. The first im- temperature ranging from 65 to 4.2 K (Fig. 3a). A con- portant observation is that the resistance of the sam- ductance dip centered at zero bias (zero-bias anomaly) ples is significantly lower than the quantum of resistance appears and grows up as the temperature is decreased. (h/2e2 ∼12.9kΩ)expectedforasinglepropagatingmode As shown in the Fig. 3a insert, the temperature depen- connected to ideal contacts. This suggests that we ob- dence of the zero-bias conductance (G ) is well de- V=0 tain high-transparency contacts and that a single NW scribed by a power law G ∝ Tα with α = 0.076. V=0 sustainsseveralpropagatingmodes, similarlytothecase In the same way the voltage dependence can also be de- 3 scaling law with an α coefficient distributed from 0.02 to 0.23andwithnoclearcorrelationtotheirRTresistance. Such scaling laws have been previously observed on many other low dimensional systems including single or multiwalledCNTs[17–19],metallicquantumwire[20,21] orsplit-gatedNWsformedonplanarGaAs/AlGaAshet- erostructures[22]. Thisbehaviourisrelatedtotheenergy cost for an electron from the 3D reservoir to tunnel into a 1D system[23]. Indeed, adding an electron to a pure 1D conductor requires changing the many body state of its collective excitations. It yields to a vanishing elec- tron tunnellingdensityof states at low energy leading to a power law dependence of the conductance on V and T. The exponent α depends principally on the electron- electron interaction strength and on the number of con- ducting channels [21]. For a clean single 1D propagating channel, electron-electron interaction is expected to give α ≈ 0.5[22]. The lower α values extracted here are sim- ilar to those obtained for multi-wall CNTs [17, 19] and indicatethatourNWsarequasi-1Dsystemswithalarge number of 1D conducting channels in parallel. In conclusion, we have developed an original fabrica- tion method to obtain buried GaAs/AlGaAs core-shell NWs with a Si δ-doping in the AlGaAs shell showing a clear quasi-1D behavior. Their geometrical, structural FIG. 3: (a) Differential conductance G = dI/dV of a single and electrical characteristics are easily tunable (as com- NW as a function of bias voltage for different temperatures. pared to the case of CNTs for instance) by adjusting The inset shows a log-log representation of G(V=0) versus the process parameters. The transport properties do not T. The dependence follows a power-law with the exponent evolve with time and the direct contacting of the NW α = 0.076. (b) GT−α versus eV/k T plot of the same data B on the growth substrate allows one to create a complex as in (a), using α=0.076. routing of connections leading to large scale integration. Thus, these new classes of materials are of broad inter- est for future applications in nanoelectronics, and are scribedbyapowerlawG ∝Vαwiththesameexponent V well-suited to study phase coherent quantum transport α = 0.076, for bias voltages larger than k T/e. Both B in one-dimension. these laws can be resumed by a scaling law GT−α = f(eV/k T) where f is the scaling function and α is a The authors thank Jean-Eric Wegrowe, Ulf Gennser, B scalingcoefficientspecifictothemeasuredNW.Thus,the for helpful discussions and Laurent Travers, Christian G(V)curvesmeasuredatdifferenttemperaturescollapse Ulysse for their technical assistance. This work was remarkably well on to a single curve GT−α vs eV/k T, partly supported by the R´egion Ile-de-France in the B as depicted in Fig. 3b. All the measured NWs exhibit a framework of C’nano IdF. [1] Y. Li, F. Qian, J. Xiang, C.M. Lieber, Mater. Today 9, 18 Phys.Lett.33,885(1978). (2006). [7] T. Mimura, S. Hiyamizu, H. Hashimoto, M. Fukuta, IEEE [2] Y.J.Doh,J.A.vanDam,A.L.Roest,E.P.A.M.Bakkers,L.P. Transactionsonelectrondevices27,2197(1980). Kouwenhoven, S. De Franceschi, Science 309, 272 (2005); C. [8] V. Umansky, M. Heiblum, Y. Levinson, J. Smet, J. Nubler, Fasth,A.Fuhrer,L.Samuelson,V.N.Golovach,D.Loss,Phys. M.J.Dolev,J.Cryst.Growth311,1658(2009). Rev.Lett.98,266801(2007). [9] J. Noborisaka, J. Motohisa, S. Hara, T. Fukui, Appl. Phys. [3] L.J. Lauhon, M.S. Gudiksen, D. Wang, C.M. Lieber, Nature Lett.87,093109(2005). 420,57(2002). [10] M.J.Tambe,S.K.Lim,M.J.Smith,L.F.Allard,S.Grade- [4] W.Lu,J.Xiang,B.P.Timko,Y.Wu,C.M.Lieber,Proc.Natl. cak,Appl.Phys.Lett.93,151917(2008). Acad.Sci.USA102,10046(2005);Y.Hu,J.Xiang,G.Liang, [11] P.Parkinson,H.J.Joyce,Q.Gao,H.H.Tan,X.Zhang,J.Zou, H.Yan,C.M.Lieber,NanoLett.8,925(2008). C. Jagadish, L.M. Herz, M.B. Johnston, Nano Lett. 9, 3349 [5] see e.g. F. Capasso, Physics of quantum electron devices, (2009). Springer series in electronics and photonics 28, Springer- [12] K.Tomioka,J.Motohisa,S.Hara,K.HirumaT.Fukui,Nano Verlag1990.. Lett.10,1639(2010). [6] R.Dingle,H.L.Stormer,A.C.Gossard,W.Wiegmann,Appl. [13] J.-C. Harmand, M. Tchernycheva, G. Patriarche, L. Travers, 4 F.Glas,G.E.Cirlin,J.Cryst.Growth301,853(2007). Balents,P.L.McEuen,Nature397,598(1999). [14] G. Patriarche, F. Glas, M. Tchernycheva, C. Sartel, L. [19] A.Bachtold,M.deJonge,K.Grove-Rasmussen,P.L.McEuen, Largeau, J.-C. Harmand, G.E. Cirlin, Nano lett. 8, 1638 M.Buitelaar,C.Schonenberger,Phys.Rev.Lett.87,166801 (2008). (2001). [15] C.Sartel,D.L.Dheeraj,F.Jabeen,J.-C.Harmand,J.Cryst. [20] E.Slot,M.A.Holst,H.S.J.vanderZant,S.V.Zaitsev-Zotov, Growth312,2073(2010). Phys.Rev.Lett.93,176602(2004). [16] We checked the insulating behavior of the burying layer by [21] L. Venkataraman, Y.S. Hong, P. Kim, Phys. Rev. Lett. 96, measuring the resistance between the bottom contact and a 076601(2006). topcontactdepositedonasamplesurfaceareawithnoemerg- [22] Y.Jompol,C.J.B.Ford,J.P.Griffiths,I.Farrer,G.A.C.Jones, ing NW. In that case, the resistance is always higher than D.Anderson,D.A.Ritchie,T.W.Silk,A.J.Schofield,Science 10MΩ. 325,597(2009). [17] K. Liu, P. Avouris, R. Martel, W.K. Hsu, Phys. Rev. B 63, [23] V.V.Deshpande,M.Bockrath,L.I.Glazman,A.Yacoby,Na- 161404(2001). ture464,209(2010). [18] M. Bockrath, D.H. Cobden, A.G. Rinzler, R.E. Smalley, L.