carbon nanotube y junctions: growth and properties · 242 l.p. biro et al. / diamond and related...

Diamond and Related Materials 13(2004) 241–249

0925-9635/04/$ - see front matter� 2003 Elsevier B.V. All rights reserved.doi:10.1016/j.diamond.2003.10.014

Carbon nanotube Y junctions: growth and properties

L.P. Biro *, Z.E. Horvath , G.I. Mark , Z. Osvath , A.A. Koos , A.M. Benito , W. Maser ,a, a a a a b b´ ´ ´ ´ ´Ph. Lambinc

Research Institute for Technical Physics and Materials Sciences, P.O. Box 49, H-1525 Budapest, Hungarya

Instituto de Carboquimica (CSIC), CyMiguel Luesma Castan 4, E-50015 Saragossa, Spainb ´Laboratoire de Physique du Solide, Facultes Universitaires Notre-Dame de la Paix, B-5000 Namur, Rue de Bruxelles 61, Belgiumc

Abstract

An overview of the available data on carbon nanotube Y junctions is given. The structural models and the transport calculationson the Y junctions are reviewed followed by the various methods, which successfully produced these junctions and the availableexperimental transport and tunneling data of the grown junctions. In the discussion section, the common and different features ofthe various growth methods are analyzed and some particularities of branched nano-objects are outlined, which have to be takenaccount in the interpretation of the tunneling data.� 2003 Elsevier B.V. All rights reserved.

Keywords: Carbon nanotubes; Y junctions; Growth; Structure; Electronic transport; Tunneling

1. Introduction

The increasing number of papers on novel, complexcarbon nanotube type nanoarchitectures like multiwallw1x and singlewall Y junctionsw2x, simple w3x andmultiple coiled nanotubesw4,5x clearly shows that thesp hybridization of carbon makes possible a large2

variety of tubular, carbon based nanostructures. Thesenew structures are potential building blocks of an ‘all-carbon’ nanoelectronics. Such a circuitry, in which theinterconnects to the macroscopic world are significantlyreduced in number, will fully exploit the nanometer sizeof its components. To be able to achieve this, networksof interconnected carbon nanotubes have to be construct-ed. For constructing a network the simplest basic build-ing elements one can use are Y or T junctions eitherachieved during nanotube growthw6–8x, or by laterprocessing steps like functionalization and interconnec-tions through chemical bondsw9,10x.The purpose of the present paper is to overview the

available data on the growth and properties of the carbonnanotube Y junctions produced during nanotube growth,the junctions produced by chemical interlinking are not

*Corresponding author. Tel.:q36-1-3922681; fax: q36-1-3922226; http:yywww.mfa.kfki.huyintynano.

E-mail address: [email protected](L.P. Biro).´

within the scope of the present paper. The paper isorganized as follows: first the structural models and thetransport calculations on the Y junctions are reviewedfollowed by the various methods, which successfullyproduced these junctions and the available experimentaltransport and tunneling data of the grown junctions. InSection 4 the common and different features of thevarious growth methods are analyzed and some partic-ularities of branched nano-objects are outlined, whichhave to be taken account in the interpretation of thetunneling data.

2. Structural models and calculated transportproperties

2.1. Structure

The first structural models for symmetric carbonnanotube Y junctions based on theoretical calculationsw11,12x were proposed shortly after the discovery ofmultiwall carbon nanotubes by Iijimaw13x. Both modelsare based on the insertion of non-hexagonal(n-H) rings(at least six heptagons) in the hexagonal network in theregion where the three branches of the Y are joinedtogether. All the subsequent structural modelsw14–19xfollow the same construction principle of conserving thesp hybridization of the carbon network, differing only2

242 L.P. Biro et al. / Diamond and Related Materials 13 (2004) 241–249´

Fig. 1. Structural models of carbon nanotube Y junctions.(a) Sym-metric, armchair Y junction as proposed by Scuseriaw11x, the sixheptagons are highlighted in black;(b) asymmetric, zig-zag Y junc-tion after w18x, n-H rings highlighted in black.

Fig. 2. Triangulene spacers as proposed inw15x to join three metalliczig-zag carbon nanotubes into an Y junction.(a) Triangulene withatom centeredC axis; (b) ring centeredC axis.3 3

in the kind, number and placement of the n-H rings.These variations make possible the constructions ofvarious symmetric and asymmetric model junctionsw18xand various angles from Y to T shapesw14x. A Yjunction is named symmetric if the three carbon nano-tubes joining each other in the Y have identical chiralityand the distribution of the n-H rings around the Y issymmetric. Such a junction will be constituted fromidentical branches oriented at 1208, like in Fig. 1a.Whenever one of the above conditions is not fulfilledthe junction will be asymmetric, a possible example isshown in Fig. 1b. The case of asymmetric junctions issomewhat more complex as in this case various combi-nations of metallic and semiconductor tubes can bebuilt. Even an asymmetric Y built only of semiconductor

carbon nanotubes, but with different diameters, mayexhibit rectification effects due to the different gapvalues associated with the different diameters leading tothe formation of a system similar to semiconductorheterojunctions in the Yw20x.

2.2. Conductance

Nonconducting bias windows may appear in the Yjunction region, even for symmetric junctions built ofmetallic carbon nanotubes due to the number of then-H rings and due to the way in which they are arrangedw15x. Treboux and co-workers have calculated the prop-erties of a series of spacer elements in the junctionregions, which they call ‘triangulenes’, Fig. 2. Accordingto their model the complete Y junction is treated asbeing composed of three semiinfinite, metallic(12, 0)zig-zag carbon nanotubes joined by the triangulene unitsand the necessary n-H rings to conserve the sp connec-2

tivity of the carbon lattice. According to their calcula-tions if the triangulene spacer has an atom centered ontheC symmetry axis, Fig. 2a, it exhibits no conduction3

243L.P. Biro et al. / Diamond and Related Materials 13 (2004) 241–249´

gap, only an electron hole asymmetry, but when thetriangulene spacer has a ring centered on theC axis,3

Fig. 2b, a characteristic gap is found in the conductancew15x. Every third increase of the size of triangulenespacer(every increase is done by increasing the numberof the hexagons along the edge of the spacer by one)results in a spacer with ring centeredC axis. The3

existence of the conduction gap is associated with LDOSdifferences arising in the two kinds of spacers, thecentral region of the triangulene spacer with ring cen-teredC axis loses the metallic character.3

2.3. Rectification

The rectification in the Y junction is a somewhatcontroversial issue. Andriotis et al.w18x calculated thequantum conductivity for a large number of Y junctionsusing the Landauer formalismw21x based on a Green’sfunction method. Not only the nanotube branches of theY junctions are taken into account, but also the effectof the metallic leads. The effect arising due to themetallic leads are introduced by calculating the selfenergy term for those carbon atoms which are situatedon the interface plane between the nanotube and its Ni(0 0 1) lead w18x. Two main groups of structures arestudied:(a) junctions with no change of chirality and(b) change of chirality on branching. The authorsconclude that in accordance with Treboux’s findings therectification and switching depend strongly on the sym-metry of the spacer connecting the three branches and,to a lesser degree, on chiralityw18x. They find: (i)perfect rectification is obtained for symmetric carbonnanotube junctions consisting of zig-zag stems;(ii)asymmetric Y junctions show no rectification;(iii ) Yjunctions containing armchair stems, although havingasymmetricI–V characteristics, exhibit small leakagecurrents for positive bias;(iv) perfect rectification andswitching properties are intrinsic properties of the Yjunctions exhibiting specific symmetry and consistingof zig-zag stems. Using a similar calculation method, ina recent work, Meunier et al.w19x bring arguments thatthe rectification in a symmetric junction is not anintrinsic property of the branching, but it is solely dueto the properties of the interface between the nanotubebranches and the metallic leads. The rectification isattributed to Schottky-type interface states and the mag-nitude of the effect is found to be strongly dependenton the detailed atomic geometry of the contactw19x. Inaddition, it seems now well established that the elec-tronic properties of a nanotubeylead interface dependstrongly on contamination, especially with oxygenw22x.From the point of view of carbon nanotube based

nanoelectronic circuitry, both kinds of Y junctions areneeded: ones that show no rectification for interconnectsand ones that show rectification for being used as activeelements.

3. Experimental results

3.1. Growth of Y junctions

3.1.1. Electric arcThe first observation of carbon nanotube Y junctions

was reported in an electric arc experiment. The junctionsfound by transmission electron microscopy(TEM) wereshort branched, multiwall onesw23x. The arc dischargewas operated under He atmosphere at 660 mbar and acurrent density of 220 Aycm was used at a 27 V DC2

bias. A hollow, cylindrical graphite rod of 0.64 cm witha hole of 0.32-cm diameter was used as anode, no fillingwas present in the hollow core of the anode. The depositfrom the cathode was collected and analyzed by TEM,junctions of various shapes L, T and Y were found. TheHRTEM images show that in the branches the graphiticlayers are oriented parallel with the tube axis, while inthe junction region they have ‘saddle-like’ shape.Recently in a DC electric arc experiment carried out

in 660 mbar He, also using a hollow anode, of 6-mmdiameter, drilled out to have a cylindrical, axial hollowof 3.5 mm in diameter and filled with a nickelyyttriummixture (resulting in an overall graphiteynickelyyttriumcomposition of 90y7y3 wt.% over the drilled-out length)single and few wall Y junctions were found by scanningtunneling microscopy(STM) in the cathodic deposittogether with straight multiwall carbon nanotubes andnanotube knees with diameters typically approximately10 nm w24x. The Y junctions found are asymmetric,usually have a stem of larger diameter, the ratio of thediameter of the stem to the diameter of the branches isin the range of 1.5–4, the apparent diameters of thestems as measured from the height of the tubes over thesubstrate are in the range of 0.3–1 nm. Taking intoaccount that the measured values of the apparent heightsin STM measurements are affected by several distortingfactors w25–27x, which cause the reduction of the realdiameter, these values strongly suggest single or fewwall branches. The observed Y junctions usually showgrowth instabilities and sometimes kinks, too, beforebranching, a typical example of such a Y junction isshown in Fig. 3, some of the growth instabilities aremarked by white arrows.STM and scanning tunneling spectroscopy(STS)

measurements were reported recently on nanotube sam-ples prepared by the electric arc method. Neither theexact location from which the sample was collectedwithin the reaction chamber, nor the arc dischargeparameters were given in detailw28x. An asymmetric Yjunction was measured in topographic mode and bySTS. The apparent heights of the three branches are:0.45, 0.4 and 0.35 nm, respectively. The same observa-tion is valid concerning the real diameters as in theprevious paragraphw25–27x. The diameter valuesobtained from the STS data indicate tubes with surpris-


Fig. 3. Topographic STM image of an arc grown asymmetric Y junc-tion w24x shown in 3D. Some of the growth instabilities causing minorchanges in the growth direction before the Y branching occurred aremarked by white arrows.

Fig. 4. Some remarkable experimental achievements by CVD growthof Y junctions.(a) Multiwall, symmetric Y junctions with well devel-oped central holloww8x; (b) Y junction with the triangular amorphousparticle at the joining of the branches(the particle is shown in detailin the right-upper inset), central inset shows the pearl-shaped particleat the end of the branchw31x (scale bar 100 nm); (c) double Y junc-tion grown using the pyrolysis of thiophene over NiySiO catalyst2

w30x; (d) SEM image of an area showing several multi-junction tubesgrown on scratched Si using Fe powder as catalystw33x.

ingly small diameters in the range 0.5–0.7 nmw28x,which clearly must be single wall ones. From spatiallyresolved spectroscopic data, by plotting the dIydV quan-tity as a function of bias voltage and position along thetube, the authors conclude the existence of metallic-semiconductor junctions in the region where the threenanotube building up the Y are joined togetherw28x.

3.1.2. Catalytic growthIn the past 3 years most of the works reporting the

growth of the Y junctions were using some variation ofthe catalytic vapor deposition(CVD) method. Thequality of the grown junctions shows wide scattering.Using microwave plasma enhanced catalytic vapor

deposition and Pd catalyst, partially Pd filled, multiwallcarbon nanotubes were grown on a porous Si supportw29x, in methaneyhydrogen atmosphere. Some of thetubes show Y junctions. The authors propose a rootgrowth mechanism for the formation of Y junctions inwhich the molten catalyst particles on the top of twoneighboring carbon nanotubes are fused together andthe growth continues like a single tube.The formation of Y branched carbon nanotubes and

of multiple branchings in a proportion of 70% is reportedby Satishkumar et al.w8x. Nickelocene is used as bothcatalyst source and carbon feedstock in combinationwith hydrogen bubbled through thiophene in argoncarrying gas, the growth apparatus consists of a twofurnace system, the nickelocene is sublimed at 623 Kin the first furnace, while the growth of the Y junctionstakes place at the outlet of the second furnace kept at

1273 K. The Y junctions are constituted of multiwall,relatively straight tubes, in general of similar diametersin the range of 40 nm, with a well developed and clearlyvisible central hollow(Fig. 4a). As revealed by HRTEMthe Y junctions have a ‘fish-bone’ structure, i.e. thegraphitic layers are not oriented parallel with the axisof the nanotubes constituting the junction, but ratherunder a certain angle. It is remarkable that even giventhis structure, the STS spectroscopy measurements per-formed on one branch of a junction show a symmetricdIydV, while at the joining point an asymmetric char-acteristic is foundw8x. The authors attribute the forma-tion of the Y junctions to the catalytic effect of thethiophene present in the gas mixture. In a subsequentwork w30x from the Rao group, the cobaltocene, ferro-cene, Ni- and Fe-phtalocyanines and Ni(Fe)ySiO were2

found to yield Y junctions under identical conditions asabove. Multiple branching was observed, too,(Fig. 4c),but the junctions continue to show fish-bone structure.Y junctions with a shorter and two longer straight

arms that can reach up to 10mm in length, and oftenexhibiting multiple branching were grown using MgOsupported Co catalyst in a HyCH gas mixture at a2 4

temperature of 10008C w31x. The branches of the Yjunction usually have similar diameters and they areoriented at angles close to 1208. The three tubes joinedin the Y have hollow cores and in general an amorphous,triangularly-shaped particle, consisting of Ca, Si, Mgand O is found at the joining of the hollow cores(Fig.4b). It is suspected that Ca and Si are introduced asimpurities in the starting materials. It is frequently


observed that one of the two long arms of the Y junctionis capped with a pearl-shaped particle, which has thesame chemical composition as that of the triangle-shapedparticle, (Fig. 4b). Sometimes a trace amount of Cowas found on the surface of the pearl-shaped particle.HRTEM investigation revealed that far from the junctionthe branches show a tubular arrangement of the graphenelayers, with satisfactory graphitization. The authors con-clude that the growth is promoted by the pearl-shapedparticles observed at the end of the arms, which originateby splitting due to thermal fluctuations from the trian-gular particle at the junction.Y junction carbon nanotubes have been grown by

hot-filament chemical vapor deposition using a tungstenfilament heated at 22008C for which a gas mixture ofacetone and hydrogen was fed and in situ evaporatedcopper from a copper plate placed under the hot filamentwas suppliedw32x. TEM investigation revealed thatunder optimized growth conditions the fraction of Yjunctions is as high as 30%, usually a thicker stembranches in two thinner arms separated at anglesbetween 50 and 808. HRTEM reveals fish-bone structurein the junction. If no copper is supplied during growth,no nanotubes are observed.Multijunction carbon nanotube networks, Fig. 4d,

containing Y and H junctions with regular, straight arms,were grown on 600-grit sand paper scratched Si sub-strates using iron powder placed in a quartz boatupstream of the substrate as catalyst source, and thepyrolysis of methane at 11008C as carbon sourcew33x.The branches have uniform diameters in the range of30–50 nm.

3.1.3. Growth using templateThe first experiment achieving the controlled growth

of carbon nanotube Y junctions was based on the useof alumina templates with branching nanochannelsw1x.Cobalt catalyst was electrochemically deposited at thebottom of the nanochannels then the pyrolysis of acet-ylene was carried out at 6508C. The grown Y junctionshave stems of 3-mm length with a diameter of 90–100nm, followed by branches of 2mm with diameters of35–60 nm making a very sharp angle. HRTEM inves-tigation of the tube walls reveals fish-bone structure.Electric transport measurements of individual Y junc-tions placed over Au electrodes, with the branchesconnected to the same electrode showed rectifyingbehaviorw20x. Similar results were obtained for arraysof Y junctions constituted of several hundreds of indi-vidual nano-objects, while no rectification was foundfor arrays of straight carbon nanotubes measured underidentical conditions.A very similar growth procedure was carried out at

somewhat lower temperature, 5508C and adding hydro-gen to the gas mixture. Multibranched junctions were

grown in nanochannels which were purposely etched inway that gives multibranched structurew34x.

3.1.4. Growth by fullerene decompositionDuring the electric arc, or laser ablation growth of

carbon nanotubes, small carbon clusters are formed inthe hot plasma, which under favorable conditions willaggregate into straight or kinked carbon nanotubes. Onthe other hand, such carbon clusters may be formed atmuch lower temperatures than that typical for electricarc or laser ablation, for example by fullerene decom-position in the presence of transition metals at temper-atures in the range of 4508C w35x. The pure carbonbeam, if directed onto a highly oriented pyrolytic graph-ite (HOPG) surface in certain cases leads to the growthof complex nanoarchitectures, like carbon nanotube Yjunctions, regularly coiled carbon nanotubes and com-binations of these structuresw6,36,37x, (Fig. 5a). Acompletely symmetric Y junction, with arms oriented at1208 is shown in Fig. 5b. The carbon nanotube Yjunctions grown by this technique are usually single orfew wall junctions for which the possibility of the fish-bone structure can be ruled out due to the small thicknessof the tube walls. Mass spectrometry has shown that inthe low temperature pure carbon beam, there are presentnot only small C clustersw35x, but also clusters with2n

masses in the range of C and C w38x. It is117 174

speculated that these large clusters, after soft-landing onthe HOPG surface may take configurations, which pro-mote the growth of Y junctions and other tubular formsof carbon containing n-H rings, like coils and doublecoils w39x.

3.1.5. Welded junctionsExperiments carried out in situ at high temperatures

(800 8C) in a high acceleration voltage TEM(1.25 MV,JEOL ARM-1250, beam intensity off10 Aycm ),2

supported by molecular dynamic calculations, show thatdue to interlinking of dangling bonds created by electronirradiation, it is possible to weld together crossingsinglewall carbon nanotubes and to produce Y, X, T andH junctionsw40x (Fig. 6). By careful electron irradiation,it is possible to remove one arm of an X junction totransform it into a Y junction. Although the experimen-tally obtained junctions appear to a certain extent defec-tive and their atomic arrangement might deviate fromthe idealized molecular models, the observed junctionsshow the expected topologies. Tight binding moleculardynamics calculations were carried out, the simulationof two crossing tubes under irradiation at 10008Crevealed the nanotube merging process, resulting in analmost perfect molecular junction.

3.2. Electronic transport measurements

3.2.1. Transport along the junctionAlthough many various methods have been used to

grow Y junctions, much less experimental data are


Fig. 5. Top view, topographic STM images of carbon nanotube Y junctions grown from carbon clusters produced by fullerene decomposition inthe presence of transition metals.(a) Y junction and regular nanotube coil laying over it,w37x; (b) top view STM image with artificial illuminationof a completely symmetric Y junction, with arms oriented at 1208, the irregular line AB is a cleavage step, while in C a surface defect is seen,the white lines indicate the 1208 anglew36x.

Fig. 6. Carbon nanotube junctions welded in situ under the electronbeam in the TEMw40x.

available on their conduction properties. This seems tobe related to the difficulty of precisely contacting agiven nano-object. A very ingenious method to measureconduction through the Y junctions grown in the nano-channels of an alumina template give solid evidence onthe rectifying behavior of arrays of such junctions,averaging over large numbers 100–10 000 of individualjunctions connected in parallelw20x. Arrays of straighttubes under identical experimental conditions exhibitedlinearI–V characteristic. The Y junctions were measuredwith their thinner branches connected to the samegrounded Al electrode, while the stems were connectedto an Au electrode. For positive bias applied to the Auelectrode no current, or only weak leakage current ismeasured, while for negative stem polarity currents of300 mA were measured aty2 V bias for an array of100 junctions. In an experiment carried out with a singlejunction chemically removed from the growth templateand placed over preformed Au electrodes, after a periodof contact resistance reduction due to gold electromigra-tion, similar characteristic was measured between twoAu contacts as for the arrays between AlyAu contacts.The experimental findings are interpreted in the frame-work of the Anderson modelw41x for semiconductorheterojunctions. On the basis of STS data for straightmultiwall carbon nanotubesw42x showing that the gapof semiconductor tubes is proportional to the inverse oftube diameter, it is inferred that the stem has smallergap than the two arms. This will cause depletion on thebranch side of the junction. The current through the

junction can be written asw20x

I;exp(eVyhk T), (1)B


wheree is the electronic charge,V is the bias across thejunction, hs1q´ N y´ N a parameter depending on2 2 1 1

the material properties on the two sides of the junction,k the Boltzmann constant,T the absolute temperature;B

with ´ and ´ the corresponding dielectric constants1 2

and N and N the equilibrium concentrations of the1 2

holes in the stem(branch) far from the junction. It wasfound that the experimentalI–V data are in goodagreement with Eq.(1).

3.2.2. Transport across the branches and the junctionTunneling conductance measurements were carried

out on a CVD grown Y junction by positioning a tipover the arm and over the joining point of the branchesw8x. While over the arm a symmetricI–V curve wasrecorded in the branching region theI–V curve showsasymmetry, smaller currents being measured for negativepolarity than for positive polarity.STS measurements carried out under UHV on an

asymmetric single wall Y junction produced in electricarc revealed changes in the electronic properties in thevicinity of the junction w28x. Spectroscopic data wererecorded as a function of bias voltage and position alongthe tubes, the data represented as gray scale tunnelingconductance maps reveal that the region connecting thethree branches exhibits metallic-type behavior and itsdIydV curve shows peaks in the range of"1.0 to"1.25V with respect to the Fermi level. The measurementsalong the branches clearly show the gradual changetowards the spectrum characteristic for the junctionregion as the measurement points are approaching tothis region.

4. Discussion

One of the general conclusions that may be formulat-ed after the review of the experimental data is thattaking into account the very different methods whichyielded in some occasions very similar nanotube Yjunctions, the research community should accept thesenano-object as ‘regular members’ of the carbon nano-structures family. A certain grouping of the methodsmay be attempted, although, for some growth methodsthe available data are very scarce.The most well represented group is constituted by the

Y junctions grown by CVD. Most frequently, thesejunctions exhibit a fishbone structure which may suggesta growth mechanism based on the fragmentation of thecatalyst particle as proposed in Ref.w31x. On the otherhand, the experiments by Rao and co-workersw30xconvincingly demonstrate that various catalyst are suit-able for the formation of Y junctions and that thepresence of sulfur may be beneficial for the formationof these junctions by the catalytic route. However, themechanism by which sulfur promotes the branching isnot clear at the moment. The very dramatic changes

sulfur may induce in the growth of carbon nanostructureswhen transition metal catalysts are used has been pointedout by several researchersw43,44x many years ago. Morerecently, the role of sulfur seems to be very importantin the growth of doublewall carbon nanotubes(DWCNTs) w45,46x, in the absence of sulfur onlySWCNTs are produced, while minute amounts of S yieldDWCNTs. Sulfur can easily combine with carbon in thegas phase to form S-rich clusters and sulfur may enhancethe catalytic activity of transition metals, and it mayenhance the metal filling of carbon nanotubes toow44x.On the other hand, there are CVD methods, which

seem to produce Y junctions, that do not have fishbonestructurew31x. In these methods, like in Refs.w31,33x,sulfur is not introduced intentionally, but it may not beexcluded that sulfur traces may have been present alongwith Ca and Si, also introduced unintentionallyw31x.The junctions, reported by Ren and co-workers do nothave a fishbone structure, therefore they can be treatedlike the multiwall carbon nanotubes, at least, far fromthe junction. In the case of fish-bone structure it is notstraightforward to apply the models used for describingthe physical properties of carbon nanotubes, in whichthe graphene sheets are oriented along the tube axis.The fish-bone structure is more frequently encounteredin graphitic fibers. As in Ref.w33x no TEM images aregiven, one cannot decide, if it may be inferred that theabsence of sulfur promotes the growth of well graphi-tized Y junctions, but certainly this is one point worthto be investigated further.There are too few results concerning the arc growth

of Y junctions for attempting to extract general rulesfrom the results. Both multiwall and singlewall junctionshave been grown. The single or few wall junctions areattractive because their wall thickness does not allowfor fish-bony structure. It is worth investigating in moredetail, if during the arc discharge growth of single wallnanotubes, in the cathodic deposit, or in the mantlesurrounding it(which contains a large variety of differ-ent carbon nanostructuresw43x) Y junctions may beregularly produced. Usually, when the growth ofSWCNTs is aimed the cathodic deposit is disregarded.There is no doubt that, at the present moment, the

use of nanochannel templates is the most suitable wayto produce large amounts of similar Y junctions. How-ever, it may be difficult to apply to these nanostructuresthe formalism developed to describe the physical phe-nomena occurring in well-graphitized carbon nanotubes.It is encouraging that theI–V characteristic of thesenanostructures shows rectifying behavior as mentionedabove.The two ‘exotic’ methods, the e-beam welding of

straight carbon nanotubes and the growth by fullerenedecomposition, may prove useful for fundamental stud-ies, but it is doubtful that they will gain importance as


Fig. 7. Wave packet dynamical simulation of geometric effects during the tunneling through a carbon nanotube Y junction placed between theSTM tip and a conductive supportw47x. Right column: arrangement of the STM tip, Y junction and support. Left column: snapshots of thetunneling process at 3.6 fs. Note that even when the STM tip is placed off-junction a significant amount of charge flows in the junctions region,thus ‘sampling’ the localized states existing there due to the n-H rings.

practical ways of producing large amounts of Yjunctions.As already discussed in some detail concerning ideal

singlewall junctions above, a widely accepted theoreticdescription is lacking at the moment. The experimentaldata concerning the transconduction of the Y junctionsregard mainly the junctions grown in alumina templatesthat have shape and structure somewhat far from that ofthe model structures used in the calculation. Whenperforming STS measurements on the Y junctions, somecare has to be taken, due to the geometric effectsinherently introduced by the presence of the junctionw47x (Fig. 7) and due to effect of the n-H rings, whichmust be present in the junction region, on the STScurvesw48x. One may observe in Fig. 7 that even whenthe STM tip is not placed right over the center of thejunction region due to the spreading of the wave packettunneling into one of the branches of the Y junction,the STS measurement will ‘sample’ the junction region,too. This may be the mechanism by which the localizedstates caused by the n-H rings affect the STS measure-ments in Ref.w28x.

5. Conclusions

In accordance with early theoretical predictions, inthe last years, stable carbon nanotube Y junctions havebeen produced by various methods. The status of thefield is to some extent similar to that of the field ofstraight carbon nanotubes a decade ago, in the sensethat the first pieces of a challenging puzzle have beenidentified, but a lot of efforts are needed to put togetherthe whole picture. A better understanding and increasedreproductibility for the growth methods has to beachieved and very clearly, once the Y junctions are

available on a regular basis many more efforts areneeded to understand their electronic properties. It seemsto be worthwhile to undertake to these challenging tasksdue to very exciting application possibilities of Y junc-tions in nanoelectronics and in composites.

Acknowledgments

This work was supported by the EC, contract NAN-OCOMP, HPRT-CT-2000-00037 and, EU5 Center ofExcellence ICAI-CT-2000-70029 and by HungarianOTKA Grant T 043685 and by the IUAP researchproject P5y01 on ‘Quantum-size effects in nanostructu-red materials’ of the Belgian Office for Scientific,Technical, and Cultural Affairs, LPB and GIM gratefullyacknowledge financial support from FNRS in Belgium.

References

w1x J. Li, Ch. Papadopoulos, J. Xu, Nature 402(1999) 253.w2x P. Nagy, R. Ehlich, L.P. Biro, J. Gyulai, Appl. Phys. A 70´

(2000) 481.w3x S. Amelinckx, X.B. Zhang, D. Bernaerts, X.F. Zhang, V. Ivanov,

J.B. Nagy, Science 265(1994) 635.w4x L.P. Biro, G.I. Mark, A.A. Koos, J.B. Nagy, Ph. Lambin, Phys.´ ´ ´

Rev. B 66(2002) 165405.w5x D.Y. Ding, J.N. Wang, Z.L. Cao, J.H. Dai, F. Yu, Chem. Phys.

Lett. 371(2003) 333.w6x P. Nagy, R. Ehlich, L.P. Biro, J. Gyulai, Appl. Phys. A 70´

(2000) 481.w7x L.P. Biro, R. Ehlich, Z. Osvath, et al., Mater. Sci. Eng. C 19´ ´

(2002) 3.w8x B.C. Satishkumar, P. John Thomas, A. Govindaraj, C.N.R.

Rao, Appl. Phys. Lett. 77(2000) 2530.w9x F. Frehill, J.G. Vos, S. Benrezzak, et al., J. Am. Chem. Soc.

124 (2002) 13694.w10x U. Dettla-Weglikowska, J.-M. Benoit, P.-W. Chiu, R. Graupner,

S. Lebedkin, S. Roth, Curr. Appl. Phys. 2(2002) 497.


w11x G.E. Scuseria, Chem. Phys. Lett. 195(1992) 534.w12x L.A. Chernozatonskii, Phys. Lett. A 172(1992) 173.w13x S. Iijima, Nature 354(1991) 56.w14x M. Menon, D. Srivastava, Phys. Rev. Lett. 79(1997) 4453.w15x G. Treboux, P. Lapstun, K. Silverbrook, Chem. Phys. Lett. 306

(1999) 402.w16x A.N. Andriotis, M. Menon, D. Srivastava, L. Chernoztatonskii,

Appl. Phys. Lett. 79(2001) 266.w17x A.N. Andriotis, M. Menon, D. Srivastava, L. Chernozatonskii,

Phys. Rev. Lett. 87(2001) 66802.w18x A.N. Andriotis, M. Menon, D. Srivastava, L. Chernozatonskii,

Phys. Rev. B 65(2002) 165416.w19x V. Meunier, M. Buongiorno Nardelli, J. Bernholc, Th. Zacharia,

J.-Ch. Charlier, Appl. Phys. Lett. 81(2002) 5234.w20x C. Papadopoulos, A. Rakitin, J. Li, A.S. Vedeneev, J.M. Xu,

Phys. Rev. Lett. 85(2000) 3476.w21x R. Landauer, Z. Phys. B: Condens. Matter 68(1987) 217.w22x P.G. Collins, K. Bradley, M. Ishigami, A. Zettl, Science 287

(2000) 1801.w23x D. Zhou, S. Serapin, Chem. Phys. Lett. 238(1995) 286.w24x Z. Osvath, A.A. Koos, Z.E. Horvath, et al., Chem. Phys. Lett.´ ´ ´

365 (2002) 338.w25x L.P. Biro, S. Lazarescu, Ph. Lambin, et al., Phys. Rev. B 56´

(1997) 12490.w26x G.I. Mark, L.P. Biro, J. Gyulai, Phys. Rev. B 58(1998) 12645.´ ´w27x L.C. Venema, V. Meunier, Ph. Lambin, C. Dekker, Phys. Rev.

B 61 (2000) 2991.w28x Z. Klusek, S. Datta, P. Byszewski, P. Kowalczyk, W. Kozlows-

ki, Surf. Sci. 507–510(2002) 577.w29x S.H. Tsai, C.T. Shiu, W.J. Jong, H.C. Shih, Carbon 38(2000)

1879.w30x F.L. Deepak, A. Govindaraj, C.N.R. Rao, Chem. Phys. Lett.

354 (2001) 5.w31x W.Z. Li, J.G. Wen, Z.F. Ren, Appl. Phys. Lett. 79(2001)

1879.

w32x B. Gan, J. Ahn, Q. Zhang, et al., Chem. Phys. Lett. 333(2001)23.

w33x J.-M. Ting, C.-C. Chang, Appl. Phys. Lett. 80(2002) 324.w34x Y.C. Sui, J.A. Gonzalez-Leon, A. Bermudez, J.M. Saniger,

Carbon 39(2001) 1709.w35x L.P. Biro, R. Ehlich, R. Tellgmann, et al., Chem. Phys. Lett.´

306 (1999) 155.w36x L.P. Biro, R. Ehlich, Z. Osvath, et al., Diamond Relat. Mater.´ ´

11 (2002) 1081.w37x A.A. Koos, R. Ehlich, Z.E. Horvath, et al., Mater. Sci. Eng. C´ ´

23 (2003) 275.w38x R. Ehlich, L.P. Biro, C. Stanciu, Z.E. Horvath, J. Gyulai, in:´ ´

H. Kuzmany, J. Fink, R. Mehring, S. Roth(Eds.), XVthInternational Winterschool on Electronic Properties of NovelMaterials an Euroconference: Molecular Nanostructures, vol.591, Ser. American Institute of Physics Proceedings, AmericanInstitute of Physics, 2001, p. 175.

w39x L.P. Biro, G.I. Mark, A.A. Koos, J.B. Nagy, Ph. Lambin, Phys.´ ´ ´Rev. B 66(2002) 165405.

w40x M. Terrones, F. Banhart, N. Grobert, J.-C. Charlier, H. Terrones,P.M. Ajayan, Phys. Rev. Lett. 89(2002) 075505.

w41x R.L. Anderson, Solid-State Electron. 5(1962) 341.w42x C.H. Olk, J.P. Heremans, J. Mater. Res. 9(1994) 259.w43x C.-H. Kiang, M.S. Dresselhaus, R. Beyers, D.S. Bethune,

Chem. Phys. Lett. 259(1996) 41.w44x N. Demoncy, O. Stephan, N. Brun, C. Colliex, A. Loiseau, H.´

Pascard, Synthetic Metals 103(1999) 2380.w45x L. Ci, Z. Rao, Z. Zhou, et al., Chem. Phys. Lett. 359(2002)

63.w46x Z. Zhou, L. Ci, X. Chen, et al., Carbon 41(2003) 337.w47x G.I. Mark, L.P. Biro, J. Gyulai, Ph. Lambin, in: H. Kuzmany,´ ´

J. Fink, M. Mehring, S. Roth(Eds.), XVth InternationalWinterschool on Electronic Properties of Novel Materials anEuroconference: Molecular Nanostructures, vol. 633, AmericanInstitute of Physics, Melville, New York, 2002, p. 381.

w48x J.-C. Charlier, Accounts Chem. Res. 35(2002) 1063.

carbon nanotube y junctions: growth and properties · 242 l.p. biro et al. / diamond and related...

Documents