Download MEASUREMENTS OF TIIE CONDUCTIVITY OF INDIVIDUAL 10 NM

.
MEASUREMENTS OF TIIE CONDUCTIVITY OF INDIVIDUAL
10 NM CARBON NANOTTJBES
GEORGE M. WHITESIDES AND CARL S. WEISBECKER
Harvard University, Departmentof Chemistry,12 Oxford St. Cambridge,MA 02138.
ABSTRACT
Catalytically grown carbonfibers approximately10 nm in diameterand severalmicrons
long were characterizedby transmissionelectronmicroscopyand determinedto be multiplewalled nanotubes,and a techniquewas developedto measurethe conductivity of an individual
nanotube.Nanotubeswere dispersedin solventsand precipitatedonto lithographically defined
gold contactsto make a 'nano-wire' circuit. Non-contactAFM was usedto image the nanowires, and a resistanceo_fLI.4 (+ 1.0 ) MC) was measuredthrough a single nanotubeat23o C.
A resistivity of 9.5x10-) Q m was estimatedfor carbonconductingalong the axis of a fiber.
Local heatingof nanotubesappearedto occur at high current densities.The nanotubescould
sustaincurrentson the order of 10 pA per fiber, but applicationofcurrents on the order of
100 pA per fiber resultedin rapid decompositionin air and breaking of the circuit.
INTRODUCTION
A circuit was designed,consistingof two gold contactsseparatedby approximately5 pms
and connectedby a carbon nanotube,and current was passedthrough the circuit. To our
knowledge no other measurementsof the conductivity of individual carbon nanotubeswith
comparablysmall diametershave beenreported,althoughsometheoreticalpredictionshave
been advancedregardingelectronicpropertiesof nanotubes[-3]. Conductivity measurements
of much larger carbon fibers of severaltypes [4-8] and recently of carbon nanotubebundles
nanofibersis difficult to
[9,10] are also documented.Electrical resistivity in catalytically grown 'nano-wires'
were
determineby conventionalmeansbecauseof their small size [11]. The
imagedby non-contactAFM. Relatedimagesof nanotubesand nanotubebundleshave been
reportedusing other scanningprobe techniquesU2-141.
Carbon nanotubesare interestingfor the purposeof building nano-scaledevicesbecauseof
their size. Our researchgroup has an interestin the developmentof new lithographic techniques
for the patterningof surfacesI I 5] , and one of the reasonsfor developingsuch techniquesis to
find better ways of fabricating small (lessthan 30 nm ) features.Various non-classical
phenomenasuch as single electrontunneling [6], which might be exploited to build new
devices,are observedat increasinglyhigh temperaturesas featuresizesdecrease.A
complementaryapproachto building nano-electronicdevicesmight be to chemically build up
structuresfrom colloidal or molecularpiecessuch as carbonnanotubesin addition to relying on
lithographic techniques.
EXPERIMENTAL
A sampleof catalytically grown carbonnanotubeswas obtainedfrom Hyperion Catalysis
International.We imaged the carbonnanotubesby transmissionelectronmicroscopy (TEM),
and the carbon in the materialconsistedmostly of fibers (figure 1). TEM was performed on a
Phillips EM 420 microscope.The carbonnanotubeswere dispersedin ethanolor xyleneswith
sonicationand pipetted onto TEM grids from StructureProbesInc. The fibers had diametersof
approximately 10 nm and lengthsof severalpms.
263
MaterialsResearchSociety
Mat. Res.Soc. Symp. Proc.Vol.349.@1994
N-dopedsilicon (100) waferswere obtainedfrom Silicon SenseInc. Thesewaferswere
oxidizedin air at l0O0 o C for 24 hours.The oxidizedwaferswere mountedin a dual sourceebeamevaporatorcontaininga gold sourceand eithera titanium,nickel,or chromium sourceto
be used as an adhesionpromoter. l0 A of adhesionpromoter followed by 200 A of gold were
depositedon the '*'afers.The gold coatedwafers were patternedthrough a mask using
photolithographl.then the exposedgold was etchedfor 2 secin Gold EtchantTFA suppliedby
TranseneInc. The oxide layer insulatedthe gold contactsfrom eachother and from the silicon
substrate.The resistremaining on the gold contactswas strippedoff, and PMMA was then
spin-coatedonto the wafersfor e-beampatterning.
E-beampatterningwas performedon a JEOL JSM-flO ScanningMicroscope.A window
parrernwas written in the PMMA resist,and it consistedof a rectangularbox with dimensions
of l0 p.m x 20 pm. The window areawas scannedwith a 35 kV 400 pA e-beamusing a line
doseof 8 nC/cm for best results.Upon development,the PMMA completely insulatedthe gold
contactsexceptinsideof this window.
The developmentof thesewindows and the number of nano-wirespresentinside of them
were assessedby AFM on a TopometricsScanningProbeMicroscope.The fibers were imaged
in air on the AFM using low resonantfrequency 1660-00silicon tips from Topometrics.The
tips oscillatedat approximately 150 kHz, and amplitudedetectionwas usedto collect the sign_al.
Currentswere applied to the gold contactsusing a PrincetonApplied ResearchModel273
operatedin the galvanostaticmode. The current sourcewas connected
PotentiostaUGalvanostat
to the gold substratevia wires plantedinto indium solder.
RESULTS
Figure 1: Typical nanotubeshave diametersof about 10 nm and lengthsof severalpms. The
high resolutionimage on the left was recordedby Y. Lu at the Harvard MRL Central Facility.
0
0nm
Figure 2: A non-contactAFM image of a nanotubeprecipitatedonto a gold film is depicted.
Figure 3: Lithographedgold contactson an oxidized silicon substrateare connectedto wires via
indium solder.A rectangularwindow in PMMA is centeredwhere the contactsmake their
closestapproachat a separationof approximately5 pm (seearrow). E-Beam exposuresalong
the top left contactoccurredduring positioning of the e-beamfor the window exposure.The
inset of this visual image is a view through an optical microscope.
\
We mounted the carbon fibers on lithographicallydefined gold contacts-sothat we could
was used as a
measuretheir conductivity, and non-contait a'tomicforce microsco-qY-(AFM)
images
method for imaging thesefibers on a circuit. Comparisonof the AFM imageswith TEM
thicknessof fibers enoneouslyappearedto be approxl-T3jglyten times
showedthat the"lat-eral
able to
wider than the .o.r..iiv i.aled vertical thicknessin the AFM images(figure 2).y" were
and
tip
AFM
the
between
interaclion
weak
the
because
scanover individuii;6.;;;;pr"ducibly
the fibers did not causethem to move.
Dispersionsof carbon nanotubeswere sonicatedin ethanolto make dispersionsof
4xl0-3 or 4xl0-4 me/ml and precipitatedonto thermally oxidized silicon (100) substrates
puti..nJ with lithofrupn"a gbtOf6ututes(figure 3). Tha gold featureshad been previoutly ,-_,
pmz areawhere
coatedwith insulatingpoly(methyl methacrylate)(P.MMAIexcept within a20o
pMMA was subsequEn'tfy'rL*oved
using e-'beamlithography.The solventwas either allowed to
under i streamof argon-'A 1,90pA current was typically
dry in air or to
"uupoiutJquickly
A voltage greaterthan 1-0V betweenthe contactswas
contacti.
two^gold
across
uppti.O
to the depositionof fibers. The occurrenceof
.up".ltatiGrfprior
iirr
to
measuredano ouseri"eJ
in a
one or more nanotubeslying ucro^r,the gap b6tweenthe two gold contactsresulted
dispersion
dramatic and sustaineddecffiasein the u"otiug"that was measired, and drops gf 9.
The nanotubes
was
realized.
drop
potential
a
substrate"until
the
onto
r;dii;"t
;;;6"t4
them from
preciiitated onto an areaof approximately I cm2, but the PMM+ layer prevented
alea
window
The
20O
the
of
i{rside
gold
except
Ttmz.window'
making electrical contactwith
number of fibers
was small enough to be scanneEwith un'efM tip so that we'could observethe
bridging the two gold contactsinside of it'
fiber' The
-ulrti"uf4: The gold contactsin this image were thought to be bridge.dvia a single
Figure
was
diameter
inner
The
hollow
nm.
10
8
and
beiween
was
n"ighTof the nanotube
and the
assumedto be 1/3 of the total diameterbasedupon TEM imagesof similar fibers,
length of the
cross-sectionalareaof carbonwas thereforeestimatedto be 57 1+ 16 ) nmz' The
f i b e rw a s7 . 5 1 + 1 . 0) P m .
266
0 . 15
0 . 10
230C
S fo p e : 1 1 . 4 ( t 1 . 0 )
MO
0.05
o
ct)
0.00
G
=
o
J
-0.05
-0.10
.df
,lf!
."'
.4.0
(p,|')
- 0 . 15
-1
-5
5
0
Current (nA)
1.5
0
Figure 5: The voltage vs. appliedcurrent plot correspondingto the image in figure 4 is depicted.
The d. c. conductivity was measuredat room temperature(23 o C) using a galvanostatto
apply currentsin forward and reversedirectionsthrough the nanotubesthat connectedthe gold
contacts.In our initial experimentsno attemptwas madeto thermostatthe circuit or correct for
self-heatingof the nanotubes.Since the conductivity was derived from the limiting slope of a
voltage vs. applied current curve at low currents,self-heatingis not likely to have been a
problem. Often, more than one fiber or an aggregateof fibers was observedto be bridging the
contacts.Our best exampleof gold contactsbridged by a single fiber is shown in figure 4.
Aggregatesof carbon fibers were also present,but only one nanotubewas observedto be
bridging the contacts.
The voltage vs. applied current plot that we attributeto this single bridging fiber is shown in
figure 5. The resistanceof the fiber was obtainedfrom a linear fit to the plot at low magnitudes
of applied current, and it is I 1.4 ( t L0 ) MO. This measurementcorrelateswell with some
other measurementsthat we have made in which more than one connectingfiber was observed
and the measuredresistancewas correspondinglylower. By measuringthe length and crosssectionalareaof the nanotubefrom its AFM image the resistivity of carbon along the fiber axis
could be estimatedto be 9.5x10-5 Q m. This estitated resistivity is more than an order of
magnitudehigher than reportedresistivitiesin the basalplane of orderedpyrolytic graphite [ 17].
The resistivity in carbon nanotubebundlessynthesizedby arc dischargehas beenbeen
estimatedby othersto vary between t0-a and 10-5C)m igl anAto be 6.5xlO-5 O m at 300 K
[0]. The room temperatureresistivity of methane-derivedcarbonfibers was shown to be
10-) Q m or less dependingon heat treatmenttemperature[7]. The resistivity of benzene-
267
derived carbon fibers was shown to be on the order of l0-6 C)m after heat treatmentat 3500'C,
and it increasedas the diameterof the fiber decreasedbetween34 and 1.6 ttms [8].
At currents above I pA the voltage vs. applied current curve in figure 5 deviates
significantly from linearity and the measuredvoltage becomesmore uncertain. The reason for
the lowering of the resistanceat high currents may be local heating of the nanotubes.We
observed that the nanotubeswere capable of supporting on the order of l0 pA of applied
current per fiber without being damaged.Application of currents on the order of 100 pA per
fiber resulted in rapid decomposition of the fibers in air and breaking of the circuit.
We wish to thank Dr. D. Moy at Hyperion for providing the carbon fibers and for advice.
CONCLUSIONS
These carbon fibers show promise as materials for building nano-scaleelectronic devices.
They can be imaged individually and reproducibly on a circuit by non-contact AFM. Our
estimate of the rJsistivity of carbon in a-single nanotubeparallel to its axis is much higher.than
resistivitiesthat have be-enreportedfor orderedpyrolytic graphite [ 17]. The resistivity ir similar
to resistivity measurementsoTnanotubesof similar diameterin coaxial bundles [9,10]. The
resisitivity is an order of magnitudeor more higher than the resistivity of individual vapor
grown fibers with diametersgreaterthan I pm [7,8]. The contributionof contactresistanceto
6ur measurementis still unclear,but someinformation about contactresistancemay be
A four-point probe measurementof a single
obtainablefrom a. c. conductivity measurements.
nanotubewould be desirable,bui also difficult to accomplish.We will also attemptin future
experiments to better control and detect local heating that is probably occurring when we apply
high currentsto the nano-wires.
REFERENCES
N. Hamada,S. -i. Sawada,and A. Oshiyama,Phys.Rev. ktt. 68 (10), 1579(1992).
H. Ajiki, T. Ando, J. Phys.Soc.Jpn.62 (4), 1255-1266( 199_3).
J. Appl. Phys.73 (2), 494-5N (1993).
M. S. Dlesselhaus,
R. Siito, G. Dresselhaus,
J.Imai and K. Kaneko,Langmuir, 8, 1695-1691(1992).
K. Kuriyama,Phys.Rev. B, 19 (47),12415-12419(1993)-__
C. Olorian, M. Matte, and C. J. Lum, Electrochem.Soc. 138 (8), 2330-2334(1991').
J. Heremans,I. Rahim, and M. S. Dresselhaus,Phys. Rev. B 32 (10), 6742 (1985).
and M. Endo,CarbonA (l),67-72 (1986).
M.Z. Tahar,M. S. Dresselhaus,
L. Langer, L. Stockman,J. P. Heremans,V. Bayot, C. H. Olk, C. Van Haesendonck,Y.
Bruynseraede,and J-P. Issi, J. Mater. Res. 9 (q,2n,Q994).
10. S. Ni. Song,X. K. Wang, R. P. H. Chang,and J. B. Ketterson,Phys.Rev. Lett. 72 (5),697
(1994).
700
1 1 .N. M. Rodriquez,J. Mater. Res. 8(12), 3246 (1993).
t 2 . C. M. Liebeiand Z.Zhang, Appl.Phys. ktt. 62,2795 (1993)'
D. Sarid, L. D. Lamb, F. A. Tinker, J. Jiao, and
1 3 .M. J. Gallagher,D. Chen,-B.P.-Jacobsen,
D. R. Huffman, Surf. Sci. Lett. 281,L335-L340 (1993).
14. C. H. Olk and J. P. Heremans,J. Mater. Res.9 (2),259-262 (1994).
1 5 .A. Kumar and G. M. Whitesides,Science263,60 (1994).
1 6 .Q. Niu, Phys.Rev. I-ett.64 (15), 1812(1990);P. Delsing, K. K. Likharev,L. S. Kuzmin,
(1989).
andT. Claeson,ibid. 63 (17), 1861-1864
1 7 .c. A. Klein, J. Appl. Phys.33, 3346(1962);p. R. Wallace, Phys.Rev. 7l (9),622 (1947).
1.
2.
3.
4.
5.
6.
7.
8.
g.