vo
-dO
hNAMB:
Science
Pahnce
PAGE:
1
SESS:
4
OUTPUT: Tue
Ocl20
00:35:53
1998
icIi4/se-sci/se-sci/se-o~~se6990-98r
I
Synthesis
of
large
Arrays
of
Well-Aligned Carbon Nanotubes
'
on
Glass
Z.
F.
Ren,*
Z.
P.
Huang,
J.
W.
Xu,
J.
H.
Wang,
P.
Bush,
M.
P.
Siegal,
P.
N.
Provencio
Free-standing allgned carbon nanotubes have previously been grown above
7OOOC
on
mesoporous silica embedded with iron nanoparticles. Here, carbon
nanotubes allgned over areas up
to
several square centimeters were grown on
nickel-coated glass below
666OC
by plasma-enhanced hot filament chemical
vapor deposition. Acetylene (CZHB) gas was used as the carbon source and
arnmonla (NH,) gas was used as a catalyst and dilution gas. Nanotubes with
controllable diameters from
20
to
400
nanometers and lengths from
0.1
to
50
micrometers were obtalned. Using this method, large panels
of
aligned carbon
nanotubes can be made under conditions that are suitable for device fabrication.
Since the first observation
of
carbon nano-
tubes
(I),
numerous papers liave reported
studies on the yield
of
well-graphitized nano-
tubes, tlicir diameter and wall thickness (sin-
glc or multiple)
(24
growth mechanisms
pendicular
to
the glass surface. The carbon
nanotube arrays
are
fabricated by first depos-
iting
a
thin nickel layer onto display glass by
radio frequency
(rf)
magnetron sputtering
(I
7).
The carbon nanotubes are then grown
(5),
alignment
(6-8),
electron emission prop-
ertics
(9-I
I),
nanodevices
(If
13),
theoret-
ical predictions
(I4),
and potential applica-
tions
(14).
Alignment
of
the carbon nano-
hibes
is
particularly important
to
enable both
fiindanicntnl studies and applications, such
as
iiiicroclcctmiiics.
'I'licrc
WIIS
littlc
siicccss
iri
obtaining nligiimcnt
of
carbon nnnotubcs on
large areas until tlie report on the growth
of
nligncd carbon nnnotubcs on nicsoporous
sil-
icil cnnliiining iroii nanopiirticlcs via tlicriiinl
dcconiposilion of acctyicne gas in nilrogcii
gas at tcmpcraturcs above
700°C
(7).
How-
cvcr, this high growth temperature makes this
nictliod unsuitable for the fabrication
of
car-
bon nanotubcs on glass, because thc strain
point of
tlic
bcst display glass is 666°C
(15).
Recently, we liave successfblly grown
large-scale wcll-aligned carbon nanotubes on
nickcl
foils at tciiipcrntiires bclow
700°C
(16).
klcrc wc rcport
the
growth
of
large-
scale well-aligncd carbon nanotube arrays on
glass at tcmpcraturcs below 666°C. These
lowcr tctiipcmturc growth conditions arc suit-
nbic for clcctron cniission applications, such
iis
cold-catliodc flat pancl displays, wliicli
rccliiirc
cmtwii
iiiiiiatubc
ciiiittcw
yIowii
per-
cold-cntliodc
flnt
liniicl
displnys
illid vnciiiiiii
2
F.
Ren,
Z.
P.
Huang,
J.
W.
Xu,
J.
H.
Wan&
Materials
Synthesis Laboratory, Natural Sciences Complex, De-
partments
of
Physics
and
Chemistry,
and
Center for
Advance!
Photonic and Electronic
Materials,
State
University
of
New
York, Buffalo,
NY
14260-3000,
USA.
P.
Ourh,
Instrumentation
Center,
State
Univer-
sity
of
Ncw
York, Buflalo,
NY
14214,
USA.
M.
P.
Sicgal
and
P,
N.
Provencio,
Sandla
National Laboratories,
Albuquerque,
NM
87185-1421,
USA.
*To
whom
correspondence
should
be
addressed
E-
mail:
uen@acsu.buffalo.edu
on the nickel-coated display glass by plasma-
enhanced hot filament cliemical vapor depo-
sition (PE-HF-CVD)
(17).
Scanning electron microscopy
(SEM)
was
used to investigate the effect
of
various
growth conditions on the morphology
of
car-
Fig.
1.
(A)
SEM
micrograph
of
carbon nano-
tubes aligned perpendicular to
the
substrate
over large areas; growth conditions are listed in
Table
1.
(6)
Enlarged view of
(A)
along the
peeled edge showing diameter, length, straight-
ness,
and uniformity in height, diameter, and
site density.
boil nanotubcs grown
on
nickcl-cootcd
tlis-
play glass. The growth conditions used arc
listcd in Table
1.
In tlie
first
experinicnt,
NII,
was iritroduccd during
tlic
first
5
min witliout
introducing
C,H,.
During
this
tinic, plasma
etching
was
used
to
rcducc
tlic
tliickncss
of
tlic
nickcl laycr, rcsulting in
a
tliickncss
of
(40
nm. After
these
initial
5
min,
C,H,
was
introduced. Immediately
a
color change
oc-
currcd,
as
a
result
of
the growth
of
carbon
nanotubes. The growth period lasted only
IO
min. In order
to
examine the orientation and
alignment
of
the
carbon nanotubes
on
thc
glass substrates, part
of
the carbon nanotube-
covered area was peeled
off
(Fig. IA, lower
left) with tweezers to expose the glass sub-
strate. During peeling, another area was
crumpled (Fig.
IA,
lower right). and
a
long
scratch was made
on
the peclcd open area
(Fig.
IA.
lowcr IcR). Undcr visiinl
nntl
SEM
observations, tlic alignment of tlic carbon
nanotubes across
the
whole surface was
as
uniform
as
in
tlie upper part
of
Fig.
IA.
To
estimate the carbon nanotube length, an
SEM
image was taken at higher magnification
Fig.
2.
SEM surface morphology of the nickel
layers. (A) Etched by
NH,
plasma for
3
min.
(6)
Etched by
N,
plasma for
3
min.
(C)
As-sput-
tered smooth surface.
DISCLAIMER
This report was prepared
as
an account of
work
sponsored
by an agency of the United
States
Government. Neither the
United
States
Government nor any agency thereof, nor any
of
their employees, make any warranty, express or implied,
or assumes any legal liability or responsibility for the
accuracy, completeness,
or
usefulness
of
any
information,
apparatus, product, or process disclosed, or represents that
its
use would not infringe privately owned
rights.
Reference
herein
to
any
specific
commercial product, process, or
service
by
trade
name,
trademark,
manufacturer,
or
otherwise does not necessarily constitute or imply
its
endorsement, recommendation, or favoring by the United
States
Government or any agency thereof. .The views and
opinions of authors expressed herein do not necessarily
state
or
reflect
those of the United
States
Government or
any agency thereof.
DISCLAIMER
Portions of this document may be illegible
in electronic image products. Images are
produced from the
best
available original
document.
,*
JO~N'AMI~:
sziciicc 1)iirlaticc
PAGE:
2
SIXS:
4
OUTIJUT:
l'uc
~ct 20
00:35:53
IYYX
Irich4lse
-
sciise
-
sdse
-
origIse6990
-
98r
REPORTS
Tabla
1.
Growth conditions
for
nanotubes
as
shown
in
Figs.
1
through
4.
C,H,INH,IN,
(SCCM)
Filament Plasma intensity
Growth
time
current
(A)
(AIVIW)
(min)
Fig.
1,
A
and
B
01160/0
followed by
80116010
Fig.
2A
01160/0
Fig.
2B
0101296
Fig.
3,
A
and
B
0Ol16OlO
Fig.
4A
40116010
Fig.
48
01160lO
followed by
801160lO
as
8.5
8.5
8.5
8.5
7.2
8.0
8.2
0.101635172
0.131670195
0.091740166
0.101400153
0.20170011
50
0.131650190
0.1
0/400/52
0.1
01560/60
5
10
3
3
25
14
20
10
along the peeled edge (Fig.
1B).
Misalign-
ment of the carbon nanotubes on the peeled
edge is a result of the peeling operation. From
Fig. 1B. it was estimated that the nanotubes
were about 100 nm in diameter and
20
yn in
length. Given tlie growth time of
10
niin, the
growth rate was calculated to be
120
pm/
hour, which
is
about five times faster than tlie
value reported in
(7).
When the sequence of
gas introduction was reversed (that
is,
when
C,I
I,
wiis introduccd first. followcd by
NI
I,
5
iiiiii
later),
no
growth of carbon nanotubes
was observed; only amorphous carbon was
formed on the nickel surface undcr these
conditions. The amorplious carbon laycr
fornicd
in
tlie first
5
iiiin
in
C,H,
plasma
covered tlie nickel surface and prevented tlie
catalytic role of nickel,
so
that there was no
growth of carbon nanotubes. It seems that tlie
carbon nanotubes grow only wlien
NH,
is
introduced first, followed by
C2fi2,
or wlien
both
C2H2
and
NH3
are introduccd at the
sanie tinic.
Wc
coiicludc that
NH,
plays
a
ciucial catalytic role together with the nickel
laycr to promote the growth of tlic carbon
nanotubes. The catalytic role of
NH,
was
fiirther confiniied by the fact that there was
also
no
carbon nanotube growth when
NH,
was replaced by
N,
gas,
with tlie other con-
ditions iincliaiigcd.
Tlie
siirCicc
of
tlic
nickcl
hycr
tiller
tlic
iiiilitil
NI
I,
or
N,
pliisiiizi
etcli-
ing was essentially the same (Fig.
2,
A
and
B,
respectively). Tlie plasma etching conditions
are listed
in
Table
1.
For
comparison, Fig.
2C
shows the as-sputtered smooth nickel surface.
It
is
clcarly
sliowii
that both
Nli3
and
N,
plasiiia
etcliiiig roiighcii
tlic
nickcl surhcc,
1)iit
tlic
iougliiriy
of
tlic
iiickcl
siirliicc
is
1101
rcsponsiblc for tlic nucleation and growth of
carbon nanotubes.
In order to examine the effect of the thick-
ness of nickel layer on the growth of carbon
nanotubes,
C,H,
and
NH,
were introduced at
tlie same time
in
the second experiment
pa-
bie
l).
Undcr these growth conditions, no
nlnsmn
rtchino
nrriirrrd
and
flip
nirCpl
laver
remained
40
nm thick. The diameters of tlie
carbon nanotubes (Fig.
3A)
were much larger
than those shown in Fig. IB. From Fig.
3B,
we estimate that the outside diameters of the
carbon nanotubes ranged from
180
to
350
nrn
and that most of tlie carbon nanotubes were
about
250
nm in diameter. This experiment
clearly shows that nickel thickness plays
a
very important role in deterniining the diam-
eters of the carbon nanotubes. The catalytic
rolc of iiickcl is
also
clcrirly
sliowii
by
tlic
Fig.
3.
(A)
SEM
micrograph
of
carbon nano-
tubes grown as in Table
1.
The diameters are
clearly larger than those shown
in
Fig.
IB.
(B)
Enlarged view of
(A)
showing the diameters and
their distributions.
A
site density of about
IO7
tubeslmm' was estimated. The nickel cap
of
one nanotube located at the left is missing, as
indicated by the arrow. Because the nickel cap
is absent, the tube is transparent and the nano-
tiihn hnhinA
it
ic
virihlta thrntinh thn
w=ll
nickel cap
on
the tip of each nanotube (Fig.
3B).
One carbon nanotube, indicated by an
arrow in Fig.
30.
did not have
a
nickel cap.
We conclude that the carbon nanotubes were
empty and had very thin walls, because an-
other carbon nanotube
is
visible behind the
capless one through
its
wall. These
large
carbon
nanotubes may be
usehl
for applications
such as storage
of
H,
and other gases
(18).
Thcsc experiments sliow that tlic tliinncr
the nickel layer, the thinner tlie nanotubes.
To
examine hrther the effect of nickel layer
thickness on carbon nanotube growth, anotli-
er pair
of
experiments was started with a
nickel layer of only
15
nm
(Table
I).
In one
experiment, we again uscd
plasma
ctcliiiig
to
reduce tlie nickel thickness by introducing
NH,
first and introducing
C,H,
20
niin later.
SEM micrographs of carbon nanotubes
grown under tlie conditions listed
in
Table
1
(Fig.
4,
A
and
B)
show clearly that tlie diam-
eters of tlie nanotubes are dependent on tlie
nickel layer thickness.
Tlie
typical diameter
in Fig.
4A
is only about
65
nm, as compared
to
240
nm in Fig.
3B.
In addition, the align-
ment in Fig.
4A
is
not as good as in
Fig.
38.
A
comparison of Figs.
4A
and
4B
denion-
strates that
20
min of plasma etcliing reduced
the thickness of nickel layer, which
in
turn
%E
Fig.
4.
(A)
SEM
micrograph showing that thin-
ner carbon nanotubes were growii
011
tiiiiiiier
(15-nm) nickel-coated glass. The alignment
is
not as good as that in
Fig.
38.
Growth condi-
tions
for
this
sample are listed in Table
1.
(B)
SEM
micrograph showing carbon nanotubes
with diameters as low as
20
nm, grown under
the growth conditions listed in Table
1.
The
image demonstrates that when the nanotube
diameters
&?f?f@
continue to decrease,
,hair
.linnmmnt
IC
nrdiialltr
lrrct
Fig.
5.
HRTEM
images showing the interior and
wail
structures
of
a
typical thin carbon nano-
tube.
(A)
Cross-section view.
(B)
Plan view.
resulted in even thinner carbon nanotubes
with typical diameters of only about
20
nm.
The comparison
also
shows that the align-
ment starts to
worsen
drastically when the
nanotube diameter
is
duced
to
20
nm. There-
fore,
for
applications rcquiring
good
align-
ment, diameters should be larger than
50
nm.
We used high-resolution transmission
electron microscopy (HRTEM) to determine
the interior and wall structures of the carbon
nanotubes
(19).
Figure
SA
shows
a
cross-
section view of
a
typical thinner carbon nano-
tube. The outside diameter
of
this carbon
nanotube
is
nearly
30
nm. It clearly shows
that the nanotube is
a
multiwalled centrally
hollow tube. not solid fiber. The fringes
on
each side
of
the tube represent individual
cylindrical graphitic layers. This particular
carbon nanotube
is
a
structure with approxi-
mately
15
walls of graphitized carbon. Both
the angular bend in the structure and the
appearance
of
carbon walls running across
the diameter
of
the nanotube demonstrate
structural defects suggestive
of
twisting of
the nanotube structure. The lack
of
fringes
inside
the
tube, as well as
the
lighter
contrast
as
compared
to
the
nanotube walls, indicate
that the core of the structure
is
hollow.
Fiirthrr
pviripnre
of
A
hnilnw
rnrP
k
REPORTS
shown in Fig.
513.
This is
a
plan view
I
IR'TEM
imiigc
of
41
siiiglc carboti nnnotubc
structure
(19).
Here we
can
more clearly see
the hollow nature
of the
nanotube, again rep-
resented
by
the lighter contrast of the inner
core. The disorder seen in the wall fringes
circumventing the hollow center
is
most like-
ly causcd by tlie twistlike dcfects tlirougliout
the carbon nanotube length, as shown in Fig.
SA.
Tlicsc HRTEM iiiiagcs dcfinitcly show
that the structures reported in this paper are
hollow multiwalled carbon nanotubes with
defects existing along the tube. The defects
of
bending and twisting
of
the thin carbon nano-
tubes shown in Fig.
5,
A
and
U,
are consistent
with tlie SEM observation shown in Fig. 4B.
The growth incchanism of aligned carbon
nanotubes is ascribed in the literature to con-
straint
of
the pores
in
either mesoporous
sil-
ica
(7)
or
laser-etclicd tracks
(8).
However,
in
our experiments
the
alignment
of
the carbon
nanotubcs cannot be dtic
to
pores
(7)
or
etclicd
tracks
(8)
because tlicre are
no
pores
(7)
or
etched tracks
(8)
in
our
glass sub-
strates, but
is
rather due to
a
nanotube nucle-
ation process catalyzed by ammonia and
nickel. In tlie presence of ammonia, each
nickel cap efficicntly catalyzes the contiiiu-
ous synthesis
of
carbon nanotubes.
As
the
nanotubcs grow, the nickel cap remains
on
the tip
of
each. Tlic alignment and thickness
of
tlic carbon nanotubcs my he dctcrniincd
by the orientation
and
sizc, rcspcctivcly, of
the initial catalytic centers. With this method,
we can envision tlie synthesis of large panels
of
well-aligncd carbon nanotubcs for use in
many applications.
References and
Notes
1.
5.
iijima, Nature
354.
56
(1991).
2.
A.
Thess et a/, Science
273. 483 (1996).
3.
C. Journet et a/.. Nature
388.
756 (1997).
4.
I.
Liu et a!., Science
280, 1253 (1998).
5.
1.
C. Charlier.
A.
De Vita.
X.
BlasC.
R.
Car,
IbM.
275.
646 11997).
15.
L.
C. Lapp. D. M. Moffatt. W.
H.
Dumbaugh.
P.
L
Bocko, product Information. Corning. The dlsplay
giass substrates used in this paper were supplied by
Corning for testing purpose only. Information about
the detailed properties can be obtained from Corning.
The most important property of the fiat-panel dis-
play glass
is
the high strain point of 666.C. as com-
pared to the strain point of
500.
to
590.C
of
com-
mercial glasses.
~
16.
Z.
P.
Huang et ai.. In preparation.
17.
Before the deposition
of
the nickel layer. display glass
was cut into pieces measuring
10
X
5
mm and then
cleaned in acetone by ultrasonication. The cleaned
pieces were mounted on the surface of a stainless
steel resistive heater, and the whole assembly was
Introduced Into the sputtering chamber. The chamber
was then pumped down below
8
X
torr before
argon gas was introduced Into the chamber to main-
tain a working pressure of
20
to
60
millitorr. During
deposition, the substrates were either heated or kept
at room temperature. The deposition of the nickel
layer lasted only from
1.5
to
6
min and produced
nickel layers from
15
to
60
nm In thickness. After the
nickel layers were deposited. the substrates were
transferred to a chemical vapor deposition chamber
and pumped down below
6
X
lo-'
torr.
As
soon as
the chamber pressure reached
6
X
lo-'
torr, acet-
ylene and ammonia gases were introduced into the
chamber to maintain
a
working pressure of
1
to
20
torr during carbon nanotube growth. The total flow
rate of acetylene and ammonia gases was
120
to
200
standard cubic centimeters per minute (SCCM), with
a volume rat10 of acetylene to ammonia varying from
1:2
to
1:lO
in different experimental runs. After the
working pressure had been stabilized, the power to
the tungsten filament coil and that to the plasma
generator were turned
on
to
generate heat and plar-
ma. Under the present experimental setup, the tem-
perature of samples
is
estimated to be below 666.C
because there was no visually noticeable change
of
the glass due to heating. The growth period ran from
5
to
20
min. After growth, the samples were taken
out and transferred to a scanning electron micro-
scope (Hitachl
5-4000)
for examination of nanotube
alignment. diameter, length, straightness, site density
and uniformity, and
so
on. Typical samples with good
alignment were also examined by x-ray diffraction.
Raman spectroscopy, x-ray photoemission spectror-
copy and HRTEM to study the structure, crystallinity.
composition. central core diameter, and tube wall
structures.
18.
A.
Dillon et al., Nature
386,377 (1997);
G.
E.
Gadd
et
a/.. Science
277.
933 (1997).
19.
HRTEM was performed
on
a JEOL
2010
In the Earth
and Planetary Science Department at the University
of
New Mexico, Albuquerque, NM. Samples for plan
view HRTEM were prepared as follow% Given the
flexible nature
of
the nanotubes. we penetrated the
films with M-Bond
610
epoxy resin (M-Line Accesso-
ries.
0.0,
00)
to provide mechanical stiffness.
it
has
-7
very low viscosity and curing
Is
time and temperature
dependent Hydrotetrafuran (diethylene oxide) makes
torr. A silica network with relatively
%
UD
about
90%
of the comDosition of M-bond.
The
6.
W.
A:
de Heer et
ai.,
/bid.
268.
845 (1995).
7.
W.
Z.
Lieta/../b/d.
274. 1701 (1996).
in this method,
the substrate was prepared by a sol-gel process from
tetraethoxysilane hydrolysis In Iron nitrate aqueous
solution. The gel was then calcined for
10
hours
at
450.C
at
uniform pores was obtained, having iron oxide nano-
4
particles embedded in the pores. The iron oxide nano-
particles were then reduced at
55OoC
in
180
torr of
flowing
9%
HJN,
(110
cm3/min) for
5
hours to
obtain iron nanoparticles. The nanotubes grew along
the direction of the pores. Only the nanotubes grown
out
of
the vertical pores were aligned. The iron
partkles
on
the surface and
in
the inclined pores
resulted in misoriented nanotubes. The alignment
was due to the constraint of the vertically aligned
oores.
8.
M.
Terrones
et
a/., Nature
388,
52
(1997).
9.
W.
A.
de Heer.
A.
Chateiain, D. Ugarte, Science
270.
10.
A.
G.
Rlnrler
et
a/., ibid.
269,
1550
(1995).
11.
Q.
H.
Wang et ai..
Appl.
Phys.
Lett.
72, 2912 (1998).
12.
P.
G.
Collins.
A.
Zettl
H.
Bando,
A.
Thess.
R. E.
13.
S.
Frank,
P.
Poncharal.
2.
L
Wang. W.
A.
de Heer.
ibid.
14.
T. W. Ebbesen. Carbon Nanotubes: Preparation and
1179 (1995).
Smalley. Science
278.
100
(1997).
280.
1744 (1998).
~m~rfbr
IcRr
P~~CC
R"C.
w-tnn
EI
10071
cirbon nanotube film was Immersed In acetone, then
M-Bond epoxy was added slowly until a
1:l
ratio was
attained. The sample cured at
room
temperature for
48
hours. The viscosity of the epoxy Is very low when
Introduced
to
the sample,
so
It
easily Impregnates
pores and completely mixes with the acetone. Stan-
dard mechanical thinning and Ion milling (low angle,
voltage, and cunent) were used to thin the sample to
electron transparency. Most of the substrate was
removed mechanically. followed by ion milling until
the film was exposed. hen both sides were ion
milled for
15
min.
knc1er
confm-f
3)AA&55-?
20.
This material
Is
based on work supported in part by
this program by
R.
R.
Reeberk greatly appreciated.
The authors also thank
C.
Sagerman for his technical
support. Sandia
Is
a multiprognm laboratory operat-
ed by Sandia Corporation,
e
Lockheed Martin Com-
pany, for
the
US.
Department of Energy under con-
tract
DE-AC04-94AL8500.
the US. Army Research Offk
I
The management of