Highly efficient vertical growth of wall-number-selected, N-doped carbon nanotube arrays.

Abstract

We demonstrate a straightforward approach for rapid growth of wall-number selected, N-doped CNT arrays. Highly uniform nanopatterned iron catalyst arrays were prepared by tilted deposition through block copolymer nanotemplates. PECVD growth of CNTs from the nanopatterned catalysts in an NH(3) environment generated vertical N-doped CNTs with a fine-tunability of their carbon wall numbers. The optimized growth conditions produced 52 microm long N-doped CNTs within 1 min. Owing to N-doping, the wall-number selected CNTs including DWNTs and TWNTs demonstrated enhanced electro-conductivity and chemical functionality. This remarkably fast growth of highly uniform N-doped CNTs, whose material properties and chemical functionalizability are reinforced by N-doping, offers a new area of large-scale nanofabrication, potentially useful for diverse nanodevices.

Full text

Page 1

Highly Efficient Vertical Growth of
Wall-Number-Selected, N-Doped Carbon
Nanotube Arrays
Duck Hyun Lee, Won Jong Lee,* and Sang Ouk Kim*
Department of Materials Science and Engineering, KAIST 373-1, Guseong-dong,
Yuseong-gu, Daejeon 305-701, Republic of Korea
Received October 29, 2008; Revised Manuscript Received January 15, 2009
ABSTRACT
Wedemonstrateastraightforwardapproachforrapidgrowthofwall-numberselected, N-dopedCNTarrays. Highlyuniformnanopatternediron
catalyst arrays were prepared by tilted deposition through block copolymer nanotemplates. PECVDgrowth of CNTs fromthe nanopatterned
catalysts in an NH3environment generated vertical N-doped CNTs with a fine-tunability of their carbon wall numbers. The optimized growth
conditions produced 52 µm long N-doped CNTs within 1 min. Owing to N-doping, the wall-number selected CNTs including DWNTs and
TWNTsdemonstratedenhancedelectro-conductivityandchemical functionality. ThisremarkablyfastgrowthofhighlyuniformN-dopedCNTs,
whosematerialpropertiesandchemicalfunctionalizabilityarereinforcedbyN-doping,offersanewareaoflarge-scalenanofabrication,potentially
useful for diverse nanodevices.
Carbon nanotubes (CNTs) are of broad technological interest
in electronics, photonics, energy devices, and other applica-
tions.1-3However, establishing a straightforward process for
mass production of uniform CNTs with desired structures
and properties has been a long-standing challenge.4-6In
particular, it is highly desired to precisely control over the
numbers of walls and diameter of CNTs, which are decisive
parameters for the physical properties of CNTs.3-5In this
respect, the preparation of monodisperse catalyst array having
a narrow size distribution is generally considered an effective
pathway to produce well-defined CNTs, since the number
of walls and diameter of the produced CNTs are closely
related to the catalyst size.6,7
Catalytic chemical vapor deposition (CVD) has become
a standard growth method for vertical or parallel CNT
arrays.8Plasma-enhanced CVD (PECVD) is particularly
attractive, since the high-energy plasma environment enables
low-temperature growth, high CNT alignment, and compat-
ibility with conventional Si processes. Nevertheless, CVD
methods suffer from low catalyst activity. Only a small
fraction of the catalyst particles produce CNTs; the majority
remain as undesired impurities due to contamination by
amorphous carbon.9For this reason oxygen-assisted CVD
approaches have been developed. Oxygen-containing com-
pounds, such as water (H2O),7alcohol (CH3OH),10oxygen
gas (O2),11and carbon monoxide (CO),12has been included
in the feed stream to remove undesired amorphous carbon
contamination and promote highly efficient CNT growth.
Ammonia (NH3) has often been used as an environmental
gas for PECVD growth of CNTs. The pretreatment of
catalyst particles with NH3plasma results in a thin nitride
layer on the catalyst surface, which suppresses catalyst
deactivation by decomposing excessive carbon.13As a
constituent of the source gas, NH3is an effective source for
the N-doping of CNTs. N-doping may greatly enhance the
electrical conductivity of CNTs. When electron-rich nitrogen
substitutes for carbon in a graphitic layer, the band gap of
the CNT reduces to form metallic CNTs.14,15Moreover,
N-doping may provide an opportunity for chirality control
and easy chemical functionalization.15,16Despite these nu-
merous advantages, CNT growth in an NH3environment has
rarely provided a sufficiently high yield.13,17-19
Here, we introduce a method for highly efficient and size-
selective growth of N-doped CNTs via PECVD growth from
a nanopatterned catalyst array in an NH3environment. Figure
1a schematically illustrates the catalyst nanopatterning
process via tilted deposition of catalyst through a block
copolymer nanoporous template (pore diameter, 21 nm;
center-to-center distance between neighboring pores, 35
nm).19-23Because of shadowing by the sidewall of the block
copolymer template, the size of the deposited iron particle
decreased with the deposition angle (θ). Figure 1b shows
the relationship between the average diameter of the catalyst
particles and the deposition angle. The deposited iron
thickness (t) was 0.7 nm at a tilting angle of 0°, but it
* To whom correspondence should be addressed. E-mail: (W.J.L.) wjlee@
kaist.ac.kr; (S.O.K.) sangouk.kim@kaist.ac.kr. Phone: (S.O.K.) +82-42-
869-3339. Fax: (S.O.K.) +82-42-869-3310.
NANO
LETTERS
2009
Vol. 9, No. 4
1427-1432
10.1021/nl803262s CCC: $40.75
Published on Web 03/12/2009
 2009 American Chemical Society

Page 2

decreased with tilting angle (t ∝ cos4θ) for a fixed deposition
time of 70 s.24On average, the catalyst diameter decreased
by 0.35 nm per 1° of tilting angle variation under these
deposition conditions. This ultrafine tunability of catalyst size
enabled the highly selective growth of double-walled or
triple-walled CNTs. We note that catalyst nanopatterning
using block copolymer nanotemplates can be complemented
by large-scale patterning to generate a hierarchically ordered
catalyst array.19
The SEM images of as-deposited, hexagonally packed iron
particles (Figure 1c) show that they have an anisotropic shape
as a result of the tiled deposition through a block copolymer
nanotemplate (tilting angle: 12.1°; deposition time: 70 s).
After a high-temperature annealing, the particles agglomer-
ated to become isotropic particles with a reduced diameter
of 11 nm (Figure 1d). The size of the agglomerated particles
was highly monodisperse over a substrate size of 1 × 1 cm2.
Figure 1e shows catalyst particles deposited with the highest
tilting angle of 29.1°. Because of the large tilting angle, the
particles had highly anisotropic shapes. They became iso-
tropic particles with an average diameter of 4.8 nm after
annealing (Figure 1f). We note that, despite the large
variation of catalyst size with tilting angle, the areal density
of the catalyst particles that is determined by the pore density
in the block copolymer template maintained the same value
of approximately 9.5 × 1010cm-2.
The PECVD growth of CNT arrays from the nanopatterned
catalysts under an NH3environment attained highly oriented
N-doped CNT growth with an extremely high growth rate.
Figure 2a shows a tilted cross-sectional SEM image of a
vertical CNT array grown under the optimized growth
conditions. The optimized growth conditions yielded 52 µm
tall CNT arrays in 1 min and 100 µm tall CNT arrays in 6
min (Supporting Information, Figure S1). To the best of our
knowledge, this is the highest growth rate for N-doped CNTs
and the first demonstration that a nitrogen containing element,
NH3, can bring about such a rapid growth. At our optimized
growth conditions, an environmental gas consisting of hyd-
rogen (H2) and NH3(4:1 v/v) was streaming at a flow rate
of 100 sccm. NH3plasma is known to have higher bombard-
ment energy and etching rate than H2plasma because of the
high atomic weight of nitrogen.17This high-energy plasma
can selectively etch amorphous carbon contamination and
thereby promote carbon diffusion into bulk catalyst particles.
Figure 2b shows a high magnification SEM image of the
bottom of a vertical CNT array. The CNTs, with diameters
Figure 1. (a), Schematic illustration of tilted catalyst deposition through a block copolymer nanoporous template. As the tilting angle (θ)
of E-beam evaporation increases, the resulting size of the catalyst particles decreases due to shadowing from the template sidewall. (b)
Average diameter of heat-treated catalyst particles plotted against tilting angle of evaporation. High-magnification SEM image of nanopatterned
iron catalyst particles before (c,e) and after (d,f) high-temperature annealing. The tilting angles of evaporation were θ ) 12.1° for (c,d) and
θ ) 29.1° for (e,f), respectively.
1428
Nano Lett., Vol. 9, No. 4, 2009

Page 3

of 5.1 nm, were grown from catalyst particles whose diameter
was 8.0 nm, resulting in a diameter ratio of approximately
2/3.
Figure 2c,d presents high-resolution TEM images of
N-doped CNTs grown from catalyst particles, whose average
diameters were 8.3 and 5.0 nm, respectively. The CNTs in
Figure 2c had an average diameter of 5.1 nm, and the
majority (71%) was triple-walled. In contrast, the CNTs in
Figure 2d had an average diameter of 3.1 nm, and the
majority (87%) was double-walled. The relative population
of CNTs with one, two, three, and more than three walls is
plotted against the catalyst deposition angle in Figure 2e
(tilting angle 12.1∼29.1°). When the tilting angle was less
than 12.1° (catalyst diameter >11 nm), the majority of CNTs
had multiwalled (four or more walls) structures. Above this
angle, the relative population of each type of CNTs varied
with the tilting angle. We note that the majority of CNTs
produced at the highest tilting angle of 29.1° (catalyst
diameter: 4.8 nm) consisted of double-walled CNTs rather
than single-walled CNTs. This is attributed to the preferential
growth of double-walled CNTs in an NH3 environment
(Supporting Information, Figure S2). N-doping of CNTs is
known to degrade the crystallinity of graphene layers.23In
particular, the incorporation of a pyridine-type nitrogen atom
accompanied by a vacant site may significantly lower the
crystallinity.25For double-walled CNTs with two graphene
layers, the formation of vacant sites is not fatal to the growth
of CNTs. In contrast, for single-walled CNTs with only one
graphene layer, a high density of vacant sites may result in
the collapse of the CNT structure.
Figure 2f shows the nitrogen XPS peaks for the N-doped
CNTs grown in an NH3content of 20 vol % in the envi-
ronmental gas. Generally, nitrogen incorporation in CNTs
can be distinguished by four types of XPS peaks: pyridinic
nitrogen (NP), pyrrolic nitrogen (NPYR), quaternary nitrogen
(NQ), and nitrogen oxide (NOX1, NOX2).27For our samples,
peak fitting resolved four different peaks representing pyri-
dinic nitrogen (NP, 398 eV), quaternary nitrogen (NQ, 400.8
eV), and two nitrogen oxides (NOX1, 402.5 eV, and NOX2,
405.6 eV). The peak for pyrrolic nitrogen (NPYR, 399 eV),
which may generate bamboolike structures in CNTs, was
absent.27
The growth rate of CNT arrays strongly depended on the
NH3fraction in the environmental gas (H2/NH3). Figure 3a
shows the CNT growth rate plotted against the NH3fraction
in the environemntal gas. The inset SEM images compare
the vertical CNT arrays grown at various NH3fractions. The
growth temperature was 750 °C and the total flow rate of
Figure 2. (a) Low- and (b) high-magnification SEM images showing cross-sectional views of a vertical CNT array. High-resolution
transmission electron microscopy (HRTEM) images of (c) triple-walled and (d) double-walled CNTs grown from nanopatterned catalyst
particles. (e) The relative population of SWNTs, DWNTs, TWNTs, and MWNTs plotted against the catalyst particle size. (f) Nitrogen XPS
peaks for N-doped CNTs showing the peaks for pyridinic N (NP), quaternary N (NQ), and N-oxides (NOX).
Nano Lett., Vol. 9, No. 4, 2009 1429

Page 4

environmental gas was fixed as 100 sccm for all composi-
tions.18The iron catalyst particles deposited with the same
tilting angle of 22.5° were used. The variation of CNT growth
rate clearly demonstrates the enhancement of CNT growth
by including NH3. The NH3 content of 20% yielded the
highest growth rate. When the NH3content higher than 20
wt %, the growth rate gradually decreased with NH3content.
This is due to the etching of CNTs by NH3plasma, whose
energy is higher than H2 plasma.17We note that, in our
PECVD growth, feeding water vapor as a environmental gas
did not enhance CNT growth (Supporting Information, Figure
S3). The high energy radicals generated from water mol-
ecules under PECVD give rise to the etching of CNTs,
resulting in the significant retardation of CNT growth. Figure
3b shows the N-doping level of CNTs measured by quantita-
tive XPS analysis plotted against the NH3fraction in the
PECVD environmental gas. The N-doping level monotoni-
cally increased up to 8%, as the NH3fraction varied from 0
to 50 vol%. The N-doping level was 4.6% at the fastest
growth condition of 20 wt % NH3.
Figure 4a presents the CNT growth rate plotted against
growth temperature. The NH3content in the environmental
gas was fixed at 20 vol %, where the CNT growth rate was
highest for all growth temperatures. The growth rate varied
significantly with temperature. SEM images of CNT arrays
at the same magnification clearly demonstrate the variation
of growth rate with temperature. The maximum growth rate
of 52 µm/min was attained at a growth temperature of
590 °C. Figure 4b shows the Arrhenius plot of the growth
rate. Two temperature regions are clearly discernible. In
region I (<590 °C), the growth rate steeply increased with
temperature, and the corresponding activation energy for
CNT growth was a positive value of 166.6 kJ/mol. Mean-
while, in region II (>590 °C), the growth rate decreased with
temperature, and the activation energy was a negative value
of -49.7 kJ/mol.
Figure 3. The CNT growth rate plotted as a function of NH3content
in the environmental gas. The growth temperature and the C2H2
flow rate were fixed at 750 °C and 25 sccm, respectively. The
environmental gas flow consisting of H2/NH3) 80 sccm/20 sccm
yielded the highest growth rate of CNTs. Inset SEM images,
presented at the same magnification, compare the heights of CNT
arrays grown at various temperatures for 1 min. The scale bar is
10 µm. (b) The N-doping level of CNTs plotted as a function of
the NH3content in the environmental gas.
Figure 4. (a) The CNT growth rate plotted as a function of
temperature. The NH3content in the environmental gas was 20
vol % for all growth temperatures. The growth temperature at
590 °C yielded the highest growth rate. Inset SEM images, presented
at the same magnification, compare vertical CNT arrays grown at
various temperatures for 1 min. The scale bar is 20 µm. (b)
Arrhenius plot for CNT growth rate. In region I, carbon diffusion
was the rate-limiting step, with activation energy of 166.6 kJ/mol.
In region II, an etching reaction by high-energy plasma was
involved that resulted in negative activation energy.
1430
Nano Lett., Vol. 9, No. 4, 2009

Page 5

The catalytic CNT growth is generally considered to occur
via a three-step process. The first step is decomposition of
the carbon source at the catalyst surface, which is a highly
exothermic process. In the second step, the decomposed
carbon dissolves and diffuses through the bulk metal catalyst,
forming an intermediate carbide species.27-29The carbon
atoms are saturated into bulk γ-Fe to an extent of 3∼7%,
and the activation energy for carbon diffusion is 147 kJ/
mol.28,29This endothermic process is generally considered
the rate determining step for CNT growth. The last step is
the growth of CNTs by forming a graphene layer from the
diffused carbon atoms, which is another exothermic process.
The activation energy for temperature region I, 166.6 kJ/
mol, is close to that of carbon diffusion in bulk γ-Fe. This
implies that the CNT growth occurs through the generally
accepted three-step process in this region. In contrast, region
II shows a negative temperature dependence, signifying that
an additional process was involved in the CNT growth. This
additional process resulted from the etching reaction of CNTs
by high energy plasma (Supporting Information, Figure S4).
The etching reaction became more active at a higher tem-
perature range due to the increase of high-energy radi-
cals.30
It is well known that N-doping greatly enhance the elec-
trical conductivity and chemical functionality of CNTs.14-16
Figure 5a shows the variation of the sheet resistance of
DWNT films (NH3content in the PECVD environmental
gas: 0, 10, 20, 40%) with applied current, measured by the
van der Pauw method.31The detailed film preparation and
characterization method is described in the experimental
section of the Supporting Information. The inset shows the
optical transmittance of the DWNT films. The thicknesses
of DWNT films were tuned to have the same optical
transmittance of 40% at the wavelength of 550 nm.32The
sheet resistance of DWNT films was significantly reduced
with N-doping. The sheet resistance of the DWNT film with
an N-doping level of 8% (40% NH3 in the PECVD
environmental gas) was approximately half of that of
undoped DWNT film. The electrical resistivity of semicon-
ducting CNTs generally decreases with temperature due to
the reduction of carrier density.33The sheet resistance of
undoped DWNT film rapidly decreased with applied current
due to the Joule heating. In contrast, the resistance of
N-doped CNT films having a large amount of free carriers
showed a significantly retarded decrease.
The pyridinic or quaternary nitrogen of N-doped CNTs
can serve as a facile chemical functionalization site.15,34
Figure 5b,c shows the HRTEM images of N-doped and
undoped TWNTs, both of which were functionalized with
carboxyl group-terminated gold nanoparticles by a simple
process of aqueous solution mixing and subsequent washing
(sonication in pure water). In a weak acidic condition, gold
nanoparticles were densely and uniformly attached along the
N-doped TWNTs via the ionic interaction between negatively
charged carboxylate group and the positively charged pro-
tonated nitrogen, which is strong enough to withstand
subsequent ultrasonication. (Figure 5b). In contrast, the gold
nanoparticles could not attach to the undoped TWNT walls
Figure 5. (a) The sheet resistance of DWNT films plotted as a
function of the NH3content in the PECVD environmental gas.
The sheet resistance decreased with the NH3content. Inset shows
the optical transmittance of DWNT films used to measure the
electro-conductivities. The film thicknesses were finely tuned
to an identical transmittance value of 40% at the wavelength of
550 nm. HRTEM images of gold nanoparticle decorated (b)
N-doped TWNTs and (c) undoped TNWTs. A high density of
gold nanoparticles are uniformly attached to the N-doped
TWNTs.
Nano Lett., Vol. 9, No. 4, 2009 1431