Liquid crystal engineering of carbon nanofibers and nanotubes
-
Upload
christopher-chan -
Category
Documents
-
view
214 -
download
2
Transcript of Liquid crystal engineering of carbon nanofibers and nanotubes
Carbon 43 (2005) 2431–2440
www.elsevier.com/locate/carbon
Liquid crystal engineering of carbon nanofibers and nanotubes
Christopher Chan a, Gregory Crawford a, Yuming Gao a, Robert Hurt a,*,Kengqing Jian a, Hao Li a, Brian Sheldon a, Matthew Sousa a, Nancy Yang b
a Division of Engineering, Box D, Brown University, Providence, RI 02912, USAb Sandia National Laboratories, Livermore, CA, USA
Received 1 October 2004; accepted 26 April 2005
Available online 1 July 2005
Abstract
Four high-aspect-ratio carbon nanomaterials were fabricated by template-directed liquid crystal assembly and covalent capture.
By selecting from two different liquid crystal precursors (thermotropic AR mesophase, and lyotropic indanthrone disulfonate) and
two different nanochannel template wall materials (alumina and pyrolytic carbon) both the shape of the nanocarbon and the graph-
ene layer arrangement can be systematically engineered. The combination of AR mesophase and alumina channel walls gives plate-
let-symmetry nanofibers, whose basic crystal symmetry is maintained and perfected upon heat treatment at 2500 �C. In contrast, AR
infiltration into carbon-lined nanochannels produces unique C/C-composite nanofibers whose graphene planes lie parallel to the
fiber axis. The transverse section of these composite nanofibers shows a planar polar structure with line defects, whose existence
had been previously predicted from liquid crystal theory. Use of solvated AR fractions or indanthrone disulfonate produces plate-
let-symmetry tubes, which are either cellular or fully hollow depending on solution concentration. The use of barium salt solutions
to force precipitation of indanthrone disulfonate within the nanochannels yields continuous nanoribbons rather than tubes. Overall
the results demonstrate that liquid crystal synthesis routes provide molecular control over graphene layer alignment in nanocarbons
with a power and flexibility that rivals the much better known catalytic routes.
� 2005 Elsevier Ltd. All rights reserved.
Keywords: Carbon nanofibers; Carbon nanotubes; Mesophase pitch; Electron microscopy; Crystal structure
1. Introduction
Much of the excitement surrounding new carbon
nanomaterials can be traced to their directional proper-
ties, which arise through precise orientation of thegraphene layers [1,2] that are the anisotropic building
blocks of all sp2-hybridized carbon forms [3]. A long-
term goal in carbon synthesis is to develop techniques
for systematic control of graphene layer arrangement
0008-6223/$ - see front matter � 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.carbon.2005.04.033
* Corresponding author. Tel.: +1 401 863 2685; fax: +1 401 863 9120.
E-mail address: [email protected] (R. Hurt).
in order to fabricate materials and nanomaterials with
crystal structures preprogrammed for specific applica-
tions [3].
In 1995, Rodriguez et al. published an article entitled:
‘‘Catalytic Engineering of Carbon Nanostructures’’ [4],describing the synthesis of three nanofiber types: ‘‘tubu-
lar’’ nanofibers (multi-wall nanotubes), ‘‘platelet’’
nanofibers, whose graphene layers lie perpendicular to
the fiber axis; and ‘‘herringbone’’ nanofibers with a tilted
layer arrangement [4]. More recent work has also dem-
onstrated the related cup-shaped nanofibers also with
titled layer arrangement [1,5]. These platelet, herring-
bone, and cup-shaped nanofibers are inferior to thetubular nanofibers in mechanical strength and conduc-
tivity, but contain exposed graphene edge sites that
Fig. 1. Four high-aspect-ratio carbon nanomaterials fabricated here
by liquid crystal assembly and covalent capture. The lines show the
local mean orientation of the graphene planes.
Fig. 2. Two liquid crystalline systems used in nanofiber and nanotube
synthesis. Panel A: AR mesophase, a thermotropic LC that shows
surface anchoring states that vary systematically with substrate [11];
Panel B: indanthrone disulfonate, a lyotropic LC whose rod-like
aggregates in aqueous solution anchor parallel to substrates (Panel C),
and align as concentration increases (Panel D). The parallel alignment
of the rods on substrates leads to edge-on alignment of the molecular
disks at the periphery of the carbon artifact.
2432 C. Chan et al. / Carbon 43 (2005) 2431–2440
make them attractive for complementary applications
such as catalysis [6,7], substrates for covalent functionali-
zation, and Li intercalation electrochemistry [8]. The
catalytic route derives its flexibility from the ability of
catalytic nanoparticles to decompose vapor-phase or-
ganic species and precipitate graphitic carbon from par-ticular crystal facets with defined orientations that
depend on catalyst formulation and reaction conditions.
In recent years, the catalytic approach has been exten-
sively applied for synthesis of high-aspect-ratio nanocar-
bons due to its flexibility in structure control, relatively
mild synthesis conditions, and good process scalability.
The goal of the present article is to demonstrate that
liquid crystal routes offer similar flexibility. By control-ling the interactions between discotic molecules and
template surfaces, liquid crystal assembly can be directed
to produce a variety of new high-aspect-ratio carbon
nanomaterials with varying graphene layer arrange-
ments. The first attempt to fabricate nanocarbons in this
way used AR mesophase, the discotic naphthalene poly-
mer, in combination with anodic alumina templates to
produce ‘‘orthogonal’’ carbon nanofibers with graphenelayers perpendicular to the fiber axis [9]. This structure
reflects the original (uncarbonized) discotic assembly
[9], which, of all possible structures, is the only one that
achieves the desired edge-on molecular anchoring on the
inner alumina surfaces while also avoiding elastic direc-
tor strain [9]. These nanofibers have the same basic crys-
tal symmetry as the catalytic platelet nanofibers, but the
graphene layers are smaller and ‘‘meander’’ around themean orientation in a manner that is typical of low-tem-
perature carbons derived from, or through, mesophases
[3]. Recently Konno et al. [10] reported platelet structure
nanofibers from PVC and PVA polymer precursors in
the presence of anodic alumina templates. They also re-
port that tunnel-etch aluminum (metal) substrates give
nanofibers with parallel orientation [10], demonstrating
the ability to set fiber structure through selection ofthe template wall material. This particular switch was
not predictable since the only report of anchoring state
on Al metal is edge-on (reported for an aluminum foil
[11]), which is inconsistent with the observed parallel
alignment. More work is needed to understand the inter-
actions of polyaromatic compounds with complex metal
surfaces, which may differ greatly in surface oxidation
state or presence of adsorbed species.The present paper describes the synthesis and struc-
ture of four high-aspect-ratio carbon nanomaterials fab-
ricated by liquid crystalline (LC) precursors (see Fig. 1).
It will be seen that the carbon crystal structure is a direct
consequence of the selection of LC precursor and the
channel wall composition, which together direct the ori-
entation of graphene layers at the channel wall. In the
case of nanoscale materials, the wall alignment can bethe deciding factor that establishes the bulk crystal
structure of the fibrous carbon form.
2. Experimental
2.1. Materials
This study uses both a thermotropic and a lyotropic
liquid crystal precursor for nanocarbon synthesis (see
Fig. 2). AR mesophase (HP grade, Mitsubishi Gas
C. Chan et al. / Carbon 43 (2005) 2431–2440 2433
Chemical) is a well-known carbon precursor made by
the polymerization of naphthalene. It has a distribution
of molecular weights spanning 400–2000 Daltons with a
mean of approximately 700 Daltons. It softens between
250 and 300 �C into a homogeneous discotic nematic
liquid crystal phase that has been described as primarilythermotropic, but with some lyotropic nature due to
the broad distribution of molecular weights [12]. In an
attempt to make hollow forms (tubes), experiments were
also carried out with soluble fractions of AR mesophase
in pyridine and quinoline solvents. The resulting solu-
tions are rich in polyaromatic material but no longer
exhibit liquid crystallinity both due to dilution and to
the lower mean molecular weight of the soluble fractioncompared to the parent material.
The second LC precursor is an aqueous solution of
the ammonium salt of indanthrone disulfonate. This is
a new carbon precursor used first only recently in our
laboratory to fabricate ordered all-edge surface carbon
thin films [13]. This LC precursor is synthesized by
introducing sulfonic acid groups on the periphery of
indanthrone (also ‘‘indanthrene’’), a polyaromatic dyeof planar discoid shape (see Fig. 2). In such amphiphilic
discotic molecules, the disk peripheries are hydrophilic,
but the polyaromatic faces remain hydrophobic, and
in aqueous solution the planar molecules stack face-
to-face to provide favorable local environments for both
faces and edge groups [14,15] (see Fig. 2). In indan-
throne disulfonate solutions, this face-to-face stacking
is extensive, leading to rod-like aggregates with aspectratio near 200 having approximate diameters of 1.5 nm
and mean lengths of 300 nm [15]. Proton dissociation
imparts negative charge to the aggregates, which aids
in their dispersion [15], and above 4–5% these solutions
form lyotropic liquid crystalline phases in which the rod-
like aggregates align by self-exclusion and electrostatic
repulsion [15]. Dried films of indanthrone disulfonate
have a density of 1.69 g/cm3 by picnometry [16] andXRD analysis reveals crystallinity with a d-spacing of
0.336 nm in the p-stacks and molecular order parame-
ters from 0.86 to 0.91 [16]. For the present work, indan-
throne disulfonate was acquired as an ammonium salt
solution from the firm Optiva (South San Francisco),
which uses the solutions for fabrication of thin film or-
ganic polarizers.
2.2. Procedures
The syntheses involved solution infiltration into com-
mercial nanochannel alumina (Whatman Anodisc 47),
which has been used extensively in the past as a template
for fabrication of nanofibers/tubes from polymers
[17,18], metals [17], organic molecules [19] and carbons
[20,21]. The goal here is to use nanochannels not onlyfor shape control, but also to direct the molecular struc-
ture of the material through polyaromatic/alumina sur-
face interactions. For some syntheses, a thin CVD
carbon coating was applied to the inner and outer alu-
mina surfaces using reactant mixtures of acetylene
(100 sccm) and ammonia NH3 (400 sccm) at 900 �C for
12 hours. Most nanocarbon samples were prepared by
spontaneous capillary infiltration at 300 �C followedby slow heating (4 �C/min) to 700 �C and a 10 min iso-
thermal hold time. Low temperatures were chosen for
the initial carbonization in order to gain insight into
the initial orientation of the graphene layers in the LC
state with only limited possibilities for high-temperature
rearrangement. Some nanocarbon samples were further
heat treated at 2500 �C in a custom rapid thermal
annealing device described elsewhere [22], whose peakheating approaches 1000 K/s. The alumina templates
were removed by 0.4 M NaOH etching as described pre-
viously [9] and the samples characterized for structure
by electron microscopy and X-ray diffraction.
3. Results and discussion
3.1. Platelet-symmetry nanofibers
Figs. 3 and 4 show new characterization results on the
platelet-symmetry LC-derived nanofibers first reported in
[9]. These nanofibers have already been shown by
HRTEM to have a platelet morphology (edge-on orienta-
tion with the graphene layers in most fibers lying approx-
imately perpendicular to the fiber long axis [9]) and arethus held together with axial Van der Waals forces. Here,
they show evidence of the expected low rigidity (Fig. 3A
and B) as well as transverse fissures after handling and
reheating (Fig. 3C). The group of nanofibers in Fig. 3
shows that most have fissures lying strictly perpendicular
to the fiber axis, but some fibers show tilted fissures
strongly suggesting tilted graphene layers (Fig. 3C). The
tilted fibers do not have herringbone structure, which re-quires a central kink, but rather a platelet structure in
which an asymmetric tilt relative to the fiber axis is super-
imposed on all the layers uniformly. The origin of the tilt
is unknown, but it is possible that the two closely related
anchoring states (strictly perpendicular vs. tilted) have
similar surface free energies, or liquid crystal ‘‘anchoring
energies’’, and thus small asymmetric stresses associated
with volatile release or carbonization shrinkage couldcause tilt in a minority of the fibers.
Arrays of these fibers show both top and bottom car-
bon films due to wetting of the alumina template face.
One film is typically of submicron thickness, while the
other film (often on the infiltration side) has a thickness
that depends upon the amount of excess AR in the ori-
ginal formulation. The thick film can be mostly removed
by scraping prior to carbonization, and then both thinfilms can be removed by controlled air oxidation at
520 �C for 10–40 min.
Fig. 3. Platelet-symmetry or ‘‘orthogonal’’ carbon nanofibers [9]
fabricated by AR mesophase infiltration into nanochannel alumina
followed by heating. A: free standing array, B: polished transverse
section of array, C: SEM of the 200 nm diameter fibers showing
numerous transverse fissures after handling and re-heating in 1% H2/
He in the presence of iron nitrate. Some of the nanofibers have tilted
fissures (‘‘T’’) suggesting tilted graphene layers, while most are strictly
perpendicular (‘‘P’’). Note that both perpendicular and tilted types are
seen by HRTEM.
2434 C. Chan et al. / Carbon 43 (2005) 2431–2440
Fig. 4 shows XRD and HRTEM analysis of the
platelet-symmetry nanofibers before and after 2500 �Cheat treatment. The initial fibers prepared at 700 �Cshow short meandering graphene layers typical of alow-temperature carbon from a liquid crystal precursor
[3,9]; a structure that reflects the molecular positions in
the liquid crystalline phase at the point of solidification
[9]. After 2500 �C heat treatment, the essential platelet
structure is preserved in the interior of the fiber and
the short, meandering fringes are greatly lengthened
and straightened. We observe both strictly perpendicu-
lar and tilted arrangements by HRTEM in both the
low-temperature and high-temperature samples. Heat
treatment is thus not the primary cause of the tilt. The
existing data set is not sufficient to state with statistical
significance whether heat treatment changes the relative
proportions of the two fiber types.
Application of the Warren equation [23] to theXRD 002 line broadening yields about 2 nm for the
700 �C sample and 43 nm for the 2500 �C sample. Near
the fiber edge, there is a clear evidence of surface
reconstruction similar to that seen by Lim et al. [24]
during annealing of catalytically produced platelet
nanofibers. The driving force for this surface recon-
struction is believed to be the minimization of surface
energy by elimination of free edge sites (danglingbonds). Without removal of the reconstructed zone
by etching or oxidation, the potentially large number
of active edge sites are likely to be unavailable for most
chemical reaction processes.
3.2. C/C-composite nanofibers
We hypothesized previously [9] that the platelet struc-ture is selected because it represents the minimum free
energy state for a discotic liquid crystal in a confined
nanocylinder, where the preferred anchoring state is
edge-on (planar). It should therefore be possible to fab-
ricate alternate nanofiber types by switching the anchor-
ing state to face-on (homeotropic), provided the new
anchoring state is strong enough to overcome the elastic
strain that is unavoidable in that configuration. Our ap-proach here is to coat the inner wall surface with a mate-
rial known to produce face-on (homeotropic) anchoring
on flat substrates. Only a few materials have been re-
ported to produce face-on anchoring of AR mesophase:
Pt, Ag, mica, polyimide, and carbon [11]. Here, we chose
to work with carbon, which can be deposited on the in-
ner alumina surfaces by well-known CVD methods [25].
Fig. 5 shows C/C composite nanofibers produced byinfiltrating AR-mesophase into nanochannel alumina
pre-coated with a thin (approx. 5 nm) CVD carbon
layer, then followed by a second carbonization at
700 �C. Fig. 5 clearly shows that CVD pre-coating does
indeed switch the graphene layer arrangement from per-
pendicular to parallel with respect to the fiber axis (see
panel C). Since the CVD layer is much thinner than
the channel radius (5 nm vs. 100 nm), the confinementgeometry is not significantly affected, and the switch
must reflect the new wall anchoring state. It is interest-
ing that the CVD and mesocarbon layers are not clearly
distinct in panel C. This is not entirely unexpected;
although their synthesis routes are different, both have
been subject to similar peak temperatures, which plays
an important role determining graphene layer diameter
and defect density. These CVD/mesocarbon compositenanofibers appear to be straighter and thus stiffer (Fig.
5A and B) than the platelet symmetry nanofibers
Fig. 4. Crystal structure characterization of as-prepared (700 �C) and annealed (2500 �C) platelet symmetry liquid-crystal-derived nanofibers. High
temperature treatment preserves and perfects the platelet symmetry (either strictly perpendicular or tilted) in the fiber interior, but also leads to
surface reconstruction to avoid free edges at the fiber periphery. This particular example image shows a tilted structure, but the high-temperature
annealed samples also contain strictly perpendicular fiber types, also of high crystallinity.
Fig. 5. Carbon/carbon composite nanofibers made by AR mesophase infiltration of CVD-carbon-lined alumina templates. A. FE-SEM image
showing fully dense nanofibers of uniform diameter, B. TEM image showing straight and uniform nanofibers, C. comparison of unfilled carbon
nanotubes (top) with the composite nanofibers (bottom). The dominant alignment of graphene layers in the mesocarbon interior is parallel to the
fiber axis following the parallel alignment in the outer CVD carbon ring.
C. Chan et al. / Carbon 43 (2005) 2431–2440 2435
2436 C. Chan et al. / Carbon 43 (2005) 2431–2440
(Fig. 3A and B), which is likely related to the dominant
parallel orientation of the graphene layers.
These CVD/mesocarbon composite nanofibers show
a fascinating transverse texture on fracture surfaces
(Fig. 6). Each of the fracture surfaces examined shows
concentric structure in an outer annular zone, givingway to parallel alignment in the center. Many of the fi-
bers have non-circular cross sections and in these cases
the molecular planes in the center region appear to lie
along the major dimension of the transverse section.
The non-circularity may be due to anisotropic carbon-
ization shrinkage. The apparent graphene layer struc-
ture is illustrated in Fig. 6B. It is important to note
that the region of concentric alignment includes not onlythe CVD layer, but also a significant portion of the inte-
rior mesophase-derived carbon. Considering the meso-
carbon alone, its texture is a known liquid crystal
confinement pattern referred to as ‘‘planar polar with
line defects’’ (PPLD), where the line defects are of
strength +1/2 (+p disclinations). The existence of PPLD
has been predicted [26,27] but not to our knowledge ob-
served experimentally prior to this study. Theoreticalpredictions of the PPLD structure include the Monte
Fig. 6. Transverse structure of the CVD/mesocarbon composite
nanofibers. A: Typical fracture surfaces, B: Sketch of the graphene
layer orientational pattern, which corresponds to a known liquid
crystal confinement pattern classified as Planar Polar with Line Defects
(PPLD). Note that the line defects are not at the CVD/mesocarbon
interface (approx. 5 nm from edge) but rather within the mesocarbon.
Carlo study by Chiccoli et al. [26], where PPLD is ob-
served in submicron cavities when homeotropic surface
anchoring is strong (as might be expected for AR on car-
bon [28]). The molecular dynamics study of Bradac et al.
[27] predicts a variety of transverse textures in cylindri-
cally confined LCs depending on anchoring strengthand elastic constants. The structures include planar po-
lar (PP), planar radial (PR), escape radial (ER) and
PPLD. The PPLD structure is predicted to be favored
for certain conditions when surface anchoring is strong.
It may be said that strong anchoring favors a uniform,
defect-free outer layer, and since this geometry (homeo-
tropic anchoring in a cylinder) requires defects due to
infinite curvature at the center, the defects are internal-ized as two +p disclinations.
Several related structures have been observed experi-
mentally. If the PPLD line defects are brought to the
periphery one obtains the simple planar polar structure,
which as been experimentally observed by Crawford
et al. [29] in <0.4 lm cavities filled with the rod-like
liquid crystal 5CB. Fathollahi and White [30] observed
the relaxation of the flow-induced mesophase micro-structures in a uniform set of 720 lm diameter (L/D >
25) capillaries subjected to identical flow conditions.
Upon annealing at 300 �C, the microstructure relaxes
to a pair of +p disclinations with a radial orientation
of the discotic planes near the outer tube surface. This
structure can be mapped onto that of Fig. 5 by a 90�rotation of all layers. This related structure [30] is not
expected here since it would require edge-on anchoringon nanochannel walls. It is a candidate structure for
the uncoated alumina templates, but is not observed.
Rather at the nanoscale, the edge-anchored discs prefer
instead to flip out of plane to produce the platelet struc-
ture (Fig. 3), which avoids the +p disclinations and
indeed all elastic strain. We offer the following explana-
tion for the Fathollahi and White texture. At larger
length scales in their supramicron cavities, rearrange-ment times are longer and the low curvature reduces
the driving force for the out-of-plane flip. As a result,
it does not occur over experimental time frames. Instead
the flow-induced alignment parallel to the cavity axis re-
mains metastable and annealing produces only rear-
rangements within the transverse plane—a limited
two-dimensional free energy minimization that leads to
the observed texture. Fig. 7 shows that 2500 �C annealingimproves crystallinity of the composite nanofibers, while
maintaining the preferred parallel orientation. Both car-
bon components show large increases in fringe length.
3.3. Platelet-symmetry nanotubes
In order to explore whether liquid crystalline routes
can be used to make hollow forms (tubes), precursorswith very high volatile yields or solvent fractions were
sought whose vaporization would produce hollow
Fig. 7. HRTEM fringe images of the C/C composite nanofibers after
rapid heating to 2500 �C. The parallel orientations are maintained in
both regions and the overall crystallinity greatly increased.
C. Chan et al. / Carbon 43 (2005) 2431–2440 2437
spaces within the nanochannels. Most attractive are
solution systems, where controlled solvent evaporation
early in the heating process may lead to thin adherent
organic films on the inner channel walls, which can be
converted by carbonization into unique carbon nano-
tube varieties. Our first attempt to make carbon tubesused soluble fractions of AR mesophase in pyridene
and quinoline solvents. The procedure was otherwise
the same as for the melt-processed platelet nanofibers.
Figs. 8 and 9 show that evaporation of the solvent pro-
duces open-ended tubes (Fig. 8C)—some fully hollow
and some with cellular structure. When high-concentra-
tion (7.5 wt.% in quinoline) solutions are employed, the
Fig. 8. Cellular carbon nanotubes (A,B,C) and fully hollow carbon nanotub
structure in (D) is the result of lower polyaromatic concentration (1 wt.%) i
cellular structure is especially pronounced (Fig. 8A).
The hollow regions clearly appear to be vesicles, some
of which show nearly regular spacing, and the overall
structure can be described as an one-dimensional solid
cellular foam. Increasing the drying time at a tempera-
ture at 80 �C (below the 113 �C boiling point of quino-line) had little effect on the foam structure (Fig. 8B).
Reducing the AR concentration to 1 wt.% greatly sup-
presses this foam structure and produces instead hollow
nanotubes (Fig. 8D). These tubes have thin wall, how-
ever, which show some inter-tube fusion after template
removal leading to some structures much larger than
the 200 nm channels. The 1D cellular foam thus appears
to arise at high concentration where saturated condi-tions readily occur during drying leading to precipita-
tion within and across the bulk channel rather than at
a later stage as a thin film drying on inner wall surfaces.
Fig. 9 shows perpendicular graphene layer alignment
in the thin wall sections of the cellular tubes. These are
platelet-symmetry tubes (structure C in Fig. 1) and are
hollow relatives of the platelet-symmetry nanofibers
(structure A in Fig. 1). The formation mechanisms arenot necessarily identical, however. The solvated pitch
fractions are isotropic both in solution and upon drying
and thus their initial assembly is not governed by liquid
crystal theory. We wish here only to propose two possi-
bilities: (1) the isotropic fractions transform into meso-
phase during carbonization and align edge-on by the
es (D) produced by solution processing of AR mesophase. The hollow
n quinoline solution compared to 7.5 wt.% for (A).
Fig. 9. Perpendicular graphene layer arrangement in the thin wall
sections of cellular carbon nanotubes produced from high-concentra-
tion solutions of AR mesophase in pyridine.
2438 C. Chan et al. / Carbon 43 (2005) 2431–2440
same anchoring mechanism as the original whole AR,
and (2) the alignment occurs by a non-LC mechanism
driven by maximization of p stacking. Even without
an LC phase, mobile polyaromatic compounds below
their decomposition temperatures will typically stack
to maximize p–p bonds, which are the strongest non-
covalent interactions in the system. Liquid crystalline
phases show long-range order, but at these nanometriclength scales, only short range ordering is necessary to
produce the molecular structure in Fig. 9, which can
then captured by carbonization. Indeed polyaromatic
p-stacking with edge-on orientation is a common assem-
bly pattern seen in organic thin films and in the ultrafine
channels of mesoporous silica [31]. It requires liquid
crystalline phases only when the alignment must propa-
gate over long length scales (�10 nm).A more well-defined synthesis route to platelet-sym-
metry tubes uses capillary infiltration of ammonium
indanthrone disulfonate aqueous solutions, which form
true liquid crystalline phases at concentrations above
4 wt.%. The starting solutions lose most of their mass
upon drying, leading to the formation of a thin film
(2–12 molecular layers) on the inner surfaces of the
100 nm radius cylindrical template channels.Fig. 10 shows that infiltration of indanthrone disulf-
onate solution followed by drying and 700 �C treatment
produces monodisperse carbon tubes of 100 nm radius
and 60 lm length, which form free standing ordered ar-
rays upon template etching (see Fig. 10A). Using
12 wt.% indanthrone disulfonate, the tubes show a cellu-
lar structure with hollow cavities separated by internal
membranes. The cellular structure can be suppressed al-most entirely by reducing solvent concentration from
12% to 2%. The lower concentration produces tubes
with thinner walls and almost no internal structure.
High resolution fringe images reveal the same platelet-
symmetry in these thin-walled carbon nanotubes (Fig.
10B). The tube wall structure consists of short (2–3
nm) graphene layers, similar to AR-derived carbons pre-
pared at the same temperature. This crystal structureimplies that the rods orient parallel to the channel axis
during drying. This is not unexpected, as the rod length
is comparable to the channel diameter, making parallel
orientation much more favorable, especially in the
curved thin liquid films that coat the inner nanochannel
walls during drying. The dried solid film of rod-like
supramolecules is then covalently captured by thermal
polymerization with accurate translation of the molecu-lar order into an arrangement of linked graphene layers.
Note that the success of this covalent capture scheme
could not be predicted a priori, as many thermal carboni-
zation processes destroy supramolecular order in the
organic precursor and/or alter the overall form of the
carbon body though re-softening and volatile product
release [3]. In separate thermogravimetric experiments,
we found that bulk samples of indanthrone disulfonatebegin to decompose at around 300 �C, at a point where
the material retains most of its optical anisotropy [13].
As heating continues, sharp features remain intact
through 700 �C indicating an all-solid-state carboniza-
tion path, and the final bulk carbons show multi-domain
anisotropy [13].
These platelet symmetry tubes are mechanically sta-
ble when made from 7% or 12% solution concentration.These solutions retain a reasonably low viscosity allow-
ing nanotube micropatterns to be written by pro-
grammed injection using capillary tube pens [32]. In
contrast, the fully hollow tubes made from 2% solution
are very fragile outside the template. More work is
needed on processing and stability of the thin-walled
hollow variety, but it is likely that their primary use will
be inside the alumina template rather than free standing.
3.4. Carbon nanoribbons
Finally, we report the ability to form long, continu-
ous carbon nanoribbons by a slight modification of
the synthesis procedure. Dipping the filled nanochannel
membrane in 10 wt.% BaCl2 solutions prior to drying
and carbonization leads not to tubes but to ribbons orstrips of 60 lm in length, 200 nm in width with rectangu-
lar cross section (Fig. 11). In separate experiments, we
Fig. 10. Large, thin-walled carbon nanotubes formed by capillary infiltration of indanthrone disulfonate solutions into nanochannel alumina
followed by thermal covalent capture at 700 �C and template removal. A: intact tube array, showing the cellular nature of the tubes made when
solution concentration is high (12 wt.%). B: high-resolution TEM image shown the perpendicular graphene layer orientation in the walls of 2 wt.%
tubes. C: sketch of the platelet-symmetry tubes with the molecular disk size exaggerated for visibility.
C. Chan et al. / Carbon 43 (2005) 2431–2440 2439
observed rapid formation of fibrous precipitates when
indanthrone disulfonate solutions were injected throughfine glass tubes into a BaCl2 solution bath. We therefore
believe that in the template synthesis, the divalent bar-
ium ion cross-links the negatively charged aggregates
and reduces their solubility, leading to early precipita-
tion within the nanochannels instead of deposition on
the inner wall surfaces during drying. HRTEM shows
these ribbons to have a more random crystal structure
suggesting that the rod-like molecular aggregates were
Fig. 11. Carbon nanoribbons or strips fabricated by barium chloride
precipitation of indanthrone disulfonate within the channels of anodic
alumina templates. A: precipitation chemistry and sketch, B/C:
example ribbons after carbonization and template removal.
not highly oriented in the solution at the point of precip-
itation prior to drying.In summary, we believe the tubular structure (Fig. 10)
first forms during drying in the form of tubular precur-
sor film that assembles on the inner channel walls as a
curved solvent meniscus recedes. Addition of barium
salts bypasses this mechanism by forcing the organic
precursor out of solution in the filled channel before
drying occurs. In this case, the precursor is not forced
to precipitate on the curved channel wall, but is free toadopt its characteristic preferred precipitate morphol-
ogy, which is not tubular but ribbon-like.
4. Conclusions
Liquid crystals provide a powerful and flexible route
to new high-aspect-ratio carbon nanomaterials. The pre-
cursor/template pair can be intelligently selected to
establish desired graphene layer arrangements at the
carbon/template interface, and at the nanoscale, this
alignment propagates inward a sufficient distance to dic-
tate the overall structure of the material. To date wehave demonstrated platelet-symmetry nanofibers, C/C-
composite nanofibers with graphene layers parallel to
the fiber axis, platelet-symmetry tubes both cellular
and open, and carbon nanoribbons with a lesser degree
of crystalline order. The platelet and composite nanofi-
bers are similar, but not identical to, the well-known
platelet and tubular nanofibers synthesized by catalytic
routes. The LC-derived version offer the followingadvantages: (1) their degree of crystallinity can be varied
2440 C. Chan et al. / Carbon 43 (2005) 2431–2440
from extremely low (including quenched or partially car-
bonized mesophase) to high by adjusting the tempera-
tures of formation and annealing, (2) they can be
easily grown in well-ordered arrays, (3) they are free of
metallic catalyst residues. The need for a sacrificial tem-
plate, however, is a significant disadvantage for bulksynthesis. For this reason, the LC-derived nanofibers
are most attractive in high-value, array-based devices.
More work is needed on the mechanical and thermal
stability of the platelet symmetry tubes both within the
template and free standing. More work is also needed
to identify surface treatment procedures for the edge-
on forms to remove reconstructed layers and access
the potentially abundant active sites.
Acknowledgements
This work was supported by the National Science
Foundation through a Nanoscale Interdisciplinary Re-
search Team (NIRT) Grant at Brown University,
CMS-0304246, and by the Electric Power Research
Institute, Dr. A. Mehta project manager. The authors
would like to thank Daniel Morris for the Fe-doped
fiber images, Michael Paukshto of Optiva Inc. for theindanthrone samples, and Essie Yamoah and Bevan
Weissman for technical contributions in the laboratory.
References
[1] Harris PJF. Carbon nanotubes and related structures. Cam-
bridge: Cambridge University Press; 2001.
[2] Ajayan PM. Carbon nanotubes. In: Nalwa HS, editor. Hand-
book of nanostructured materials and nanotechnology. New
York: Academic Press; 2000 [Chapter 8].
[3] Hurt RH, Chen ZY. Liquid crystals and carbon materials. Phys
Today 2000;53(3):39–44.
[4] Rodriguez NM, Chambers A, Baker RTK. Catalytic engineering
of carbon nanostructures. Langmuir 1995;11:3862–6.
[5] Endo M, Kim YA, Hayashi T, Fukai Y. Structural characteriza-
tion of cup-stacked-type nanofibers with an entirely hollow core.
Appl Phys Lett 2002;80(7):1267–9.
[6] Gao R, Tan CD, Baker RTK. Ethylene hydroformylation on
graphite nanofiber supported rhodium catalysts. Catal Today
2001;65(1):19–29.
[7] Bessel CA, Laubernds K, Rodriguez NM, Baker RTK. Graphite
nanofibers as an electrode for fuel cell applications. J Phys Chem
B 2001;105(6):1115–8.
[8] Yoon SH, Park CW, Yang H, Korai Y, Mochida I, Baker RTK,
et al. Novel carbon nanofibers of high graphitization as anodic
materials for lithium ion secondary batteries. Carbon
2004;42:21–32.
[9] Jian K, Shim HS, Schwartzman A, Crawford GP, Hurt RH.
Orthogonal carbon nanofibers by template-mediated assembly of
discotic mesophase pitch. Adv Mater 2003;15(2):164–7.
[10] Konno H, Sato S, Habazaki H, Inagaki M. Formation of platelet
structure carbon nanofilaments by a template method. Carbon
2004;42:2756–9.
[11] Jian K, Shim HS, Tuhus-Dubrow D, Bernstein S, Woodward C,
Pfeffer M, et al. Liquid crystal surface anchoring of mesophase
pitch. Carbon 2003;41(11):2073–83.
[12] Hu Y, Hurt RH. Thermodynamics of carbonaceous mesophase:
II. General theory for nonideal solutions. Carbon
2001;39(6):887–96.
[13] Jian K, Xianyu H, Eakin J, Gao Y, Crawford GP, Hurt RH.
Orientationally ordered and patterned discotic films and carbon
films from liquid crystal precursors. Carbon 2005;43(2):407–15.
[14] Iverson IK, Casey SM, Seo W, Tam-Chang SW, Pindzola BA.
Controlling molecular orientation in solid films via self-organi-
zation in the liquid-crystalline phase. Langmuir 2002;18(9):
3510–6.
[15] Dembo A, Ionov A, Lazarev P, Manko A, Nazarov V. Lyotropic
dye-water mesophases formed by rod-like supramolecules. Mol
Cryst Liquid Cryst Sci Technol Section C: Mol Mater
2001;14(4):275–90.
[16] Lazarev P, Lokshin K, Nazarov V. X-ray diffraction by large area
organic crystalline nano-films. Mol Cryst Liquid Cryst Sci
Technol Section C: Mol Mater 2001;14(4):303–11.
[17] Yu BZ, Li MK, Lu M, Li HL. Morphologies and optical
properties of poly(2,5-diethoxyphenylene) nanofibril arrays. Appl
Phys A: Mater Sci Process 2003;76(4):593–8.
[18] Martin CR. Nanomaterials: A membrane-based synthetic
approach. Science 1994;266(5193):1961–6.
[19] Zhao L, Yang W, Ma Y, Yao J, Li Y, Liu H. Perylene nanotubes
fabricated by the template method. Chem Commun
2003;19:2442–3.
[20] Kyotani T, Tsai L, Tomita A. Formation of ultrafine carbon tubes
by using an anodic aluminum oxide film as a template. Chem
Mater 1995;7(8):1427–8.
[21] Hulteen JC, Chen HX, Chambliss CK, Martin CR. Template
synthesis of carbon nanotubule and nanofiber arrays. Nanostruct
Mater 1997;9(1–8):133–6.
[22] Shim HS, Hurt RH. Thermal annealing of chars from diverse
organic precursors under combustion-like conditions. Energy
Fuels 2000;14(2):340–8.
[23] Short MA, Walker Jr PL. Measurement of interlayer spacings and
crystal sizes in turbostratic carbons. Carbon 1963;1(1):3–9.
[24] Lim S, Yoon SH, Mochida I. Surface modification of carbon
nanofiber with high degree of graphitization. J Phys Chem B
2004;108(5):1533–6.
[25] Che G, Lakshmi BB, Martin CR, Fisher ER, Ruoff RS. Chemical
vapor deposition based synthesis of carbon nanotubes and
nanofibers using a template method. Chem Mater
1998;10(1):260–7.
[26] Chiccoli C, Pasini P, Semeria F, Berggren E, Zannoni C.
Computer simulations of cylindrically confined nematics. Mol
Cryst Liquid Cryst Sci Technol Section A: Mol Cryst Liquid Cryst
1996;290:237–44.
[27] Bradac Z, Kralj S, Zumer S. Molecular dynamics study of nematic
structures confined to a cylindrical cavity. Phys Rev E
1998;58(6):7447–54.
[28] Hurt RH, Krammer G, Crawford G, Jian K, Rulison C.
Polyaromatic assembly mechanisms and structure selection in
carbon materials. Chem Mater 2002;14:4558–65.
[29] Crawford GP, Allender DW, Doane JW. Surface elastic and
molecular anchoring properties of nematic liquid crystals confined
to cylindrical cavities. Phys Rev A 1992;45(12):8693–708.
[30] Fathollahi B, White JL. Polarized-light observations of flow-
induced microstructures in mesophase pitch. J Rheol 1994;38(5):
1591–607.
[31] Kim TW, Park IS, Ryoo R. A synthetic route to ordered
mesoporous carbon materials with graphitic pore walls. Angew
Chem Int Ed 2003;42(36):4375–9.
[32] Sousa ME, Chan C, Jian K, Gao Y, Yang N, Hurt R, et al.
Micro-patterned carbon nanotube arrays using pen-writable
lyotropic liquid crystals. Soc Inf Display Digest Tech Papers
2004;35:936–9.