Additive-free carbon nanotube dispersions, pastes,
gels, and doughs in cresols
Kevin Chioua, Segi Byuna, Jaemyung Kima, and Jiaxing Huanga,1
a
Department of Materials Science and Engineering, Northwestern University, Evanston, IL 60208
Edited by David A. Weitz, Harvard University, Cambridge, MA, and approved April 23, 2018 (received for review January 8, 2018)
cresol
| carbon nanotubes | dough | solution processing | viscoelasticity
C
arbon nanotubes (CNTs) have been found to be attractive
for applications due to their excellent electrical, thermal, and
mechanical properties (1–5). Some types of nanotubes are already mass-manufactured in the ton scale in the form of powders
(3, 6–8). As with other industrial materials, powders are often
used with solvents during processing, such as in the forms of
dispersions, pastes, gels, or doughs, so that they can be made
into the desirable geometries and structures (9). Solvent-based
strategies that can disperse and process CNTs without contaminating their functional surface or leaving residues would be very
useful for their applications. Some types of solvents have been
discovered that can produce relatively high-concentration dispersions of CNTs, such as super acids (10), ionic liquids (11), and
N-cyclohexyl-2-pyrrolidnone (12). However, most common solvents for nanotubes, such as N-methyl-2-pyrrolidone (NMP)
(13), dimethylformamide (DMF) (13), and 1,2-dichrolobenzene
(14), can only directly disperse some types of nanotubes at very
low concentrations [e.g., typically <0.02 wt% for single-walled
CNTs (SWCNTs)]. Here, we report that cresols, a group of industrial chemicals for a number of applications (15), including
for making household cleaning agents, are generic solvents for
unfunctionalized CNTs of various types. They can process CNTs
at concentrations up to tens of weight percent, resulting in a
continuous transition from dilute dispersions, thick pastes, and
free-standing gels to an unprecedented playdough-like state, as
the CNT loading increases. These states exhibit polymer-like
rheological and viscoelastic properties (16), which are not attainable with other common solvents, suggesting that the nanotubes are indeed disaggregated and finely dispersed in cresols.
Cresols can be removed after processing by heating or washing,
without altering the surface of CNTs. As demonstrated below,
the four nanotube/cresol states are highly processable and can be
readily used in a broad array of materials-processing techniques
to form desirable structures and composite materials.
www.pnas.org/cgi/doi/10.1073/pnas.1800298115
Results and Discussion
CNTs in m-Cresol. Earlier works in the field of conjugated polymers
have found that m-cresol is capable of dissolving some of the most
hard-to-process conducting polymers such as polyaniline, and it
interacts with the polymer chains through an effect called secondary doping (17–19). This inspired us to investigate the use of mcresol as the processing solvent for CNTs, which can be viewed as
highly conjugated polymers as well. Indeed, we found that powders
of both SWCNTs and multiwalled CNTs (MWCNTs) can be well
dispersed in m-cresol after sonication or grinding without the need
for any surface functionalization. As shown in the scanning electron microscopy (SEM) images (Fig. 1 A and D for MWCNTs and
SWCNTs, respectively), initially the nanotubes were heavily agglomerated and entangled in the powders, but they became well
separated after casting from the corresponding m-cresol dispersions (Fig. 1 B, C, E, and F for MWCNTs and SWCNTs, respectively). These results suggest that the interaction between mcresol and the surface of CNTs must be sufficiently strong to allow
the agglomerated nanotubes to disperse. Proton NMR (1H-NMR)
spectroscopy was used to probe the nature of such interaction. As
shown in Fig. 1G, in the presence of SWCNTs and MWCNTs, the
phenolic hydroxyl proton peak shifted upfield by 0.10 ppm, while
other proton peaks remained unchanged. This shift is a result of
increased electron density on the phenolic hydroxyl proton, indicating charge-transfer interaction with the nanotubes, as is found
for other Lewis acid type of solvents for CNTs (20, 21).
Sonicating or grinding CNTs in m-cresol does not induce
chemical changes to either the solvent or the nanotubes. This
is illustrated with SWCNTs due to their higher spectroscopic
Significance
Carbon nanotubes can now be produced in the ton scale in the
form of powders, but they need to be further processed, usually by solution-based routes, into disaggregated and more
usable forms for applications. There has been extensive effort
to search and design solvents that can disperse nanotubes,
which can also be easily removed afterward. Here, we report
that m-cresol and its liquid mixtures with other isomers, which
are already manufactured for other industrial purposes, are
such solvents. They can disperse carbon nanotube powders of
many types at unprecedentedly high concentrations, rendering
them polymer-like rheological and viscoelastic properties, and
high processability. This makes carbon nanotube powders immediately usable by current materials-processing techniques
for creating desirable structures or composites.
Author contributions: J.H. designed research; K.C., S.B., and J.H. performed research; S.B.
and J.K. contributed new reagents/analytic tools; K.C., S.B., and J.H. analyzed data; and
K.C. and J.H. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
Published under the PNAS license.
1
To whom correspondence should be addressed. Email: jiaxing-huang@northwestern.edu.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
1073/pnas.1800298115/-/DCSupplemental.
Published online May 14, 2018.
PNAS | May 29, 2018 | vol. 115 | no. 22 | 5703–5708
ENGINEERING
Cresols are a group of naturally occurring and massively produced
methylphenols with broad use in the chemical industry. Here, we
report that m-cresol and its liquid mixtures with other isomers are
surprisingly good solvents for processing carbon nanotubes. They
can disperse carbon nanotubes of various types at unprecedentedly high concentrations of tens of weight percent, without the
need for any dispersing agent or additive. Cresols interact with
carbon nanotubes by charge transfer through the phenolic hydroxyl proton and can be removed after processing by evaporation or washing, without altering the surface of carbon nanotubes.
Cresol solvents render carbon nanotubes polymer-like rheological
and viscoelastic properties and processability. As the concentration of nanotubes increases, a continuous transition of four states
can be observed, including dilute dispersion, thick paste, freestanding gel, and eventually a kneadable, playdough-like material.
As demonstrated with a few proofs of concept, cresols make powders of agglomerated carbon nanotubes immediately usable by a
broad array of material-processing techniques to create desirable
structures and form factors and make their polymer composites.
Fig. 1. (A–F) MWCNTs (A–C) and SWCNTs (D–F)
before and after ultrasonication in m-cresol. (A and
D) The nanotubes in their initial powder form were
highly entangled and agglomerated and became
well separated after being processed in m-cresol.
(B and C) SEM images of MWCNTs cast from m-cresol. (E and F) SEM (E) and AFM (F) images of SWCNTs
cast from m-cresol. The line scan in F shows that the
height of the nanotube is ∼1 nm, consistent with the
diameter of SWCNTs. (G) The 1H-NMR spectra show
hydroxyl proton of m-cresol shifted upfield in the
presence of either SWCNT (1, blue trace) or MWCNT
(2, red trace). The CNT samples for NMR were uniformly dispersed (G, Inset). (H) FTIR spectra showing
that m-cresol itself does not degrade during ultrasonication with or without SWCNTs. (I) No obvious
change in the Raman spectra of pristine SWCNTs and
those cast from m-cresol, suggesting that they were
not damaged during sonication. The cast SWCNTs
were dried and rinsed with water before taking
Raman spectra.
sensitivity to structural changes. The Fourier-transform infrared
(FTIR) spectra in Fig. 1H show that m-cresol itself does not degrade after ultrasonication with or without SWCNTs. As a relatively weak acid, m-cresol does not induce permanent chemical
changes to the nanotube surface and can be removed by evaporation or washing. The Raman spectra of the pristine SWCNTs
and a dried SWCNT film casted from m-cresol dispersion do not
show obvious difference (Fig. 1I), suggesting that they are not
damaged during processing. The absence of new bands between
400 and 1,000 cm−1, where m-cresol shows strong Raman signals
(22), indicates that they have been successfully removed.
Among the three isomers of cresols, m-cresol is a liquid at room
temperature; therefore, it was used for most of the experiments in
this work. While o- and p-cresol are solid at room temperature,
they can also process CNTs at molten state or when blended with
m-cresol at room temperature (SI Appendix, Fig. S1). This suggests that even the unrefined, crude grade of cresols, which is a
liquid mixture of the three isomers, can be directly used for industrial scale processing of CNTs. Indeed, UV-visible near-IR
(UV-Vis-NIR) spectra of SWCNTs dispersed in a ternary isomer
mixture of cresol showed characteristic bands of well-dispersed
nanotubes (SI Appendix, Fig. S2A), which was confirmed by
TEM studies (SI Appendix, Fig. S2B). Industrial grades of cresols
often contain phenolic impurities, and it was found that adding an
additional 10 wt% of phenol into the ternary mixture did not
negatively affect the stability of the nanotube dispersions (SI Appendix, Fig. S2A). The impurity tolerance and ease of removal
make cresols the ideal type of nonreactive solvents for the solution
processing of CNTs. Below, we demonstrate that cresol solvents
render CNTs polymer-like rheological and viscoelastic properties
and processability, making them immediately usable by already
available material-processing techniques to create desirable structures and form factors and make composites.
5704 | www.pnas.org/cgi/doi/10.1073/pnas.1800298115
Four States of MWCNTs in m-Cresol. m-cresol alone can disperse
and process CNTs up to tens of weight percent, which has been
unprecedented (SI Appendix, Table S1). Since MWCNTs are the
more common type of mass-produced CNTs and are much more
affordable and available, they were chosen as the model material
for most of the work below unless otherwise mentioned. The
photos in Fig. 2 A–E show the as-received MWCNT powders and
the corresponding dilute dispersion, paste, gel, and a playdoughlike state as the concentration in m-cresol increased. Dilute dispersions are typically made by sonication and can remain stable
for at least many months (e.g., a photo of a 1-y-old sample is
shown in SI Appendix, Fig. S3). The other higher-concentration
states are typically made by grinding. Transitions between the four
states are accompanied by threshold-like changes in their electrical, rheological, and viscoelastic properties. For example, the
transition from a dilute dispersion to a thick paste was accompanied by the onset of electrical conductivity ∼3 mg/mL (Fig. 2F),
which can be attributed to the formation of a percolated nanotube
network, establishing a continuous electrically conductive pathway
throughout the volume. At higher concentrations, increased density of the MWCNT network resulted in significant changes in
rheological and viscoelastic properties. For example, the transition
from a thick paste to a self-standing gel was marked by its inability
to free flow ∼40–50 mg/mL, after which its viscosity increased
significantly (Fig. 2G). This rheological transition was similar to
the observations in a previous study of extensively oxidized CNTs
in water, which can be attributed to continuous entanglement of
nanotubes (16). At concentrations >100 mg/mL, a viscoelastic,
kneadable playdough-like material was obtained, which was highly
cohesive and exhibited resistance to compression as characterized
by rapidly increased compression modulus (Fig. 2H).
The continuous transition between these four highly processable polymer solution-like states suggests that the nanotubes were
dispersed and outstretched in m-cresol, forming a cohesive network that densifies at increasing concentrations. If the nanotubes
Chiou et al.
Fig. 2. Four continuous states of MWCNTs in mcresol exhibiting polymer solution like rheological
and viscoelastic properties. (A–E) The nanotube
powders (A) can be processed in m-cresol to yield
dilute dispersion (B), thick paste (C), self-standing gel
(D), and finally kneadable dough (E). (F–H) The
transitions between these states are characterized
by a threshold-like increase in electrical conductivity
(F), viscosity (G), and compression modulus (H), due
to the formation and gradual densification of a 3D
network of dispersed nanotubes.
Dilute Dispersion and Langmuir-Blodgett Assembly. Both SWCNTs
and MWCNTs can disperse at higher concentrations in m-cresol
than in other common solvents such as NMP and DMF (SI
Appendix, Fig. S5). The m-cresol dispersion can be directly applied to Langmuir–Blodgett (LB) assembly for making monolayer thin films. Successful LB monolayer assembly requires
high-quality nanotube dispersions without other surface-active
materials to disrupt their packing on water surface, which is
challenging for additive-based CNT dispersions. Since m-cresol
can gradually dissolve in water, it dissipated into the subphase
after spreading the nanotubes on the water surface, leaving clean
nanotubes on the water surface (Fig. 3A). The water-supported
monolayers could be further densified by closing two barriers,
yielding a positive surface pressure (Fig. 3B), which could then
be transferred to a substrate by dip-coating (Fig. 3 B, Inset). Fig.
3C is a low-magnification SEM overview of a MWCNT film on
glass slide collected at a surface pressure of 30 mN/m, which
appears to be continuous, uniform, and cohesive. Since many of
the starting MWCNTs were curled, twisted, or even kinked (Fig.
1C) and could not lay flat, the near-monolayer thickness of the
film (Fig. 3D) also confirms that the heavily agglomerated
MWCNTs in the starting powders (Fig. 1A) indeed have been
well separated in m-cresol. Strong van der Waals attraction at
the tube–tube junctions contributed to the continuity and cohesiveness of the MWCNT monolayer.
Transferring the nanotube monolayer onto soft plastic substrates such as poly(ethylene terephthalate) formed a flexible
transparent conductor (Fig. 3E). Sheet resistance and optical
transparency of the nanotube coating could be fine-tuned further
by precisely controlling the number of deposited layers, as well as
the packing density within each monolayer. For example, a sheet
resistance of 90 kΩ/square was obtained at 72% of optical transparency, which is already comparable to the best values obtained
with films made from MWCNT powders (23). As shown in Fig. 1,
using m-cresol as a processing medium did not damage the surface
Chiou et al.
of nanotubes or leave hard-to-remove residues, which resulted in
satisfying conductivity of the LB films without the need for extensive further annealing steps. Similarly, LB assembly of SWCNT
monolayers has been achieved (SI Appendix, Fig. S6).
Thick Paste, Blade Coating, and Screen-Printing. Increasing the loading of MWCNTs up to 40 mg/mL resulted in a more viscous paste,
which exhibited relatively high viscosity and shear thinning behavior (Fig. 4A) with yield stress in the range of 1–10 Pa (Fig. 4B),
making it suitable to use by brushing or painting. To make a
continuous film by these techniques, the paste must be sufficiently
cohesive so that the coating does not break up under the shear
during spreading or crack by the capillary action during drying.
Therefore, the nanotubes need to be interconnected throughout
the paste without extensively segregated domains (see SI Appendix,
Fig. S4 and related discussion). Fig. 4C illustrates blade coating of
ENGINEERING
were still agglomerated as in their powders, the corresponding
high-concentration products would not be cohesive due to segregated domains, resulting in poor processability (see schematic
illustrations in SI Appendix, Fig. S4 and related discussion in the
legend and below). These four states have been observed for all of
the CNTs tested (e.g., unfunctionalized single-walled or multiwalled tubes of various sizes). As demonstrated by the examples
below, m-cresol indeed offers unprecedented versatility for processing CNTs for existing and new applications.
Fig. 3. Monolayers of MWCNTs from dilute dispersion by LB assembly.
(A) After spreading, a uniform semitransparent film is observed over the
entire area of the trough. (B) Isothermal compression of the monolayer increases its surface pressure, indicative of higher nanotube density. B, Inset
shows a dip-coated film on glass. (C and D) SEM image showing the film is a
continuous, uniform, paper-like monolayer (C) made of a network of
nanotubes (D). (E) Sheet resistances and the corresponding transparencies of
MWCNT layers on PET substrate made by repetitive dip-coating.
PNAS | May 29, 2018 | vol. 115 | no. 22 | 5705
greatly shortens the baking time needed for solvent removal. Fig. 5
shows a proof-of-concept experiment, where PMMA powders were
directly mixed with the paste by mortar and pestle (also see SI
Appendix, Fig. S8). The product was rolled into a flexible and highly
plastic sheet, which sustained >800% of tensile strain. Upon thermal curing at 150 °C, the sheet hardened (Fig. 5B) due to partial
removal of m-cresol. At 1 wt% loading of MWCNTs in PMMA,
the Young’s modulus of the composite (1.46 GPa) increased by
24% in comparison with a similarly processed PMMA sheet
(1.17 GPa). SEM observation confirmed that the MWCNTs had
been finely dispersed in the PMMA matrix (Fig. 5C). Such soft–
hard transition is critical for industrial forming techniques, which
turn materials into desirable geometries and form factors. The
additive-free CNT pastes in cresols could be useful for accelerating
the development and manufacturing of polymer nanocomposites.
Fig. 4. Blade coating of MWCNTs from the thick paste. (A) The MWCNT
paste exhibits shear-thinning behavior at all of the concentrations tested,
which is typical for polymer dissolved in good solvents. From II to VI, the
concentrations are increased from 10, 20, 30, 40, to 50 mg/mL. Pure mcresol (I) does not exhibit this behavior. (B) The yield stress of the paste
increases relatively slowly as the nanotube concentration increases, until it
reaches the range of the gel state (V). (C–E ) Blade-coating (C) creates a
continuous and uniform nanotube film on glass after drying (D), which is
free of cracks (E, SEM image) that are typically seen for coatings made with
other solvents, such as NMP (SI Appendix, Fig. S7). (F ) Patterns of interdigitated electrodes screen-printed on paper.
the paste. The oven-dried coating on glass is continuous and free
of cracks over the entire area (Fig. 4D). SEM images show that it is
made of an interwoven, continuous, and high-density network of
nanotubes (Fig. 4E). As a comparison, a coating casted with NMP
at the same concentration resulted in discontinuous islands (see
SEM image in SI Appendix, Fig. S7). Segregated MWCNTs in
NMP resulted in an incohesive paste that could not maintain a
continuous layer after blading and during drying. Similar to blade
coating, industrial screen-printing can directly use the MWCNT
paste to generate functional patterns. A proof-of-concept demonstration of interdigitated electrode patterns printed on paper is
shown in Fig. 4F. Blade coating is commonly used to make electrodes for energy storage devices from slurries, which often use
CNTs as conductive binder for active materials (8). Highly cohesive, additive-free pastes with well-dispersed nanotubes are
readily compatible with these slurry-processing techniques and
could directly benefit this large-scale application of CNTs.
MWCNT Pastes for Polymer Composites. Polymer nanocomposite is
another area that uses a very large scale of CNTs (3, 7). The paste
state offers a number of potential advantageous for manufacturing.
To start, the paste can be easily mixed with powders of polymers,
which is one of the most common forms of industrial polymers.
Moreover, m-cresol itself is a known solvent for many commodity
polymers such as poly(methyl methacrylate) (PMMA), nylons,
polyethylene terephthalate, polystyrene, and phenolic resins (24),
which helps the blending process. Using the paste also drastically
reduces the amount of solvent needed for manufacturing and
5706 | www.pnas.org/cgi/doi/10.1073/pnas.1800298115
Gel and 3D Printing. Above 40 mg/mL, the MWCNT network in
m-cresol was sufficiently dense to hinder free flow, leading to a
freestanding gel. As the nanotube concentration increased, the gel
became more solid-like with increased storage modulus (Fig. 6A).
The loss modulus increased more slowly than the storage modulus,
rendering the gel a sufficient level of liquid character for extrusion
type of processing (Fig. 6B). Therefore, the MWCNT gel could
deform and reconnect easily. Fig. 6C shows a MWCNT gel being
extruded to form self-supporting fibers through a 0.5-mmdiameter needle. Since the gel is cohesive, extrusion can be continuously operated even with finer needles (e.g., 0.1-mm diameter). In contrast, fiber extrusion cannot be performed with
other solvents such as NMP in similar range of concentrations.
Instead, jetting (for 0.5-mm needle) and clogging (for 0.1-mm
needle) occurred due to jamming of the nozzles by blobs of segregated nanotubes (SI Appendix, Fig. S4 and related discussion).
This again reflects that the nanotubes were uniformly dispersed by
m-cresol and outstretched like polymers in the gel, rendering it
suitable rheological properties for continuous, unhindered extrusion. This gel is immediately usable for programmed and automated printing (Fig. 6D, inner diameter of 0.1 mm). As a proof of
concept, a cup-shaped structure was 3D-printed from the gel (Fig.
6E). The base of the cup was made of two criss-cross layers of
close-packed fibers, and the side was made of vertically stacked
rings. After drying, the cup structure shrank slightly isotropically
Fig. 5. MWCNTs/polymer nanocomposite by direct mixing using the
paste. (A) Photos showing MWCNT/PMMA composite sheet made by direct mixing of polymer powders with the paste, followed by cold rolling
(also see SI Appendix, Fig. S8). (B) The uncured composite sheet (1.0 wt%
MWCNTs) is highly ductile and hardens upon curing at 150 °C for 2 h. (C )
SEM image shows well-dispersed MWCNTs embedded within the nanocomposite.
Chiou et al.
them polymer-like processability. Cresols work generically for
unfunctionalized CNTs of many types and can be conveniently
removed from the final products without negatively altering their
pristine properties. Cresols are abundantly produced, relatively
inexpensive, and quite stable to handle at room temperature and
ambient atmosphere. These advantages make cresols an ideal
class of processing solvents for CNTs, especially for their massproduced powder form. It should help to overcome many aspects of the processability problems of CNTs, which has been
one of the greatest hurdles preventing their widespread industrial applications. The surprise that solvents with such
simple molecular structures work so well is also likely to inspire
many more discoveries about the interactions between organic
molecules and graphitic surface, as well as in new material and
engineering technologies based on CNTs and other graphitic
nanostructures.
Materials and Methods
Fig. 6. Extrusion and 3D printing using the MWNCTs gel. (A and B) The gel
shows increasingly solid-like behavior as nanotube concentration increases,
based on the results of storage moduli (A) and loss moduli (B) measurements. (C and D) MWCNT gel can be continuously extruded from a needle
(inner diameter of 0.5 mm; C), which allows patterning of nanotubes using a
programmable stage (inner diameter of 0.1 mm; D). (E and F) A 3D printed
cup made of MWCNTs (E), which maintains its shape after drying (F).
Materials. CNT powders of various types, sources, and levels of purities from
three vendors were tested, and all dispersed well in m-cresol and its liquid
mixtures with other isomers. These included: (i) CoMoCAT MWCNTs (98%
carbon content), CoMoCat SWCNTs (90% carbon content, 90% semiconducting), and double-walled CNTs [90% carbon content, made by
chemical vapor deposition (CVD)] were obtained from Sigma-Aldrich; (ii)
SWCNTs (P2, 90% purity) and carboxylic functionalized SWCNTs (P3, 90%
purity) were made by arc-discharge and obtained from Carbon Solution Inc.;
and (iii) graphitized MWCNTs (TNGM2; 99.9% purity, approximate lengths
of 50 μm), low-density SWCNTs (TNSR; 95% purity, approximate lengths of
5–30 μm, 0.027 g/cm3), high-density SWCNTs (TNST; 95% purity, 0.14 g/cm3),
short SWCNTs (TNSSR; 95% purity, approximate lengths of 1–3 μm), and
short MWCNTs (TNSM2; 95% purity, approximate lengths of 0.5–2 μm) were
all made by CVD and obtained from TimesNano.
P2 SWCNTs and MWCNTs (CoMoCat) were used for demonstrating LB
assembly (Fig. 2 and SI Appendix, Fig. S6). The results of the pastes, gels, and
MWCNT Dough. The last state of MWCNT/m-cresol is a viscoelastic
dough (>100 mg/mL), which can be kneaded or rolled without
fracture. In contrast to a gel (Fig. 7B), when kneaded on paper,
the dough did not leave any stain mark (Fig. 7A and SI Appendix,
Fig. S9). This is due to the strong attraction between the nanotubes in the densely woven 3D network, which prevents them from
leaving residues on paper. Control experiments were also done
with other solvents, such as NMP, at similar concentrations.
However, the resulting mixtures were too fragile to manipulate
and broke into pieces upon kneading. Since the nanotube/m-cresol
dough was kneadable and stain-free (Fig. 7A), it must be highly
cohesive and free of mechanically weak boundaries between segregated grains of CNTs (see SI Appendix, Fig. S4 and related
discussion), as seen in the starting powders (SEM images in Fig. 1
A and D). As with a bread dough, the MWCNT dough could be
cut into pieces and rejoined when pressed together or molded into
arbitrary shapes without altering its viscoelastic properties. Fig. 7C
shows a thick film cold-rolled from the dough, which was still soft
and plastic (Fig. 7D) and could be reshaped by using a mold (Fig.
7E). The MWCNT doughs could be hardened to fix their shapes
after heating at >200 °C to remove m-cresol. The hardened
structures could then be returned to the soft dough state by absorbing m-cresol. The playdough-like processability should open
up opportunities to fabricate arbitrarily shaped 3D solids of neat
CNTs for a range of electronic, thermal, and energy applications.
Conclusion
Cresol-based CNT dispersions, pastes, gels, and doughs exhibit
polymer-like rheological and viscoelastic properties, rendering
Chiou et al.
ENGINEERING
but maintained its shape, resulting in a stiff solid object that could
be further handled (Fig. 6F).
Fig. 7. Playdough-like MWCNTs/m-cresol solid. (A and B) Kneading a
nanotube dough (>100 mg/mL) only leaves traces of solvent on paper (A),
while a stiff nanotube gel (<100 mg/mL) leaves extensive stains of nanotubes
(B; also see SI Appendix, Fig. S9). (C–E) The dough can be transformed into
arbitrary geometries, such as a freestanding strip by cold rolling (C and D)
and other arbitrary shapes defined by a mold (E).
PNAS | May 29, 2018 | vol. 115 | no. 22 | 5707
doughs shown in the work were demonstrated with CoMoCat MWCNTs as
the model material, although other types of MWCNTs work as well.
Other chemicals were purchased from Sigma-Aldrich and used as received,
including m-cresol (99%), o-cresol (99%), p-cresol (98%), toluene (99.9%), phenol
(>99%), DMF (99.8%), NMP (anhydrous, 99.5%), PMMA [200,000 molecular
weight (Mw)], and methyltrichlorosilane (99%). Ternary isomer mixture of cresol
(>99% wt, 1:1:1 ratio) was purchased from Fisher Scientific and used as received.
LB Assembly and Transparent Conductive Thin Films. Powders of MWCNTs or
SWCNTs were first mixed with m-cresol by using a mortar and pestle, then
sonicated in pulse mode (2 s on/2 s off cycles for a total of 1 h) by using a
Qsonica Q125 sonicator rated at 125 W, equipped with a 1/4-inch standard
tapered tip at 90% power. After sonication, the dispersion was subject to
exhaustive high-speed centrifugation at 11,000 rpm for 1 h by using an
Eppendorf 5804 desktop centrifuge (equivalent to a relative centrifugal
force of 15,557 × g). The supernatant was recovered and used. Samples for
making transparent conductors were first purified by a nonoxidative route,
including washing in 3 M HCl at 65 °C for 4 h, followed by baking in a
muffled furnace at 250 °C for 1 h.
All parts of the LB system (Nima Technology) were thoroughly cleaned
with acetone before use. By using a glass syringe, 1 mL of m-cresol dispersion
(SWCNT or MWCNT) was carefully spread onto the air–water interface. A
tensiometer with a Wilhelmy plate was used to monitor surface pressure
while closing the barriers. At surface pressures of ∼40 mN/m for SWCNTs and
30 mN/m for MWCNTs, monolayer films were dip-coated onto a substrate
(typically glass slides) with a pull speed of 2 mm/min. The obtained LB films
were annealed at 150 °C for 30 min before subsequent LB deposition to
produce multilayered films.
Blade-Coating and Screen Printing. MWCNT paste in m-cresol (100 mg/mL) was
made by direct mixing using a mortar and pestle, then diluted to 40 mg/mL
and hand-ground further to yield a spreadable thick paste. Glass slides were
first silanized with 5 wt% methyltrichlorosilane in toluene for 10 min and then
washed thoroughly by using toluene followed by acetone. Two strips of
Kapton tapes were attached to the sides of the silanized glass slide as spacers
to control the thickness of the coating. Approximately 0.3 mL of MWCNT paste
was deposited onto the shallow trough created by the Kapton tapes. A razor
blade was used to drag the paste to coat the slide. The coating was left to dry
at 150 °C for 2 h. Control experiments were done by using NMP instead of
m-cresol as the solvent at the same nanotube concentration. Screen-printing
was done on paper through a mask by using a paste of 10 mg/mL.
Polymer Composite. To make MWCNT/PMMA nanocomposite, a MWCNTs/mcresol paste (40 mg/mL) was ground directly with powders of PMMA
(200,000 Mw) by using a mortar and pestle for 10 min. The composite was
then flattened by cold rolling, which turned flexible and rubbery after being
air-dried (Fig. 5 and SI Appendix, Fig. S6). Curing at 150 °C for 2 h significantly hardened the piece and fixed its shape.
into a syringe and manually extruded from needles with diameters of
0.1 and 0.5 mm, which can be fitted onto a 3D printer (Hyrel 30M). Printed
3D structure can be removed from the glass substrate after being air-dried
for 12 h, which can be further hardened by baking to remove m-cresol.
MWCNT Dough. MWCNT/m-cresol dough was made by directly mixing using a
mortar and pestle at a concentration of 300 mg/mL or higher. The mixture was
then diluted to 150 mg/mL and ground further to yield a dough-like material,
which was kneaded to the shape of a ball. Kneading or rolling a nanotube
dough does not stain the substrate, while doing so with a gel or paste would
result in significant staining. A kneaded dough was sandwiched between two
stainless steel foils and cold rolled to a film with final thickness of 200 μm,
which can be cut into various shapes with a razor blade or cookie cutters.
Characterization. Dispersions of carbon materials in m-cresol were drop-casted
onto silicon wafers and dried at 200–250 °C, before SEM (FEI Nova 600 system)
and AFM (Park Systems XE-100, tapping mode). UV/vis spectra were taken with
an Agilent 8453 UV/Vis spectrometer. NIR spectra were taken by using a PerkinElmer LAMBDA 1050 spectrometer. TEM images were taken with a JEOL
ARM300F GrandARM transmission electron microscope. Drop-cast SWCNTs
were air-dried and rinsed with water and ethanol before Raman spectroscopy
measurement (WITec Alpha 300; 532-nm excitation). FTIR spectra were recorded on a PerkinElmer Instrument spectrometer (Spectrum Spotlight 300).
The 1H-NMR spectra were acquired on a 400-MHz Agilent DD MR-400 NMR
system. The samples were prepared by adding 100 μL of SWCNT or MWCNT
dispersions in m-cresol in 1 mL of CDCl3. The nanotubes were found to be
stably dispersed in the entire duration of NMR experiments. Transparency of
the LB films was measured by using an Agilent 8453 UV/Vis spectrometer.
Sheet resistance of the films was obtained by using an in-line four-point probe
equipped with a Keithley 2400 source meter. Viscoelastic and rheological
properties were measured by using an Anton Paar MCR 502 rheometer using a
cone-on-plate configuration. The cone has a 25-mm diameter with a 5° gap
angle. Viscosity vs. concentration measurements in Fig. 2G were measured
with a rotation speed of 1°/s. Yield stresses were obtained by using a Herschel–
Bulkley regression included in the Anton Paar software package. Shear-thinning viscosities of Fig. 4A were measured with a linear ramping shear rate
between 0.01 and 100 rad/s. Storage and loss moduli were measured simultaneously by using the same rheometer setup at an amplitude of 1%. Tensile
and compression tests were done on a Bose electroforces 5500 tester. The
composite films were cut into dog-bone shapes and pulled at a rate of
0.05 mm/s until failure. Only the results from samples that failed in the middle
were considered. Gel and dough samples for compression tests were first
molded into cylindrical shapes and carefully transferred to the tester. Compression was done at 0.005 mm/s until the sample ruptured. The slope of the first
linear region of the stress-strain curve was taken as the compression modulus.
Three-Dimensional Printing. MWCNTs/m-cresol gel was made by direct mixing
using a mortar and pestle at a concentration of 120 mg/mL The resulting
mixture was diluted to 80 mg/mL and ground further. The gel was loaded
ACKNOWLEDGMENTS. We thank Dr. J. Luo for providing some CNT samples.
K.C. is a NSF Graduate Research Fellow. S.B. acknowledges a visiting student
fellowship supported by Basic Science Research Program through the National
Research Foundation of Korea funded by the Ministry of Education, Science,
and Technology. J.H. was supported by an earlier Guggenheim Fellowship,
part of which was applied to purchase some materials.
1. Iijima S (1991) Helical microtubules of graphitic carbon. Nature 354:56–58.
2. Schnorr JM, Swager TM (2011) Emerging applications of carbon nanotubes. Chem
Mater 23:646–657.
3. De Volder MFL, Tawfick SH, Baughman RH, Hart AJ (2013) Carbon nanotubes: Present
and future commercial applications. Science 339:535–539.
4. Liu C, Cheng HM (2013) Carbon nanotubes: Controlled growth and application. Mater
Today 16:19–28.
5. Lin GD, et al. (2015) Multi-walled carbon nanotubes as novel promoter of catalysts for
certain hydrogenation and dehydrogenation reactions. Sci China Chem 58:47–59.
6. Endo M, Hayashi T, Kim YA (2006) Large-scale production of carbon nanotubes and
their applications. Pure Appl Chem 78:1703–1713.
7. Huang JQ, Zhang Q, Zhao MQ, Wei F (2012) A review of the large-scale production of
carbon nanotubes: The practice of nanoscale process engineering. Chin Sci Bull 57:157–166.
8. Li Y (2017) The quarter-century anniversary of carbon nanotube research. ACS Nano 11:1–2.
9. Fayed ME, Otten L (1997) Handbook of Powder Science & Technology (Springer,
Boston), 2nd Ed, p 1465.
10. Davis VA, et al. (2009) True solutions of single-walled carbon nanotubes for assembly
into macroscopic materials. Nat Nanotechnol 4:830–834.
11. Fukushima T, et al. (2003) Molecular ordering of organic molten salts triggered by
single-walled carbon nanotubes. Science 300:2072–2074.
12. Bergin SD, et al. (2009) Multicomponent solubility parameters for single-walled carbon nanotube-solvent mixtures. ACS Nano 3:2340–2350.
13. Furtado CA, et al. (2004) Debundling and dissolution of single-walled carbon nanotubes in amide solvents. J Am Chem Soc 126:6095–6105.
14. Bahr JL, Mickelson ET, Bronikowski MJ, Smalley RE, Tour JM (2001) Dissolution of small diameter single-wall carbon nanotubes in organic solvents? Chem Commun (Camb) 193–194.
15. Mukhopadhyay AK (2005) Industrial Chemical Cresols and Downstream Derivatives
(Marcel Dekker, New York), p xix, p 202.
16. Shaffer MSP, Windle AH (1999) Analogies between polymer solutions and carbon
nanotube dispersions. Macromolecules 32:6864–6866.
17. Macdiarmid AG, Epstein AJ (1994) The concept of secondary doping as applied to
polyaniline. Synth Met 65:103–116.
18. Cao Y, Smith P, Heeger AJ (1992) Counterion induced processibility of conducting
polyaniline and of conducting polyblends of polyaniline in bulk polymers. Synth Met
48:91–97.
19. Cao Y, Smith P (1993) Liquid-crystalline solutions of electrically conducting polyaniline. Polymer 34:3139–3143.
20. Ramesh S, et al. (2004) Dissolution of pristine single walled carbon nanotubes in superacids by direct protonation. J Phys Chem B 108:8794–8798.
21. Engtrakul C, et al. (2005) Protonation of carbon single-walled nanotubes studied
using 13C and 1H-13C cross polarization nuclear magnetic resonance and Raman
spectroscopies. J Am Chem Soc 127:17548–17555.
22. Yu HY, Yan P (2012) Determination of m-cresol and p-cresol in industrial cresols by
Raman spectrometer. Adv Mat Res 468–471:1104–1109.
23. Yu L, Shearer C, Shapter J (2016) Recent development of carbon nanotube transparent conductive films. Chem Rev 116:13413–13453.
24. Mark JE (2009) Polymer Data Handbook (Oxford Univ Press, Oxford), 2nd Ed, pp vii, 1250.
5708 | www.pnas.org/cgi/doi/10.1073/pnas.1800298115
Chiou et al.