Journal of Colloid and Interface Science 260 (2003) 89–94

www.elsevier.com/locate/jcis

Production of aqueous colloidal dispersions of carbon nanotubes

Linqin Jiang, Lian Gao,∗ and Jing Sun
State Key Laboratory of High Performance Ceramics and Superfine Microstructure, Shanghai Institute of Ceramics, Chinese Academy of Sciences,
1295 Dingxi Road, Shanghai 200050, People’s Republic of China
Received 6 August 2002; accepted 8 November 2002

Abstract
Stable homogeneous dispersions of carbon nanotubes (CNTs) have been prepared by using sodium dodecyl sulfate (SDS) as dispersing
agent. To our knowledge, it is the first report to quantitatively characterize colloidal stability of the dispersions by UV–vis spectrophometric
measurements. When the sediment time reaches 500 h, the supernatant CNT concentration drops as much as 50% for the bare CNT
suspension, compared to 15% with the addition of SDS. Furthermore, after 150 h, no precipitation is found for CNT/SDS dispersions,
exhibiting an extreme stability. Zeta potential, auger electron microscopy, and FTIR analysis are employed to investigate the adsorption
mechanism in detail. It has been concluded that the surfactant containing a single straight-chain hydrophobic segment and a terminal
hydrophilic segment can modify the CNTs–suspending medium interface and prevent aggregation over long periods. The morphology of the
CNT dispersions is observed with optical microscopy. An intermediate domain of homogeneously dispersed nanotubes exhibits an optimum
at 0.5 wt% CNTs and 2.0 wt% SDS.
 2003 Elsevier Science (USA). All rights reserved.

Keywords: Carbon nanotubes; SDS; Dispersion; Surface chemistry; FTIR; Optical microscopy

1. Introduction The possibility of producing an aqueous C60 suspension

The discovery of carbon nanotubes (CNTs) and carbon-
nanotube-based materials has inspired scientists for a range
of potential applications [1–4]. The carbon nanotubes have
high mechanical strength (strength and flexibility), making
them ideal reinforcing fibers in nanocomposites. Therefore,
it is generally expected that CNTs can be used as additives to
reinforce composite materials, such as epoxy [5], petroleum
pitch [6], PMMA [7], and alumina composites [8]. However,
fabrication of homogeneous nanocomposites with carbon
nanotubes remains a technical challenge [9]. When added
to an alumina composite by the hot press and hot extrusion
method [8], the mechanical strength of the alumina com-
posite is not enhanced, a fact attributed to aggregation (or
inhomogeneous dispersion) of CNTs in the composite mate-
rials. CNTs always form aggregates owing to very strong van
der Waals interactions. To obtain stable dispersions of CNTs
in water indeed is a significant problem, and it is also a pre-
requisite for its application as additives for reinforcement of
composite materials.
* Corresponding author.
E-mail address: liangaoc@online.sh.cn (L. Gao).
0021-9797/03/$ – see front matter  2003 Elsevier Science (USA). All rights reserved.
doi:10.1016/S0021-9797(02)00176-5
has been reported [10], and suspensions have been obtained
by means of chemical modification, by using colloidal sur-
factants, phospholipids, or polyvinyl pyrrolidine (PVP) as
solubilizing or/and dispersing agents [11–13], or with the
help of supramolecular complexes “host-guest” formation
with γ -cyclodextrin [11–14]. Dispersing agents, such as
surface-active agents, have been used to disperse fine parti-
cles of hydrophobic materials in aqueous solution [15,16].
In general, there are three principles for dispersing fine
particles in water [17]: (i) the repulsion between the par-
ticles with their zeta potentials, (ii) the steric hindrance of
the adsorption layer, and (iii) the reduction of hydrophobic
linkages among dispersed particles. Commercially available
original CNTs are generally too long to be dispersed in so-
lution and have no well-defined chemical function groups
to modify [18]. According to previous reports [19], surfac-
tants can modify the particles–suspending medium interface
and prevent aggregation over long time periods. However,
how surfactant molecules help dispersion of CNTs is still
not clear [20–22]. It is essential to have a more complete un-
derstanding of the dispersion mechanism.
In this paper, an attempt to prepare aqueous dispersions
of CNTs in the presence of sodium dodecyl sulfate (SDS) is
90 L. Jiang et al. / Journal of Colloid and Interface Science 260 (2003) 89–94

made, and this resulted in the generation of stable homoge-

neous dispersions. UV–vis spectrophometric measurements
are used to quantitatively characterize colloidal stability of
the dispersions as a new method. The stabilization mecha-
nisms of the CNT dispersions are discussed by zeta poten-
tial, FTIR, and auger electron spectroscopy (AES) studies.
It is expected that the results can help provide a guide-
line for choosing suitable dispersants. Optical microscopy
is employed to characterize the aggregate state of the CNT
suspension on a micrometer scale. A phase diagram of the
CNT/SDS aqueous system allows the optimum amount for
obtaining homogeneous dispersion to be deduced. Fig. 1. TEM micrograph of raw CNTs.

2. Materials and methods

2.1. Materials

CNTs were kindly provided by Shenzhen Nanotech Port

Ltd. Co. (Shenzhen, China) and were used as received.
According to thermogravimetric analysis (STA 449C, NET-
ZSCH, Selb, Germany), CNTs contained about 45 wt%
water, after being purified by the standard acid treatments.
A TEM micrograph (Model 200CX, JEOL, Tokyo, Japan)
of raw CNTs is shown in Fig. 1. Distilled water was used
Fig. 2. UV–vis spectrum of CNTs in aqueous solution.
in all studies. The concentrations of the CNTs (wt%) in this
paper are given as the weight concentration of the pristine
CNTs relative to water.

2.2. Techniques

2.2.1. Stability evaluation

Usually the stability of the ceramic suspension is de-
termined by measuring the sediment volume versus the
sediment time. However, this method was not suitable for
the CNT dispersion. The dispersion was too dark to discern
the sediment volume clearly. In this paper, the stability of
the 0.2 wt% suspension was determined by measuring the
supernatant concentration of 50-ml suspensions versus the
sediment time. In aqueous solution, the absorption of CNTs
appeared at 253 nm (Fig. 2). With increasing sediment time,
the absorbance of CNTs in the supernatant aqueous solution
was diminished at 253 nm. This phenomenon was used as a
basis for the determination of CNT concentration. A linear
calibration curve was constructed at a wavelength of 253 nm
for CNT dispersions with no additives in the range of 0.01–
0.02 wt% (Fig. 3). For SDS-adsorbed CNT dispersions, the
same concentration of SDS aqueous solution was used as
reference to eliminate the absorbance of SDS in the sus-
pensions. Thus, a standard curve was determined similarly.
These solutions were examined with a UV–vis scanning
spectrophotometer (UV-3101PC, SHIMADZU, Japan). The
concentration of the supernatant was determined by fitting
the absorbance to the calibration curve.
Fig. 3. The calibration curve of CNTs in aqueous solution without SDS at
λ = 253 nm.
2.2.2. Zeta potential measurements
The zeta potential of the dilute suspensions (0.01 wt%)
was measured with a Zetaplus analyzer (Zetaplus, Brook-
haven, NY, USA). Each sample was ultrasonicated for 1 h
prior to analysis. The ionic strength was maintained at
10−3 M using NaCl.
2.2.3. FTIR and AES spectroscopy
In a typical experiment, 0.02 g of CNTs were added to
0.1 g dm−3 SDS aqueous solution, followed by ultrason-
ication to disperse the CNTs. The value of pH had been
previously adjusted to 9 using analytical grade NaOH. The
suspensions were dried at 140 ◦ C for 24 h and then used
 L. Jiang et al. / Journal of Colloid and Interface Science 260 (2003) 89–94
for infrared spectroscopy analyses. Infrared measurements
were made on an FTIR spectrophotometer (Bio-Rad FTS-
185, Hercules, CA). AES measurements were done on a
scanning auger microscope (Microlab-310F, VG Scientific
Ltd., England). The reported AES spectra in this paper were
the averages of those collected in at least five different parts
of the same sample.
2.2.4. Optical microscopy measurements
Suspensions containing a known weight fraction of CNTs
were prepared by dispersing CNTs in the aqueous solution
without and with SDS, respectively. After ultrasonication for
1 h, a drop of the suspension was carefully sucked up with
a burette and dipped on a slide to settle. The morphology
of the suspension was examined with a polarization optical
microscope (AMPLIVAL Pol. D, Carl Zeiss, Tena, German).
3. Results
3.1. Colloidal stability of the CNT dispersions
A new method introduced in this paper can be used to
calculate quantitatively and accurately the suspension con-
centration with increasing sediment time. Figure 4 illustrates
the CNT concentration of the supernatant suspension ver-
sus the sediment time without and with SDS, respectively.
For the bare CNT suspension, the fastest settling occurs. At
500 h, the supernatant CNT concentration drops as much
as 50%. Additionally, without the dispersant, the CNT con-
centration still reveals a decreasing tendency after 500 h.
In contrast, less precipitation is observed in the presence of
SDS. The supernatant CNT concentration only falls 15% in
this system at 500 h. As the sediment time exceeds 150 h,
no precipitation is found for CNT/SDS dispersions at room
temperature. Thus, it can be concluded that the CNT disper-
sions prepared with SDS are very stable.
Figure 4 also shows that the initial concentration of
CNT dispersions calculated by UV–vis is 0.2 wt%, which is
consistent with the concentration of the suspension prepared
in the experimental section. That is, the carbon nanotubes
Fig. 4. CNT concentrations vs time for 0.2 wt% CNT suspensions
(a) without SDS and (b) with 0.1 wt% SDS at pH 9.
91
are well dispersed by ultrasonication. It has been reported
[18,19] that ultrasonication is an external mechanical energy
that helps the particles to overcome the attractive van der
Waals force at contact.
3.2. Characterization of the CNT dispersions
Zeta potentials of the CNT dispersions as a function of
pH without and with SDS are given in Fig. 5. The ζ values
are calculated from the particle velocities by means of the
Helmholtz–Smoluchowski equation [23],
ζ = 4πµη/D,
where µ is the electrophoretic mobility and η and D are
respectively the viscosity and the dielectric constant of the
liquid in the boundary layer. It is evident that the CNTs are
negatively charged, which is consistent with the previously
reported result [22]. In addition, adsorption of SDS on CNTs
reflects the change in the zeta potential below pH 8, whereas
in alkaline solutions SDS adsorption is not significant
because the zeta potentials of CNTs with or without SDS are
almost the same except around pH 11. This means that CNTs
purified by acid treatment contain some acidic sites which
can be dissociated in the alkaline solutions. With the addition
of the SDS, the maximum ζ potential is as low as −40 mV,
in contrast to −30 mV for the bare CNT suspension.
A decrease of the measured zeta potential is caused by SDS
for all pH values studied, which means that SDS can adsorb
onto the CNTs throughout the entire experimental range.
The effect of SDS is attributed to the electrostatic repulsion
provided by adsorbed surfactants which stabilize the CNTs
against van der Waals attraction [24].
Figure 6 illustrates AES spectra of the CNTs at pH 9
without and with SDS coatings. The blank CNTs exhibit a
strong peak of C at 216 eV. Simultaneously, a very weak
peak of O at 464 eV is observed probably due to nanotube
defects [25]. It has been reported that the relevant functional
groups already exist in the original carbon nanotube ma-
terial, possibly originating from oxidation of atomic-scale
defects, created during CNT growth [26], upon air exposure.
Fig. 5. Zeta potential of CNT suspension vs pH (a) without SDS and (b) with
0.005 wt% SDS.
92 L. Jiang et al. / Journal of Colloid and Interface Science 260 (2003) 89–94
Fig. 6. AES spectra of CNTs (a) without SDS and (b) with SDS coatings at
pH 9.
Fig. 7. FTIR spectra of CNTs (a) without SDS and (b) with SDS at pH 9.
After SDS adsorption, the S peak at 98 eV and the Na peak at
942 eV appear. Moreover, the C peak at 216 eV is weakened
and the O peak is enhanced. This is considered as evidence
for the strong adsorption of SDS at the surface of the CNTs.
To investigate the adsorption mechanism of SDS on
CNTs, FTIR spectra are used for the first time. Curve a
in Fig. 7 is the FTIR spectrum of the pure SDS at pH 9.
Curve b is that obtained by subtracting the CNT contribu-
tion from the spectrum of the SDS-adsorbed CNTs at the
same pH. The strong bands at 2920 cm−1 and 2852 cm−1 of
curve a are due to asymmetrical and symmetrical stretching
of –CH2 –, respectively. The band at 2958 cm−1 is assigned
to the asymmetrical stretching of –CH3 . The symmetrical
and asymmetrical bending vibrations of –CH3 are repre-
sented at 1379 cm−1 and 1469 cm−1 [27]. From curve b, the
asymmetrical and symmetrical stretching of –CH2 – remains
constant. The relative intensity of the asymmetrical stretch-
ing of –CH3 at 2957 cm−1 decreases significantly, and the
bands assigned to the bending vibration of –CH3 are elimi-
nated, indicating the interfacial interaction of –CH3 with the
nanotubes.
3.3. Morphology of the CNT dispersions
Images taken with optical microscopy are shown in
Figs. 8a–8d. Unlike electron microscopy, optical microscopy
is unable to resolve the tubes, but nevertheless it enables the
Fig. 8. (a)–(d) Optical micrographs taken for the same CNT concentrations
and different amounts of SDS. (e) Phase diagram of the CNT/SDS system:
" represents image b; 1 represents image c; Q represents image d.
aggregate state of the suspension to be directly characterized
on a micrometer scale [19,24]. Figure 8a shows the image of
bare CNT suspensions. Without SDS, large and dense clus-
ters of the initial material are still found after sonication. The
CNT suspensions stabilized with SDS, however, show a dra-
matic change in their aggregation state (Figs. 8b–8d). From
Figs. 8b and 8d, a few carbon aggregates can be imaged by
optical microscopy, meaning that most of the materials have
been dispersed, but the suspensions are still heterogeneous.
Varying the SDS and CNT concentrations, the homogeneous
and single-phase suspensions are obtained (Fig. 8c). So, the
phase diagram is obtained by varying the CNT/SDS pro-
portions (Fig. 8e). In this phase diagram, a region resulting
from homogeneous dispersions is obtained. An intermedi-
ate domain of homogeneously dispersed nanotubes exhibits
an optimum at about 0.5 wt% CNTs and 2.0 wt% SDS. Any
amount lower or higher than this region will lead to large ag-
gregation. Insufficient surfactant cannot produce an efficient
coating that will induce electrostatic repulsion and counter-
balance the van der Waals attractions [28]. At higher SDS
concentrations, the clusters become larger and denser. Such
a behavior can presumably be ascribed to the reduction of
electrostatic repulsion forces between CNTs due to the large
ionic strength and the increasing concentration of surfactant
 L. Jiang et al. / Journal of Colloid and Interface Science 260 (2003) 89–94
aggregates, known as micelles [28], in the aqueous solu-
tion. Micelles cannot fit in between two CNT bundles that
are close to each other. As a result, the osmotic pressure
of the micelles around bundles creates an effective attrac-
tion. In classical colloidal suspensions [29] and in multiwall
nanotube dispersions [19], this attraction is known as a de-
pletion attraction. Thus, the optimum is reached when the
SDS concentration is slightly below the critical micellar
concentration (CMC = 2.49 wt%) [30] at which surfactant
molecules start to self-aggregate into micelles [31]. Van-der-
Waals-induced aggregation at low SDS concentration and
depletion-induced aggregation at high SDS concentration
delimit an intermediate domain of homogeneously dispersed
nanotubes.
4. Discussion
With the aid of SDS, the formation of homogeneous CNT
dispersions in an aqueous medium is no surprise. However,
a more complete understanding of the interfacial chemistry
and the dispersion mechanism can provide a guideline for
better choices of dispersant types.
TEM micrographs of raw CNTs (Fig. 1) display the long
tubes attached together; that is, even after ultrasonication the
CNTs used here is shaped like a bundle. The magnitude of
the ζ potential for the SDS-adsorbed CNTs is higher than
that for the bare CNTs. Electrostatic repulsion between the
negatively charged cluster surfaces plays an essential role in
the stabilization of the CNT clusters in water with SDS.
The interfacial chemistry and the dispersion mechanism
of CNTs adsorbed with SDS can be discussed further by
using AES and FTIR spectra. AES spectra show the peak
of adsorbed SDS, which suggests the strong adsorption
between the CNT surface and SDS. SDS is a cationic
surfactant containing a sulfate hydrophilic segment and a
hydrocarbon hydrophobic segment. Naturally the adsorption
of the surfactant has an effect on the wetting behavior and
on the interfacial adhesion, which in return has an impact on
the surfactant’s ability to disperse the nanotubes [20].
The adsorption mechanism can be described as follows:
The sulfate group ionized in a very large pH range gives a
negative charge to SDS. Electrostatic repulsion forces exist
between the CNTs and the hydrophilic sulfate groups of SDS
as they are both negatively charged. Thus, the alkane groups
of SDS prefer to interact with the CNT bundle surface,
namely, the surfactant interacts with CNTs through the
hydrophobic segment [31], which is proved by FTIR spectra.
The bands assigned to the group of –CH3 are weakened or
eliminated, which is due to the interaction of –CH3 with
the CNTs. Below pH 8, SDS adsorbs on CNTs due to
hydrophobic interaction. However, adsorption of SDS will
be prevented due to electrostatic repulsion between CNTs
and negatively charged SDS in alkaline solutions.
On the other hand, ionized sulfate groups cause higher
negative surface charge, which is proved by the ζ poten-
93
tial measurements. The repulsive forces between the same
charge particles stabilize the SDS-adsorbed CNT suspen-
sions. Long alkane groups also provide external steric re-
pulsion to help the particles overcome the attractive van
der Waals force at contact. As a conclusion, the surfac-
tant containing a single straight-chain hydrophobic segment
and a terminal hydrophilic segment can modify the CNTs–
suspending medium interface and prevent aggregation over
long time periods.
5. Conclusions
Stable aqueous colloidal dispersions of CNTs are ob-
tained with the aid of SDS. UV–vis spectroscopy is used
to quantitatively characterize the stability of the dispersions
for the first time. The CNT/SDS dispersion exhibits extreme
stability, with the supernatant CNT concentration decreas-
ing only 15% compared with a decrease of 50% for the
bare CNTs. The interaction between CNTs and SDS through
the hydrophobic segment causes a higher negative surface
charge and steric repulsion, which improves the stability of
the CNT/SDS dispersion. It has been concluded that the sur-
factant containing a single, long, straight-chain hydrophobic
segment and a terminal hydrophilic segment proves to be a
suitable dispersant for the stable CNT dispersion. Large dif-
ferences exist between the morphology of the suspensions
without and with SDS. The optimum amounts to obtain a
stable homogeneous CNT dispersion are 0.5 wt% CNTs and
2.0 wt% SDS.
As our UV–vis spectrophometric measurement is easy
to characterize quantitatively and accurately the stability
of the CNT dispersion, it would contribute to the study of
production of stable CNT dispersions. The investigation of
the stabilization mechanism can help provide a guideline for
choosing suitable dispersants to prepare homogeneous CNT
dispersions.
References
[1] S. Ijjima, Nature 354 (1991) 56.
[2] A.G. Rinzler, J.H. Hafner, P. Nikolaev, L. Lou, S.G. Kim, D. Tomanek,
P. Nordander, D.T. Cobert, R.E. Smalley, Science 269 (1995) 1550.
[3] W.A. de Heer, A. Chatelain, D. Ugarte, Science 270 (1995) 1179.
[4] P.G. Collins, A. Zettl, H. Bando, A. Thess, R.E. Smalley, Science 278
(1997) 100.
[5] L.S. Schadler, S.C. Giannaris, P.M. Ajayan, Appl. Phys. Lett. 73
(1998) 3842.
[6] R. Andrews, D. Jacques, A.M. Rao, T. Rantell, F. Derbyshire, Y. Chen,
J. Chen, R.C. Haddon, Appl. Phys. Lett. 75 (1999) 1329.
[7] Z. Jia, Z. Wang, C. Xu, J. Liang, B. Wei, D. Wu, S. Zhu, Mater. Sci.
Eng. A 271 (1999) 395.
[8] T. Kuzumaki, K. Miyazawa, H. Ichinose, K. Ito, J. Mater. Res. 13
(1998) 2445.
[9] H.D. Wagner, O. Lourie, Y. Feldman, R. Tenne, Appl. Phys. Lett. 72
(1998) 188.
[10] W.A. Scrivens, J.M. Tour, K.F. Creek, L. Pirisi, J. Am. Chem. Soc. 116
(1994) 4517.
94 L. Jiang et al. / Journal of Colloid and Interface Science 260 (2003) 89–94

[11] D.M. Guldi, R.E. Huie, P. Neta, H. Hungerbühler, K.-D. Asmus, [21] G.L. Hwang, K.C. Hwang, J. Mater. Chem. 11 (2001) 1722.
Chem. Phys. Lett. 223 (1994) 511. [22] M. Sano, J. Okamura, S. Shinkai, Langmuir 17 (2001) 7172.
[12] Y.N. Yanakoshi, T. Yagami, K. Fukuhara, J. Chem. Soc. Chem. Com- [23] H.E. Ries, Nature 226 (1970) 72.
mun. (1994) 517. [24] B. Vigolo, A. Pénicaud, C. Coulon, C. Sauder, R. Pailler, C. Journet,
[13] H. Hungerbühler, D.M. Guldi, K.-D. Asmus, J. Am. Chem. Soc. 115 P. Bernier, P. Poulin, Science 290 (2000) 1331.
(1993) 3386. [25] Y. Fan, M. Burghard, K. Kern, Adv. Mater. 14 (2002) 130.
[14] K.I. Prigadarsini, H. Mohan, J.P. Mittal, Fullerene Sci. Technol. 3 [26] Y. Xia, Y. Ma, Y. Xing, Y. Mu, C. Tan, L. Mei, Phys. Rev. B 61 (2000)
(1995) 479. 11088.
[15] H.W. Sophie, P. Philip, J. Colloid Interface Sci. 243 (2001) 306. [27] K. Nakanshi, P.H. Solomon, Infrared Absorption Spectroscopy,
[16] E. Kunio, J. Colloid Interface Sci. 241 (2001) 1. Holden–Day, Oakland, CA, 1977.
[17] K. Meguro, K. Esumi, Nenryo Kyokai Shi. (1986) 902. [28] R.J. Hunter, Foundations of Colloid Science, Oxford Univ. Press,
[18] M. Sano, A. Kamino, J. Okamura, S. Shinkai, Science 293 (2001) Oxford, 1989.
1299. [29] S. Asakura, F. Oosawa, J. Chem. Phys. 22 (1954) 1255.
[19] J.-M. Bonard, T. Stora, J.-P. Salvetat, F. Maier, T. Stöckli, C. Duschl, [30] Z. Shen, G.T. Wang, Colloids and Surfaces Chemistry, Chemical
L. Forró, W.A. Heer, A. Châtelain, Adv. Mater. 9 (1997) 827. Industry Press, Peking, 1991.
[20] X. Gong, J. Liu, S. Baskaran, R.D. Voise, J.S. Young, Chem. Mater. 12 [31] R.J. Hunter, Introduction to Modern Colloid Science, Oxford Univ.
(2000) 1049. Press, Oxford, 1993.