Colloids and Surfaces A: Physicochem. Eng.

Aspects 404 (2012) 25–35

Contents lists available at SciVerse ScienceDirect

Colloids and Surfaces A: Physicochemical and

Engineering Aspects
journal homepage: www.elsevier.com/locate/colsurfa

Investigating the effectiveness of PEO/PPO based copolymers as dispersing

agents for graphitic carbon black aqueous dispersions
S. Yasin, P.F. Luckham ∗
Department of Chemical Engineering & Technology, Imperial College, South Kensington, Post Code: SW7 2AZ, London, United Kingdom

a r t i c l e i n f o a b s t r a c t

Article history: The dispersability of graphitic carbon black, selected as a model for carbon nanotubes, has been investi-
Received 6 October 2011 gated using a combination of rheological, conductivity and atomic force microscopy (AFM) techniques.
Received in revised form 4 April 2012 The effectiveness of three PEO/PPO based non-ionic dispersants, namely polyethylene oxide polypropy-
Accepted 6 April 2012
lene oxide ABA copolymers (synperonic PE/F 103 with 2 × 16 ethylene oxide units and PE/F 108 with
Available online 12 April 2012
2 × 148 ethylene oxide units) and NPE1800 (nonyl phenyl polypropylene oxide-polyethylene oxide with
27 ethylene oxide units), is reported. Adsorption isotherms were determined for these dispersants. The
Keywords:
adsorption isotherms of PE/F 103 in comparison with PE/F 108 revealed that in molar terms (␮mol/m2 )
Graphitic carbon dispersions
Rheology
the adsorption decreases for PE/F 108 with more ethylene oxide units, indicating that adsorption is
Surfactant adsorption governed by size of the PEO (polyethylene oxide) chain length. Also, the synperonic PE series which has
Conductivity polypropylene oxide as an anchor group and does not contain any aromatic ring in their anchoring group,
AFM gave lower adsorption amounts (in moles) as compared to the NPE 1800 which contains an aromatic ring
(nonyl phenyl) in its anchoring group as well as propyleneoxide. The relative viscosity-effective volume
fraction ˚ curves are compared with the theoretical curves for the hard sphere dispersions calculated
using Krieger–Dougherty equation. For the graphitic carbon black studied here, in an aqueous medium,
neither PE/F 103, nor PE/F 108 showed good agreement with the Krieger–Dougherty equation; the vis-
cosities were all much higher than that predicted by that equation. Whilst NPE 1800 produced dispersions
of lower viscosity and the viscosity values showed a good agreement with the Krieger–Dougherty equa-
tion. The results achieved from oscillatory measurements showed that PE/F 103 and PE/F 108 dispersants
showed a frequency cross-over of G and G at lower volume fractions. Also they produced dispersions of
high electrical conductivity, suggesting that these systems are aggregated. In the AFM force spectroscopy
measurements, the interactions between the adsorbed layers of PE/F103 were initially attractive, whilst
somewhat surprisingly the PE/F 108 adsorbed layers only showed repulsive interactions on approach
and separation. NPE 1800 stabilised systems exhibited much lower viscosity and elastic modulus than
the PE/F stabilised dispersions; produced dispersions of lower electrical conductivity and showed repul-
sive interactions in AFM, suggesting that these systems are much more stable than the carbon black
dispersions bearing adsorbed PE/F polymers.
The results indicated that synperonics (PE/F 103 and PE/F 108) are not a good dispersants for graphitic
carbon black and by consequence for carbon nanotubes; whilst NPE 1800 is a suitable dispersant for these
dispersions and could be a good dispersant for carbon nanotubes.
© 2012 Elsevier B.V. All rights reserved.

1. Introduction problem has motivated considerable recent research to develop

Carbon nanotubes are nanometre scale wires with the potential
of benefiting mankind in many ways because of their extraordinary
mechanical, electrical, optical and thermal properties [1–4]. Car-
bon nanotubes pack themselves into ropes which further aggregate
and this aggregation is an obstacle to many applications [5]. This
∗ Corresponding author. Tel.: +44 0 2075945583; fax: +44 0 2075945636.
E-mail addresses: s.yasin@imperial.ac.uk (S. Yasin), p.luckham01@imperial.ac.uk
(P.F. Luckham).
methods to disperse the nanotubes more effectively [6].
The use of ultrasound has achieved some success in dispersing
carbon nanotubes, but this method tends to damage the inter-
nal structure of the tubes [7]. Physical adsorption of polymeric
layers that will act as a steric stabilizer is an alternative [8,9].
Graphitic carbon black possesses largely a uniform surface, like
carbon nanotubes, composed of aromatic carbon rings owing to
crystalline nature of the two materials. Graphitic carbon is chemi-
cally inert and stable, hence does not react with the environment.
Also graphitic carbon black particles in a solution/suspension, like
carbon nanotubes have a tendency to pack themselves and settle

0927-7757/$ – see front matter © 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.colsurfa.2012.04.001
26 S. Yasin, P.F. Luckham / Colloids and Surfaces A: Physicochem. Eng. Aspects 404 (2012) 25–35
Table 1
Characteristics of polymers/dispersants [10].
Dispersant Molecular weight (kg/mol)
PE/F 103 4.7
PE/F 108 16.2
NPE 1800 2.2
after some period. Therefore, keeping graphitic particles dispersed
in any medium using polymers for longer periods is of consider-
able interest. Since there are considerable safety issues in working
with carbon nanotubes, graphitic carbon black was selected as a
model. The model is based on the hypothesis that a good disper-
sant for graphitic carbon black will also be a good dispersant for
carbon nanotubes due to the similarity of the surface structure and
properties. In previous studies, Miano and co-workers [10,11] have
investigated the dispersion of a non graphitic, amorphous carbon
black in water by three classes of dispersants; an ABA polyethy-
lene oxide-polypropylene oxide-polyethylene oxide dispersant;
a nonyl phenol ethoxylate dispersant and a novel nonylphenol-
polypropylene oxide-polyethylene oxide dispersant. The studies
concluded that the later dispersant showed best dispersion prop-
erties (although the nonyl phenol ethoxylates dispersants also
showed similar dispersing behaviours) and both were considerably
better dispersants than the ABA polyethylene oxide-polypropylene
oxide-polyethylene oxide PPO-PEO polymers. This is surprising
since PPO-PEO polymers (often called by their trade name of
Pluronics) are often used to disperse carbon nanotubes [12]. The
present work is an extension of the Miano’s work to a graphitic
carbon black (Monarch 1000).
Among the wide range of dispersants available, three PEO/PPO
based non-ionic dispersants named polyethoxylated PE/F 103, PE/F
108 and nonyl phenyl polypropylene oxide-polyethylene oxide
were selected. The anchoring chain, polypropylene oxide, is same
for PE/F 103 and PE/F 108 except in NPE 1800 where the anchoring
chain consists of a nonyl phenyl group and polypropylene oxide.
Whilst the size of stabilising chain, i.e. polyethylene oxide is dif-
ferent for all dispersants. Therefore, effect of the size of stabilising
chain is investigated by keeping the anchoring chain same in syn-
peronics (PE/F 103 and PE/F 108). As well effect of a nonyl phenyl
in NPE 1800 is studied on dispersion properties. The quality of
carbon black dispersions is evaluated based on rheological mea-
surements, electrical conductivity measurements and AFM (atomic
force microscopy) operating in the so-called force spectroscopy
mode.
2. Experimental
2.1. Materials
Graphitic carbon black (Monarch 1000 jet black in colour) with
a density of 1.8 g/cm3 was kindly supplied by Cabot Chemical Cor-
poration, UK. PE/F 103, PE/F 108 and NPE 1800 were supplied by ICI
Speciality Chemicals, now known as Croda Chemicals. Characteris-
tics of dispersants are given in Table 1.
2.2. Surface area measurements
In order to determine the maximum amount of dispersant
adsorbed onto the carbon black particles, the surface area of the
particles was determined using two different methods. First, the
surface area was calculated by nitrogen adsorption method (BET)
[13]. Using this method the surface area of Monarch 1000 was
found to be 301 m2 /g with a porosity of 0.96 cm3 /g. Nitrogen gas
is a simple diatomic molecule, able to access the smallest pores of
the carbon black and is much smaller than any of the polymeric
No. of ethylene oxide units No. of Propylene oxide units Nonyl phenyl
2 × 16 56 No
2 × 148 56 No
27 13 Yes
dispersants. Therefore, the surface area based on gas adsorption
method is not a good estimate for the area available to the disper-
sants. Thus a wet surface area based on methylene blue adsorption
method was determined as well; using this method the surface area
of the particles was found to be 163 m2 /g [10,11,14]. As expected,
the surface area based on methylene blue adsorption is much less
than that measured by nitrogen adsorption method. Therefore, the
wet surface area is more accurate estimate for the available surface
area of the particles for adsorption of dispersants. Hence, the wet
surface area is used to estimate the amount of polymer adsorbed
per square metre of surface for present work.
2.3. Adsorption measurements
Carbon black (Monarch 1000) dispersions of 0.15 weight per-
cent were prepared using dispersants of different concentrations.
After 30 min vigorous stirring using a Silverson mixer, all disper-
sions were shaken for at least two days to establish adsorption
equilibrium. All solutions were then centrifuged in order to sepa-
rate the carbon black and determine the equilibrium concentration
of the polymer. However it was noticed that centrifugation alone
did not give a clear supernatant at the top of the centrifuge tubes
and that some carbon black particles were still present in solu-
tion. Therefore, micro filters (Millex Millipore syringe driven filter
unit, 0.2 ␮m) were used to filter any remaining carbon black from
the solution and hence a clear supernatant solution was obtained.
The concentration of polymer remaining in the supernatant was
determined using an UV/vis spectrometer. In case of PE/F polymeric
dispersants a colourimetric method was used by forming a coloured
complex of iodine (I2 ) and potassium iodide (KI) [10,11]. For each
10 ml supernatant solution 0.25 ml of a freshly prepared complex-
ing solution consisting of 1 g of I2 and 2 g of KI in 100 ml of distilled
water) was added. The absorbance values were then determined
using an UV/vis spectrometer at a wavelength of 460 nm which was
compared with the calibration curves. The equilibrium concentra-
tions of NPE 1800 were obtained by measuring the absorbance of
the phenyl ring at 274 nm.
2.4. Rheology measurements
Aqueous dispersions of Monarch 1000 (8–22 wt%) were pre-
pared using the surfactant solutions of known concentrations. The
dispersion was first stirred manually and then in a Silverson mixer
for about one hour followed by ball milling for 16 h. After screening
the balls, the dispersions were taken for rheological experiments.
Bead milling for more than 16 h did not result in any change in the
viscosity.
Steady state and oscillatory measurements were carried out
with a PAAR UDS rheometer using concentric cylinder geome-
try. The rheometer is capable of varying the shear rates in a
range of 0.001–1000 s−1 and oscillatory strain frequencies in a
range of 0.001 to 100 Hz. The oscillatory measurements were more
complex, requiring as a first step, determination of a linear vis-
coelastic region. This was obtained by performing a strain sweep
test whereby the amplitude of the strain was ramped in order to
determine the critical strain where G (elastic modulus) and G (loss
modulus) begin to be dependent on the strain and start to decrease
 S. Yasin, P.F. Luckham / Colloids and Surfaces A: Physicochem. Eng. Aspects 404 (2012) 25–35
with increase in strain [15]. Typically the linear viscoelastic region
was less than 5% strain.
2.5. Conductivity measurements
In order to obtain information concerning the structure of the
dispersions, conductivity of the dispersions were determined at a
fixed concentration of carbon black (1 wt%). The concept here being
that the more aggregated the dispersions are, the more connected
the particles in the dispersions are and would therefore show a
higher conductivity. Dispersions of Monarch 1000 were prepared
with dispersants at their optimum concentrations as obtained by
adsorption isotherms and without dispersant (blank) for the rela-
tive conductivity measurements.
2.6. AFM (atomic force microscopy)
Given that polymers are used as stabilisers in many indus-
trial suspensions, it is beneficial to measure the interaction forces
between surfaces coated with these polymers. An in-house AFM
was used for all measurements reported here. The in-house AFM
had no imaging capability. However, the instrument has certain
advantages over a commercial instrument for force measurements
(principally it is easier to exchange fluids and easier to control
approach rates). The instrument was placed on an anti vibration
table in a basement laboratory to minimize vibrations.
The main constituents of the AFM are comprised of a V shaped
cantilever, a piezoelectric ceramic tube, a laser diode and a pho-
todiode PSD (position sensitive detector). In our instrument, the
cantilever remains in a fixed position while the piezo tube can move
in a vertical direction. The particle of interest is attached on the can-
tilever tip. The cantilever with a particle attached is mounted onto
a metal rod fixed right under the laser light. The sample surface
(carbon rod in this study) is placed in the measuring cell which in
turn is attached onto the piezoelectric ceramic tube. Further details
of the design and operation of the AFM are given elsewhere [16,17].
When a probing tip or a particle attached onto the cantilever
comes close to another particle surface, an interaction between
the surfaces is detected. The interaction forces sensed by the can-
tilever tip cause it to bend. Any attractive interactions bend the
cantilever down whilst repulsive interactions cause it to bend up.
The degree of deflection of the cantilever depends upon magnitude
of the interaction forces. Therefore, the interaction forces can be
measured in AFM by measuring the deflection of cantilever. The
deflection can be measured by optical beam technique based on
reflection of the laser beam from the cantilever. The reflected beam
is detected by a position sensitive detector (PSD). Lab view, a soft-
ware supplied by National Instruments, UK was used to record the
data. The force–distance curves were constructed from data of the
PSD voltage versus piezo motion. Many curves were obtained for
each dispersant, however one typical cycle is presented here for
brevity.
In AFM, generally smooth micrometer sized particles are
attached to the cantilever. The measured force is a function of dis-
tance, D, between two spheres of radius, R1 , and R2 . The force is
related to the total interaction energy per unit area according to
the Derjaguin approximation, [18]:
 R R 
1 2
F(D) = 2 E(D)
R1 + R2
Hence, for R1 = R2 = R
F(D)
E(D) =
R
27
Fig. 1. Adsorption isotherm of Monarch 1000 (0.15 wt%) using PE/F 103 and PE/F
108 in molar units. Solid line is Langmuirian fit using best fitting values of K and  m .
Symbols are:
 Using PE/F 103 (K = 62.6 m2 /␮mol,  m = 0.26 ␮mol/m2 ).
 Using PE/F 108 (K = 148 m2 /␮mol,  m = 0.11 ␮mol/m2 ).
For two equal sized spheres the van der Waals attractive energy is
given by [19]:
AR
E(D) = − (3)
12D
where A is the Hamaker constant that depends upon the properties
of two approaching bodies and the characteristics of medium.
The above equation was used to compute the theoretical van
der Waals forces to be compared with the experimentally deter-
mined attractive forces. The Hamaker constant for graphitic carbon
black and water were taken to be 47.0 × 10−20 J and 11.0 × 10−20 J,
respectively [20].
2.7. Sample preparation
Monarch 1000 particles are only 10 nm in radius thus are too
small to be used in the AFM experiments. Therefore spherical glassy
carbon black (2–12 micron size and 99.95% pure) was used to model
Monarch 1000. Carbon black particles tend to form cluster and so it
was very hard to pick a single particle. Trials were made to pick the
smallest cluster (a few micrometers to 20 ␮m) and glued onto the
cantilever. A very small amount of glue was used to avoid contam-
ination of the particle. The glued particle was kept for 24 h to be set
at room temperature. The carbon surface was made by sprinkling
the carbon black particles onto a glued glass rod.
Measurements were performed by placing the sample surface in
a cleaned Petri dish (measuring cell) filled with 40 ml water or any
solution of interest. After bringing the particle on the cantilever
and the sample surface in close proximity of ≈2 mm, 4 ml water
was replaced with 4 ml of 0.1 wt% polymeric stock solution using a
syringe so that the overall polymer concentration was ≈0.01 wt%.
The system was left for overnight to attain equilibrium.
3. Results and discussion
(1)
3.1. Adsorption results
Adsorption isotherms of the surfactants are shown in
Figs. 1 and 2. The adsorption isotherms are Langmuirian, high
(2)
affinity type adsorption isotherms typical of a polymer and a
28 S. Yasin, P.F. Luckham / Colloids and Surfaces A: Physicochem. Eng. Aspects 404 (2012) 25–35

Table 2
Maximum amount adsorbed of dispersants obtained from adsorption isotherms on
molar and weight basis.

Dispersant  max (␮mol/m2 )  max (mg/m2 )

PE/F 103 0.24 1.10

PE/F 108 0.11 1.80
NPE 1800 1.80 4.00

 max = maximum amount adsorbed.

The amount of dispersant adsorbed on the carbon black particles

Fig. 2. Adsorption isotherm of Monarch 1000 (0.15 wt%) using NPE 1800 in molar
units. Solid line is Langmuirian fit using best fitting values of K = 7.9 m2 /␮mol and
 m = 2.1 ␮mol/m2 .
can be determined from these Langmuirian curves these data are
presented in Table 2.
It is also apparent from the data presented in Table 2 that size
of the stabilising chain plays a significant role in determining the
amount of dispersant adsorbed. This can be seen best by compar-
ing the adsorption of the PE/F 103 with PE/F 108 where the anchor
group is constant, while size of the stabilising part of polymer
changes. Here, one can see that increasing the stabilising molecu-
lar weight decreases the number of adsorbing molecules. This effect
has also been observed by others for the adsorption of these poly-
mers onto various surfaces [21–24], but not as far as we are aware
for the graphitic carbon black. It seems that by having a large sta-
bilising group makes it hard for other molecules to come and pack
tightly on the surface.
However, although the stabilising ethylene oxide plays a role,
the anchoring chain is the principal part which determines the
amount adsorbed (in molar terms). The adsorption of NPE 1800
containing an aromatic ring in its structure observed is significantly
higher and is likely to be due to ␲–␲ interactions of the aromatic
ring with the graphitic rings of the carbon black, Table 2.

3.2. Rheological results

Fig. 3. Linearized plot of Langmuirian isotherm of PE/F 103 and PE/F 108 adsorbing 3.2.1. Steady shear tests
onto Monarch 1000 carbon black (0.15 wt%). Symbols are: Having established the optimum dispersant concentration from
 Using PE/F 103. the adsorption isotherms, all rheology data were obtained at that
 Using PE/F 108.
optimum concentration (i.e. at the dispersant concentration corre-
sponding to the plateau of adsorption isotherms).
surfactant adsorption to colloidal dispersions. The data were fitted The dispersions show shear thinning behaviour (viscosity
to linear form of the Langmuir equation, i.e. decreases with the increase of shear rate) which becomes more evi-
dent at higher volume fractions. Fig. 5 best illustrates this behaviour
Ce Ce 1
= + (4) where the viscosity of dispersions of varying volume fractions of
 m Km carbon black prepared by using NPE1800 as the dispersant is plot-
The linearity of the plot of Ce / and Ce shows that adsorption ted a function of the shear rate. Two linear regions were found
isotherms obey the Langmuir equation as shown in Figs. 3 and 4. by plotting the viscosity versus the shear rate, i.e. low shear limit

Fig. 5. The dependence of viscosity versus shear rate representing the shear thinning
behaviour of carbon black dispersions of different volume fraction using NPE1800
Fig. 4. Linearized plot of Langmuirian isotherm of NPE 1800 adsorbing onto (3.8 mg/m2 ). Symbols are:
Monarch 1000 carbon black (0.15 wt%).  ˚ = 0.05; + ˚ = 0.09;  ˚ = 0.11; 䊉 ˚ = 0.12;  ˚ = 0.13.
 S. Yasin, P.F. Luckham / Colloids and Surfaces A: Physicochem. Eng. Aspects 404 (2012) 25–35
Fig. 6. High shear rate (672 1/s) viscosity of carbon black (Monarch 1000) disper-
sions versus volume fractions of solids in water. Symbols are: – Krieger–Dougherty
equation (˚p = 0.65 and [] = 3.0);  PE/F 103 (1.1 mg/m2 );  PE/F 108 (1.8 mg/m2 );
 NPE 1800 (3.8 mg/m2 ).
viscosity and high shear limit viscosity. The viscosity decreases
markedly at high shear rates probably due to alignment of the par-
ticles in shear bands [15]. The viscosity measurements in the linear
region helps in collecting and treating the data independent of flow
conditions, thus theoretical treatment of viscosity versus volume
fraction can be used.
Figs. 6 and 7 show the relative high and low shear viscosity data
versus volume fraction in aqueous medium. The relative viscos-
ity may also be calculated from the Krieger–Dougherty equation
shown as a continuous line on these graphs [25].
 []p

r = 1+ (5)
p
Initially exponent of the Krieger–Dougherty equation was fixed
by making a general assumption for values of ˚p (the theoreti-
cal maximum volume fraction) and the intrinsic viscosity [], and
then extrapolating the (r )−1/[]˚p versus ˚ data to zero. From the
experimental maximum packing volume fraction ˚m , the expo-
nent of Krieger–Dougherty equation can be adjusted until the
best fit of experimental data with the Krieger–Dougherty equation
is obtained. In this way, Prestdige and Tadros [26] extrapolated
(r )−0.5 versus ˚ to zero assuming ˚p [] = 2 but actual quoting
˚p = 0.7 and [] = 2.5. Goodwin and Ottewill [27] plotted (r )−0.662
versus ˚ for the fit of their experiments assuming ˚p = 0.604 and
Fig. 7. Low shear rate (1/s) viscosity of carbon black (Monarch 1000) dispersions
versus volume fractions of solids in water. Symbols are: – Krieger–Dougherty equa-
tion (˚p = 0.6 and [] = 2.5);  PE/F 103 (1.1 mg/m2 );  PE/F 108 (1.8 mg/m2 );  NPE
1800 (3.8 mg/m2 ).
29
Fig. 8. Adsorbed layer thickness ı of dispersant onto the particle surface.
[] = 2.5. Faers [28] reported ˚p = 0.67 and 0.6 for high and low
shear viscosity regions respectively for latex suspensions while
Kruif et al. [29] plotted relative viscosity in high and low shear
region taking maximum packing fraction 0.71 and 0.63 respectively
for sterically stabilized hard sphere silica suspensions in cyclohex-
ane. Krieger [24] reported ˚p = 0.632 and [] = 3.13 in low shear
region for hard sphere systems. Here we estimate ˚p ·[] = 1.625 by
selecting ˚p = 0.65 and [] = 2.5 for the high shear viscosity region
and ˚p ·[] = 1.8 by selecting ˚p = 0.6 and [] = 3 for the zero shear
viscosity region.
Once ˚m , referred to as the maximum core volume fraction of
carbon black particles, is found by extrapolating the (r )−1/[]˚p
versus ˚ data to zero, the true volume fraction ˚ consisting of the
core solid volume fraction plus the contribution of the adsorbed
dispersant layer can be estimated. Fig. 8 shows the adsorbed poly-
mer layer thickness on the particle surface. By comparing ˚p with
˚m the adsorbed layer thickness can be calculated, as given in Eq.
(6).
 1/3 
˚p
ı=R −1 (6)
˚m
The contribution from the adsorbed polymer layer to the total
hydrodynamic volume fraction of the dispersion should also be
considered. Therefore ˚ may be converted into the effective vol-
ume fraction ˚ constituting the core volume of particles plus the
hydrodynamic volume of the adsorbed polymer layer through Eq.
(7).
   3
 ı
˚ = 1+ ˚ (7)
R
High and low shear viscosity values do not correspond to the
Krieger–Dougherty equation when plotted against the volume frac-
tion of core of the carbon black (˚) using any of the dispersants as
can be seen from Figs. 6 and 7. However by taking into account
the estimated value of the thickness of the adsorbed polymer layer
and plotting the viscosity versus the effective volume fraction, ˚ ,
a closer correspondence of the viscosity to the Krieger–Dougherty
was observed, Figs. 9 and 10.
Even so, for the graphitic carbon black studied here, in an
aqueous medium, neither PE/F 103, nor PE/F 108 showed good
agreement with the Krieger–Dougherty equation; the viscosities
were all much higher than that predicted by the equation. These
high viscosities were due to strong flocculation of the particles.
This was somewhat surprising as Miano et al. [11] found that both
polymers were reasonably good stabilizers for amorphous carbon
30 S. Yasin, P.F. Luckham / Colloids and Surfaces A: Physicochem. Eng. Aspects 404 (2012) 25–35
Fig. 9. High shear rate viscosity of carbon black (Monarch 1000) dispersions) versus
effective volume fractions of solids in water. Symbols are: – Krieger–Dougherty
equation (˚p = 0.65 and [] = 2.5);  PE/F 103 (1.1 mg/m2 );  PE/F 108 (1.8 mg/m2 );
 NPE 1800 (3.8 mg/m2 ).
black. Therefore, determination of true value of the adsorbed layer
thicknesses for graphitic carbon black (Monarch 1000) was not pos-
sible due to strong flocculation. The viscosity values of dispersions
prepared by PE/F 108 were lower than PE/F 103 which may be
attributed to the larger stabilising chain of PE/F 108 providing bet-
ter stabilisation as compared to PE/F 103. However, the stabilising
layer thickness was not sufficient in case of PE/F 108 to prevent
aggregation as is shown by the high viscosities.
However, NPE 1800 proved to be a good stabilizer for crystalline
graphitic carbon black as the dispersions had considerably lower
viscosities and could be reasonably well fitted by the Dougherty
Krieger equation, see Figs. 9 and 10. It is significant to note that
this dispersant has some aromatic functionality in its anchoring
group (nonyl phenyl polypropylene), whilst the PE/F 103 and
the PE/F 108 polymers only have propyleneoxide moieties in
their anchoring group, we have already seen that the presence of
the aromatic ring increases the amount of dispersant adsorbed,
here we see that this increased amount adsorbed gives rise to
more stable dispersions with a lower viscosity. In this regard it
is significant that Lin and Xing [30] investigated the adsorption
affinity of aromatic hydrocarbons to carbon nanotubes which
increased with increasing number of aromatic groups, in an
order of cyclohexanol < phenol < phenylphenol< naphthol. They
concluded that ␲–␲ interactions were supportive for adsorption
Fig. 10. Low shear rate viscosity of carbon black (Monarch 1000) dispersions) versus
effective volume fractions of solids in water. Symbols are: – Krieger–Dougherty
equation (˚p = 0.6 and [] = 3.0);  PE/F 103 (1.1 mg/m2 );  PE/F 108 (1.8 mg/m2 );
 NPE 1800 (3.8 mg/m2 ).
Fig. 11. Storage modulus, solid symbols and loss modulus, open symbols of carbon
black (Monarch 1000) dispersion using NPE1800 (3.8 mg/m2 ) versus the frequency
at different volume fractions. Symbols are:
 Storage modulus, volume fraction 0.144;  loss modulus volume fraction 0.144.
 Storage modulus, volume fraction 0.138;  loss modulus volume fraction 0.138.
䊉 Storage modulus, volume fraction 0.11;  loss modulus volume fraction 0.11.
affinity of aromatic rings onto the carbon nanotube surfaces and
the interactions increase with increasing aromaticity. It is likely
that similar effects are occurring for the present dispersions.
3.2.2. Oscillatory tests
Viscoelastic properties of the carbon black aqueous dispersions
using PE/F 103 and PE/F 108 were also investigated by performing
oscillatory shear measurements. Initially strain sweep tests were
run to ensures that the experiments were performed in the linear
viscoelastic region, where there is no structure breakdown of the
dispersion, this was followed by a frequency sweep test [31,32].
In a frequency sweep test, such as that shown in Fig. 11, shows
the variation of G with frequency at different volume fractions of
carbon black in presence of NPE 1800 in water. G increases rapidly
with the frequency at lower volume fractions while it becomes less
pronounced at higher volume fractions. At high volume fraction
the dispersions behave like a nearly elastic material so G becomes
independent of frequency at higher volume fractions, because the
inter-particle distance decreases as volume fraction increases and
PEO chains come more frequently into contact, resulting in an
increased elastic response [33].
Figs. 12 and 13 show some general features of the dispersions
prepared using PE/F 103, PE/F 108 and NPE 1800. Effect of the
Fig. 12. Storage modulus of aqueous carbon black suspensions versus volume frac-
tion of carbon black (Monarch 1000). Symbols are:  PE/F 103 (1.1 mg/m2 );  PE/F
108 (1.8 mg/m2 );  NPE 1800 (3.8 mg/m2 ).
 S. Yasin, P.F. Luckham / Colloids and Surfaces A: Physicochem. Eng. Aspects 404 (2012) 25–35
Table 3
Relative conductivity values of dispersion of Monarch 1000 prepared with and with-
out dispersants using optimum concentrations.
Dispersant Relative conductivity (conductivity of the
dispersion/conductivity of the supernatant)
No dispersant 56.0
PE/F 103 40.0
PE/F 108 33.0
NPE 1800 9.00
volume fraction on G and G at a frequency of 1 Hz can be seen
from these figures. For NPE 1800, at lower volume fractions G and
G increase very slowly as the average separation between the par-
ticles is high and so the adsorbed polymer layers on the particles
do not come into contact very often. On increasing the volume
fraction G and G increase as inter-particle distance is reduced
also the adsorbed layers on the particle surfaces come into con-
tact more often, generating stronger repulsive interaction which
increases the elastic behaviour of the dispersions [33]. The cross-
over of the elastic and viscous modulii is a consequence of these
particle interactions, so that the dispersion shows a predominantly
elastic response at high volume fractions as the polymer layers
stabilising the dispersion come into overlap and repel. In case of
PE/F 103 and PE/F 108, G is greater than G even at lower volume
fractions indicating that the particles are flocculated.
4. Conductivity results
In Table 3 the relative conductivity values (i.e. ratio of the con-
ductivity of the dispersion to the conductivity of the supernatant)
for the dispersions are reported. Since carbon is a conducting
material, the conductivity of the dispersions is a reflection of the
connectivity between the particles. Thus one would expect a highly
flocculated dispersion to show a higher relative conductivity than
a dispersed suspension. The relative conductivity of the blank dis-
persion (i.e. a dispersion without any dispersant) was found to be
much higher than the other dispersions as shown in Table 3. This
is because in absence of any dispersant the particles are aggre-
gated. In the case of PE/F 103 and PE/F 108, although the relatively
conductivity is lower than for the blank sample but still show
significant values. These values indicate that there is still some
aggregation of the carbon black giving rise to conductivity values
for the dispersions. This is again a reflection that these polymers
are not good dispersants for the carbon black. It is interesting
to note that the conductivity for the PE/F103 is higher than for
the PE/F108, an expected result based on the higher molecular
Fig. 13. Loss modulus of aqueous carbon black suspensions versus volume fraction
of carbon black (Monarch 1000). Symbols are:  PE/F 103 (1.1 mg/m2 );  PE/F 108
(1.8 mg/m2 );  NPE 1800 (3.8 mg/m2 ).
31
weight of the PE/F108. While for NPE 1800 the relative conduc-
tivity was much lower, supporting the rheological data which also
suggest that the polymer is a far better dispersant than the PE/F108
and the PE/F 103. The complete surface coverage by anchor chain
(nonyl phenyl-polypropylene oxide) and sufficient adsorbed poly-
mer layer thickness around the particles resulted in a well dispersed
suspension with lower conductivity values.
4.1. AFM results
4.1.1. Interaction forces in the absence of dispersant
To confirm the role of the various dispersants on stabilising the
carbon black particles the interactions between carbon black parti-
cles bearing adsorbed dispersant molecules were measured using
atomic force microscopy. Therefore initially the interaction forces
between two carbon black particles were measured and the results
presented in Fig. 14.
Upon approach, there was no interaction until the particles were
separated by 18 nm and then an attraction was observed. These
attractive forces resulted in the deflection of cantilever towards
the sample surface. At a distance of ∼5 nm, the cantilever sponta-
neously jumps into contact (linear region of force–distance curve
where one can see lower data points). These attractive forces can
be observed until the cantilever and the sample surface were mov-
ing linearly at zero separation. On separation, a strong attraction
was observed between the surfaces. That strong attraction was
due to strong adhesion between surfaces as can be seen in Fig. 14.
Eventually, surfaces snapped apart at a distance of roughly 9 nm.
Thereafter attraction is still observed until distance ∼25 nm. After
that no further interactions were noted. The slight hysteresis in
force–distance curves on approach and separation may be a con-
sequence of strong adhesion between the surfaces causing the
cantilever to twist before separation. Overall we can conclude that
in the absence of polymer, attractive forces were present between
the surfaces on approach and separation.
Fig. 15 is a plot of theoretical values calculated from Eq.
(3), and the experimentally determined attractive interactions.
The theoretical van der Waals forces were considerably shorter
range/weaker than experimental attractive forces as can be seen
in Fig. 15. The origin of these long range attractive forces is not
well understood. However it must be recalled that carbon black
surfaces are hydrophobic in nature furthermore long ranged attrac-
tive forces between hydrophobic surfaces have been determined
by various authors over the last 30 years or so. Recently it has been
proposed that very small nanobubbles occur on the hydrophobic
surfaces and that these may be responsible for long range forces.
When these air bubbles come in contact with other surface, their
contact results in bridging of air between the surfaces hence draw
them together through capillary effects [34–37]. The main conclu-
sion was that strong attractive forces were observed in absence of
the polymer/dispersant.
4.2. Interaction forces in presence of the dispersant
In order to establish how the dispersants modify the interaction
forces between surfaces, it is necessary to conduct the experiments
in polymeric solutions rather than in pure water.
In AFM (atomic force microscopy), the study of the repul-
sive interactions as a consequence of attachment of the polymer
molecules on the particle surface suffers from an uncertainty of
true zero separation distance. Adsorbed dispersant between the
surfaces in constant compliance region, which is generally taken
to be at zero separation. However it is reasonable to take an
assumption that the separation distances between beginning of the
repulsive interactions and the high compression (hard contact) of
polymer layers is equal to twice the compressed adsorbed polymer
32 S. Yasin, P.F. Luckham / Colloids and Surfaces A: Physicochem. Eng. Aspects 404 (2012) 25–35

Fig. 14. Force–distance interaction profile between carbon black particle and carbon surface in water.  For approach  For separation

layer thickness. While actually a thin polymer layer is sandwiched
between two surfaces and the presence of that thin polymer layer
would correspond to an additional thickness which is 0.3–1.8 nm
in this study. So these separation distances would be relative to
this thickness. Musoke and Luckham [38] calculated this addi-
tional thickness in the range of 1–2 nm for a series of synperonics.
This additional thickness was estimated on the basis of adsorption
isotherms and a similar thickness is likely in these experiments.
In Figs. 16 and 17, force–distance profile curves are plotted on
approach and separation in presence of the PE/F 103 and the PE/F
108 respectively.
Fig. 16 shows that in presence of the PE/F 103, strong attractive
forces at a separation of 10 nm were observed, weaker than in
the absence of polymer. On separation, again adhesion was noted
before surfaces spontaneously snapped apart and long range (at
distance of 17 nm) attractive forces were observed. Nestor et al.
[39] observed similar behaviour for glass surfaces bearing an
adsorbed layer of inulin surfactant (INUTEC SP1) of concentration
1 × 10−5 mol/dm3 . The weak repulsive forces were generated
due to adsorption of small amount of polymer and these weaker
repulsive forces were converted into stronger repulsive forces
by further increasing the concentration of polymer. These attrac-
tive interactions are consistent with the flocculated state of the
carbon black when an adsorbed layer of PE/F103 is present as the
rheological data suggested.
In case of the PE/F 108 (Fig. 17), strong repulsive forces were
observed on approach (13 nm) and on separation (22 nm). These
repulsive forces are steric forces generated by the combination of
the compression of polymer layers on approach so decreasing the
entropy and increasing the concentration of polymer between two
surfaces hence the osmotic pressure (enthalpic effects). In the cur-
rent experiment, the polypropylene oxide group of these polymers
adsorbs onto the surface and polyethylene oxide extends into the
solvent. On approach, the polyethylene oxide layers outside the
surface compress and there may be some interlocking of polymer
layers. On separation, these polyethylene oxide layers simply relax
back to its uncompressed configuration rather than desorbing from
the surface giving rise to longer ranged (22 nm) repulsive forces due
to extra stretching as a consequence of entanglement of polymer
layers. It is interesting to note that there is no sign of any attractive
force at a separation distance of 18 nm even though steric interac-
tions do not occur until the surfaces are roughly 13 nm apart. This
is because adsorbed copolymer, which removes the hydrophobic
nature of surface, makes it hydrophilic once more [38,40].
Fig. 18 shows the force–distance profile for suspension pre-
pared in presence of the NPE1800. In presence of the NPE 1800,

Fig. 15. Comparison of the theoretical and experimental force–distance interaction profile between carbon black particles on approach in water. Filled squares are for
experimental approach curve and solid line is the theoretical curve based on Eq. (3).
 S. Yasin, P.F. Luckham / Colloids and Surfaces A: Physicochem. Eng. Aspects 404 (2012) 25–35 33

Fig. 16. Force–distance interaction profile between carbon black particle and carbon surface bearing an adsorbed layer of PE/F 103 in water.
 For approach
 For separation

Fig. 17. Force–distance interaction profile between carbon black particle and carbon surface bearing an adsorbed layer of PE/F 108 in water.
 For approach
 For separation

Fig. 18. Force–distance interaction profile between carbon black particle and carbon surface bearing an adsorbed layer of NPE 1800 in water.
 For approach
 For separation
34 S. Yasin, P.F. Luckham / Colloids and Surfaces A: Physicochem. Eng. Aspects 404 (2012) 25–35
on approach, repulsive interactions were observed which started
at a separation distance of 10 nm. On separation, again repulsive
interactions were observed at the separation distance of 12 nm.
Again a slight hysteresis was observed in force–distance curves on
approach and separation.
Based on these AFM data we would expect the NPE 1800 and the
PE/F108 dispersants to act as good stabilisers for the carbon black
dispersions, while the PE/F103 to act as a poor stabiliser as an attrac-
tion was observed for the polymeric dispersant. Similar results
showing poor stabilising capability of the PE/F103 for the carbon
black dispersions is shown earlier from the rheological data. How-
ever rheological and conductivity experiments, reported earlier,
also suggest that PE/F108 is also a poor dispersant (at least com-
pared to NPE1800). This discrepancy could be due to two factors:
firstly the AFM although sensitive, may not be sensitive enough
to pick up a weak long range attraction; secondly the different
carbon black particles used in the AFM and rheology studies may
give rise to slightly different levels of adsorption and hence interac-
tion profiles. It is known that the PE/F103 and the PE/F108 are good
dispersants for polystyrene latex [21] and that both give repulsive
profiles when adsorbed to hydrophobic glass surfaces [38], suggest-
ing that this class of polymers are on the edge of being able to act
as effective dispersants.
5. Conclusions
The size of the EO stabilising chains (polyethylene oxide units)
regulates the number of molecules adsorbed on the carbon black
surfaces. The larger stabilizing chain (larger number of EO units in
case of PE/F 108) makes it harder for other molecules to come and
pack tightly onto the surface and thus adsorption amount decreases
(in molar terms). The presence of the nonyl phenyl in anchoring
chain of the NPE 1800 provided higher adsorption amounts as com-
pared to the synperonics.
PE/F type copolymers, also known as synperonics (Uniqima
now Croda Chemicals) proved to be ineffective stabilizers for the
graphitic carbon black. This was evidenced by rheological mea-
surements; conductivity values measurements and AFM (except
PE/F 108). The carbon black used in AFM measurements was differ-
ent from the monarch 1000 (graphitic carbon black) used for other
experiments and different behaviour of PE/F 108 may be attributed
to these different kinds of carbon black. While, NPE 1800 containing
aromatic group in its structure proved to be an effective stabilizer
for the graphitic carbon black dispersions. Miano [10,11] also found
that dispersants containing nonyl phenyl (aromatic ring in their
anchoring group) were better than synperonics. Synperonics are
commonly used to disperse carbon nanotubes, due to similar struc-
ture of graphitic carbon black and carbon nanotubes, we would
suggest that the dispersants containing an aromatic ring in their
anchor group would be good candidates for further investigation
in this regard [12,41–44].
Acknowledgements
We would like to thank Cabot Chemical Corporation for a kind
donation of carbon black; Croda (formerly Uniqima) for donating
the PE/F103, PE/F108 and NPE 1800.
One of the authors, S. Yasin, is thankful for financial bursary from
University of Engineering & Techonology, Lahore, Pakistan.
References
[1] K. Rupesh, B. Suryasarathi, Carbon nanotubes based composites – a review, J.
Miner. Mater. Charact. Eng. 4 (2005) 31–46.
[2] D. Ken, A. Robert, T. Lang, S. Vicki, D. Rodger, F. Gavin, A. Andrew, Carbon nano-
tubes: a review of their properties in relation to pulmonary toxicology and
workplace safety, Toxicol. Sci. 92 (2006) 5–22.
[3] Y.R. Rozen, L. Szleifer, Utilizing polymers for shaping the interfacial behaviour
of carbon nanotubes, R. Soc. Chem. 2 (2006) 24–28.
[4] J.H. Hafner, C.L. Cheung, C.M. Leiber, Growth of nanotubes for probe microscopy
tips, Nature 398 (1999) 761–762.
[5] D.W. Schaefer, J.M. Brown, D.P. Anderson, J. Zhao, K. Chokalingam, D. Tomlin,
J. Ilavsky, Structure and dispersion of carbon nanotubes, Appl. Cryst. 36 (2003)
553–557.
[6] R. Rastogi, R. Kaushal, S.K. Tripathi, A.L. Sharma, I. Kaur, L.M. Bharadwaj, Com-
parative study of carbon nanotube dispersion using surfactants, J. Colloid
Interface Sci. 328 (2008) 421–428.
[7] L.K. Lu, Y.K. Lago, M.L.H. Green, P.J.F. Harris, S.C. Tsang, Mechanical damage of
carbon nanotubes by ultrasound, Carbon 34 (1996) 814–816.
[8] J. Mewis, A.J.B. Spaull, Rheology of concentrated dispersions, Adv. Colloid Inter-
face Sci. 6 (1976) 188.
[9] J. Kaldasch, B. Senge, Shear thickening in polymer stabilized colloidal suspen-
sions, Colloid Polym. Sci. 287 (2009) 1481–1485.
[10] F. Miano, A. Bailey, P.F. Luckham, Th.F. Tadros, Adsorption of nonyl phenol
propylene oxide-ethylene oxide surfactants on carbon black and the rheology
of resulting dispersions, Colloids Surf. 62 (1992) 111–118.
[11] F. Miano, A. Bailey, P.F. Luckham, Th.F. Tadros, Adsorption of poly (ethylene
oxide)-poly (propylene oxide) ABA block copolymers on carbon black and the
rheology of the resulting dispersions, Colloids Surf. 68 (1992) 9–16.
[12] G. Ciofani, V. Raffa, V. Pensabene, A. Menciassi, P. Dario, Dispersion of multi-
walled carbon nanotubes in aqueous pluronic F127 solutions for biological
applications, Fuller. Nanotub. Carbon Nanostruct. 17 (2009) 11–25.
[13] S. Brunauer, P.H. Emmett, E. Teller, Adsorption of gases in multimolecular lay-
ers, Physik. Chem. 407 (1938) 309–319.
[14] J.J. Kipling, R.B. Wilson, Adsorptive properties of polymer carbons. Part 2 –
determination of pore sizes, Trans. Faraday Soc. 56 (1960) 562–569.
[15] J.D. Ferry, Viscoelastic properties of polymer solutions, J. Res. 41 (1948)
53–61.
[16] G.J.C. Braithwaite, A. Howe, P.F. Luckham, Interactions between poly (ethylene
oxide) layers adsorbed to glass surfaces probed by using a modified atomic
force microscope, Langmuir 12 (1996) 4224–4237.
[17] G.J.C. Braithwaite, P.F. Luckham, Effect of molecular weight on the interactions
between poly (ethyleneoxide) layers adsorbed on to glass surfaces, J. Chem.
Soc. Faraday Soc. 93 (1997) 1409–1415.
[18] B.V. Derjaguin, L. Landau, Theory of stability of strongly charged lyophobic sols
and of the adhesion of strongly charged particles in solutions of electrolytes,
Acta Physicochim. URSS 14 (1941) 633–662.
[19] E.M. Lifshitz, The theory of molecular attractive force between solids, Soviet
Phys. 2 (1956) 73–74.
[20] J. Visser, On Hamaker constants, a comparison between Hamaker constant and
Lifshitz–van der Waals constants, Adv. Colloid Interface Sci. 3 (1972) 331–363.
[21] M.A Faers, P.F. Luckham, Rheology of polyethylene oxide-polypropylene oxide
block copolymer stabilized lattices and emulsions, Colloids Surf. 86 (1994)
317–327.
[22] J.M. Corkill, J.F. Goodman, J.R. Tate, Adsorption of non-ionic surface-active
agents at the graphon/solution, Trans. Faraday Soc. 62 (1966) 979.
[23] R.H. Ottewill, in: M.J. Schick (Ed.), Nonionic Surfactants, Dekker, NY, 1967, p.
627, CH. 19.
[24] B. Kronberg, Thermodynamics of adsorption of nonionic surfactants on latexes,
Colloid Surf. 12 (1983) 113.
[25] I. Krieger, Rheology of monodisperse latices, Adv. Colloid Interface Sci. 3 (1972)
111–136.
[26] C. Prestidge, Th.F. Tadros, Viscoelastic properties of aqueous concentrated
polystyrene latex dispersions containing grafted poly (ethylene oxide) chains,
J. Colloid Interface Sci. 124 (1988) 660–665.
[27] J.W. Goodwin, R.H. Ottewill, Properties of concentrated colloidal dispersions, J.
Chem. Soc. Faraday Trans. 87 (1991) 357–369.
[28] M.A. Faers, Polymers: their effect on the rheology of concentrated colloidal
dispersions, Thesis, Imperial College London, 1994.
[29] C.G. Kruif, E.M.F. Van Iersel, A. Vrij, Hard sphere colloidal dispersions: viscos-
ity as a function of shear rate and volume fraction, J. Chem. Phys. 83 (1985)
4717–4725.
[30] D. Lin, B. Xing, Adsorption of phenolic compounds by carbon nanotubes: role
of aromaticity and substitution of hydroxyl groups, Environ. Sci. Technol. 42
(2008) 7254–7259.
[31] C.L. Barrie, P.C. Griffiths, R.J. Abbott, I. Grillo, E. Kudryashov, C. Smyth, Rheol-
ogy of aqueous carbon black dispersions, J. Colloid Interface Sci. 272 (2004)
210–217.
[32] J.B. Kayes, D.A. Rawlins, Adsorption characteristics of certain polyoxyethylene-
polyoxypropylene block copolymers on polystyrene latex, Colloid Polym. Sci.
257 (1979) 622.
[33] S.D. Park, D.H. Han, D. Teng, Y. Kwon, Rheological properties and dispersion
of multi-walled carbon nanotube (MWCNT) in polystyrene matrix, Curr. Appl.
Phys. 8 (2007) 482–485.
[34] R.M. Pashley, P.M. McGuiggan, B.W. Ninham, Attractive forces between
uncharged hydrophobic surfaces: direct measurements in aqueous solution,
Science 229 (1985) 1088.
[35] P. Attard, J.L. Parker, P.M. Claesen, Bubbles, cavities and long ranged attraction
between hydrophobic surfaces, J. Phys. Chem. 98 (1994) 8468.
[36] P. Attard, Thermodynamic analysis of bridging bubbles and a quantitative com-
parison with the measured hydrophobic attraction, Langmuir 16 (2000) 4455.
[37] P. Attard, Nanobubbles and the hydrophobic attraction, Adv. Colloid Interface
Sci. 104 (2003) 75.
 S. Yasin, P.F. Luckham / Colloids and Surfaces A: Physicochem. Eng. Aspects 404 (2012) 25–35
[38] M. Musoke, P.F. Luckham, Interaction forces between polyethylene oxide-
polypropylene oxide ABA copolymers adsorbed to the hydrophobic surfaces, J.
Colloid Interface Sci. 277 (2004) 62–70.
[39] J. Nestor, J. Esquena, C. Solans, P.F. Luckham, M. Musoke, B. Lavecke,
K.Th. Booten, F. Tadros, Interaction forces between particles stabilized by a
hydrophobically modified inulin surfactant, J. Colloid Interface Sci. 311 (2007)
430–437.
[40] B.T. Naden, Polymer adsorption to titania surface studied by adsorption
isotherm, rheology and atomic force microscopy, PhD Thesis Imperial College,
London, 2008.
[41] L. Vaisman, H.D. Wagner, G. Marom, The role of surfactants in dispersion of
carbon nanotubes, Adv. Colloid Interface Sci. 128–130 (2006) 37–46.
35
[42] R. Shvartzman-Cohen, E. Nativ-Roth, E. Baskaran, Y. Levi-Kalisman, I. Szleifer,
R. Yerushalmi-Rozen, Selective dispersion of single-walled carbon nanotubes
in the presence of polymers: 251658240 the role of molecular and colloidal
length scales, J. Am. Ceram. Soc. 126 (2004) 14850–14857.
[43] A.N.G. Parra-Vasquez, I. Stepanek, V.A. Davis, V.C. Moore, E.H. Haroz, J. Shaver,
R.H. Hauge, R.E. Smalley, M. Pasquali, Simple length determination of single-
walled carbon nanotubes by viscosity measurements in dilute suspensions,
Macromolecules 40 (2007) 4043–4047.
[44] B. White, S. Banerjee, S. O’Brien, N.J. Turro, I.P. Herman, Zeta-potential mea-
surements of surfactant-wrapped individual single-walled carbon nanotubes,
J. Phys. Chem. C 111 (2007) 13684–13690.