Colloids and Surfaces

A: Physicochemical and Engineering Aspects 161 (2000) 243 – 257

www.elsevier.nl/locate/colsurfa

Interaction of binders with dispersant stabilised alumina

suspensions
Asad U. Khan, Brian J. Briscoe, Paul F. Luckham *
Particle Technology Group, Department of Chemical Engineering and Chemical Technology, Imperial College of Science,
Technology and Medicine, London, UK

Abstract

The rheological response of selected aqueous alumina suspensions, stabilised with a polyelectrolyte or with an
organic polyvalent salt dispersant, and including poly(vinyl) alcohol (PVA) as a binder, are described in this study.
The polymer dispersant was composed of an ammonium salt of poly(methacrylate) and the organic polyvalent
compound was a sodium salt of an aromatic sulphate. The results show that the addition of PVA, without any
included dispersant does not significantly influence the rheology of the system. However, in the presence of a
dispersant the rheology is greatly affected. At a given concentration of the dispersant, the viscosity, storage and loss
moduli all increase, as the PVA concentration is increased. Also, for a given concentration of the PVA, it is observed
that the viscosity, storage and loss moduli values increase as the concentration of the dispersant is increased. It is
argued that at low PVA concentrations, an excess concentration of the unadsorbed dispersant causes flocculation of
the particles in the suspension by a reduction of the repulsive electrostatic (double layer) effect. In contrast, at higher
concentrations of the PVA the flocculation of the suspension is promoted via a depletion mechanism. © 2000 Elsevier
Science B.V. All rights reserved.

Keywords: PVA; Suspensions; Binders

1. Introduction In this paper, we utilise these ideas to explore the

The rheological properties of concentrated sus-
pensions are complex and much of the early work
was phenomenological in nature. Hunter [1–3]
and Goodwin [4 – 7] were really the first to relate
rheologically determined parameters, the yield
value in the first instance and latter the elastic
modulus, to the microstructure of the suspension.
* Corresponding author. Tel.: +44-171-594-5583; fax: +
44-171-594-5604.
E-mail address: p.luckham01@ic.ac.uk (P.F. Luckham)
rheological characteristics of ceramic suspensions.
Organic binders are an essential component for
the effective processing of many commercial high
performance ceramics. These binders are used to
provide sufficient strength to the body such that
the green bodies can be moulded and retained in
the desired shape without breaking or damage,
before and during sintering. Many ceramic form-
ing processes make use of a binder, including
those of dry pressing, slip casting, tape casting,
extrusion, roll forming, thick film printing, injec-
tion and compression moulding.

0927-7757/00/$ - see front matter © 2000 Elsevier Science B.V. All rights reserved.
PII: S 0 9 2 7 - 7 7 5 7 ( 9 9 ) 0 0 3 7 4 - X
244 A.U. Khan et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 161 (2000) 243–257
There are various organic substances that have
been used, or categorised, as potentially useful
binders for ceramics and many useful compila-
tions and descriptions of these different organic
binders have been published [8 – 17]. They include
poly(vinyl) alcohol, cellulose based materials, nat-
ural gums, starches, sodium and ammonium algi-
nates etc. The use of polymer latex systems has
also been reported recently [18].
The suitability of a binder for a given system
depends upon on a number of factors including
the particle/solvent interactions with a dispersant,
the affinity with the processing liquid, the polarity
and the interaction with the particulate material.
The ‘ideal’ binder should be compatible with the
dispersant, perhaps also function as a stabiliser,
induce no interference with the solvent quality,
function as a lubricant between the particles and
should not produce foaming on air entrainment.
Furthermore, an effective burnout profile without
the formation of a deleterious residue is also
essential. A low glass transition temperature and a
high mechanical strength to molecular weight ra-
tio are also desirable. Naturally, cost is also a
significant factor. Inevitably, the binders used in
practice do not have the properties of the ‘ideal’
binder; a compromise is required. Usually, differ-
ent substances are mixed in order to obtain, or to
approximate to the desired properties. Naturally,
the mixing of the different components greatly
complicates the rheology of the system.
The binders are introduced in a liquid phase,
except in the cases of injection and compression
moulding, and are employed in the formulation of
a particulate system. The liquid phase provides
the vehicle that is necessary for the uniform dis-
persing of the binder molecules. The dry green
strength in the cast body is developed by the
evaporation of the supernatant liquid. The resid-
ual binder remains in place within the body form-
ing the necessary interparticle bridges required to
provide a strong adhesion amongst the ceramic
particles.
The addition of the organic binder often affects
the rheology of the liquid phase. An increase in
the viscosity, as well as a change in flow charac-
teristics, from Newtonian (for water) to shear
thinning, are generally the main consequences.
The changes in the rheology of the binder/liquid
solution directly affect the behaviour of the sus-
pension formed upon the addition of particulates
to the solution.
The usefulness of a given binder for a specific
system depends very much upon the interactions
developed between the different components, e.g.
the dispersant, the particles, etc. of the suspen-
sion. In this paper, the influence of a poly(vinyl)
alcohol (PVA) addition, in the presence of two
different dispersants, upon the rheology of an
alumina AES-11 aqueous suspension is reported.
This binder and these dispersants are frequently
used in commercial practice.
The amount of dispersant and PVA used in this
paper is expressed on a dry weight of the powder
basis (dwb), which means it is equivalent to the
wt./wt. basis of the anhydrous solid.
It has been shown, for various particulate sys-
tems such as polymer latex particles and surfac-
tant micelles, that when particles are dispersed in
a dispersion medium, there are at least four kinds
of interactions operating between the different
components of the suspensions. Van der Waals
attractive forces are the most ubiquitous. The van
der Waals attractive forces are very dominant for
high surface area particulate materials, and cause
aggregation of the particles. If the particulate
material has an electrical charge induced, say
caused by the adsorption of a charged dispersant,
the aggregation of the particles in a suspension
can be suppressed through the formation of an
electrical double layer. Adsorption, or grafting, of
polymer molecules can also prevent aggregation
of the particles by the mechanism of steric stabili-
sation. The addition of a non-adsorbing polymer
to a stabilised suspension may cause depletion
flocculation, whereby the concentration of non-
adsorbing free polymer is effectively reduced in
between the particles, and the higher concentra-
tion outside the particle–particle approach zone,
exerts an osmotic pressure causing the particles to
flocculate; Russel et al. [19], Hunter [20], Napper
[21], Everett [22] and Fleer et al. [23]. Fig. 1 shows
schematically different effects which may cause
instability in a stabilised suspension (central re-
gion). The arrow direction shows the increasing of
that specific effect.
 A.U. Khan et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 161 (2000) 243–257
Fig. 1. Different forces that cause instability in a stabilised suspension. The arrow direction indicates the increasing of the effect.
2. Experimental
2.1. Materials
Alumina AES-11 (Mandoval, Surrey, UK),
99.8% pure, crystalline with a surface area of 8.14
m2 g − 1 (BET values), a mean particle size of 0.4
mm was used as received. Two dispersant systems
were studied, ‘Darvan C’, an ammonium salt of
poly(methacrylate), (25% by wt. polymer in
water of MW 10 000 – 16 000) (R.T. Vanderbilt
Company, USA), and ‘Tiron’ (Fluke Chemicals),
which is sodium salt of 4 – 5 dihydroxy-1, 3-ben-
zene disulfonic acid (C6H8Na2O8S2, MW 332.2)
also called as pyrocatechol-3, 5-disulfonic acid
disodium salt. The chemical structures of these
two dispersants are given in references [24,25].
The PVA was ‘Mowiol 10-98’ (Halow Chemical,
UK), and had an average molecular weight
245
61 000, with a degree of hydrolysis \ 98% (den-
sity=1290 kg m − 3). All the suspensions de-
scribed in this paper were composed of 40% by
volume of alumina and deionised water was used
to prepare the suspensions. The concentration of
these substances (PVA+ dispersants) are ex-
pressed in ‘dwb’, i.e. on dry weight basis of the
alumina powder. It means that if the alumina
powder used is 100 g in the suspension, 1% dwb is
equal to 1 g substance in the suspension.
2.2. Methods
All of the measurements were performed with a
Bohlin VOR rheometer (Bohlin Rheologi, Lund,
Sweden). Before performing rheological charac-
terisation, the suspensions were presheared at a
comparatively high shear rate of 960 s − 1 for
 3–4 min followed by at least a 2 min ‘rest’
246 A.U. Khan et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 161 (2000) 243–257

period in order to attempt to provide a common

and consistent shear history for the systems
[24,25]. The viscosity at low shear rates was mea-
sured first, stepping up to higher shear rates,
covering the shear rate range of  10 − 1 to 103
s − 1. All of the experiments were performed at
25°C. Initially, for the dynamic measurements a
‘strain sweep’ measurement was performed in
which the imposed strain amplitude was varied
while the frequency was maintained constant at 1
Hz. The values of the G% and G%% moduli were
measured as a function of the strain amplitude in
order to identify the so-called ‘linear viscoelastic
region’ in which the viscoelastic properties of the Fig. 2. Viscosity of alumina AES-11 40% v/v suspensions
system were relatively independent of the applied against PVA concentration % dwb (dry weight of powder
strain (deviationB 5% in G% values). After identi- basis) at three different shear rates (SR)/(1.46, 14.6 and 146
fying the linear viscoelastic region (strain range), s − 1); no specific dispersant.
the imposed strain was fixed within this region
and the deformation frequency was varied (usu- 3.2. Dispersant without binder
ally from 0.05 to 10 Hz). This is termed here the
‘oscillatory measurement’. The specific results re- Addition of a dispersant to alumina suspen-
ported in this paper are for a frequency of 1 Hz. sions causes changes in the viscosity of the sus-
pensions. This effect is shown in Fig. 3 as the
viscosity at a shear rate of 1.46 s − 1 against the
3. Results dispersant concentration (see also [25]). The opti-
mum concentration, i.e. the concentration which
3.1. Binders without dispersant gives the lowest viscosity, is 1% dwb for ‘Darvan
C’ and 0.1% (or 0.125%) dwb for ‘Tiron’. Below
The suspensions prepared from the alumina this optimum concentration, the surface coverage
powder and water, without the addition of any of the alumina particles with the dispersant
dispersant or binder, were very flocculated and molecules is not complete, and the particles aggre-
dried very quickly and were therefore not suitable gate due to the domination of van der Waals
for the production of ‘good’ quality ceramic mate- attractive forces. Above this optimum concentra-
rials. The addition of external processing agents tion value there is an excess amount of dispersant
changes the interparticle interactions appreciably,
which changes the rheology of the resulting sus-
pensions [26]. PVA is one of the most frequently
used water soluble binders and it has been re-
ported [27] that PVA molecules adsorb weakly
onto the alumina particles (adsorbed concentra-
tion 0.1 mg m2). Fig. 2 reports the viscosity
against the PVA concentration at three different
shear rates of 1.46, 14.6 and 146 s − 1 for 40% (v/v)
alumina AES-11 suspensions. From the Figure, it
appears that, for the concentration range of the
PVA investigated, the viscosities of the suspen-
sions are virtually unchanged as the PVA concen- Fig. 3. Viscosity at a shear rate of 1.46 s − 1 against dispersant
tration is changed, at a given shear rate. concentration of alumina AES-11 40% v/v suspensions.
 A.U. Khan et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 161 (2000) 243–257 247

which is not adsorbed onto the particles, remains

in the aqueous phase, functions as an electrolyte,
hence reduces the range of electrical double layer
repulsion and, therefore, increase in the viscosity
results. The fact that the optimum concentration
corresponds to the monolayer coverage of the
particles was established using adsorption
isotherms [24,25]. Fig. 3 also shows the point of
critical coagulation concentration (c.c.c) for the
‘Tiron’ case. The c.c.c. was calculated using
DLVO theory (Eq. (1)) [28].
9.85 · 104o 3k 5T 5g 4
c.c.c.= (1)
Nae 6A 2z 6
Fig. 4. Relative viscosity against PVA concentration, for alu-
where Na is Avogadro’s constant, k is Boltz- mina AES-11 40% v/v suspensions stabilised with 1% dwb
mann’s constant, o is permitivity, z is the charge ‘Darvan C’, at three different shear rates (SR)/(1.46, 14.6 and
number of the ion, T is absolute temperature, A is 146 s − 1).
Hamaker’s constant, and e charge on an electron
and g is given by Eq. (2) in which 8d is the surface ogy of the suspensions significantly. Figs. 4 and 5
potential of the particles. represent the relative viscosity, hr (hr = ho, where
hs and ho are the viscosities of the suspension and
exp[ze8d/2kT] −1
g= (2) the continuous medium (PVA+water+ non-ad-
exp[ze8d/2kT] +1 sorbed dispersant), respectively) at three different
To calculate the c.c.c., the surface potential was shear rates (1.46, 14.6 and 146 s − 1) as a function
taken equal to 50 mV which is the measured value of the PVA concentration for the suspensions
of the zeta-potential at the suspension pH ( 9). containing respectively 1% dwb ‘Darvan C’ and
The zeta-potential was measured using an elec- 0.125% dwb ‘Tiron’. These concentration values
trophoresis experiment. Hamaker constant was of the ‘Darvan C’ and ‘Tiron’ dispersants are
taken equal to 6.7×10 − 20 J K − 1. Using these those which provide the lowest viscosity, at a
values, a c.c.c. of 35.9 × 10 − 3 mol dm − 3 was given shear rate, when incorporated in a 40% v/v
estimated which corresponds to a 0.225% dwb
This concentration is above the minimum viscos-
ity concentration of 0.125% dwb which would
corresponds to the maximum stabilisation part of
the dispersant. Adding more dispersant increases
the electrolyte concentration of the bulk and
would be expected to cause of aggregation. Thus
experimentally we estimate the c.c.c. to be above
0.125% dwb of the dispersant. The estimated
value of 0.225% dwb is in accord with this. ‘Tiron’
concentration (‘Tiron’ molecule was considered as
2:1 salt and Na+ ions as counter ions).

3.3. Dispersant – binder combinations

Fig. 5. Relative viscosity against PVA concentration, for alu-
In contrast to the PVA without dispersant case, mina AES-11 40% v/v suspensions stabilised with 0.125% dwb
the addition of PVA to a dispersant stabilised ‘Tiron’, at three different shear rates (SR)/(1.46, 14.6 and 146
alumina AES-11 suspension can change the rheol- s − 1).
248 A.U. Khan et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 161 (2000) 243–257

suspension containing no added binder [24,25].

For both cases, at the lowest shear rate (1.46 s − 1)
the relative viscosity initially increases monotoni-
cally as the PVA concentration is increased from
0 to 0.5% dwb. Thereafter, the relative viscosities
at this shear rate remain virtually unchanged. The
increase in the relative viscosity is approximately
four times for ‘Darvan C’ and about an order of
magnitude for ‘Tiron’, as the PVA concentration
is increased from 0 to 0.5% dwb. At the shear rate
of 14.6 s − 1, the relative viscosity still increases
with the increase of the PVA concentration. How-
ever, the rate of the increase of the viscosity with
the increase of the PVA concentration is less than
that observed at the shear rate of 1.46 s − 1 for the Fig. 6. Relative viscosity against PVA concentration, for alu-
respective suspensions. The increase in the relative mina AES-11 40% v/v suspensions stabilised with different
viscosity at the former shear rate is 2 –3 times concentrations of ‘Darvan C’ (% dwb), at a shear rate 1.46
higher than the later shear rate when the PVA s − 1.
concentration is increased from 0 to 0.5% dwb.
There is a less significant change in the relative caused by an electrical double layer screening
viscosity at the shear rate of 146 s − 1 as the PVA effect, such as that which has been discussed
concentration is varied; the major effects are elsewhere [25]. The increase of the dispersant con-
noted at the lower shear rate. centration in the presence of PVA significantly
The relative viscosity of the alumina suspen- modifies the interactions between the alumina
sions as a function of the PVA concentration at a particles. This result is shown in Figs. 8 and 9 for
shear rate of 1.46 s − 1, is reported in Figs. 6 and ‘Darvan C’ and ‘ Tiron’, respectively. These are
7 for the different concentrations of the disper-
plots of the viscosity at a shear rate of 1.46 s − 1 as
sants ‘Darvan C’ and ‘Tiron’, respectively. For
a function of the dispersant concentration for the
the comparatively low concentrations of the dis-
various concentrations of PVA. The figures show
persants (e.g. for ‘Darvan C’ from 1 to 2% dwb
and for ‘Tiron’ from 0.10 to 0.50% dwb), the
relative viscosity generally increases gradually as
the PVA concentration is increased from 0 to
0.5% dwb, and thereafter levels off or even
slightly decreases. For the 4% dwb concentration
of the ‘Darvan C’ and 1% dwb for ‘Tiron’, the
relative viscosity does not significantly change as
the PVA concentration is increased from 0 to
0.5% dwb. However, increasing the PVA concen-
tration above 0.5% dwb slightly lowers the rela-
tive viscosity with the increase of the PVA
concentrations. As before, the influence of the
dispersants is similar.
If the concentration of the dispersant is in-
creased, in the absence of any binder, above the
optimum amount (i.e. the amount which gives the Fig. 7. Relative viscosity against PVA concentration, for alu-
lowest viscosity), the viscosity of the suspension mina AES-11 40% v/v suspensions stabilised with different
increases, due to the aggregation of the particles concentrations (% dwb) of ‘Tiron’, at a shear rate of 1.46 s − 1.
 A.U. Khan et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 161 (2000) 243–257
Fig. 8. Viscosity as a function of ‘Darvan C’ concentration (%
dwb) for different concentrations of PVA (% dwb), of alumina
AES-11 suspensions at a shear rate of 1.46 s − 1.
that, as the dispersant concentration is increased
for the two dispersants, the viscosity of the alu-
mina suspensions simply increases, for a given
PVA concentration. However, as the concentra-
tion of the PVA is increased, the rate of the
increase in the viscosity with the increasing of the
dispersant concentration decreases. In Figs. 10
and 11, for ‘Darvan C’ and ‘Tiron’ cases, respec-
tively, the viscosities of the 40% alumina AES-11
suspensions, at a shear rate of 1.46 s − 1 are plot-
ted against the PVA concentration for the differ-
ent concentrations of dispersants. The viscosity
values, at this shear rate, increase steadily as the
PVA concentration is increased. Also, for the
Fig. 9. Viscosity as a function of ‘Tiron’ concentration (%
dwb) for different concentrations of PVA (% dwb), of alumina
AES-11 suspensions at a shear rate of 1.46 s − 1.
249
Fig. 10. Viscosity against PVA concentration (% dwb), for
alumina AES-11 40% v/v suspensions stabilised with different
concentrations of ‘Darvan C’ (% dwb), at a shear rate of 1.46
s − 1.
higher concentration values of the dispersant the
viscosities of the alumina suspensions are higher
than those for the lower dispersant concentration
for a given PVA concentration. Figs. 12 and 13
describe the normalised/relative viscosity, hn
against the PVA concentration for each concen-
tration of the dispersant (corresponding to Figs.
10 and 11, respectively) at a shear rate 1.46 s − 1.
This parameter corresponds to the viscosity of the
alumina suspensions, divided by the viscosity of
the suspension having no PVA for the corre-
Fig. 11. Viscosity against PVA concentration (% dwb), for
alumina AES-11 40% v/v suspensions stabilised with different
concentrations of ‘Tiron’ (% dwb), at a shear rate of 1.46 s − 1.
250 A.U. Khan et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 161 (2000) 243–257

Fig. 13. Normalised viscosity as a function of PVA concentra-

tion, for alumina AES-11 suspensions stabilised with different
Fig. 12. Normalised viscosity as a function of PVA concentra- concentrations of ‘Tiron’ (5 dwb), at a shear rate of 1.46 s − 1.
tion, for alumina AES-11 suspensions stabilised with different Inset is plot of same variables for different ranges and on
concentrations of ‘Darvan C’ (5 dwb), at a shear rate of 1.46 different scale.
s − 1.
The results of the oscillatory strain measure-
1 ments, at a frequency of 1 Hz., for the alumina
sponding dispersant concentration suspension.
AES-11 suspensions with 1% dwb ‘Darvan C’ and
The normalised/relative viscosity value, hn as a
0.125% dwb ‘Tiron’ are shown in Figs. 15 and 16.
function of the PVA concentration, increases al-
For the ‘Darvan C’ case, the storage modulus
most linearly. However, as the dispersant concen-
gradually increases as the PVA concentration is
tration is increased successively, the gradient of increased. This increase is particularly pro-
the line (normalised viscosity vs. PVA concentra- nounced for the lower concentrations of PVA.
tion) decreases. Fig. 14 is a plot of the relative The loss modulus is always less than the storage
viscosity against the PVA concentration, for the modulus. However, for ‘Tiron’ suspensions the
0.05 and 2% dwb ‘Tiron’ concentrations. These storage modulus is only less than the loss modulus
concentrations are respectively rather below and for the PVA concentrations less than 0.20% dwb.
much above compared, to the optimum concen- The storage modulus also increases relatively
tration (0.10 – 0.125% dwb) of ‘Tiron’ that steeply in the initial stages as the PVA concentra-
which provides the minimum viscosity. Also, plot-
ted in the same Figure are the data for the opti-
mum concentration (0.125% dwb) of ‘Tiron’ for
comparison. The Figure shows that with the in-
creasing of the PVA concentration there is first a
decrease in the relative viscosity followed by a
steady increase. The effect is more pronounced for
the 0.05% dwb ‘Tiron’ concentration. This viscos-
ity-PVA concentration trend for these ‘Tiron’
concentrations is in stark contrast to the optimum
concentration trend for the ‘Tiron’ system plotted
against the PVA concentration.

1
The normalised/relative viscosity hn = hp/hpo, where hp and
hpo are the viscosities of the suspension with and without PVA
for a given dispersant concentration, respectively. Further-
more, when the PVA concentration is zero, hp/hpo = 1.
Fig. 14. Relative viscosity against PVA concentration, for
alumina AES-11 40% v/v suspensions stabilised with different
concentrations (% dwb) of ‘Tiron’, at a shear rate of 1.46 s − 1.
 A.U. Khan et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 161 (2000) 243–257
Fig. 15. G% and G%% against PVA concentration, for alumina
AES-11 40% v/v suspensions stabilised with 1% dwb ‘Darvan
, G%
C’, at a frequency of 1 Hz.
tion is increased and the slopes of the curves
decreases as the PVA concentration is further
increased. This trend (but not the absolute values)
is similar to that of the viscosity at a shear rate of
1.46 s − 1 that is shown in Figs. 4 and 5 as the
relative viscosity against the PVA concentration
curve. The loss modulus, G%%, and the dynamic
viscosity follow the same general trend as the G%
modulus as the PVA concentration is varied. A
similar trend of the storage modulus against vol-
ume fraction of the polystyrene suspensions was
Fig. 16. Dynamic viscosity, G%, G%% and dynamic viscosity
against PVA concentration, for alumina AES-11 40% v/v
suspensions stabilised with 0.125% dwb ‘Tiron’, at a frequency
, G%
of 1 Hz.
, G% 0.125%, at a frequency of 1 Hz.
251
Fig. 17. G% against PVA concentration, for alumina AES-11
40% v/v suspensions stabilised with different concentrations of
‘Darvan C’ (% dwb): ", G% 4%; , G% 2%; , G% 1.5%;
0.125%, at a frequency of 1 Hz.
observed by other researchers [6], and the chang-
ing shape of the curve (storage modulus versus
volume fraction) was attributed to the changing
geometry of the fractal structure of the particles
formed due to the flocculation.
It is seen in Figs. 17 and 18 that, for all the
adopted dispersant concentrations, the G% elastic
modulus values increase with the increasing of the
PVA concentration. However, the slopes of G%
against the PVA concentration curves decrease
(especially for the PVA concentration between 0
Fig. 18. G% against PVA concentration, for alumina AES-11
40% v/v suspensions stabilised with different concentrations of
‘Tiron’ (% dwb): , G% 0.1%; , G% 0.25%; ", G% 0.5%;
1%;
252 A.U. Khan et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 161 (2000) 243–257

Fig. 21. Storage modulus against PVA concentration, for

Fig. 19. G%% against PVA concentration, for alumina AES-11
alumina AES-11 40% v/v suspensions stabilised with different
40% v/v suspensions stabilised with different concentrations of
concentrations of ‘Tiron’, at a frequency of 1 Hz.
‘Darvan C’ (% dwb):
, 1%; , 1.5%; , 2%; ", 4%, at a
frequency of 1 Hz.

tion which is shown in Figs. 19 and 20 follows

and 0.5% dwb) as the dispersant concentration is
raised from 1 to 4% dwb for ‘Darvan C’ and from
0.1 to 1.0% dwb for ‘Tiron’. A similar trend was
observed in the computed viscosity values at a
shear rate of 1.46 s − 1 which is expressed as the
relative viscosity against the PVA concentration
for the different concentrations of dispersant
(Figs. 6 and 7). At the higher PVA concentra-
tions, the G% moduli for all of the concentrations
of the dispersants approach each other. The loss
modulus, G%%, as a function of the PVA concentra-
Fig. 20. G%% against PVA concentration, for alumina AES-11
40% v/v suspensions stabilised with different concentrations of
‘Tiron’ (% dwb):
, G%% 0.1%; , G%% 0.25%; , G%% 0.5%; ", against PVA concentration data for the same
G%% 1%;
, G%% 0.125%, at a frequency of 1 Hz.
nearly the same trend as that observed in the G%
against PVA profile for the various concentrations
of dispersant. There are a few exceptions, notably
for the 4% dwb ‘Darvan C’ and 0.05% dwb PVA,
and 1% ‘Tiron’ and 0.2% PVA suspensions, for
which there is a sudden decrease in the loss mod-
ulus value. A similar type of decrease, but to a
lesser degree, was also observed, in the G% values.
The difference between the G% and the G%% values
is almost equal to one order of magnitude
throughout with the G%% values being always less
than the G% values, for the ‘Darvan C’ concentra-
tions of 1.5% dwb and above and for ‘Tiron’
concentrations greater than 0.125% dwb. In gen-
eral, the dynamic viscosity, at a frequency of 1
Hz., and the viscosity at the shear rate of 1.46 s − 1
are observed to be similar, for suspensions of the
similar compositions.
For suspensions containing 0.05 and 2% dwb
‘Tiron’, the storage modulus as a function of PVA
concentration is shown in Fig. 21. The storage
moduli data, for 0.125% dwb ‘Tiron’, are also
plotted in the same figure for the purpose of
comparison. The storage modulus against PVA
concentration curves follow the same general
trend as that shown by the relative viscosity
concentrations of ‘Tiron’ (Fig. 14).
 A.U. Khan et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 161 (2000) 243–257
4. Discussion
The increase in the viscosity of the dispersant
stabilised suspensions at the shear rate of 1.46
s − 1, and the G%, G%% values as well as the dynamic
viscosity at the frequency of 1 Hz., with increasing
of the PVA concentration suggests that the addi-
tion of the PVA causes a flocculation of the
alumina stabilised particles with the 1% dwb
‘Darvan C’ and 0.125% dwb ‘Tiron’ dispersant
concentrations. The degree of the flocculation in-
creases as the PVA concentration is increased,
and this effect of flocculation is more pronounced
for the PVA concentration range between 0 and
0.5%. Above a 0.5% PVA concentration, there is
now little change in the degree of flocculation. As
was discussed earlier, it was found that PVA does
not adsorb, to any measurable extent, onto the
alumina AES-11 particles. Therefore, the prospect
of a flocculation of this system by a bridging
mechanism can be reasonably eliminated. The
alumina AES-11 particles, in the presence of
aqueous ‘Darvan C’ and ‘Tiron’ solutions, are
highly surface charged. This was confirmed by the
use of electrophoresis experiments [21,22]. It was
also observed, in a few test cases, that the addi-
tion of a simple salt (KNO3), in the ‘Darvan C’
and ‘Tiron’ stabilised alumina AES-11 suspen-
sions containing different concentrations of PVA,
enhances the extent of, flocculation. This indicates
that these particles are stabilised largely by a
classical electrostatic repulsion mechanism and
that this flocculation is due to the compression of
the electrical double layer giving rise to the domi-
nation of van der Waals attractive forces. How-
ever, in the case of stabilised particles in the
presence of non-adsorbing PVA, the weak floccu-
lation that is observed is unlikely to be due to van
der Waals forces as the particles are stable in the
absence of the PVA. It is therefore highly proba-
ble, based upon elimination, that the aggregation
of the ‘Darvan C’ and ‘Tiron’ stabilised alumina
suspensions is caused by a depletion flocculation
process when PVA is added to these suspensions.
It has been reported [29] that hydroxyethyl cellu-
lose (HEC) of MW 70 000 and more does not
adsorb to polystyrene latex particle, causes deple-
tion flocculation of the particles at a concentra-
253
tion starting from  0.2% by volume. These
concentrations are comparable with those values
reported here where we observe an increase in low
shear viscosity and G% at PVA concentration
0.05% dwn ( 0.1% by volume (see Figs. 4, 5, 15
and 16).
When the dispersant concentration in the alu-
mina AES-11 suspensions, without any PVA ad-
dition, is increased beyond 1% dwb for ‘Darvan
C’ and 0.125% dwb for ‘Tiron’ (the optimum
concentrations), the alumina particles commence
aggregating; this aggregation is due to the fact
that any dispersant introduced above the opti-
mum concentration level remains unadsorbed and
then functions as an electrolyte which reduces the
range and extent of the electrostatic repulsion, as
was discussed earlier in Section 2. When the PVA
is added to the alumina suspension, containing
the dispersant above the optimum concentration,
there are at least two kinds of interactions in-
volved. One causes flocculation, due to the reduc-
tion in the repulsion between the particles (excess
dispersant), and the other is the process where
‘free’ PVA brings about the flocculation probably
through a depletion mechanism. As a combined
result of these two effects, the slope of the relative
viscosity (and viscoelastic properties) against PVA
concentration curves gradually decreases as the
dispersant concentration is increased (Figs. 6 and
17 for ‘Darvan C’, and Figs. 7 and 18 for ‘Tiron’).
For the 4% ‘Darvan C’ and 1% ‘Tiron’ concentra-
tion level alumina suspensions, there is virtually
no change in the slope of relative viscosity against
PVA concentration curves. It is reasonable to
conclude that the degree of flocculation, caused
by the electrolyte effect (an excess of dispersant),
is so high that the addition of the PVA does not
greatly change the nature of the flocculation.
The results given in Figs. 8 and 9 may be
explained as follows. The slope of the viscosity, at
the shear rate of 1.46 s − 1, against the dispersant
concentration curves gradually decreases as the
PVA concentration is increased. The change (de-
crease) of the slope of the viscosity against disper-
sant concentration, as the PVA concentration is
increased from 0 to 1.5% dwb, signifies that, at
low concentrations of PVA, the aggregation of
the particles is principally brought about by the
254 A.U. Khan et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 161 (2000) 243–257
electrostatic mechanism. However, as the PVA
concentration is increased the depletion mecha-
nism predominates.
The results shown in Figs. 12 and 13 also
highlight the two mechanism of flocculation of
these alumina AES-11 suspensions. The nor-
malised viscosity, without any PVA addition, is
taken as unity. For the 1% dwb ‘Darvan C’ and
0.125% dwb ‘Tiron’ suspensions without any
PVA, the particles are well dispersed and there is
a low level of aggregation. As the PVA concentra-
tion is increased from 0 to 1.5% dwb there is an
approximate 25-fold increase for ‘Darvan C’ sus-
pensions, and  100 times for ‘Tiron’ suspen-
sions, in the viscosity; and this increase in the
viscosity is brought about primarily by the aggre-
gation of particles through a depletion mechanism
as was explained above. However, as the disper-
sant concentration is increased the increase in the
normalised viscosity with the increase of the PVA
concentration (the slope of normalised viscosity
against PVA concentration) decreases. This is due
to the fact that, at the higher ‘Darvan C’ concen-
trations, there is already some aggregation of the
primary particles brought about by the reduction
in the electrostatic stabilisation effect. Therefore,
the increase in the normalised viscosity with the
increasing of the PVA concentration is reduced.
For the highest concentration of the dispersants
(4% dwb for ‘Darvan C’ and 1% dwb ‘Tiron’)
reported in Figs. 12 and 13, the effect is a small
increase in the normalised viscosity as the PVA
concentration is changed from 0 to 1.5% dwb.
The same conclusion can be drawn from the
examination of the viscoelastic properties as was
described earlier.
For the 0.05% ‘Tiron’ concentration case, the
surface concentration of the ‘Tiron’ molecules
adsorbed onto the alumina is not sufficient to
completely cover the surface of the alumina AES-
11 particles. At this less than optimum concentra-
tion of the ‘Tiron’, the van der Waals attractive
forces are significant enough to cause the aggrega-
tion of the particles. It has also been observed
that the addition of PVA to the alumina AES-11
suspensions, without any dispersant, does not infl-
uence the rheology. As a result, it may be con-
cluded that PVA is a non-adsorbing polymer for
the present system. Therefore, one expects that
the PVA will not change the rheology of the
alumina AES-11 suspensions stabilised with
0.05% dwb ‘Tiron’. However, the results (Figs. 14
and 21) show that this is not the case. The relative
viscosity drops as the PVA concentration is in-
creased. Despite these changes, the suspensions
remain well flocculated. These changes in the rhe-
ological properties are probably due to changes
induced in the structure of the flocs, an effect that
may not be readily quantified.
For the case where the ‘Tiron’ concentration
was 2.0% dwb, which is  20 times higher than
the optimum concentration ( 0.10–0.125%
dwb), the behaviour is rather different. At this
high concentration of ‘Tiron’, the suspensions
were highly flocculated and the relative viscosity
and G% values were the highest as compared to the
other ‘Tiron’ concentrations used for the alumina
AES-11 suspensions. The addition of the PVA
slightly lowers the value of the relative viscosity
and the other rheological properties. However,
this decrease is comparatively small and the sus-
pensions remain very flocculated after the addi-
tion of PVA. This decrease in the rheological
parameters, with the increasing of the PVA con-
centration, may be due to the changes in the
consolidation of the floc structure. The effect is
again not amenable to a quantitative prediction.
To summarise the discussion of the results of
the two dispersant-PVA combinations for the alu-
mina AES-11 suspensions; the addition of the
PVA increases the degree of aggregation and this
aggregation of particles is probably brought
about by the depletion flocculation mechanism.
However, as the dispersant concentration is in-
creased above the optimum value (1% dwb for
‘Darvan C’ and 0.125% dwb ‘Tiron’) in the pres-
ence of the PVA, there are two kinds of mecha-
nisms involved which cause the flocculation of the
particles: (1) the reduction of the electrostatic
stabilisation effect caused by the excess amount of
the free dispersant which functions as an elec-
trolyte; and (2) the depletion effect of the non-ad-
sorbing PVA between the particles. At very high
concentrations of the dispersant, the reduction in
the electrostatic repulsion mechanism dominates
due to the stronger degree of flocculation of the
 A.U. Khan et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 161 (2000) 243–257
particles induced by the excess electrolyte. De-
pletion flocculation is known, in comparison, to
be a relatively weak interaction.
5. Comparison of the two dispersant – poly(vinyl)
alcohol combinations
For the two dispersants, ‘Darvan C’ and
‘Tiron’ used to stabilised the alumina AES-11
suspensions, the viscosity at a shear rate of 1.46
s − 1 for the respective dispersant’s optimum con-
centrations, as a function of the PVA concentra-
tion, is shown in Fig. 22. This figure shows that
as the PVA concentration is increased, the vis-
cosities of the suspension stabilised with the two
, ‘Tiron’ and , ‘Darvan C’) concentrations at their opti-
dispersants start increasing. This increasing of
the viscosities, with increasing of the PVA con-
centrations, is due to the alumina particles ag-
gregating, which is probably caused by a
depletion flocculation mechanism. This effect on
the viscosity, caused by changing the PVA con-
centrations, is greater in the ‘Tiron’ stabilised
system compared to the ‘Darvan C’ stabilised
suspensions. To explain this point further the
normalised/reduced viscosity, hn, (hn = hs/h0
where hs and h0 are the viscosities of the sus-
pensions with and without PVA, respectively) is
plotted against the PVA concentration in Fig.
23. It is clear from an inspection of the Figure
Fig. 22. Viscosity against PVA concentration of alumina AES-
11 40% v/v suspensions for two different dispersants (‘Darvan
C’ and ‘Tiron’) at their optimum concentrations, at a shear
rate of 1.46 s − 1.
255
Fig. 23. Normalised viscosity against PVA concentration of
alumina AES-11 40% suspensions for three different dispersant
(
mum concentration. Also normalised viscosity of PVA solu-
tions against PVA concentration.
that the normalised/reduced viscosity is the
higher for the ‘Tiron’ suspensions than for the
suspensions containing ‘Darvan C’; the degree
of depletion flocculation, with respect to an in-
crease in the PVA concentration, is greater in
the case of the ‘Tiron’ stabilised suspensions. In
Fig. 23, the normalised viscosity2 of the
aqueous–PVA solution is also plotted as a func-
tion of PVA concentration (upper X-axis). In
the case of the aqueous–PVA solution, there is
no depletion effect and the increase of the vis-
cosity of the solution is caused by the PVA–sol-
vent interactions alone. However, in the case of
suspensions there is an addition effect i.e. deple-
tion flocculation, and therefore the slopes are
different for the three systems. The upper X-axis
expresses the PVA concentration in g l − 1 of the
continuous medium for the different 40% (v/v)
suspensions containing the PVA. A comparison
of the viscoelastic properties for the two disper-
sants and the PVA combinations indicates a
similar (to viscosity) level for the PVA interac-
tions in the three systems.
2
The normalised/reduced viscosity, hn of aqueous PVA
solution is hn =hp/hw where hp and hw are the viscosities of
the PVA solution and water, respectively.
256 A.U. Khan et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 161 (2000) 243–257
6. Conclusion
In this paper, selected rheological properties of
alumina suspensions in the presence of a PVA
binder and two dispersants ‘Darvan C’ and
‘Tiron’ have been described. When the PVA was
used without any dispersant, the resulting viscos-
ity of the 40% by volume suspensions remained
virtually unaffected for the 0 – 1.5% dwb PVA
concentration range. The behaviour of many dif-
ferent concentration combinations of the PVA
and the dispersants has been described. The use of
‘Darvan C’ and ‘Tiron’ as dispersants with PVA
for the 40% (v/v) suspensions progressively in-
creases the relative viscosity, at a shear rate of
1.46 s − 1, as the PVA concentration is 0–0.5%
dwb. Above a PVA concentration of 0.5% dwb,
there is virtually no change in the relative viscos-
ity. At a higher shear rate (=146 s − 1), the rela-
tive viscosity is almost constant for the 0–1.5%
PVA concentration range. For a given PVA con-
centration, the viscosity at the shear rate of 1.46
s − 1 for ‘Tiron’ system is less than that of the
‘Darvan’ system. As the concentration of the dis-
persants is increased above the optimum value,
the viscosity of the resulting alumina suspensions
is higher than the viscosity of the next lower
concentration level of the dispersant, for a given
value of the PVA concentration. The viscoselastic
behaviour, i.e. the G%, G%% parameters and dy-
namic viscosity against the PVA concentration,
show similar trends to the corresponding viscosity
against PVA concentration behaviour. The high
viscosities noted for the ‘above optimum’ disper-
sant concentrations were consistent with the ac-
tion of an electrostatic shielding mechanism. The
excess amount of the dispersant remains in the
continuous medium and functions as an elec-
trolyte and ‘compresses’ the range of the double
layer and hence changes the rheological proper-
ties. The increase of the viscosity with the increas-
ing of the PVA concentration is due the increase
of interparticle interactions within the system and
these enhanced interactions cause an aggregation
of the particles by a depletion mechanism. The
addition of the PVA causes increase in the relative
viscosity of both the ‘Darvan C’ and ‘Tiron’
stabilised alumina suspensions. This increase of
the relative viscosity is probably due to a deple-
tion flocculation mechanism. A 0.5% dwb PVA
concentration gives the maximum interactions
which becomes constant above this PVA concen-
tration. Thus, in summary, the rheological re-
sponses of these systems may be rationalised using
established wisdom; Fig. 1 is a summary. The
most marked effect arise from the electrostatic
screening induced by excess dispersant addition
but excessive polymer addition also promote de-
pletion flocculation. In the case of both, an opti-
mal amount of material is required to provide the
most satisfactory formulation.
The system we have studied is a complex one
such as those frequently encountered industrially.
That we can explain the rheological data in terms
of the microstructure of the system is testament to
the work of Dr Goodwin and his colleagues who
have been pre-eminent in trying to explain what
rheological data mean. Without his pioneering
studies, we could not have been so bold.
References
[1] B.A. Frith, R.J. Hunter, J. Colloid Interface Sci. 57
(1976) 248.
[2] B.A. Frith, R.J. Hunter, J. Colloid Interface Sci. 57
(1976) 257.
[3] B.A. Frith, R.J. Hunter, J. Colloid Interface Sci. 57
(1976) 266.
[4] R. Buscall, J.W. Goodwin, M.W. Hawkins, R.H. Ot-
tewill, J. Chem. Soc. Faraday Trans. I 78 (1982) 2873.
[5] R. Buscall, J.W. Goodwin, M.W. Hawkins, R.H. Ot-
tewill, J. Chem. Soc. Faraday Trans. I 78 (1982) 2889.
[6] R. Buscall, P.D.A. Mills, J.W. Goowin, D.W. Lawson, J.
Chem. Soc. Faraday Trans. I 84 (1988) 4249.
[7] J.W. Goodwin, R.H. Ottewill, J. Chem. Soc. Faraday
Trans. 87 (1991) 357.
[8] J.W. Whittemore, Am. Ceram. Soc. Bull. 23 (1944) 427.
[9] E.P. McNamara, J.E. Comefora, J. Am. Ceram. Soc. 28
(1945) 25.
[10] C.C. Treischel, E.W. Emrich, J. Am. Ceram. Soc. 29
(1946) 129.
[11] A. Wild, Am. Ceram. Soc. Bull. 33 (1954) 368.
[12] T. Knapp, Am. Ceram. Soc. Bull. 33 (1954) 11.
[13] H. Thurnauer, in: W.D. Kingery (Ed.), Ceramic Fabrica-
tion Processes, Wiley, New York, 1958.
[14] S. Levine, Ceram. Age 75 (1960) 39.
[15] S. Levine, Ceram. Age 75 (1960) 25.
[16] T.A. Smith, Trans. Br. Soc. 61 (1962) 523.
[17] A.R. Teter, Ceram. Age 82 (1966) 30.
 A.U. Khan et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 161 (2000) 243–257
[18] R.W. Greenwood, Research Report, E.U., E.H.C.M. Pro-
gramme, Preparation of Concentrated Aqueous Alumina
Suspensions Suitable for Tape Casting, I.R.T.E.C.,
Faenza, Italy.
[19] W.B. Russel, D.A. Saville, W.R. Schowalter, Colloidal
Dispersion, Cambridge University Press, Cambridge,
1989, p. 258.
[20] R.J. Hunter, Foundations of Colloid Science, Oxford
Science Publications, 1991.
[21] D.H. Napper, Polymeric Stabilisation of Colloidal Dis-
persions, Academic Press, 1983.
[22] D.H. Everett, Basic Principles of Colloids, Royal Society
of Chemistry, 1989.
[23] G.H. Fleer, M.A. Cohen Sturat, J.M.H.M. Scheutjens, T.
Cosgrove, B. Vincent, Polymer at Interfaces, Chapman
and Hall, London, 1993.
257
[24] A.U. Khan, PhD Thesis, Imperial College of Science and
Technology and Medicine, London, 1997.
[25] B.J. Brisocoe, A.U. Khan, P. Luckham, J. Eur. Ceram.
Soc. 18 (1998) 2141.
[26] G.Y. Onoda, Jr., in: G.Y. Onoda, L.L. Hench (Eds.),
Ceramic Processing Before Firing, Wiley-Interscience
Publication, New York, 1978.
[27] B.C. Bonekamp, W.H. Van’T Van, M.J. Schoute, H.J.
Veringa, in: G. de With, R.A. Terpstra, R. Metselaar
(Eds.), EuroCeramic, vol. I, Elsevier Applied Science,
London, 1989, p. 1.145.
[28] D.J. Shaw, Introduction to Colloid and Surface Chem-
istry, fourth ed., Butterworth-Heinemann, Oxford, 1992,
pp. 225 – 227.
[29] W. Liang, Th. F. Thadros, P.F. Luckham, J. Colloid
Interface Sci. 160 (1993) 183.