J. Non-Newtonian Fluid Mech.

125 (2005) 101–119

The shear-induced solid–liquid transition in yield stress materials

with chemically different structures
P.H.T. Uhlherra , J. Guoa,∗ , C. Tiua , X.-M. Zhanga , J.Z.-Q. Zhoua , T.-N. Fanga,b
a Department of Chemical Engineering, Monash University, Clayton, Vic. 3800, Australia
b Department of Chemical Engineering, East China University of Science and Technology, Shanghai, China

Received 12 April 2004; received in revised form 23 September 2004; accepted 29 September 2004

Abstract

The shear-induced solid–liquid transition of viscoplastic materials has been studied extensively through various steady-shearing experiments
including steady stress sweep, stress ramp, creep and strain recovery. The results are consistent in showing a clear change in behaviour of the
materials from solid-like to liquid-like. A transition stress plateau separates the regions of solid behaviour and liquid behaviour. The slope
of the plateau reflects the uniformity of the structure, and hence the distribution of bonding strength within this structure. Depending on the
structure of the material, the yielding process of viscoplastic materials can occur over a wide or a narrow range of stress, which represents
“ductile-type” or “brittle-type” failure. Altering the concentration and extent of particle flocculation (for suspensions) or polymer chain
entanglements (for polymer gels) can vary the bonding strength and strength distribution, and therefore change the slope of the stress plateau.
The continuous solid structure exhibits creep at stresses well below the yield stress and fails at a critical strain. The yielding of the material
seems to be characterised by a critical strain rather than by a critical stress or a critical shear rate. The deformation of viscoplastic materials can
be recovered fully or partially in the solid-like region once the stress is removed. This is a significant difference from the behaviour of purely
viscoelastic materials. Strain recovery tests result in two characteristic strain values that can be used to define the two commonly used “yield
stress” values, the higher one of which is in good agreement with the traditional value of yield stress as measured by vane torsion, for example.
© 2004 Elsevier B.V. All rights reserved.

Keywords: Yield stress; Viscoplastic fluid; Suspension; Polymer gel; Creep; Strain recovery; Stress plateau

1. Introduction foods, cosmetics and paints, which must be spreadable but

Very soft (weak) solids such as concentrated mineral
suspensions, drilling muds, solutions of thickeners, many
food stuffs and cosmetics are widely processed on a large
scale. During processing, or use, conditions alternate between
sheared bulk flow in “liquid” form, and gelation or solidifi-
cation under unsheared, stagnant conditions. Solid behaviour
is often a specified requirement for these materials, as in
the “slope disposal” of mineral tailings; in oil drilling-fluids
which must be able to entrap bore-solids within a structured
matrix; long-distance pipeline of suspensions which must be
able to carry large particles without settling; in formulating
∗ Corresponding author.
E-mail address: jian.guo@eng.monash.edu.au (J. Guo).
must not flow once shear is removed.
The strength of such weak solid-like materials has been
characterised traditionally by a yield stress. Although the in-
adequacy of the rheological definition of this property has
generated worldwide debate on the existence of a true yield
stress [1–9], the concept remains useful, and the yield stress
remains an important parameter in rheological characterisa-
tion of complex fluids and in understanding their processing
characteristics in industrial applications. Yield stress is gen-
erally regarded as the transition stress between elastic solid-
like behaviour and viscous liquid-like behaviour, and is re-
lated to the presence of an internal network structure. It was
found that at small deformation up to a limit, there is a linear
zone of stress versus strain and a yield stress as the transition
point from linear to non-linear zones can be determined at

0377-0257/$ – see front matter © 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.jnnfm.2004.09.009
102 P.H.T. Uhlherr et al. / J. Non-Newtonian Fluid Mech. 125 (2005) 101–119
such limit [10]. However, many researchers have suggested
that transition typically occurs not at a single point, but in-
stead over a range of stress starting at a lower limit, corre-
sponding to progressive transition between elastic and plastic
deformation and ultimately ending at a higher limit, corre-
sponding to the transition between plastic deformation and
bulk viscous flow [11–13]. This leads to the definition of two
types of yield stress, i.e. a static and a dynamic yield stress
[14–17]. The presence of a stress plateau has been taken as
indicating the existence of a yield stress, and a redefinition
in terms of this plateau has been suggested [8]. Uhlherr et al.
[13,18] have found that, for traditional yield stress materials,
the stress plateau coincides with an unambiguous transition
from solid-like behaviour to liquid-like behaviour. However,
the presence of a stress plateau is not uniquely associated with
a yield stress but occurs for many shear-thinning polymer so-
lutions as well [8]. Yield stress is an important and much used
rheological concept, but its definition is in urgent need of clar-
ification. In this paper, the transition behaviour of traditional
yield stress materials is investigated extensively by various
steady-shearing tests. The results provide information about
the yielding behaviour of such materials and its dependence
on material structure.
Yield stress materials normally are multiphase systems.
For example, in a suspension, the solid particles are dispersed
in a liquid medium and interact with each other to create a
continuous network with a flocculated structure. The suspen-
sion flows only when the stress is large enough to break down
the network structure. Hence, the yielding process of the ma-
terial must be related to the strength of the coherent network
structure. The yield stress has been analysed in a number of
studies [19–26] in terms of surface chemistry conditions and
physical parameters of the materials, such as zeta potential,
solids concentration, particle diameter and particle size distri-
bution. However, an understanding of the role of the material
structure in the transition process of yield stress materials is
still lacking. As already stated, the transition of rheological
behaviour from solid-like to liquid-like is characterised by
a stress plateau. It is hypothesized here that the steepness of
the stress plateau reflects the uniformity of the solid structure,
and hence the distribution of “bond” strength. Depending on
the width of this distribution, the yielding process could oc-
cur over a wide or narrow range of stress, representing a
“ductile-type” or a “brittle-type” of failure. To test the above
hypothesis, systematic studies have been carried out on the
transition behavior of chemically different structured materi-
als using a range of techniques. The information gained from
this study adds to our understanding of the yielding process
of viscopalstic materials.
2. Materials and experimental work
Materials with chemically different structures have been
formulated including kaolin, TiO2 and SiO2 /TiO2 suspen-
sions, as well as Carbopol 934 gels. The different structures
of these materials are expected to influence the observed
shear-induced transition behaviour. Electrostatic coagulation
is normally considered to be responsible for the bonding of
particles in kaolin and TiO2 suspensions, with bond strength
of the network thus dependent on surface chemistry condi-
tions. The morphology of kaolin and of TiO2 pigment is very
different: TiO2 pigment contains sphere-like particles, while
kaolin is present as plate-like particles. The surface chemistry
and rheology of kaolin suspensions are more complicated
than those of TiO2 suspensions because of the presence of
heterogeneously charged edges and faces on each particle. A
suspension of SiO2 mixed with small amount of TiO2 rep-
resents a different flocculated state again. The incorporation
of small quantities of metal oxides into aqueous silica slurry
can develop strong colloidal interaction, which has a marked
effect on the flow properties leading to a significant increase
in yield stress [27]. The electrostatic attraction between the
small TiO2 particles (0.3 ␮m) and the larger SiO2 particles
(9 ␮m), which are negatively charged at neutral pH, leads
to the formation of large flocs that hold a quantity of water.
Carbopol gel is commonly used as model material in studies
of viscoplastic behaviour because of its stable and transpar-
ent nature. The strength of Carbopol gel may come from the
entanglements and confinement between microgel particles.
A grade A-HR TiO2 pigment from Tioxide Australia Pty
Ltd. was used to prepare the TiO2 suspensions with different
concentrations and pH values. The pigment is made in anatase
form, with no inorganic coating, but treated with organics.
The average crystal size is 0.15 ␮m (supplier’s data). A grade
400G silica flour, which was obtained from Unimin Australia
Ltd., was used for suspensions with small amount of R-HD2
grade TiO2 pigment, which was also from Tioxide Australia
Pty Ltd. Kaolin suspensions were prepared using a product
from Commercial Mineral Pty. Ltd.
The particle size distributions for TiO2 pigments and silica
flour were measured using a Malvern MasterSizer/E. To break
up loosely agglomerated particles, samples were dispersed by
intensive stirring and ultra-sonication. The volume average
particle size is found to be 0.3 ␮m for R-HD2 TiO2 , 0.5 ␮m
for A-HR TiO2 and 9 ␮m for 400G silica flour, respectively.
Zeta potential of the pigments was measured as a function
of pH by electrophoresis using a Malvern Zetasizer 4. The
isoelectric point (IEP) is found at around pH 2.4 for A-HR
TiO2 , pH 6.5 for R-HD2 TiO2 and pH 1.5 for silica flour.
Mixed suspension was formulated with a TiO2 /SiO2 ra-
tio of 0.12. At natural pH, the A-HR TiO2 pigment is well
dispersed in water without significant flocculation, and no
yield stress is detected for the suspension. To prepare a stock
suspension, HNO3 in deionised water was added to create
a suspending medium with low pH, and then TiO2 pigment
was dispersed in this acidic water. The pH of the resulting
suspension was further adjusted by adding more 2 M HNO3 .
The suspension was then gently agitated for at least 24 h to
ensure an equilibrium state. The stock suspension has a con-
centration of 50 wt.% solid and a pH value of 4.8. Suspensions
with different pH values and concentrations were obtained by
 P.H.T. Uhlherr et al. / J. Non-Newtonian Fluid Mech. 125 (2005) 101–119
Table 1
Details of test samples
Sample Composition
Kaolin suspension Kaolin, water
TiO2 suspension A-HR TiO2 , water
SiO2 /TiO2 suspension (1.00:0.12) SiO2 flour, R-HD2 TiO2 , water
Carbopol gels Carbopol 934, water
Separan AP-903 in glycerin solution Separan AP-903, glycerin, water
adding either HNO3 or water. In some cases, higher concen-
tration was achieved by adding more TiO2 pigment.
Carbopol gels with different concentrations were prepared
using Carbopol 934 resin from B.F. Goodrich. Unneutral-
ized Carbopol dispersions have an approximate pH value of
3.0 and have no yield stress. The yield stress of dispersions
gradually increases with the addition of NaOH and peaks at
pH 6-7. Neutralized Carbopol gels were gently stirred for at
least 24 h and then left another 24 h before characterization.
A viscoelastic solution, Separan AP-903 in glycerine, was
prepared for comparison of the behaviour in strain recov-
ery. Details of materials and the yield stress measured by the
conventional vane technique are summarized in Table 1.
A stress-controlled rheometer (DSR, Rheometric Scien-
tific Inc.) equipped with a vane with four blades was used
for all rheological tests in this study. The surface of the vane
was covered with sandpaper to prevent possible slip. Some
variability was observed in creep tests with a smooth vane,
and this was eliminated by roughening the blades. Also a
speed-controlled rheometer (Haake, RV20) was used to mea-
sure the yield stress by the vane torsion method [28]. The
shear-induced solid–liquid transition has been studied by
various steady-shearing techniques including stress sweep,
stress ramp, constant stress creep and strain recovery.
In the stress sweep test, a series of stresses was applied to
the sample in steps, and the angular deflection corresponding
to each stress was measured. The apparent shear strain was
then calculated in terms of the assumption of cylinder geom-
etry in an infinite medium. The time interval for each stress
Fig. 1. Stress–strain and stress–strain rate behavior for 51 wt.% kaolin suspension by stress sweep tests.
103
Concentration (wt.%) Yield stress (Pa)
51.0 45.0
Varied concentration and pH 5–400
70 (total solid) 78.0
Varied concentration and pH 20–200
0.8% Separan AP-903, 55.3% glycerin None
increment varied from 1 s to 10 min. In the stress ramp test,
the shear stress was increased continuously at a constant ramp
rate and the response in strain was measured. The creep test
was conducted by suddenly imposing a constant shear stress
on the sample. Then, the increase in strain was recorded as
a function of time. The recoil of material structure after re-
moving the stress was also measured by the creep and strain
recovery test. For all measurements, the sample was placed
in a container whose diameter and height were at least three
times larger than those of the vane used. Prior to each set
of measurements, the samples were stirred for 5 min using a
spatula. After inserting the vane, the samples were allowed
to rest for an hour. In this way, reproducible initial conditions
of structure were achieved.
3. Results and discussion
3.1. Stress sweep and stress ramp tests
The first set of tests was carried out using a shear-stress
sweep. The curves of shear stress versus shear strain and shear
stress versus shear rate for kaolin suspension and Carbopol
934 gel are shown in Figs. 1 and 2, respectively. Fig. 3 shows
typical plots of shear stress against strain and against strain
rate for 50 wt.% TiO2 suspension using the stress-ramp tech-
nique with different ramp rates. Results for all the different
materials using both these testing techniques exhibit consis-
tent behaviour. The curves of stain–stress show that, at lower
104 P.H.T. Uhlherr et al. / J. Non-Newtonian Fluid Mech. 125 (2005) 101–119

Fig. 2. Stress–strain and stress–strain rate behaviour for 0.5 wt.% Carbopol 934 gel by stress sweep tests.

stress values, the strain is a unique function of stress and inde-
pendent of the rate of stress increment. A linear relation with
a slope approximately equal to unity can be seen in a log–log
plot of stress versus strain, which implies a Hookean solid-
like behaviour. In contrast, at larger stress values, the strain
is not a unique function of stress, but depends on the rate
of stress increase or time between increments. This is con-
sistent with the unbounded deformation of a viscous fluid
under shear. A stress plateau separates these two regions.
Also, it can be seen from the curves of strain rate versus
stress that, at large stress values, all results collapse onto
a single curve, indicating that the shear stress is a unique
function of strain rate. This is conventionally accepted as
fluid behaviour and the existence of a unique viscosity. For
small values, the shear stress varies with the strain rate and
the rate of stress increase. Again, there is a stress plateau
between the two regions. Clearly, The stress plateau repre-
sents the transition from elastic solid-like behavior to viscous
liquid-like behavior, which is associated with the yielding
process of viscoplastic materials. In terms of material struc-
ture, the initial stage represents the elastic deformation of
the network bonds. When the network bonds are increasingly
stretched, some weaker bonds start to break and local viscous
flow occurs. When the stress increases to a certain level, the

Fig. 3. Stress–strain and stress–strain rate behaviour for 50 wt.% TiO2 suspension by stress ramp tests.
 P.H.T. Uhlherr et al. / J. Non-Newtonian Fluid Mech. 125 (2005) 101–119 105

Fig. 4. Development of apparent strain in creep test for kaolin suspension.

majority of the bonds break rapidly in a narrow range of
stress, and then the material begins to flow as a fully viscous
liquid. The stress plateau reflects the breaking process of the
majority of bonds and thus, the slope of the plateau can be
interpreted in terms the uniformity of bond strength. A large
slope where the breaking process occurs over a wide range
of stress, the structure deforms gradually and can be likened
to “ductile-type” failure. Conversely, a small slope reflects
a “brittle-type” sudden failure of the material structure. The
structure of the material will determine the type of yielding
process, reflected in the slope of the stress plateau. Thus,
the slope of stress plateau can be viewed as representing the
uniformity of bond strength. The effect of material structure
on the slope of the stress plateau is discussed further below.
3.2. Creep test
The yielding behaviour of viscoplastic materials has also
been explored using a creep test. A constant stress was ap-
plied to the sample (by means of the roughened, four-bladed
vane), and the increase in strain with time was recorded.
The creep profiles of kaolin suspension, SiO2 /TiO2 suspen-
sion and Carbopol gel at different shear stresses are shown
in Figs. 4–6, respectively. The deformation of a suspension

Fig. 5. Development of apparent strain in creep test for SiO2 /TiO2 suspension.
106 P.H.T. Uhlherr et al. / J. Non-Newtonian Fluid Mech. 125 (2005) 101–119

Fig. 6. Development of apparent strain in creep test for Carbopol 934 gel.

during creep can generally be classified into three regions.
There is an initial elastic response when stress is applied
to the sample, followed by a creep process. After a certain
time, the creep rate increased slowly at first followed by a
sharp upturn, indicating that some viscous component was
involved. A steady strain rate is finally developed correspond-
ing to a flow condition in the material. The region of creep
is shortened with increasing applied stress. Once the stress
reaches a certain level, this region almost disappears. The
results prove an important fact that the yielding process of
the material is time-dependent. A prolonged creep time can
cause the failure of the material structure at lower stresses,
even at well below the conventional “yield stress”. The strain
generated by creep appears to be cumulative, i.e. the rate
of structure regeneration is slower than the rate of structure
breakdown even at very low creeping rate. Thus, the struc-
ture of materials seems to fail at a certain strain regardless of
the level of applied stress. However, the time leading to such
structure failure and catastrophic flow could be very long
when the applied stress is significantly lower than the con-
ventional “yield stress”. The time-dependent behaviour of the
yielding process is directly responsible for the diverse results
observed in the determination of the yield stress using dif-
ferent techniques because of different time frames involved
[14]. In addition, results show that continuous flow exists at
stresses well below the yield stress, but at an extremely slow

Fig. 7. Variation in strain rate over time at different stress for kaolin suspension.
 P.H.T. Uhlherr et al. / J. Non-Newtonian Fluid Mech. 125 (2005) 101–119 107

Fig. 8. Variation in strain rate over time at different stress for Carbopol 934 gel.

shear rate. The shear rate could be of order 10−5 to 10−7 ,
which implies a time frame of weeks or even months. At
such a large time scale, it is appropriate to classify the flow
as “creeping” rather than “viscous flow”. The classification
of this flow causes the controversy over the existence of yield
stress [11,14].
Figs. 7 and 8 show the progression of strain rate over time
in the creep tests. It can be seen that for a very short period
after applying the stress, there is a stress-insensitive region
corresponding to the initial elastic deformation. The strain
rate then decreases steadily as creeping progresses. Depend-
ing on the applied stress, the strain rate could fall dramati-
cally by several orders of magnitude, indicating that the ma-
terial may be trending towards the cessation of flow. The
declining but at an extremely slow shear rate. The declining
course of strain rate shortens with increase in stress. After a
certain time, the strain rate suddenly increases and then ap-
proaches a constant value. This is a clear indication of failure
of the material structure that leads to viscous flow. The re-
sults again point to the fact that there exists neither a critical
stress nor a critical strain rate for causing the catastrophic
flow of materials. The structure failure only occurs once the
accumulated strain reaches a certain value. Therefore, using
a critical strain to characterize the structure failure/yielding
of viscoplastic materials may be more appropriate than a
yield stress. This has been already suggested, for example,
by Ketz et al. [29] and Tempel in Houwink and Decker’s
book [11].

Fig. 9. Critical apparent strain under different stresses for different materials.
108 P.H.T. Uhlherr et al. / J. Non-Newtonian Fluid Mech. 125 (2005) 101–119

Fig. 10. Creep and recovery profile of Carbopol 934 gel subjected to a stress of 4 Pa.

The yield point of a material at different shear stresses can
be determined graphically from the creep profiles shown in
Figs. 4–6 by extrapolating the two straight lines correspond-
ing to solid-like and liquid-like behaviours. The intersection
of the two lines represents the yield point. Fig. 9 plots the vari-
ation of the critical strain corresponding to this yield point
with shear stress for kaolin suspension, silica/titanium diox-
ide suspension and Carbopol 934 gel. It appears that the crit-
ical strain is a weak function of stress for the two particulate
suspensions, while it appears to be independent of the stress
for Carbopol 934 gel. In all cases, the critical strain is of or-
der unity. It must be recalled that strain was calculated for a
cylinder in an infinite medium and is therefore an apparent
value. Different stress levels simply cause catastrophic flow
after different times, once the solid structure has deformed a
characteristic amount.
3.3. Creep and strain recovery
The solid-like behaviour of yield stress materials was fur-
ther studied by carrying out creep and strain-recovery mea-
surements. Figs. 10–12 show the progress of strain build-up
under a constant stress and the course of strain recovery after
the stress was removed for 0.5 wt.% Carbopol 934 gel, which

Fig. 11. Creep and recovery profile of Carbopol 934 gel subjected to a stress of 45 Pa.
 P.H.T. Uhlherr et al. / J. Non-Newtonian Fluid Mech. 125 (2005) 101–119 109

Fig. 12. Creep and recovery profile of Carbopol 934 gel subjected to a stress of 60 Pa.

has a conventional yield stress value of 62 Pa. A constant
shear stress in the range 4–60 Pa was maintained on samples
for periods of 5, 15, 30, 60 and 120 min. The creep curves
generally consist of instantaneous elastic, retarded elastic and
viscous responses. It is obvious that the development of creep
strain depends on both applied stress and creep time. The
strain recovery shows two stages, an instantaneous recov-
ery followed by a gradual recovery. At a stress well below
the yield stress (e.g. 4 Pa), the strain can be recovered fully,
which is typical of purely elastic solid behaviour. At a stress
of 45 Pa, the results show viscoelastic behaviour. Increasing
the time of stress application, increases the extent of viscous
creep of the material showing otherwise solid-like behaviour.
Strain recovery is incomplete and its extent decreases with in-
creasing creep time under applied stress. At a stress of 60 Pa,
approaching the conventional yield stress (62 Pa), the results
show typical purely viscous liquid behaviour. Strain increases
with increasing time under stress and remains at the final
largest value when the stress is removed. There is clearly a
fundamental difference in behaviour between the low stress
and high stress situations, which is entirely consistent with
conventional definitions of the solid and liquid states.
Figs. 13 and 14 show the profiles of creep and recovery for
TiO2 suspension and SiO2 /TiO2 suspension. In both cases,

Fig. 13. Creep and recovery profile of TiO2 suspension subjected to different stresses.
110 P.H.T. Uhlherr et al. / J. Non-Newtonian Fluid Mech. 125 (2005) 101–119

Fig. 14. Creep and recovery profile of SiO2 /TiO2 suspension for different applied stresses.

the recovery behavior is quite different from that of Carbopol
gel. The recoverable strain is remarkably small. It is seen that
the instantaneous recovery of strain after removal of stress is
much smaller than instantaneous elastic deformation on ap-
plication of stress. Even for applied stresses well below the
yield stress, the strain is not recovered fully after the removal
of stress although the retarded strain and viscous flow are neg-
ligibly small compared with the instantaneous elastic strain
jump. Full strain recovery is only observed for extremely
small stresses. This indicates that the relaxation time for the
suspensions is very short and the elastic deformation quickly
relaxes internally.
3.4. Strain recovery ratio
The extent of strain recovery can be expressed by a strain
recovery ratio, which is defined as the ratio of recoverable
strain to total strain. It is found that the strain recovery ratio
is only dependent on the total strain attained at the end of the
creep period. Fig. 15 illustrates the variation of strain recov-
ery ratio with total strain for different materials. Recovery
data on semi-logarithmic coordinates for each material lie on
a straight line irrespective of applied stress and creep time.
The decrease in strain recovery ratio from 100 to 0% repre-
sents a change of the materials’ behaviour from elastic solid

Fig. 15. Strain recovery ratio in terms of strain for different materials.
 P.H.T. Uhlherr et al. / J. Non-Newtonian Fluid Mech. 125 (2005) 101–119 111

Fig. 16. Determination of yield stresses from strain recovery ratio for Carbopol gel.

to viscous liquid. Due to the difference between the struc-
tures of suspensions and carbopol gel, the variation of strain
recovery with total strain is quite different. The network in
suspensions is formed by the aggregation and flocculation of
particles with relatively inflexible and short range structure,
while the network structure in carbopol gel is mainly formed
by closely-packed microgel particles with dangling free ends
of polymer strands, which entangle with neighboring micro-
gel particles forming a flexible, longer range structure. There-
fore, the structure of Carbopol gel can sustain a larger elas-
tic deformation than that of kaolin suspension before plastic
deformation occurs. They represent the two different types
of failure: “ductile” and “brittle”. Recovery behaviour of a
viscoelastic liquid, 0.9 wt.% Separan AP-903 in 55.3 wt.%
glycerin, is also included in Fig. 15 for comparison. It can be
seen that the viscoelastic behaviours of yield stress materials
and viscoelastic liquid is quite different. A full recovery of
strain can always be attained for yield stress materials. How-
ever, only a partial recovery in strain can ever be achieved for
the viscoelastic liquid.
Two critical values of strain can be identified if a straight
line is used to represent the recovery curve. Strain recovery is
expected to be complete if the total strain in creep is less than
the lower critical value; whereas little or no strain recovery is
observed if the total strain exceeds the higher critical value.
The curve of strain recovery ratio versus strain can be com-
bined with the curve of stress versus strain. Thus, two yield
stresses can be determined from these critical strains repre-
senting full recovery and no recovery. Figs. 16 and 17 show
two examples. It can be seen that, for “brittle” kaolin suspen-
sion, the lower value of yield stress determined from 100%
recovery is well away from the stress plateau perhaps close to
the limit of elastic deformation, while, for “ductile” Carbopol
gel, the two yield stresses determined by 100 and 0% recov-

Fig. 17. Determination of yield stresses from strain recovery ratio for kaolin suspension.
112 P.H.T. Uhlherr et al. / J. Non-Newtonian Fluid Mech. 125 (2005) 101–119
ery fall near the limits of the stress plateau. The higher value
of stress is in good agreement with the traditional value of
yield stress as measured using vane torsion. The two stresses
obtained from recovery ratio can be interpreted as the upper
limit of solid-like behaviour and the lower limit of liquid-
like behaviour, representing the two yield stresses, static and
dynamic [14–17].
3.5. Stress plateau
The transition of rheological behaviour of viscoplastic
materials from solid-like to liquid-like is characterized by a
stress plateau in a plot of shear stress versus strain. Yielding
of the material can occur over a wide range of stress, which
leads to a stress plateau with a large slope, or over a narrow
range of stress, which results in a stress plateau with a small
slope. Therefore, the slope of the stress plateau may reflect
the uniformity of the structure. It is expected that the nature
of inter-particle forces will determine the yielding process of
materials, and thus, the slope of the stress plateau. Variation
of the stress plateau with bonding strength of material was in-
vestigated using TiO2 suspensions and Carbopol gels, which
are chemically very different materials and hence structures.
The bonding strength of the materials was controlled by vary-
ing concentration, extent of particle flocculation or polymer
chain entanglements.
As a colloidal suspension, the rheological properties of
TiO2 suspension are directly associated with the surface
chemistry condition of the particles. The TiO2 pigment used
(A-HR) in this study was organically treated. However, there
is no detailed information on what organic species was added
in the manufacturing stage. Thus, the surface chemistry con-
dition of the pigment used is unclear. However, the zeta poten-
tials of the pigment provide a qualitative measure of the mag-
Fig. 18. pH dependence of zeta potential and yield stress for organically treated TiO2 suspension.
nitude and sign of the surface charges on the colloidal parti-
cles in suspension. The isoelectric point (IEP) of pH 2.4 deter-
mined by microelectrophoresis measurement is significantly
different from the IEP of pH 5.5 for un-coated anatase TiO2
particles [30]. This indicates that particles are highly nega-
tively charged and have dense anionic ions on the surface,
hence the interaction between particles should be governed
by the electrostatic forces. The total inter-particle potential
energy between particles is expected to be the sum of van der
Waals attraction and electrical double layer repulsion accord-
ing to DLVO theory of colloid stability [31,32]. Increasing
the strength of H+ (decreasing the pH value of suspension)
compresses the electrical double layer, thus, reducing the re-
pulsive forces and resulting in the flocculation of the particles.
At the isoelectric point, repulsive forces disappear and van
der Waals attractive forces dominate the particle interaction,
hence, the suspension is expected to be in a strongly floccu-
lated state and has a maximum yield stress. Conversely, at the
pH values where zeta potential is very large, the suspension
is expected to be well dispersed, exhibiting negligible yield
stress. The variations of zeta potential and yield stress of TiO2
suspensions with decreasing pH are shown in Fig. 18. The pH
value where a maximum yield stress occurred is quite close
to the isoelectric point determined by electrokinetic measure-
ments. This confirms that pH corresponding to the maximum
yield stress represents the condition of net zero surface charge
on the TiO2 particles, and hence only van der Waals attractive
forces are present in the suspension. At pH values away from
the isoelectric point, the yield stress decreases rapidly as the
electrostatic forces become more important. The results of
yield stress were obtained using the vane torsion technique
with 50 wt.% TiO2 suspensions.
Colloidal forces play a significant role in the yielding be-
havior of TiO2 suspensions. For a given particle size and
 P.H.T. Uhlherr et al. / J. Non-Newtonian Fluid Mech. 125 (2005) 101–119 113

Fig. 19. Effect of pH on transition behaviour of 50 wt.% TiO2 suspensions.

distribution, the van der Waals forces are essentially con-
stant, whilst the electrostatic forces are highly sensitive to
the pH of the medium. A variation in the electrostatic forces
with changing pH can lead to a dramatic change in the inter-
action between particles, which in turn profoundly alters the
yielding behavior of the suspension.
Fig. 19 shows the strain–stress curves of 50 wt.% TiO2 sus-
pensions with different pH values at a ramp rate of 0.1 Pa/s,
while Fig. 20 extracts only the stress plateau of the curves in
Fig. 19. Clearly, the slope of the stress plateau decreases as
pH decreases towards the isoelectric point. This means that
the stress range of the yielding process becomes narrower.

Fig. 20. Stress plateau of 50 wt.% TiO2 suspensions with different pH values.
114 P.H.T. Uhlherr et al. / J. Non-Newtonian Fluid Mech. 125 (2005) 101–119
A narrow stress range for the yielding process implies more
uniform bonding strength within the network, hence, result-
ing in a sudden “brittle-type” structure failure. Conversely, a
wide stress range for the yielding process implies a progres-
sive and partial breakdown of particle bonds, which is caused
by a wide distribution of bonding strength.
It should be noted that the meaning of words “brittle”
and “ductile” used here is quite different from the common
definition in solid mechanics. Solid materials are normally
classified as either being ductile or brittle according to their
stress–strain characteristics. Generally say, any material that
can be subjected to large strain before it ruptures is called
a ductile material, while materials that exhibit little or no
yielding before failure are referred to as brittle materials. The
criterion of determining the ductility of material is the strain
variation. For the yield stress materials discussed here, the
“ductile-type” failure means that beyond the elastic deforma-
tion, the material can still be subjected a wider range of shear
stress before the structure is completely broken down and
material flows as a viscous fluid. If the structure of material
breaks down almost instantaneously once the stress reaches a
certain value, this type failure may be called a “brittle-type”
failure. Thus, the criterion of determining the brittleness of
material would be the stress range in the yielding process,
which is reflected by the slope of the stress plateau.
A comparison of the variations with pH of the slope of the
stress plateau and the yield stress is plotted in Fig. 21. The
results indicate that reducing pH of the suspension towards
the iso-electric point can increase not only the bond strength
(position of the stress plateau), but also the uniformity of
bond strength (slope of the stress plateau). This is because
the distribution of surface charge on particles may be non-
uniform. To improve the dispersability in various media, the
TiO2 pigment is commonly treated with different organics
Fig. 21. Variation of slope of stress plateau with pH for 50 wt.% TiO2 suspensions.
in the micronization stage of production. Cheever [33] stud-
ied the adsorption of some organic surfactants on rutile TiO2
surfaces using the electron microscopy and found that the
organics adsorb in patches or on preferred sites on the par-
ticle surface rather than as uniform monolayers. Hence, the
electrostatic repulsive forces acting on particles could also
be not uniform, resulting in non-uniform bond distribution.
With increasing concentration of H+ , the anionic charges on
particles due to the presence of organics gradually decrease
towards zero when the isoelectric point is reached. Hence,
the electrostatic repulsive forces gradually disappear and the
van der Waals attractive forces eventually dominate the in-
teraction between particles, resulting in more uniform bond
strength. As a result, the structure of TiO2 suspension fails
more suddenly, in a “brittle” manner.
At a well-defined condition of surface chemistry and given
particle size and distribution, the volume fraction of solids de-
termines the concentration of particle–particle bonds and the
density of network junctions, and thus, the strength of the
network. Therefore, the yield stress of suspensions increases
sharply with an increase in the particulate volume fraction.
In this study, effects of solids concentration have been inves-
tigated over a wide range of volume fraction at pH values of
2.4 and 4.4, corresponding to the conditions of the isoelec-
tric point and a weakly flocculated network. These constant
surface chemistry conditions enable the study to focus on the
purely structural effects on the network strength and bonding
uniformity of flocculated suspensions.
Fig. 22 shows the variation of the slope of the stress plateau
and vane yield stress with increasing volume fraction at differ-
ent surface chemistry conditions. Varying the volume fraction
of suspension can change not only the bonding strength of the
network, but also the uniformity of bonding strength. It can
be seen that yield stress increases with increasing volume
 P.H.T. Uhlherr et al. / J. Non-Newtonian Fluid Mech. 125 (2005) 101–119
Fig. 22. Effect of volume fraction on slope of stress plateau and yield stress of TiO2 suspensions.
fraction of suspension, but the slope of the stress plateau
varies in the opposite direction. It can be assumed that the
bonds between particles in suspensions are heterogenous in a
less concentrated suspension. Some particles may have more
neighbours and more bonds, thus, stronger local bonding
strength. On the other hand, other particles may have weaker
bonding strength because of fewer bonds with neighbours.
When the network of a suspension is subjected to a shearing
force, the parts with weaker bonding strength will be bro-
ken down first, hence exhibiting a gradual yielding process.
However, with increasing particulate volume fraction, parti-
cles are more crowded together and the number of neighbours
that bond with any given particle in the network is effectively
increased. Thus, the local bonding strength is increased and
the interaction between particles becomes more homogenous.
An electron photograph of anatase TiO2 pigment shown in
another study [28] has suggested that the particles are dis-
persed in water as small flocs. The results from the analysis
of the particle size distribution also indicate that aggregation
was present in our TiO2 pigment. If this is the case, the system
can be viewed as a series of weakly interconnected aggregate
structures comprising strongly interacting primary particles
[24]. The forces holding these aggregates together are vari-
able. The weak linkages between flocs can be broken down
under a lower stress than those bonds within the aggregate.
However, with increasing solids concentration, particles are
becoming more crowded together and the number of such
weak links diminishes gradually.
Fig. 23 plots the slope of the stress plateau against yield
stress for different pH conditions for TiO2 suspensions of
various concentrations. It is clearly indicated that the surface
chemistry determines the magnitude of the slopes. Decreas-
115
ing pH towards the isoelectric point leads to a significant
change in the “brittleness” of the structure. With the same
values of yield stress, the strongly flocculated structures, cor-
responding to the isoelectric point, fail more abruptly than
those that are weakly flocculated.
Carbopol resins are high molecular weight polyacrylic
acid, cross-linked with polyalkenyl polyether. Upon disper-
sion in water, the tightly-coiled molecule begins to hydrate
and partially uncoil. This initial action only leads to a slight
change in the rheological properties. A maximum viscosity
and yield stress can only be achieved once the molecules are
completely uncoiled. The most common way is to neutralize
the Carbopol molecules using inorganic bases such as NaOH.
Neutralization ionizes the molecules and generates negative
charges along the backbone of the polymer. These negative
charges generate electrostatic repulsion and hence, cause the
molecules to uncoil more into an extended structure. It was
commonly recognized that the unique rheological properties
of Carbopol gels are due to interchain entanglement in the
early studies [34,35]. More recently, Carbopol gel has been
described as microgel dispersion, which consists of highly
cross-linked (interior), enormously swollen and deformable
particles [29,36–39]. The particles are similar to sponges full
of water, consist of a matrix of macromolecular coils and a
“filler” of included water. They become close packed beyond
an overlap concentration. In the close packed configuration,
the particles are sterically confined by their neighbors. In ad-
dition, the exterior of particles is covered with dangling free
ends of gel strands, and these interact strongly with their cor-
responding neighbors on adjacent microgel particles to form
a network. Beyond the overlap concentration, microgel dis-
persions display typical viscoplastic behavior and possess a
116 P.H.T. Uhlherr et al. / J. Non-Newtonian Fluid Mech. 125 (2005) 101–119

Fig. 23. Comparison of uniformity of bonding strength at different surface chemistry conditions for TiO2 suspensions of various concentrations.

yield stress. This hypothesis seems to be more favorable for
the explanation of yield stress possessed by Carbopol gels.
Our studies have found that Carbopol gels exhibit a critical
yield strain in the same order of magnitude as that shown
by some mineral suspensions [13]. Complete or partial strain
recovery can be achieved if the applied stress is below the
yield stress [18]. The findings support the picture of micro-
gel ordering and flow in a close-packed lattice, which was
suggested in the cellular model [29,40].
Figs. 24 and 25 show the effects of neutralization on the
yield stress and slope of the stress plateau of Carbopol gels,
respectively. It can be see that the yield stress of Carbopol gel
is very sensitive to pH, but the slope of the stress plateau is al-
most unaffected. When the gel is neutralized, the yield stress
increases sharply. This is because as the carboxyl groups are
progressively ionized, the intra-molecular repulsion of the
charges extends the coils and results in dramatic swelling
of Carbopol particles. Hence, the particles become closer

Fig. 24. Effect of pH on yield stress for Carbopol 934 gels.

 P.H.T. Uhlherr et al. / J. Non-Newtonian Fluid Mech. 125 (2005) 101–119 117

Fig. 25. Effect of pH on slope of stress plateau for Carbopol 934 gels.

packed as neutral pH is approached. In terms of the cellular
model [29,40], the yield stress can be defined microscopi-
cally as the stress required to move the swollen and packed
particles past one another in the “cellular lattice”. The closer
packed the particles, the larger the stress required to deform
the structure permanently. As a result, neutralized Carbopol
gels have a higher yield stress. However, the distribution of
bonding strength of the network seems not to improve sig-
nificantly with neutralization, as suggested by the lack of
dependence of stress-plateau slope on pH.
Effect of resin concentration on the yielding behaviour has
also been studied for neutralised Carbopol gels. The yield
stress increases steadily with increasing concentration, while
the slope of the stress plateau decreases with increasing con-
centration (see Fig. 26). Increasing the concentration of Car-
bopol resin effectively increases the coil density in the gel.

Fig. 26. Effect of concentration on slope of stress plateau and yield stress.
118 P.H.T. Uhlherr et al. / J. Non-Newtonian Fluid Mech. 125 (2005) 101–119
Fig. 27. Comparison of slope of stress plateau for TiO2 suspensions and Carbopol 934 gels.
Thus, the swellable volume of each particle decreases, and
the size of particles in the gel is eventually reduced. The
particles become harder and less deformable. This may con-
tribute to the increased uniformity of structure, and hence
more uniform bonding strength. In addition, the chain entan-
glements between particles caused by the dangling ends at the
surface of the particles increase significantly with increasing
concentration, which also enhances the interaction between
particles. Therefore, with increasing the concentration, the
gel would become harder, tougher and more brittle.
TiO2 suspension and Carbopol gel have chemically dif-
ferent structures. It is interesting to compare their yielding
behaviors. Fig. 27 shows a comparison of the slopes of the
stress plateau for TiO2 suspensions and Carbopol gels. The
pH of TiO2 suspensions is at the isoelectric point, while Car-
bopol gels are fully neutralized. Hence, the network structures
of both materials have maximum strength. It is apparent that
the uniformity of the bonding strength for the two materials is
significantly different. This implies different types of yield-
ing processes occurring in TiO2 suspensions and Carbopol
gels. TiO2 suspensions consist of undeformable particles and
are flocculated strongly by van der Waals attractive forces.
They show more brittle behavior, and the failure of the struc-
ture occurs in a narrow stress range. By contrast, Carbopol
gels exhibit some ductility due to deformable particles and
entanglements of polymer chains between particles, which
can withstand relatively larger strain before breakdown of
the network occurs.
4. Conclusion
There exists a clearly defined stress plateau separating an
elastic region and a viscus flow region for the viscoplastic
materials studied. This plateau represents the transition in
rheological behaviour from solid-like to liquid-like. The fail-
ure of the materials’ structure can occur at a stress well below
the conventional yield stress if sufficient time is allowed, and
appears to occur at a critical strain rather than at a critical
stress. A clear distinction in viscoelastic behaviour between
viscoplastic materials and viscoelastic fluids is exhibited
through their creep and strain recovery behaviour. Complete
strain recovery can always be achieved for viscoplastic ma-
terials if the applied stress is sufficiently small, while the de-
formation of viscoelastic material can only be recovered par-
tially even at the smallest applied stress. Strain recovery tests
for viscoplastic materials result in two “critical” strain values
that are found to correspond to the top and bottom limits of
the stress plateau. Two yield stresses, “static” and “dynamic”,
can therefore be defined in terms of these strain values.
The slope of the stress plateau reflects the uniformity of
the structure, and hence the distribution of bonding strength.
Depending on the inter-particle forces, the yielding process
can occur over a wide or a narrow range of stress, which im-
plies a “ductile-type” or “brittle-type” of failure, respectively.
The structures of TiO2 suspensions, which are dominated by
electrostatic forces, are more “brittle” than those of Carbopol
gels, which are determined by the deformability of particles
and entanglements between particles.
The uniformity of bonding strength varies with concentra-
tion, and extent of flocculation. For TiO2 suspensions, either
adjusting pH towards the isoelectric point or increasing solids
concentration can enhance the bonding strength and make
bonds more uniform, hence, resulting in a more brittle struc-
ture. For Carbopol gels, concentration is the dominant factor
controlling bond strength uniformity. Higher concentration
leads to less ductile structure.
Acknowledgment
The authors are grateful for financial support from the
Australian Research Council through the ARC large grant
scheme.
 P.H.T. Uhlherr et al. / J. Non-Newtonian Fluid Mech. 125 (2005) 101–119
References
[1] H.A. Barnes, K. Walters, The yield stress myth, Rheol. Acta 24
(1985) 323–326.
[2] J.P. Hartnett, R.Y.Z. Hu, Technical note: the yield stress—an engi-
neering reality, J. Rheol. 33 (1989) 671–679.
[3] H.A. Barnes, A review of the slip (wall depletion) of polymer
solutions, emulsions and particle suspensions in viscometers: its
cause, character, and cure, J. Non-Newtonian Fluid Mech. 56 (1995)
221–251.
[4] H.A. Barnes, The yield stress—a review or ‘␲␣υ␶␣
␲␧␫’—everything flows? J. Non-Newtonian Fluid Mech. 81
(1999) 133–178.
[5] H.A. Barnes, A brief history of the yield stress, Appl. Rheol. 9
(1999) 262–266.
[6] H.A. Barnes, Q.D. Nguyen, Rotating vane rheometry: a review, J.
Non-Newtonian Fluid Mech. 98 (2001) 1–14.
[7] J. Schurz, The yield stress: an empirical reality, Rheol. Acta 29
(1990) 170–171.
[8] I.D. Evans, Letter to the editor: on the nature of the yield stress, J.
Rheol. 36 (1992) 1313–1316.
[9] R.D. Spaans, M.C. Williams, At last, a true liquid-phase yield stress,
J. Rheol. 39 (1995) 241–246.
[10] D. De Kee, R.P. Chhabra, Elastic modulus and yield stress of sus-
pension, Rheol. Acta 33 (1994) 238–240.
[11] R. Houwink, H.K. de Decher, Elasticity, Plasticity and Structure of
Matter, third ed., Cambridge University Press, London, 1971.
[12] A.M. Kraynik, Comments on ER fluid rheology, in: Proceedings of
the Second International Conference on ER Fluids, Lancaster, PA,
USA, 1990.
[13] P.H.T. Uhlherr, X.-M. Zhang, J. Guo, J.Z.-Q. Zhou, C. Tiu, The
shear-induced solid–liquid transition in polymer gels and concen-
trated suspensions, in: Proceedings of the Third Pacific Rim Confer-
ence on Rheology, Vancouver, Canada, 2001.
[14] D.C.-H. Cheng, Yield stress: a time-dependent property and how to
measure it, Rheol. Acta 25 (1986) 542–554.
[15] R.T. Bonnecaze, J.F. Brady, Yield stresses in electrorheological flu-
ids, J. Rheol. 36 (1992) 73–115.
[16] D.M. Husband, N. Aksel, The existence of static yield stresses in
suspensions containing noncolloidal particles, J. Rheol. 37 (1993)
215–235.
[17] E.A. Toorman, Modelling the thixotropic behaviour of dense cohe-
sive sediment suspensions, Rheol. Acta 36 (1997) 56–65.
[18] P.H.T. Uhlherr, Z.Q. Zhou, Stress plateau and yield stress for soft
solids, in: Proceedings of the Second Pacific Rim Conference on
Rheology, Melbourne, 27–31 July, 1997, p. 285.
[19] R.J. Hunter, S.K. Nicol, The dependence of plastic flow behaviour
of clay suspensions on surface properties, J. Colloid Interface Sci.
28 (1968) 250–259.
[20] P.N. Mishra, D.A. Severson, T.C. Owens, Rheological study of con-
centrated silica suspensions, Chem. Eng. Sci. 25 (1970) 653–663.
[21] B.A. Firth, R.J. Hunter, Flow properties of coagulated colloidal sus-
pensions. III. The elastic floc model, J. Colloid Interface Sci. 57
(1976) 266.
119
[22] W.B. Russel, Review of the role of colloidal forces in the rheology
of suspensions, J. Rheol. 24 (1980) 287–317.
[23] Y.-K. Leong, N. Katiforis, D.B. Harding, T.W. Healy, D.V. Boger,
Role of rheology in colloidal processing of zirconia, J. Mater. Pro-
cess. Manuf. Sci. 1 (1995) 445–453.
[24] P.C. Kapur, P.J. Scales, D.V. Boger, T.W. Healy, A theoretical frame-
work for the yield stress of suspensions loaded with size distributed
particles, AIChE J. 43 (1997) 1171–1179.
[25] S.B. Johnson, A.S. Russell, P.J. Scales, Volume fraction effects in
shear rheology and electroacoustic studies of concentrated alumina
and kaolin suspensions, Colloids Surf. A: Physicochem. Eng. Aspects
141 (1998) 119–130.
[26] Z.-W. Zhou, M.J. Solomon, P.J. Scales, D.V. Boger, The yield stress
of concentrated flocculated suspensions of size distributed particles,
J. Rheol. 43 (1999) 651–671.
[27] R.R. Horsley, R.L. Jones, R.J. Snow, The influence of inorganic
impurities on the rheological behaviour of silica based slurries, in:
Proceedings of the IX International Congress on Rheology, Mexico,
1984.
[28] Q.D. Nguyen, D.V. Boger, Direct yield stress measurement with the
vane method, J. Rheol. 29 (1985) 335–347.
[29] R.J. Ketz Jr., R.K. Prud’homme, W.W. Graessley, Rheology of con-
centrated microgel solutions, Rheol. Acta 27 (1988) 531–539.
[30] H.-G. Yang, C.-Z. Li, H.-C. Gu, T.-N. Fang, Rheological behavior
of titanium dioxide suspensions, J. Colloid Interface Sci. 236 (2001)
96–103.
[31] B. Derjaguin, L. Landau, Theory of the stability of strongly charged
lyphobic sols of the adhesion of strongly charged particles in so-
lutions of electrolytes, Acta Physiochem. URSS 14 (1941) 633–
662.
[32] E.J.W. Verwey, J.T.G. Overbeek, Theory of Stability of Lyophobic
Colloids, Elsevier, Amsterdam, 1948.
[33] G.D. Cheever, E.G. Bobalek, Adsorption of some organic surfactants
on rutile surfaces, Ind. Eng. F 3 (1964) 89.
[34] W.H. Fischer, W.H. Bauer, S.E. Wiberley, Yield stress and flow prop-
erties of carboxypolymethylene–water systems, Trans. Soc. Rheol. 5
(1961) 221–235.
[35] B. Testa, J.C. Etter, Application of rheology to the study of inter-
actions between carbopol macromolecules as well as to the semi-
quantitative determination of the ionic strength of their dispersion,
Pharm. Acta Helv. 48 (1973) 378–388.
[36] N.W. Taylor, E.B. Bagley, Dispersions of solutions. A mechanism
for certain thickening agents, J. Appl. Polymer Sci. 18 (1974)
2747–2761.
[37] N.W. Taylor, E.B. Bagley, Rheology of dispersions of swollen gel
particles, J. Polymer Sci. 13 (1975) 1133–1144.
[38] J.O. Carnali, M.S. Naser, The use of dilute solution viscometry to
characterize the network properties of carbopol microgels, Colloid
Polymer Sci. 270 (1992) 183–193.
[39] G.P. Roberts, H.A. Barnes, New measurements of the flow-curves
for carbopol dispersions without slip artefacts, Rheol. Acta 40 (2001)
499–503.
[40] H.M. Princen, Rheology of foams and highly concentrated emul-
sions, J. Colloid Interface Sci. 91 (1983) 160–175.