03.07.2013 Views

Dynamic properties of shear thickening colloidal suspensions

Dynamic properties of shear thickening colloidal suspensions

Dynamic properties of shear thickening colloidal suspensions

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Rheol Acta (2003) 42: 199–208<br />

DOI 10.1007/s00397-002-0290-7<br />

Young Sil Lee<br />

Norman J. Wagner<br />

Received: 22 June 2002<br />

Accepted: 20 November 2002<br />

Published online: 5 February 2003<br />

Ó Springer-Verlag 2003<br />

Y. S. Lee Æ N. J. Wagner (&)<br />

Department <strong>of</strong> Chemical Engineering<br />

and Center for Composite Materials,<br />

University <strong>of</strong> Delaware, Newark,<br />

DE19716, USA<br />

E-mail: wagner@che.udel.edu<br />

Introduction<br />

When subjected to increasing <strong>shear</strong> stress, concentrated<br />

<strong>colloidal</strong> <strong>suspensions</strong> can exhibit a steep rise in viscosity<br />

(Lee and Reder 1972; H<strong>of</strong>fman 1974, 1997; Barnes<br />

1989). This <strong>shear</strong> <strong>thickening</strong> phenomenon can damage<br />

processing equipment and induce dramatic changes in<br />

suspension microstructure, such as particle aggregation,<br />

which results in poor fluid and coating qualities. On the<br />

other hand, this behavior can be exploited in the design<br />

<strong>of</strong> damping and control devices, whereby the fluid can<br />

limit the maximum rate <strong>of</strong> flow through a highly nonlinear<br />

response (Laun et al. 1991; Helber et al. 1990).<br />

It has been demonstrated that reversible <strong>shear</strong><br />

<strong>thickening</strong> in concentrated <strong>colloidal</strong> <strong>suspensions</strong> is due<br />

to the formation <strong>of</strong> jamming clusters bound together by<br />

hydrodynamic lubrication forces, <strong>of</strong>ten denoted by the<br />

term ‘‘hydroclusters’’ (Bossis and Brady 1989; Farr et al.<br />

1997; Foss and Brady 2000; Catherall et al. 2000). The<br />

microstructure <strong>of</strong> <strong>shear</strong>ing <strong>suspensions</strong> has been studied<br />

ORIGINAL CONTRIBUTION<br />

<strong>Dynamic</strong> <strong>properties</strong> <strong>of</strong> <strong>shear</strong> <strong>thickening</strong><br />

<strong>colloidal</strong> <strong>suspensions</strong><br />

Abstract The transient <strong>shear</strong> rheology<br />

(i.e., frequency and strain dependence)<br />

is compared to the steady<br />

rheology for a model <strong>colloidal</strong> dispersion<br />

through the <strong>shear</strong> <strong>thickening</strong><br />

transition. Reversible <strong>shear</strong> <strong>thickening</strong><br />

is observed and the transition<br />

stress compares well to theoretical<br />

predictions. Steady and transient<br />

<strong>shear</strong> <strong>thickening</strong> are observed to<br />

occur at the same value <strong>of</strong> the average<br />

stress. The critical strain for<br />

<strong>shear</strong> <strong>thickening</strong> is found to depend<br />

inversely on the frequency at fixed<br />

applied stress for low frequencies<br />

(high strains), but is limited to an<br />

apparent minimum critical strain at<br />

higher frequencies. This minimum<br />

critical strain is shown to be an artifact<br />

<strong>of</strong> slip. Lissajous plots illustrate<br />

the transition in material<br />

<strong>properties</strong> through the <strong>shear</strong> <strong>thickening</strong><br />

transition, and the energy<br />

dissipated by a <strong>shear</strong> <strong>thickening</strong><br />

suspension is analyzed as a function<br />

<strong>of</strong> strain amplitude.<br />

Keywords Shear <strong>thickening</strong> Æ<br />

Suspension rheology Æ Colloid Æ<br />

Dispersions Æ Dilatancy<br />

by rheo-optical experiments (D’Haene et al. 1993;<br />

Bender and Wagner 1995), neutron scattering (Laun et<br />

al. 1992; Bender and Wagner 1996; Newstein et al. 1999;<br />

Maranzano and Wagner 2001a, 2002) and stress-jump<br />

rheological measurements (Kaffashi et al. 1997). The<br />

onset <strong>of</strong> <strong>shear</strong> <strong>thickening</strong> in steady <strong>shear</strong> can now be<br />

quantitatively predicted (Maranzano and Wagner<br />

2001a, 2001b) for suspension <strong>of</strong> hard-spheres and electrostatically<br />

stabilized dispersions.<br />

Although a significant number <strong>of</strong> experimental and<br />

simulation studies have addressed <strong>shear</strong> <strong>thickening</strong> in<br />

steady <strong>shear</strong> flow, only limited work has addressed the<br />

viscoelastic <strong>properties</strong> <strong>of</strong> a <strong>shear</strong> <strong>thickening</strong> fluid (Laun<br />

et al. 1991; Boersma et al. 1992; Raghavan and Khan<br />

1997; Mewis and Biebaut 2001). This is partly a consequence<br />

<strong>of</strong> the difficulty <strong>of</strong> studying <strong>shear</strong> <strong>thickening</strong>,<br />

which is a highly nonlinear response that is <strong>of</strong>ten triggered<br />

by relatively high stress levels.<br />

Nonlinear oscillatory <strong>shear</strong> experiments are useful for<br />

characterizing the onset <strong>of</strong> <strong>shear</strong> <strong>thickening</strong> as well as for


200<br />

determining the time scales required to generate the<br />

<strong>shear</strong> <strong>thickening</strong> response. Of interest for technological<br />

applications is the time scale and the minimum strain<br />

required to generate the hydrocluster microstructure<br />

underlying the <strong>shear</strong> <strong>thickening</strong> response. For example,<br />

electrostatically stabilized dispersions have been investigated<br />

in both steady and dynamic oscillatory <strong>shear</strong><br />

flows for possible damping applications. Laun et al.<br />

(1991) reported the dynamic strain hardening behavior<br />

<strong>of</strong> a polymer latex dispersion. Their oscillatory test<br />

protocol increased the strain amplitude at a given frequency,<br />

which leads to a <strong>shear</strong> <strong>thickening</strong> rheology.<br />

The critical strain for dynamic <strong>shear</strong> <strong>thickening</strong> (cc) was observed to decrease with increasing frequency<br />

(x), but eventually plateaued at higher frequencies.<br />

The low frequency behavior was interpreted in terms<br />

<strong>of</strong> the steady <strong>shear</strong> behavior, where a critical <strong>shear</strong> rate<br />

( _c dynamic<br />

c _c steady<br />

c ) must be achieved to thicken. The high<br />

frequency limiting value suggested that a minimum<br />

<strong>shear</strong> strain ( 50%) is necessary in each half cycle to<br />

cause the dispersion to switch to the high viscosity state.<br />

Similar conclusions for low frequencies were reached by<br />

Boersma et al. (1992), who investigated monodisperse<br />

silica particles suspended in a mixture <strong>of</strong> glycerol and<br />

water. They reported ‘‘flow blockage’’ in oscillatory<br />

testing, which was also related to steady <strong>shear</strong> <strong>thickening</strong><br />

at low frequencies. Intermediate frequencies yielded a<br />

weaker frequency dependence, but no plateau value.<br />

Finally, at high frequencies the critical deformation for<br />

<strong>shear</strong> <strong>thickening</strong> was found to be independent <strong>of</strong> particle<br />

volume fraction, but again scaled inversely with<br />

frequency. This latter behavior was attributed to the<br />

solid-like response for a sample that is fully in the<br />

hydrocluster state. Note that these experiments were<br />

also performed on a controlled strain rheometer and<br />

the critical strain amplitudes were <strong>of</strong> order O(10 -2 ) at the<br />

highest frequencies.<br />

Studies <strong>of</strong> near hard sphere dispersions (Bender<br />

1995) also confirm the agreement between steady <strong>shear</strong><br />

<strong>thickening</strong> and the low frequency dynamic oscillatory<br />

response. Raghavan and Khan (1997) observed similar<br />

congruence at low frequencies, as well as a high frequency<br />

limiting critical strain (cc O(1)) for fumed silica<br />

dispersions in poly(propylene glycol). Recently,<br />

Mewis and Biebaut (2001) also observed dynamic<br />

<strong>shear</strong> <strong>thickening</strong> in sterically stabilized <strong>colloidal</strong> <strong>suspensions</strong>.<br />

They observed that the peak <strong>shear</strong> stress at<br />

the onset <strong>of</strong> <strong>shear</strong> <strong>thickening</strong> in oscillatory flow corresponds<br />

to the same steady <strong>shear</strong> stress measured at<br />

the onset <strong>of</strong> <strong>shear</strong> <strong>thickening</strong>, with no evidence <strong>of</strong> a<br />

limiting critical strain down to <strong>shear</strong> strains <strong>of</strong> order<br />

0.5. Notably, Mewis and Biebaut (2001) also investigated<br />

the <strong>shear</strong> thickened state by parallel superposition,<br />

observing a viscoelastic liquid response in the<br />

<strong>shear</strong> thickened state, not the solid response suggested<br />

by Boersma et al. (1992).<br />

In summary, the literature data to date suggests that<br />

the onset <strong>of</strong> strain hardening at low frequencies for<br />

concentrated <strong>suspensions</strong> in oscillatory <strong>shear</strong> flow can<br />

be interpreted in terms <strong>of</strong> the onset <strong>of</strong> steady <strong>shear</strong><br />

<strong>thickening</strong>. However, there is contradictory evidence as<br />

to whether a critical strain is required for oscillatory<br />

strain hardening, and as to whether a third, solid-like<br />

regime exists at higher frequencies. This issue is relevant<br />

for the design <strong>of</strong> devices based on the <strong>shear</strong><br />

<strong>thickening</strong> response (Helber et al. 1990). The goal <strong>of</strong><br />

this work is to relate the nonlinear viscoelastic <strong>properties</strong><br />

to the steady <strong>shear</strong> response for a <strong>shear</strong> <strong>thickening</strong><br />

fluid, and to determine if a minimum critical<br />

strain is necessary for <strong>shear</strong> <strong>thickening</strong>. This is achieved<br />

by rheological investigation <strong>of</strong> a model dispersion. Of<br />

particular interest is the critical strain amplitude required<br />

for <strong>shear</strong> <strong>thickening</strong> in dynamic <strong>shear</strong> flow and<br />

its dependence on frequency. Finally, Lissajous plots<br />

are constructed to illustrate the ‘‘switching’’ from liquid<br />

to solid observed during deformation, and to determine<br />

the energy dissipation’s depend on strain amplitude in<br />

a <strong>shear</strong> <strong>thickening</strong> fluid.<br />

Experimental<br />

Sample preparation and characterization The <strong>colloidal</strong> silica investigated<br />

here was obtained from Nissan Chemicals (MP4540), which<br />

is provided as an aqueous suspension (pH=8.5 at 25 °C) with a<br />

particle concentration <strong>of</strong> about 40 wt%. The particle size distribution<br />

has been characterized with dynamic light scattering and<br />

TEM. Figure 1 shows a transmission electron micrograph <strong>of</strong> the<br />

suspension; which is observed to contain a minor fraction <strong>of</strong><br />

smaller particles. The average particle diameter (z-average) was<br />

determined to be 446±8.4 nm by dynamic light scattering, which<br />

agrees with the TEM measurements <strong>of</strong> the large particle fractions.<br />

The solution density <strong>of</strong> the particles has been obtained by measuring<br />

the density <strong>of</strong> the suspension as a function <strong>of</strong> weight fraction<br />

<strong>of</strong> the particles. The weight fraction <strong>of</strong> silica was determined<br />

gravimetrically after drying the sample at 180 °C for 5 h using a<br />

convection oven. The density <strong>of</strong> the silica calculated from this<br />

method is 1.78 g/cc. The zeta potential has been determined to be<br />

)32 mV from electrophoresis measurements (Brookhaven Zeta-<br />

PALS) at pH=8.5 and CSALT=0.045 mmol/l. This suspension was<br />

concentrated by tabletop centrifugation. The sediment was resuspended<br />

using a vortex mixer after adding <strong>of</strong> small amount <strong>of</strong> the<br />

supernatant liquid. Dilution with the mother liquor provided a<br />

series <strong>of</strong> aqueous silica <strong>suspensions</strong>. The suspending fluid was also<br />

replaced with ethylene glycol (EG) by repeated centrifugation and<br />

resuspension with a vortex mixer. This process has been repeated<br />

four times to prepare a second series <strong>of</strong> dispersions <strong>of</strong> the same<br />

particles in ethylene glycol.<br />

Rheological measurements The experiments were performed primarily<br />

in a stress-controlled rheometer (SR-500, Rheometrics) at<br />

25 °C with cone-plate geometry having a cone angle <strong>of</strong> 0.1 radian<br />

and a diameter <strong>of</strong> 25 mm. Complementary measurements were<br />

performed on a Rheometrics ARES controlled strain rheometer. A<br />

parallel plate geometry was also used with varying gap size to<br />

characterize slip. To prevent adhesive slip between the sample and<br />

the rheometer plates, parallel plates <strong>of</strong> diameter 25 mm were covered<br />

with emery cloth (NORTON, E-Z FLEX METALITE K224)<br />

using double stick tape. The gaps explored varied between 0.05 mm


and 1.5 mm. A solvent trap prevented evaporation <strong>of</strong> the solvent.<br />

Three types <strong>of</strong> dynamic experiments have been performed: stress or<br />

strain sweeps at constant frequency, frequency sweeps with a<br />

constant imposed stress, and time transient measurement <strong>of</strong> strain<br />

response with fixed stress amplitude and frequency. To remove<br />

loading effects, a pre<strong>shear</strong> <strong>of</strong> 1 s -1 was applied for 60 s prior to<br />

further measurement. All measurements presented here were<br />

reproducible.<br />

Results and discussion<br />

Result <strong>of</strong> stress and frequency sweeps<br />

Figure 2 shows the steady <strong>shear</strong> viscosity and the complex<br />

viscosity as a function <strong>of</strong> the steady <strong>shear</strong> stress or<br />

average dynamic <strong>shear</strong> stress, respectively, for the<br />

aqueous suspension at volume fractions <strong>of</strong> u=0.55 (a)<br />

and 0.60 (b). Shear <strong>thickening</strong> is evident in both steady<br />

and dynamic measurements. To confirm the reversibility<br />

<strong>of</strong> the <strong>shear</strong> <strong>thickening</strong> behavior, the complex viscosity<br />

was measured for both ascending and descending stress<br />

sweeps, with good agreement.<br />

Shear <strong>thickening</strong> is known to be a stress controlled<br />

phenomena; hence one might expect to be able to superimpose<br />

steady <strong>shear</strong> viscosity and complex dynamic<br />

viscosity when plotted against applied stress. The aver-<br />

Fig. 1 Transmission electron microscopy <strong>of</strong> <strong>colloidal</strong> silica obtained<br />

from Nissan Chemicals (MP4540) at a magnification <strong>of</strong><br />

40,000<br />

201<br />

age dynamic <strong>shear</strong> stress applied to the fluid during oscillatory<br />

testing is obtained by integrating the absolute<br />

value <strong>of</strong> the applied stress over one cycle as<br />

sd ¼ x<br />

Z 2p=x<br />

jsdjdt ¼<br />

2p 0<br />

2s0<br />

ð1Þ<br />

p<br />

where sd=s0cosxt and s0 is the maximum stress. Note<br />

that the root mean square value, i.e.,<br />

Fig. 2a,b Reversible <strong>shear</strong> <strong>thickening</strong> behavior <strong>of</strong>: a 55 vol.%; b<br />

60 vol.% <strong>colloidal</strong> silica dispersed in water for both steady and<br />

dynamic <strong>shear</strong>ing plotted against the average applied dynamic<br />

<strong>shear</strong> stress


202<br />

sRMS d ¼<br />

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi<br />

R<br />

x 2p=x<br />

2p 0 s2 ddt q<br />

¼ sdffiffi p 1:11sd, could be used<br />

2<br />

with equal superposition to within the accuracy <strong>of</strong> our<br />

measurements. However, the data do not superimpose<br />

when plotted against the maximum applied stress, or the<br />

<strong>shear</strong> rate. At the point <strong>of</strong> <strong>shear</strong> <strong>thickening</strong>, the dynamic<br />

viscosity is found to be comparable to the steady <strong>shear</strong><br />

viscosity measured at the same average stress. Deviations<br />

between the complex and steady <strong>shear</strong> viscosities<br />

are evident at both low and high <strong>shear</strong> stresses, as will be<br />

discussed below.<br />

Using the model developed by Maranzano and<br />

Wagner (2001a, 2001b), the critical <strong>shear</strong> stress for <strong>shear</strong><br />

<strong>thickening</strong> (sc) can be predicted from independent<br />

measurements <strong>of</strong> the particle size, concentration, surface<br />

potential, and ionic strength. The equation for the critical<br />

<strong>shear</strong> stress for electrostatically stabilized dispersions<br />

is<br />

sc ¼ 0:024 kBT ðjaÞW 2<br />

ð2Þ<br />

a 2 lb<br />

where l b is the Bjerrum length defined by l b ” e 2 /<br />

(4p 0k BT), a is the radios <strong>of</strong> the particle, and Y s=w se/<br />

k BT is the dimensionless surface potential. The theoretical<br />

predictions for s c are found to be in good agreement<br />

with the measured values (Table 1). In this table h m is<br />

the characteristic separation distance in the incipient<br />

hydrocluster state, which is determined directly from<br />

j )1 , the Debye length by h m=1.453/j.<br />

The extended Cox-Merz rule equates the steady and<br />

dynamic viscosities at equivalent <strong>shear</strong> rate and frequency.<br />

As our data is highly nonlinear, this relationship<br />

is observed not to hold. For materials with slow relaxing<br />

microstructure, a modified Cox-Merx rule has been<br />

proposed by Doraiswamy et al. (1991) known also as the<br />

‘‘Delaware-Rutgers’’ rule. The basis <strong>of</strong> this approach is<br />

that slowly relaxing materials respond to the highest<br />

applied strain rate applied during the dynamic measurement.<br />

Consequently dynamic and steady <strong>properties</strong><br />

overlay when the highest <strong>shear</strong> rate experienced during<br />

the oscillation is taken as the effective steady <strong>shear</strong> rate<br />

ðc 0x _c). Raghavan and Khan (1997) use this rule to<br />

correlate the steady <strong>shear</strong> <strong>thickening</strong> to strain <strong>thickening</strong><br />

for their fumed silica suspension. In Fig. 3 we compare<br />

data obtained from a frequency sweep experiment with<br />

imposed maximum stress <strong>of</strong> 10 Pa to steady <strong>shear</strong> data<br />

in the sprit <strong>of</strong> the ‘‘Delaware-Rutgers’’ rule. It is<br />

apparent that the ‘‘Delaware-Rutgers’’ rule applies in<br />

Table 1 Comparison <strong>of</strong> experimental and theoretical critical <strong>shear</strong><br />

stress foraqueous silica particle <strong>suspensions</strong><br />

Volume<br />

fraction /<br />

ja hm<br />

(nm)<br />

Theoretical critical<br />

<strong>shear</strong> stress sc (Pa)<br />

0.55 8.4 39 37 20±0.5<br />

0.60 9.4 35 41 46±0.5<br />

Experimental critical<br />

<strong>shear</strong> stress sc (Pa)<br />

the region <strong>of</strong> <strong>shear</strong> thinning for the suspension <strong>of</strong> 60%<br />

silica particle, but that the <strong>shear</strong> <strong>thickening</strong> transitions<br />

do not superimpose. Closer inspection <strong>of</strong> the data <strong>of</strong><br />

Raghavan and Khan (Fig. 10 <strong>of</strong> their work) shows a<br />

similar level <strong>of</strong> disagreement upon <strong>shear</strong> <strong>thickening</strong>.<br />

Consequently neither extended Cox-Merz nor the Delaware-Rutgers<br />

rule correlates dynamic and steady viscosities<br />

in the <strong>shear</strong> <strong>thickening</strong> regime. This is further<br />

confirmed by Mewis and Biebaut (2001), who report<br />

qualitative differences in the parallel superposition viscoelastic<br />

measurements between <strong>shear</strong> thinning and<br />

<strong>shear</strong> <strong>thickening</strong> parts <strong>of</strong> the flow curve. Thus, although<br />

the ‘‘Delaware-Rutgers’’ rule applies to the <strong>shear</strong> and<br />

frequency thinning regions <strong>of</strong> the flow curve, it fails to<br />

correlate the viscosity in the <strong>shear</strong> thickened state or the<br />

onset <strong>of</strong> the <strong>shear</strong> <strong>thickening</strong> transition.<br />

This result is consistent with the concept that the<br />

<strong>shear</strong> <strong>thickening</strong> transition is stress controlled, as demonstrated<br />

in Fig. 2 above. As the <strong>shear</strong>-<strong>thickening</strong><br />

transition is relatively fast for our materials, and the<br />

response is in the nonlinear regime, neither the extended<br />

Cox-Merz rule nor the Delaware-Rutgers approach<br />

would be expected to hold in the <strong>shear</strong> <strong>thickening</strong><br />

regime, as demonstrated.<br />

Shown in Fig. 4 is the <strong>shear</strong> <strong>thickening</strong> behavior as a<br />

function <strong>of</strong> strain amplitude for the aqueous suspension<br />

at 60 vol.% silica. The strain reported here is the measured,<br />

peak strain amplitude (this experiment is<br />

Fig. 3 Application <strong>of</strong> Cox-Merz and Delaware-Rutgers rules for<br />

60 vol.% aqueous silica dispersion


Fig. 4 Strain <strong>thickening</strong> behavior <strong>of</strong> 60 vol.% aqueous silica<br />

dispersion as a function <strong>of</strong> dynamic frequency<br />

performed on a stress controlled instrument). From this<br />

data, the critical strain for strain <strong>thickening</strong> was found<br />

to depend inversely on the dynamic frequency for low<br />

frequencies, as shown in Fig. 5. The frequency dependence<br />

<strong>of</strong> the critical amplitude has been reported for<br />

various systems. Laun et al. (1991) and Raghavan and<br />

Khan (1997) observed transitional behavior, namely that<br />

at low frequencies the critical amplitude decreases inversely<br />

with the frequency, while at very high frequencies<br />

(x>100 rad/s) the critical amplitude approaches a<br />

constant value as shown in Fig. 5. In the present experiments<br />

the frequency for the onset <strong>of</strong> <strong>shear</strong> <strong>thickening</strong><br />

is around 30 rad/s and the minimum critical strain was<br />

found to be 2. Figure 5 also shows the critical strain<br />

for the ethylene glycol suspension <strong>of</strong> 62 vol.% silica. The<br />

overall behavior is similar to that <strong>of</strong> the aqueous silica<br />

suspension, but the minimum critical strain <strong>of</strong> the ethylene<br />

glycol based suspension is lower ( 0.7).<br />

The low frequencies data reported in Fig. 5 can be<br />

understood within the context <strong>of</strong> the model presented by<br />

Maranzano and Wagner (2001a, 2001b). As shown, the<br />

dynamic oscillatory measurement show <strong>shear</strong> <strong>thickening</strong><br />

at the same value <strong>of</strong> the average applied stress as the<br />

steady <strong>shear</strong> measurements. During the oscillation, <strong>shear</strong><br />

<strong>thickening</strong> is observed to occur when s>sc. One can<br />

thus estimate the strain required by<br />

sc<br />

cc ; ð3Þ<br />

gcx where g c is the <strong>shear</strong> viscosity at the critical point<br />

for <strong>shear</strong> <strong>thickening</strong> and s c is the critical stress, both<br />

determined from the steady <strong>shear</strong> data. The lines in<br />

Fig. 5 are calculated from the measured steady <strong>shear</strong><br />

data. The predictions work well for our data (Fig. 5a)<br />

and that <strong>of</strong> Raghavan and Khan (1997) and Mewis and<br />

Biebaut (2001) (Fig. 5d) at low frequencies. However,<br />

deviations from this prediction become evident at higher<br />

frequencies. The data <strong>of</strong> Boersma et al. (1992) (Fig. 5b)<br />

may be limiting toward the predictions, whereas the data<br />

<strong>of</strong> Laun et al. (1991) does not have the corresponding<br />

steady <strong>shear</strong> data for making the prediction. However,<br />

note that Laun’s data qualitatively displays the predicted,<br />

low frequency behavior.<br />

Wall slip has been postulated to influence the high<br />

frequency results (Boersma et al. 1992). Figure 6 is a<br />

schematic drawing <strong>of</strong> the postulated deformation with<br />

slip between parallel plates, where the slip distance at the<br />

wall has been defined as D slip and h is the gap size. Thus,<br />

the measured or apparent strain is the sum <strong>of</strong> the true or<br />

real strain in the sample and the strain due to slip as<br />

cc;app ¼ cc;real þ 2Dslip<br />

: ð4Þ<br />

h<br />

Thus, at the point <strong>of</strong> <strong>shear</strong> <strong>thickening</strong> cc,app(=Dapp/h)<br />

is the apparent critical strain for <strong>shear</strong> <strong>thickening</strong> and<br />

cc,real (=Dreal/h) is the real critical strain for <strong>shear</strong><br />

<strong>thickening</strong>. The apparent critical strain has been measured<br />

using 25-mm parallel plates where the gap size is<br />

varied from 1.5 mm to 0.05 mm. Figure 7 shows the<br />

measured apparent critical strain for <strong>shear</strong> <strong>thickening</strong><br />

for silica dispersed in both water and ethylene glycol at<br />

25 °C. The lines correspond to a fit <strong>of</strong> Eq. (4) to determine<br />

the slip distance. The fits yield slip distances <strong>of</strong><br />

0.44 mm for the ethylene glycol based <strong>shear</strong> <strong>thickening</strong><br />

fluid and 1.45 mm for the water based <strong>shear</strong> <strong>thickening</strong><br />

fluid. However, with the use <strong>of</strong> roughened plates the<br />

critical strain for <strong>shear</strong> <strong>thickening</strong> obtained during strain<br />

sweep experiments were found to be nearly independent<br />

<strong>of</strong> gap size, as shown in Fig. 7, demonstrating that the<br />

slip was substantially reduced.<br />

This analysis can explain why the measured critical<br />

strain for <strong>shear</strong> <strong>thickening</strong> limits at high frequencies. We<br />

can model the critical strain shown in Fig. 5 by accounting<br />

for a slip distance at the fixture walls as<br />

c c;app ¼ sc<br />

g cx<br />

203<br />

2Dslip<br />

þ : ð5Þ<br />

h<br />

In the comparison, we take h 1.25 mm, the edge<br />

gap, to be a characteristic distance for the cone-plate<br />

geometry used for these measurements. In Fig. 8 the<br />

lines representing the predictions <strong>of</strong> Eq. (5) are in<br />

good agreement with the observed critical strain as a<br />

function <strong>of</strong> frequency. For comparison, we used our<br />

slip measurements and the reported steady <strong>shear</strong> data<br />

to predict the critical strains reported by Raghavan<br />

and Khan (1997) for fumed silica in polypropylene<br />

glycol. Their data fits well with their steady <strong>shear</strong> data


204<br />

Fig. 5a–d Critical strain for <strong>shear</strong> <strong>thickening</strong> plotted as a function<br />

<strong>of</strong> frequency for: a 60 vol.% aqueous and 62 vol.% ethylene glycol<br />

based silica dispersions; b silica in glycerol/water by Boersma et al.<br />

(1992); c latex in ethylene glycol by Laun et al. (1991); d fumed<br />

silica in PPG by Raghavan and Khan (1997) and silica in octanol<br />

by Mewis and Biebaut (2001). The lines are predictions <strong>of</strong> Eq. (3)<br />

using the measured steady <strong>shear</strong> rheology when available<br />

and our slip distance characterized for a silica suspension<br />

in ethylene glycol. According to our experiments<br />

with roughened plates, the slip can be greatly<br />

reduced (i.e., Fig. 7), such that the critical strain for<br />

<strong>shear</strong> <strong>thickening</strong> no longer exhibits a plateau value for<br />

high frequencies (Fig. 8). This shows that the high<br />

frequency limiting value <strong>of</strong> the critical strain observed<br />

in previous experiments with smooth tooling is an<br />

artifact <strong>of</strong> adhesive failure and slip.<br />

The experimental analysis <strong>of</strong> slip demonstrates that<br />

the high frequency plateau in the critical strain can be<br />

explained as wall slip. The slip distance is expected to<br />

depend on the <strong>properties</strong> <strong>of</strong> solvent, which to first<br />

order might be expected to correlate with the wetting<br />

<strong>of</strong> the tool by the solvent. Indeed, we observe more<br />

slip for water than ethylene glycol, which qualitatively<br />

follows the expected wetting behavior (Vidal 2002)<br />

(contact angles on alumina are reported to be 75.9°<br />

for water and 50.4° for EG). Interestingly, this analysis<br />

confirms the postulate <strong>of</strong> Boersma et al. (1992),<br />

who surmised that their high frequency behavior, for


Fig. 6 Schematic illustration <strong>of</strong> the displacement <strong>of</strong> a suspension<br />

between parallel plates with wall slip<br />

Fig. 7 Apparent critical strain for <strong>shear</strong> <strong>thickening</strong> as a function <strong>of</strong><br />

gap size between parallel plates. Silica suspension in water (filled<br />

squares), ethylene glycol (open circles) with cone and plate and<br />

ethylene glycol (filled circles) with emery cloth covered parallel<br />

plates<br />

which the critical strain for <strong>shear</strong> <strong>thickening</strong> does not<br />

plateau (Fig. 5b), was due to a solid-like response<br />

without slip.<br />

Lissajous plots<br />

205<br />

Fig. 8 Measured and predicted critical strain as a function <strong>of</strong><br />

angular frequency for silica suspension in water (open squares),<br />

ethylene glycol (open circles) measured with cone and plate,<br />

ethylene glycol (filled circles) measured with emery cloth covered<br />

parallel plates and fumed silica suspension in polypropylene glycol<br />

(open triangles) from Raghavan and Khan (1997)<br />

To understand further the dynamical nature <strong>of</strong> the <strong>shear</strong><br />

thickened state, the dynamic oscillatory measurements<br />

were analyzed as flows. The energy dissipated by the<br />

<strong>shear</strong> <strong>thickening</strong> fluid can be obtained by integrating the<br />

area contained in a plot <strong>of</strong> stress vs strain for a dynamic<br />

oscillatory test. The flow mechanism can be understood<br />

from the shape <strong>of</strong> the resulting stress-strain curve (Lissajous<br />

plot). At 5 Pa <strong>of</strong> maximum imposed stress and<br />

0.1 rad/s <strong>of</strong> the frequency, an elliptical hysteresis loop is<br />

recorded as shown in Fig. 9. Figures 9, 10, and 11 show<br />

the data taken at stress amplitudes <strong>of</strong> 5 (linear viscoelastic<br />

region), 50 Pa (strain thinning nonlinear region),<br />

and 500 Pa (strain <strong>thickening</strong> nonlinear region). For this<br />

sample the critical maximum stress (s 0) for stress <strong>thickening</strong><br />

in a dynamic test at x=0.1 rad/s was around<br />

s 0=80 Pa and the critical stress for <strong>shear</strong> <strong>thickening</strong> in<br />

steady <strong>shear</strong> was 46 Pa, which approximately corre-<br />

sponds to sd ¼ 2s0<br />

p .<br />

The area enclosed by the ‘‘loop’’ can be interpreted as<br />

viscous damping. The angle between the primary axis <strong>of</strong><br />

the ellipse and the horizontal axis indicates the elastic<br />

modulus; a zero degree orientation is Newtonian. As seen,<br />

the <strong>colloidal</strong> dispersion at s\ p<br />

2 sd yields a cycle that has<br />

both elastic and viscous character, indicating viscoelasticity.<br />

In order to explore the difference in mechanical


206<br />

Fig. 9 Lissajous plot for aqueous silica dispersion at 60 vol.% and<br />

a frequency <strong>of</strong> 0.1 rad/s and the applied stress amplitude <strong>of</strong> 5 Pa<br />

which is relatively low<br />

Fig. 10 Lissajous plot for aqueous silica dispersion at 60 vol.%<br />

and a frequency <strong>of</strong> 0.1 rad/s and the applied stress amplitude <strong>of</strong><br />

50 Pa which is close to the point <strong>of</strong> <strong>shear</strong> <strong>thickening</strong><br />

<strong>properties</strong> between <strong>shear</strong> thinning and <strong>thickening</strong><br />

region, the angular frequency was held fixed at 0.1 rad/s<br />

while the maximum stress was increased stepwise. As<br />

seen, the area enclosed increases with increasing stress<br />

amplitude, indicating an increase in viscous dissipation.<br />

Above maximum imposed stresses <strong>of</strong> 50 Pa the loops<br />

Fig. 11 Lissajous plot for aqueous silica dispersion at 60 vol.%<br />

and a frequency <strong>of</strong> 0.1 rad/s and the applied stress amplitude <strong>of</strong><br />

500 Pa, which is well above the critical stress for <strong>shear</strong> <strong>thickening</strong><br />

deviate strikingly from an elliptical shape. Further<br />

increases in stress amplitude above the critical stress<br />

converge to a highly non-elliptical shape. This can be<br />

interpreted as the superposition <strong>of</strong> a primarily fluid<br />

response for low stresses in the cycle with a primarily<br />

elastic response for stresses exceeding the critical stress<br />

for <strong>shear</strong> <strong>thickening</strong>. Although qualitative, the shape<br />

analysis immediately distinguishes this fluid from other,<br />

complex behaviors (such as a yielding fluid), and signals<br />

the onset <strong>of</strong> <strong>shear</strong> <strong>thickening</strong> in the dispersion. Notice<br />

that, on the time scale <strong>of</strong> the oscillation, the fluid is<br />

<strong>thickening</strong> and ‘‘melting’’, such that the material<br />

response time for <strong>shear</strong> <strong>thickening</strong> is substantially faster<br />

than the experiment’s frequency.<br />

The normalized strain (c/cmax) is plotted as a function<br />

<strong>of</strong> the normalized applied stress (s/s0) (sinusoidal frequency<br />

x=0.1 rad/s) on a period (Fig. 12). Table 2<br />

gives the normalizing factors for Fig. 12. Below the<br />

transition stress for <strong>shear</strong> <strong>thickening</strong> (46 Pa) the distortion<br />

<strong>of</strong> stress-strain curve is increasing with stress amplitude.<br />

However, above the transition stress for <strong>shear</strong><br />

<strong>thickening</strong>, the Lissajous diagrams show the same pattern<br />

with increasing imposed maximum stress. Note that<br />

the maximum strain limits at high stress (Table 2), which<br />

is in agreement with the slip analysis if the sample itself<br />

exhibits primarily a solid response. This result is also in<br />

good agreement with the observation <strong>of</strong> Mewis and<br />

Biebaut (2001), who observed a unique, viscoelastic<br />

master curve for their <strong>shear</strong> <strong>thickening</strong> dispersions using<br />

parallel superposition. Note, however, that this pattern<br />

and the parallel superposition spectrum observed by


Fig. 12 Scaled Lissajous plot obtained for 60 vol.% aqueous silica<br />

dispersion at x=0.1 rad/s and stress amplitude <strong>of</strong> 1, 5, 10, 50, 100,<br />

500, and 1000 Pa<br />

Table 2 Scaling values for Fig. 12<br />

Angular frequency<br />

x (rad/s)<br />

Maximum<br />

stress s0 (Pa)<br />

Maximum<br />

strain c0<br />

0.1 1 2.8·10 )4<br />

0.1 5 4.8·10 )3<br />

0.1 10 1.3·10 )2<br />

0.1 50 16<br />

0.1 100 16<br />

0.1 500 17<br />

0.1 1000 18<br />

Mewis and Biebaut (2001) must also reflect the large<br />

amount <strong>of</strong> slip observed in the <strong>shear</strong> <strong>thickening</strong> state.<br />

Analysis <strong>of</strong> the <strong>shear</strong> <strong>thickening</strong> fluid subjected to a<br />

dynamic frequency <strong>of</strong> 0.1 rad/s with stress amplitudes<br />

varying from 1 to 1000 Pa shows (Fig. 13) that the<br />

energy dissipated is an increasing function <strong>of</strong> strain<br />

amplitude (peak value <strong>of</strong> measured strain). The energy<br />

dissipated during a cycle (U d) is given by the area<br />

enclosed by the Lissajous plot (Yziquel et al. 1999) (i.e.,<br />

Ud ¼ H sdc). The following phenomenological relation is<br />

<strong>of</strong>ten used to relate dissipation energy to strain amplitude<br />

(Citerne et al. 2001):<br />

Ud ¼ Jc n 0 : ð6Þ<br />

where J and n are material constants. The constant J is<br />

referred to as the damping constant and n is called the<br />

damping exponent. The energy dissipated per volume<br />

per cycle for our silica suspension is proportional to the<br />

strain raised to the second power when the fluid is deformed<br />

in linear viscoelastic region, as expected. However,<br />

above the linear viscoelastic region, the damping<br />

exponent is very large <strong>of</strong> c0 (n 15). This abrupt change<br />

in behavior signals a change in microstructure and is<br />

taken to be a signature <strong>of</strong> the hydroclustered state.<br />

Conclusions<br />

207<br />

Fig. 13 Energy dissipated per cycle per volume as a function <strong>of</strong><br />

measured strain at x=0.1 rad/s for 60 vol.% aqueous silica<br />

dispersion obtained from integrating the area enclosed in the<br />

Lissajous plots<br />

Reversible <strong>shear</strong> <strong>thickening</strong> is measured in both oscillatory<br />

and steady <strong>shear</strong> flow for a charge stabilized<br />

<strong>colloidal</strong> suspension in two solvents. The Delaware-<br />

Rutgers rule can correlate suspension rheology prior to<br />

<strong>shear</strong> <strong>thickening</strong>, but is violated in the <strong>shear</strong> thickened<br />

state. Instead, a superposition <strong>of</strong> steady and the dynamic<br />

viscosities at the point <strong>of</strong> <strong>shear</strong> <strong>thickening</strong> regime is<br />

achieved if plotted vs the average stress magnitude. The<br />

critical strain required for dynamic <strong>shear</strong> <strong>thickening</strong><br />

scales inversely with the frequency for small frequency,<br />

while for high frequency, slip leads to an apparent plateau.<br />

Consequently, measurements <strong>of</strong> steady <strong>shear</strong><br />

<strong>thickening</strong> and the slip enable predicting both the onset<br />

<strong>of</strong> <strong>shear</strong> <strong>thickening</strong> in oscillatory flow, as well as the<br />

behavior <strong>of</strong> the strain amplitude with frequency. The


208<br />

damping characteristics <strong>of</strong> a <strong>shear</strong> <strong>thickening</strong> suspension<br />

as calculated from the Lissajous plots show a dramatic<br />

increase in viscous dissipation upon <strong>shear</strong> <strong>thickening</strong>,<br />

which is due to the jamming inherent in the hydroclustered<br />

state.<br />

References<br />

Barnes HA (1989) Shear-<strong>thickening</strong> (‘‘dilatancy’’)<br />

in <strong>suspensions</strong> <strong>of</strong> nonaggregating<br />

solid particles dispersed in<br />

Newtonian liquids. J Rheol 33:329–366<br />

Bender JW (1995) Rheo-optical investigations<br />

<strong>of</strong> the microstructure <strong>of</strong> model<br />

<strong>colloidal</strong> <strong>suspensions</strong>. Thesis, Newark<br />

Bender JW, Wagner NJ (1995) Optical<br />

measurement <strong>of</strong> the contributions <strong>of</strong><br />

<strong>colloidal</strong> forces to the rheology <strong>of</strong><br />

concentrated suspension. J Colloid<br />

Interface Sci 172:171–184<br />

Bender JW, Wagner NJ (1996) Reversible<br />

<strong>shear</strong> <strong>thickening</strong> in monodisperse and bi<br />

disperse <strong>colloidal</strong> dispersion. J Rheol<br />

40:899–916<br />

Boersma WH, Laven J, Stein HN (1992)<br />

Viscoelastic <strong>properties</strong> <strong>of</strong> concentrated<br />

<strong>shear</strong>-<strong>thickening</strong> dispersion. J Colloid<br />

Interface Sci 149:10–22<br />

Bossis G, Brady JF (1989) The rheology <strong>of</strong><br />

Brownian <strong>suspensions</strong>. J Chem Phys<br />

91:1866–1874<br />

Catherall AA, Melrose JR, Ball RC (2000)<br />

Shear <strong>thickening</strong> and order-disorder<br />

effects in concentrated colloids at high<br />

<strong>shear</strong> rates. J Rheol 44:1–25<br />

Citerne GP, Carreau PJ, Moan M (2001)<br />

Rheological <strong>properties</strong> <strong>of</strong> peanut butter.<br />

Rheol Acta 40:86–96<br />

D’Haene PD, Mewis J, Fuller GG (1993)<br />

Scattering dichroism measurements <strong>of</strong><br />

flow-induced structure <strong>of</strong> a <strong>shear</strong> <strong>thickening</strong><br />

suspension. J Colloid Interface Sci<br />

156:350–358<br />

Doraiswamy D, Mujumdar AN, Tsao I,<br />

Beris AN, Danforth SC, Metzner AB<br />

(1991) The Cox-Merz rule extended: a<br />

rheological model for concentrated<br />

<strong>suspensions</strong> and other materials with a<br />

yield stress. J Rheol 35:647–685<br />

Farr RS, Melrose JR, Ball RC (1997)<br />

Kinetic theory <strong>of</strong> jamming in hardsphere<br />

startup flows. Phys Rev E<br />

55:7203–7211<br />

Foss DR, Brady JF (2000) Structure, diffusion<br />

and rheology <strong>of</strong> Brownian <strong>suspensions</strong><br />

by Stokesian dynamics<br />

simulation. J Fluid Mech 407:167–200<br />

Helber R, Doncker F, Bung R (1990)<br />

Vibration attenuation by passive stiffness<br />

switching mounts. J Sound Vib<br />

138:47–57<br />

H<strong>of</strong>fman RL (1974) Discontinuous and<br />

dilatant viscosity behavior in concentrated<br />

<strong>suspensions</strong>. II. Theory and<br />

experimental tests. J Colloid Interface<br />

Sci 46:491–506<br />

H<strong>of</strong>fman RL (1997) Explanations for the<br />

cause <strong>of</strong> <strong>shear</strong> <strong>thickening</strong> in concentrated<br />

<strong>colloidal</strong> <strong>suspensions</strong>. J Rheol<br />

42:111–123<br />

Kaffashi B, O’Brien VT, Mackay ME,<br />

Underwood SM (1997) Elastic-like and<br />

viscous-like components <strong>of</strong> the <strong>shear</strong><br />

viscosity for nearly hard sphere,<br />

Brownian <strong>suspensions</strong>. J Colloid Interface<br />

Sci 187:22–28<br />

Laun HM, Bung R, Schmidt F (1991)<br />

Rheology <strong>of</strong> extremely <strong>shear</strong> <strong>thickening</strong><br />

polymer dispersions (passively viscosity<br />

switching fluids). J Rheol 35:999–1034<br />

Laun HM, Bung R, Hess S, Loose W, Hess<br />

O, Hahn K, Ha¨ dicke E, Hingmann R,<br />

Schmidt F, Lindner P (1992) Rheological<br />

and small angle neutron scattering<br />

investigation <strong>of</strong> <strong>shear</strong>-induced particle<br />

structures <strong>of</strong> concentrated polymer dispersions<br />

submitted to plane Poiseuille<br />

and Couette flow. J Rheol 36:743–787<br />

Lee DI, Reder AS (1972) The rheological<br />

<strong>properties</strong> <strong>of</strong> clay <strong>suspensions</strong>, latexes<br />

and clay-latex systems. TAPPI Coating<br />

Conference Proceedings, pp 201–231<br />

Acknowledgments Dr. Eric Wetzel <strong>of</strong> Army Research Laboratory<br />

(MD) is acknowledged for helpful discussions. This work has been<br />

supported through the Army Research Laboratory CMR program<br />

(Grant No. 33-21-3144-66) through the Center for Composite<br />

Materials <strong>of</strong> the University <strong>of</strong> Delaware. Funding for equipment<br />

from the NSF-MRI (CTS-9977451) is gratefully acknowledged.<br />

Maranzano BJ, Wagner NJ (2001a) The<br />

effect <strong>of</strong> interparticle interactions and<br />

particle size on reverse <strong>shear</strong> <strong>thickening</strong>:<br />

hard-sphere <strong>colloidal</strong> dispersions.<br />

J Rheol 45:1205–1222<br />

Maranzano BJ, Wagner NJ (2001b) The<br />

effect <strong>of</strong> particle size on reverse <strong>shear</strong><br />

<strong>thickening</strong> <strong>of</strong> concentrated <strong>colloidal</strong><br />

dispersion. J Chem Phys 114:<br />

10,514–10,527<br />

Maranzano BJ, Wagner NJ (2002) Flowsmall<br />

angle neutron scattering<br />

measurements <strong>of</strong> <strong>colloidal</strong> dispersion<br />

microstructure evolution through the<br />

<strong>shear</strong> <strong>thickening</strong> transition. J Chem<br />

Phys 117:10291–10302<br />

Mewis J, Biebaut G (2001) Shear <strong>thickening</strong><br />

in steady and superposition flows effect<br />

<strong>of</strong> particle interaction forces. J Rheol<br />

45:799–813<br />

Newstein MC, Wang H, Balsara NP,<br />

Lefebvre AA, Shnidman Y, Watanabe<br />

H, Osaki K, Shikata T, Niwa H,<br />

Morishima Y (1999) Microstructural<br />

changes in a <strong>colloidal</strong> liquid in the <strong>shear</strong><br />

thinning and <strong>shear</strong> <strong>thickening</strong> regimes.<br />

J Chem Phys 111:4827–4838<br />

Raghavan SR, Khan SA (1997) Shear<strong>thickening</strong><br />

response <strong>of</strong> fumed silica<br />

suspension under steady and oscillatory<br />

<strong>shear</strong>. J Colloid Interface Sci 185:57–67<br />

Vidal M (2002) Surface energy study.<br />

http://www.mse.gatech.edu/research/<br />

surf/surt99/vidal.htm<br />

Yziquel F, Carreau PJ, Tanguy PA (1999)<br />

Non-linear viscoelastic behavior <strong>of</strong><br />

fumed silica <strong>suspensions</strong>. Rheol Acta<br />

38:14–25

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!