Viscosity of dilute Na-montmorillonite suspensions in electrostatically stablecondition under low shear stress

Y. Tsujimoto, M. Kobayashi , Y. Adachi

Graduate School of Life and Environmental Sciences, University of Tsukuba, 1-1-1, Tennnoudai, Tsukubashi, Ibaraki 305-8572, Japan

Abstract We have studied the rheological behavior of dilute suspensions of electrostatically dispersed Na-montmorillonite to elucidate the importance of the electroviscous effect under the condition of zerostress limits. The viscosities of the suspensions were measured using an original spiral-type viscometerthat consisted of two measuring cylinders connected with a one-meter-long capillary tube. Extremelydilute suspensions, whose volume fractions ranged from 2.0 104to 2.0 103, were used for the experi-ments ionic strength, in accordance with a manifestation of the electroviscouseffect.The ionic strength of the suspensions was controlled to be less than 0.001 M. Thus, a fully developedelectrical double layer was formed. It was confirmed that the measured viscosities of the suspensionsincreased with a decrease in their he viscosity increased markedly owing to the secondary electroviscous effect under the limit ofthe salt-free condition. To analyze this effect, we focused on the effective radii of the suspended mont-morillonite particles. We estimated their effective radii by fitting the DoughertyKrieger equation to theviscosity data. We found that the obtained effective radii depended significantly on the ionic strengthin the low Peclet number regime, which was similar to the theoretical dependence of the effective radiidetermined from the interaction balance between the particles governed by hydrodynamic, electrostatic,and diffusive energies. Thus, the increase in the effective radius with a decrease in the ionic strength givesrise to the expansion of the electric repulsive force, resulting in an increase of the viscosity of this claysuspension. In addition, the magnitudes of the effective radii were greater than the Debye length. Thisindicates that even when the distance between the montmorillonite particles is greater than the Debyelength, the electrical repulsive force affects the interaction between the particles.

1. Introduction Montmorillonite is a clay mineral that has been used in manyindustries and has received many interests in terms of practicalapplication [1]. Rheological behavior of this clay suspensionis important for various applications such as oil drilling andis also related to adsorption or transportation of pollutants in underground water. It is known that the unit of this clay is quitethin layer with a thickness of 10 angstrom. The layer consists oftwo different sheets. One is silica tetrahedral sheet located outside of unit clay sheet. The other is alumina octahedral sheet holdbetween two silica sheets. Isomorphous substitution from Al3+toMg2+in the alumina sheet yields permanent negative charges atthe face of the layer. On one hand, the broken bonds located atthe edge of the layer have a capacity to adsorb H+or OH. As aresult, such charging property gives rise to different types of flocstructure as well as colloidal stability [14]. These charging andflocculationdispersion properties of montmorillonite affect therheological characteristics of this clay suspension. Understandingmacroscopic properties of montmorillonite suspensions in termsof microscopic colloidal interactions has been regarded as a majorissue. The flow properties and solgel transition of montmorillonitesuspensions are dependent on ionic strength, solid contents andparticle size [46]. Abend and Lagaly carried out rheological mea-surements of montmorillonite suspensions as functions of thesefactors and revealed that the existence of two types of gelation,i.e., attractive gel and repulsive gels [4]. These two types of gel areformed by different mechanisms. One is the so-called cardhousestructure due to electrostatic attractive interaction between edgeand face of clay, which provides attractive gel [1,3]. Laxton andBerg investigated clay yield stress relating to zeta potential ascolloidal parameter using laponite suspensions and reported thatyield stress is due to the development of microstructure throughelectrostatic attraction between the oppositely charged edge andface of clay particles [7]. The other mechanism is induced by thedevelopment of the electrical double layer (EDL) around the clayparticles. The EDL induces the electrostatic repulsive force thatcauses gelation through the formation of the lamellar crystal likestructure for the repulsive gel. Under electrostatically stabilizedcondition, the increase of viscosity and the appearance of yieldstress occur with decreasing ionic strength. Our previous measure-ments on the yield stress of montmorillonite suspension using avane spindle with torque sensor also revealed the increase of resis-tance against the deformation of suspensions with decreasing ionicstrength [8]. The experimental condition was at higher pH whereedgeface interaction is negligible. A simple model assuming par-allel array of thin clay layer successfully explained the dependenceof yield stress on the volume fraction of clay and the ionic strengthon the basis of the repulsive force. According to their assumption,the estimated mean distance between clay layers at

1% volumefraction of clay is about 100 nm and is thus significantly longerthan the thickness of EDL. They concluded that the electrostaticrepulsive force is effective for long-range over the Debye length,and controls the appearance of the yield stress and gelation.The effects of charging properties of clay particles and thedevelopment of EDL are also reflected in the viscosity of sus-pensions. Socalled electroviscous effects of colloidal suspensioncan be ascribed to the additional distortion energy of fluid dueto the formation of EDL around colloidal particles. The deforma-tion of EDL against fluid distortion of shear causes the primaryelectroviscous effect, which appears as an increment of the intrin-sic viscosity that was also affected by the shape of particles. Thiseffect was investigated in detail by many scientists, in particu-lar by Booth [9]. The secondary electroviscous effect is caused bythe additional energy dissipation generated by the overlapping ofEDL when two charged particles are conveyed to a close positionby an external fluid motion. It was treated theoretically by Rus-sel [10,11] and has been regarded as one of the major factor ofnon-Newtonian behavior. Krieger and Eguiluz pointed out that thesecondary electroviscous effects are extremely important in mod-ifying the rheological behavior of a colloidal suspension where therelative viscosity at a given shear rate can vary by several orders ofmagnitude [12]. Additionally, they demonstrated that this effect isrelated to shear thinning behavior. Buscall analyzed this rheologicalbehavior of concentrated lattices based on the Dou ghetryKriegerequation (DK equation) [13]. Furthermore, Takamura and van deVen incorporated the particle interaction parameter into the DKequation [14]. They revealed that the parameter was given as afunction of Peclet number and the secondary electroviscous effectsbecame negligible at high shear rate. For montomorillonite suspen-sions, there are a few experimental studies of the electroviscouseffects. Egashira found that the intrinsic viscosity of montmoril-lonite is 190, which is much larger than that of the theoretical valueof spherical particles [15] but he did not refer to electroviscouseffects. Tamaki et al. measured the viscosity of dilute suspensionof montmorillonite at high pH using the Ostwald viscometer andfound that the magnitude of these electroviscous effects increaseswith a decrease of ionic strength, especially in the region of very lowionic strength, c.a. NaCl concentration less than 1.0 102M [16].However, experimental data for the dependence of the secondaryelectroviscous effect of montmorillonite suspension on the appliedshear stress are still lacking. Therefore, it is desirable to analyze thesecondary electroviscous effect of dilute montmorillonite suspen-sion in the limit of zero shear stress. Kobayashi et al. developed aviscometer enabling us to measure the viscosity under zero shearlimits and suggested the rheological behavior is non-Newtonianeven if the suspension is so dilute that it is usually regarded asNewtonian fluid [17,18]. However, they did not mention the elec-

troviscous effects in dispersed state under low shear stress. Thus,in this study, we analyze the electroviscous effects in dilute Na-montmorillonite suspensions under low shear stress on the basisof experimental results obtained by using this original viscometeras well as standard capillary viscometer. 2. Theory 2.1. Viscosity of colloidal suspension The viscosity of colloidal suspension is expressed as a functionof volume fraction, as follows,

where _r, _sand _0denote the relative viscosity, the viscosity ofsuspension and the viscosity of solvent, respectively.K1 is determined by the interaction between a single particle and surrounding fluid, which is corresponding to the intrinsic viscosity. On the otherhand, K2 is determined on the basis of the hydrodynamic inter-action between two particles. The increment of K1and K2 can berelated to as the primary and the secondary electroviscous effects,respectively. Not only volume fraction but also Peclet number, Pe,affects the viscosity of suspensions. This is the dimensionless groupgiven by the timescale for diffusive motion relative to that for con-vection of flow and defined as follows [10,11]

Where _ indicates the shear stress, and 2a indicates the hydro-dynamic diameter. Boltzmann constant k = 1.38 1023J/K and absolute temperature T = 298 K, respectively. Russel reported that the rheological behavior of dilute sus-pensions composed of monodispersed sphere is affected by the interaction between particles governed by viscous, electrostaticand van der Waals forces, etc. The balance among these factorsis qualitatively described by a characteristic length which is called the effective collision radius (effective radius). He worked out the-oretical treatment on the secondary electroviscous effect assuming

the increase of the effective radius, aeffwith developing EDL. Then,he gave the relative viscosity as

means a coefficient determining the intrinsic viscosity that canbe calculated by Booth theory for the primary electroviscous effect.However, because the intrinsic viscosity for montmorillonite sus-pension is affected by the shape of particles, it does not strictly agree with theoretical values for sphere [16]. Because Eq. (3) can be applied at only lower Pe, we adopt theother model, DK equation, by which van de Ven et al. and Buscallanalyzed the electroviscous effects,

Dimana

Max indicates packing volume fraction. We use Eqs. (4) and (5) toestimate the experimental values of aeffby fitting that to our exper-imental data assuming _max= 0.74 regardless of shear rate arguedby van de Ven [19].

2.2. Effective radius The effective radius aeffis regarded as the distance inducing thedistortion of the particle trajectory as shown in Fig. 1 and maybee stimated from the balance between the electrostatic repulsive fac-tor and factors inducing collision [11]. When Pe is so small that the Brownian motion is dominant, the theoretical aeffis determined by the balance between electrical repulsive energy, Ve(r), and thediffusive energy, kT, as follows

Dimana

r ant indicates the distance from a center of the particle andzeta potential, respectively. The thickness of EDL, 1, so-called the Debye length, is

determined as a function of the ionic concentrationnias follows,

Determining these coefficients, we assumed that elementary charge e = 1.6 1019C, the valence of electrolyte solution for NaC lfor z = 1, the dielectric constant of vacuum 0= 8.85 1012C2/J m, relative dielectric constant of water r= 78.54, respectively. Theeffective radius is

On the contrary, when the Peclet number is larger than one,which implies the shear stress contribute to the interaction, themean effective radius can be estimated from the balance betweenthe hydrodynamic force and electrostatic repulsive force [20]. Thehydrodynamic force is calculated from the Stokes drag as follows

The electrostatic repulsive force is

Balancing these two forces we get

Therefore, we can calculate the effective radius based on Eqs. (8)and (11). In this study, we regard aeffcalculated by Eqs. (8) and(11) as the theoretical effective radius at lower Pe (_1) and higherPe, respectively. We compare these values with the experimentaleffective radii by Eqs. (4) and (5) mentioned in Section 2.1 3.2. Material Montmorillonite suspensions were prepared from commer-cially available Kunipia-F (Kunimine Industry Co. Ltd.). Coarsecomponents, mainly silica sands, contained in purchased mate-rial were removed by sedimentation. Surface cation was replacedby sodium ion by dispersing the clay particles into the 2.0 M NaClsolution. Salt free stock dispersion was prepared by repeatedly dia-lyzing

against distilled water until the electric conductivity wascome down to 1.5 _S/cm. After these treatments, we freeze-driedthis sample. Viscosity measurements were performed for the sample madefrom the freeze-dried montmorillonite properly suspended with the aqueous solution of sodium chloride with predetermined con-centration. The volume fraction of clay particles was adjusted from2.0 104to 2.0 103, assuming the density of montmorilloniteis 2.68 g/cm3[8] with background NaCl concentration in the range0 to 103M. Although we fully reduced the conductivity of montmorillonite sample during the dialysis, the electric conductivity ofsuspension was affected by volume fraction of montmorillonitesuspensions (Fig. 3). To do rheological experiment under fully developed EDL, ionic strength was set to be lower than 103M.The value of pH in the suspension was not purposely controlled,but was confirmed to be between 5.5 and 6.5 throughout experiments. Zeta potential was estimated from Smoluchowski equationin advance by measuring the mobility of montmorillonite particles(Table 1). Also, the average stokes diameter estimated by dynamic light scattering (DLS) measurement is 275 nm. We regarded thisvalue as a representative diameter of montmorillonite particle. 3.3. Measurements of viscosityFor the viscosity measurement using the spiral viscometer(Fig. 2), we started to record the elapsed time when the difference ofwater/suspensions levels come down to 12.8 mm as an initial point

The required time for the suspension level moves by1.6 mm, which corresponds to the unit scale of the cylinders, was measured. That is, viscosity of suspension was simply determined by the temporal variation of h based on Eq. (14). On the other hand,for the measurements under high shear stress, CannonFenske cap-illary viscometers with inner diameter of 0.540.74 mm were usedto compare the viscosity with that obtained from the spiral vis-cometer. All

experiments were performed in an air-conditionedroom where the temperature was controlled at 20.0.
20.0.

4.Results Measurements of viscosity of Na-montmorillonite suspensionwere done using the spiral viscometer and general Cannon-Fenskeviscometer. For the data obtained by the spiral viscometer, we plot-ted ln{h(t)/h0} against time with different ionic strengths in Fig. 4.In this figure, we draw the linear lines as eyeguide based on the Eq.(14). When the experimental data are plotted linearly, the suspen-sion can be regarded as Newtonian fluid. Thus, the results in Fig. 4 implied that as ionic strength became higher, the montmorillonite suspension varied from Newtonian to non-Newtonian. The obtained results are shown as plots of relative viscosityagainst volume fraction in Fig. 5. Regardless of viscometer, the rel-ative viscosity of suspensions increases with a decrease of ionicstrength. This tendency is remarkable for the results obtained byspiral viscometer, meaning that the influence of the secondary elec-troviscous effect clearly occurs under low shear stress. The obtained relative viscosities are represented against the applied shear stressin Fig. 6. Especially below ionic strength of 105M, the incrementof the relative viscosity with decreasing shear stress is significant.Such appearance of so called shear thinning phenomena was confirmed in the investigations for colloidal suspensions [1214]. Based on the particle radius, zeta potential (Table 1) and theobtained viscosities, we calculated and plotted the effective radiusas a function of ionic strength in Fig. 7. The values of salt-free sus-pension were calculated as ionic strength of 5 106M. For highionic strength, the theoretical effective radius was constant at bothlow and high Peclet number. However, as ionic strength became

lower, the theoretical effective radius obtained by Eq. (8) was largerthan that obtained by Eq. (11). On the other hand, the dependenceof the effective radius obtained from fitting Eq. (4) to experimen-tal data in Fig. 5(a) at low shear on ionic strength is remarkablethan that from fitting Eq. (4) to data in Fig. 5(b) at high shear.This tendency qualitatively agreed with the theoretical

prediction.However the magnitude of aeffis larger than theoretical values. Also,these experimental values are larger than the Debye length.

5. Discussion Here in, we discuss our analysis of the electroviscous effect at lowshear stresses. As shown in Fig. 6, we found that the electroviscouseffect was manifested distinctly at low shear stresses. The electri-cal repulsive force is significant even at distances greater than theDebye length, as had also been pointed out by Sakairi et al. Thus,we analyzed this sharp increase in the viscosity owing to the elec-troviscous effect by focusing on the effective radii of the suspendedmontmorillonite particles.

As shown in Fig. 7, the measured magnitudes of aeffdiffered fromthose determined theoretically. We surmised that this discrepancy between the theoretically determined values of aeffand those measured experimentally was due to the method used to estimate the zeta potential and the radius of the montmorillonite particle. In this study, we estimated the zeta potential of the particles from theirelectrophoretic mobility using the Smoluchowski equation. Mor-rison has shown that the Smoluchowski equation is valid when
a _ 1,

regardless of the shape of particles [21]. Duran et al. havealso estimated the zeta potential by using the Smoluchowski equa-tion to calculate the interaction energy and have shown that thedependence of the interaction energy on the pH of the suspension,as determined from the zeta potential of the suspended particles,could

explain, to some degree, the rheological behavior of montmorillonite suspensions [22]. On the other hand, Chassagne et al.have argued that the actual zeta potential is three times larger thanthat calculated using the Smoluchowski equation if one takes intoaccount the Stern layer conductance [23]. They developed a new formula and technique to characterize the electrokinetic behavior of spheroids and applied it to determine the zeta potentialof kaolinite. We also calculated the zeta potential by a different method that involved comparing the theoretically determined val-ues of aeffto those measured experimentally. For an ionic strengthof less than 105M, if the zeta potential was a few times larger thanthat obtained from the electrophoretic mobility using the Smolu-chowski equation, the theoretical values of aeffwere similar to themeasured values. However, for higher ionic strengths and larger Pe values, it was several hundred times larger than that obtainedby the Smoluchowski equation. Because the difference in the zetapotentials was excessively large, we could not explain the quantitative differences between the theoretically determined values ofaeffand those measured experimentally. We calculated the theoretical effective radii of the montmoril-lonite particles using the value of hydrodynamic radius, a, whichwas measured using DLS measurement. On the contrary both Eq.(8) and (11) assume that the suspended particles are spherical. That is, the influence of the thin sheet-like shape of the montmorillonite particles is not reflected in the theoretical values of the effectiveradius. Thus, developing a theory of electroviscosity taking intoaccount the shape of the particles in future, we may explain the differences in the theoretically determined effective radii and thosemeasured experimentally. Tamaki and co-workers had also pointedout the effect the shape of the particles had on the electroviscouseffect, by comparing the data of montmorillonite under high shearrate and that of Latex under the same conditions [16]. According to Eq. (11), the effective radii depend on shear stress(i.e., the Peclet number) and a. In this study, we analyzed the influ-ence of Pe on the effective radii of the suspended particles. As shownin Fig. 8, we plotted the theoretical effective radii, which were obtained from Eq. (11), as a function of Pe.

Whereas, the values of aeffestimated using Eq. (8), which are listed in Table 2, were inde-pendent of Pe. As also shown in Fig. 8, the dependence of aeffon the Peclet number is significant at ionic strengths lower than 105M,even if the values of aeffare replaced by those listed in Table 2 fora small Pe. The dependence of the increase in the viscosity due tothe electroviscous effect on the Peclet number and ionic strength issimilar to that of the effective radii on these parameters. The aeff obtained by fitting are plotted in Fig. 9. While these values are a few times larger than the theoretical values shown in Fig. 9, the dependence of the effective radii on the Peclet number was qualitatively similar to the dependence of the theoretical ones. Thus, underlow shear stress and for low ionic strengths, the distance distortingthe collision trajectory of the particles can be expanded owing tothe electric repulsive force, which induces greater energy dissipa-tion and the increase in the viscosity. Our results demonstrated thatthe effective radius is more important than the Debye length when analyzing the increase in the viscosity of montmorillonite suspe-sions due to the electroviscous effect. Moreover, the influence ofthe effective radii, which was significant, indicated that, as was alsosuggested in our previous study, long-range interactions betweenthe montmorillonite particles induce a yield stress, even at low volume fractions, when the distance between the montmorillonite particles is greater than the Debye length.

6. Conclusions We have investigated that the rheological behavior of dilutesuspensions of electrostatically dispersed Na-montmorillonite toreveal the importance of electroviscous effects under the condition of zero stress limits using an original spiral type viscometer. Weconfirmed that the measured viscosities increased with a decreasein the ionic strength, in accordance with an appearance of electroviscous effects. A remarkable enhancement of viscosity rise due tothe secondary electroviscous effect in low shear stress was con-firmed. To analyze these effects,

we estimated the effective radiiby fitting the DK equation to our viscosity data. We found thatthe experimental effective radii depended significantly on the ionicstrength in the low Peclet number regime, which is qualitativelyagreed with that determined theoretically. Also, the magnitude ofthe experimental effective radius was larger than the Debye length.This indicates that even if the distance between montmorilloniteparticles is large compared to the Debye length, the electrical repul-sive force affects the interaction among montmorillonite particlesand thus induces that the increment of the viscosity of this claysuspension