Rheo-optical Behaviors and Stability of a Silica Particle Suspension Coated with Silane Coupling Agents

JOURNAL OF COLLOID AND INTERFACE SCIENCE ARTICLE NO.

205, 397– 409 (1998)

CS985651

Rheo-optical Behaviors and Stability of a Silica Particle Suspension Coated with Silane Coupling Agents Jae-Dong Lee and Seung-Man Yang1 Department of Chemical Engineering, Korea Advanced Institute of Science and Technology, 373-1, Kusong-dong, Yusong-gu, Taejon 305-701, Korea Received January 9, 1998; accepted May 8, 1998

The rheo-optical behaviors and suspension stability of silica particles coated with silane coupling agents were investigated experimentally. Mono-dispersed silica particles were synthesized by the sol– gel method and the particles were coated with silane coupling agents such as vinyltriethoxy silane (VTES) and g-methacryloxypropyl triethoxy silane (MPTES). Although all the suspensions of identical particle volume fraction exhibited similar rheological behaviors at high shear rates, only the stabilized suspensions coated with either VTES or MPTES displayed smooth shear thinning rather than abrupt change in the shear viscosity, as is typical of suspensions with no surface treatment. The present study showed that the MPTES coating was very effective in enhancing the phase stability compared to the VTES coating. The flow-induced dichroism for the MPTEScoated suspensions did not experience sign change, while those of the non-stabilized suspensions changed sign as the shear rate was increased. The VTES-coated suspensions underwent a transition from a stable to an unstable state as the particle volume fraction increased. Specifically, the rheological behaviors of the VTES-coated suspensions of particle volume fraction 0.35 were similar to those of the MPTES-coated suspensions. When the volume fraction exceeded 0.45, however, the effect of VTES coating diminished. Finally, the stress-optical rule proposed by Bender and Wagner was adopted to model the stabilized suspensions considered here. The present results indicated that the stress-optical coefficient could be predicted successfully by the proposed stress-optical rule if the contribution from the hydrodynamic interaction is considered separately from the thermodynamic contribution. © 1998 Academic Press Key Words: stabilization; silica dispersion; shear thinning; rheooptics; dichroism; silane coupling agents.

INTRODUCTION

In the present work, the rheo-optical behaviors and the flow-induced microstructure were considered for a highly concentrated suspension of silica particles coated with silane coupling agents. In general, a highly concentrated suspension exhibits considerable deviations from Newtonian behaviors such as shear thinning and shear thickening although the continuous medium is a Newtonian fluid. These non-Newtonian 1

To whom correspondence should be addressed.

behaviors, which are greatly affected by the microstructure and phase stability of the particle suspensions, are of practical significance in many application areas. For example, the flowinduced microstructure and the corresponding rheological behaviors are closely relevant to the processing of the composite materials in which the particle phase acts as a reinforcing additive. Frequent encounter of the particle suspensions has brought enormous investigations of the microstructure evolution under a flow and their stability (1–18). Generally, the particles dispersed in a continuous medium cause the viscosity increment via the particle–particle interactions. Since the pioneering work by Einstein (3) to estimate the viscosity increment by the presence of small particles in a dilute suspension, theoretical and computational studies have followed to consider a concentrated suspension. Among them, Batchelor (1, 2) calculated successfully the first correction to the Einstein’s theory to include the particle–particle interactions by considering separately the non-hydrodynamic (Brownian) and the hydrodynamic-interaction contributions. Although Batchelor’s result provided a great advance in the rheology of a particle suspension, its validity is limited to a comparatively low or semi-dilute suspension. Consequently, Batchelor’s theory failed to predict the existence of either shear thinning or shear thickening behavior, which is typical of a highly concentrated suspension. Recently, Bossis and Brady have developed a computational strategy (the so-called “Stokesian” dynamics) for the particle–particle interactions. Their computational results based on the Stokesian dynamics confirmed that a highly concentrated suspension could exhibit either shear thinning or shear thickening behavior depending upon the Peclet number of the flowing system. The Peclet number is the ratio of transport rates contributed from Brownian diffusion and hydrodynamic convection or the ratio of time scales associated with diffusion and convection. However, since the computational simulation of Bossis and Brady considered an ideal system only with the Brownian and hydrodynamic-interaction contributions and excluded any other inter-particle interactions, the application of these results leaves some restrictions in a practical suspension. In addition to the sophisticated theoretical and computational works, semi-empirical models have been also proposed

397

0021-9797/98 $25.00 Copyright © 1998 by Academic Press All rights of reproduction in any form reserved.

398

LEE AND YANG

by several groups (6 – 8). Of practical importance is the socalled Krieger–Dougherty model which takes into account the pair of spheres, each experiencing rotation, translation, and coalescence. Meanwhile, Ball and Richmond (8) derived a differential form of the Einstein-type equation starting from the assumption that the effects of particle presence in a concentrated suspension were the sum of contributions from each particle. It is noteworthy, however, that the Krieger–Dougherty model and the Ball–Richmond equation are identical to each other if the maximum packing fraction and the intrinsic viscosity are inserted into the corresponding constitutive equations. Further, de Kruif et al. (9) developed the viscosity equation as a function of the shear rate and the particle volume fraction in a “hard”-sphere suspension. Recent advances in the optical techniques enable us to measure directly the optically anisotropic properties such as dichroism and birefringence induced by an imposed flow. In particular, the phase modulation technique developed by Fuller’s group has been applied to a variety of flow problems including the transient behaviors of particle suspensions, polymer solutions, and ordered micellar solution in a shear flow (19 –21). The pioneering phase modulation technique makes it possible to measure simultaneously the flow-induced birefringence and the dichroism (21). The rheo-optical properties, flow-induced birefringence and dichroism, are directly related to the microstructural evolutions in the particle dispersions, polymers, liquid crystals, micellar solutions, and other complex fluids (20 –31). Moreover, the microscopic rheo-optical properties can be related to the macroscopic rheological properties through the well-known stress-optical rule (henceforth SOR), which is a simple linear relationship between the stress and the optical anisotropy tensors. Although the SOR has been proved valid and useful in many systems, the breakdown of SOR is also encountered in either multi-component systems or phase separated dispersions (31). Especially, for high shear rates, the classical SOR is not valid when the flow becomes sufficiently strong that the system cannot sustain its stability any longer. This is particularly true of concentrated particle suspensions. Thus, the linear SOR was modified to consider a concentrated suspension at a strong flow regime (23, 24). In this case, the stress field in the suspension is affected by the separate contributions from the hydrodynamic interaction and the thermodynamic origin, and the linear SOR reflects only the thermodynamic contribution. It can be also expected that the hydrodynamic-interaction contribution is predominant in a strong flow regime compared with the thermodynamic contribution. Consequently, the deviation from the classical SOR may be significant in a strong flow. In practice, the prototype model dispersions that are both stable and structurally uniform have a great deal of merit for theoretical and experimental analyses (9, 11, 17, 22–26). Starting with the preparation of a mono-dispersed particle suspension, its stabilization is essentially important, especially in a hard-sphere colloidal suspension. Among many methods for

the preparation of mono-dispersed particles (32, 33) and their stabilization (34, 35) is the novel sol– gel process, which is composed of hydrolysis and condensation of silicone alkoxide. To improve the suspension stability, the particle surface was modified through the chemical adsorption with various surface active substances such as octadecyl alcohol (34), silane coupling agents (35), and nonionic or polymeric surfactants (36). The surface modification with the chemical adsorption creates the proper steric repulsion between the neighboring particles and gives rise to the so-called hard-sphere behaviors in an organic solvent. The silane coupling agents having organic alkyl group and inorganic silicone in their molecules are the promising substances for catalysts, composites, tire reinforcements, etc. The silane coupling agents bring easy and fast reactions to the final modification and provide effective bridging between the organic and inorganic interface (37). The surface coating and the stabilization with the silane coupling agents not only makes the coated particle suspension behave like a hard-sphere colloid but also makes its refractive index match that of the ethanol/toluene mixture (35). Meanwhile, the sol– gel process has some limitation to the particle content (or volume fraction) and usually produces a dilute suspension. To increase the solid volume fraction in the dispersion, the growth method through successive addition of alkoxide was developed by Bogush and Zukoski (33). Most of material processes for the particle suspension accompany flow, which inevitably changes its microstructure and thereby modifies the phase stability. However, the interrelation between the microstructural evolution induced by an external field and the suspension stability has not been fully understood, in spite of its practical significance. This is the primary thrust of the present study. Specifically, we considered in the present paper the flow-induced microstructural changes and the stabilization effects due to the coated layers. First, the monodispersed silica particle suspensions were prepared by the sol– gel process using an alkoxide. The particle surfaces were coated with silane coupling agents such as vinyltriethoxy silane (VTES) and g-methacryloxypropyl triethoxy silane (MPTES). Then, the flow-induced dichroism in a cone and plate geometry was measured using a rheo-optical apparatus to examine the suspension stability. The rheo-optical stability results were also confirmed by a shear rheometry. Finally, the modified SOR was applied to estimate the correct stress-optical coefficients by separating the thermodynamic contribution from the rheological responses in the suspensions stabilized with the silane coupling agents. EXPERIMENTAL

1. Materials and Preparation of Suspensions Mono-dispersed silica particles were synthesized by the sol– gel method proposed by Stober et al. (32) through hydrolysis and condensation of the alkoxide tetraethylorthosilicate

SILICA PARTICLE SUSPENSION

TABLE 1 Carbon and Hydrogen Contents and Hydrogen-to-Carbon Molar Ratios Determined from Elemental Analysis (Also Included Is the Estimated Surface Coverage of Silane Coupling Agents on the Silica Surface.)

Silica with no treatment Silica with MPTES treatment Silica with VTES treatment

Carbon (wt%)

Hydrogen (wt%)

Molar ratio, H/C

Estimated surface coverage of silane coupling agents (molecules/nm2)

3.9442

1.2311

3.745

—

4.5552

1.3563

3.572

3.708

4.9503

1.2606

3.069

21.36

399

Fuller (20). Fig. 1 shows the schematic of two optical trains that can produce a linearly polarized light. The optical arrangements are composed of a He–Ne laser (L, wavelength of 632.8 nm), a polarizer (P, 0°), an analyzer (A, 245°), two quarter wave plates or a single one (QWP, 0°), a photoelastic modulator with retardance dm (PEM, 45°), and a photodiode detector. The phase modulation at the PEM was designed to generate the sinusoidally modulated retardance of dm 5 A sin(vt), in which v was a reference frequency of 50.3 kHz. The amplitude A was measured as A 5 506 from J 0 ( A) 5 0 through the blank test in Alignment II. To determine the values of Bessel functions J 1 ( A) and J 2 ( A), the flow cell in the optical Alignment II was replaced by an additional QWP. The result was given as J 1 ( A) 5 2.15 and J 2 ( A) 5 1.79. Thus, the ratio was given as J 1 /J 2 5 1.201, which was very close to the theoretical value of 1.2. The reduced dichroism (d0) and its orientation angle (x2) of a given sample in the shear field were determined by taking the

(TEOS, Si(OC2H5)4, Aldrich) in ethanol (Oriental Chem.). The shape and size were controlled by the reaction catalyst, ammonium hydroxide (NH4OH, Aldrich). Then, the synthesized silica particle surface was coated with silane coupling agents such as vinyltriethoxy silane (VTES, (C2H5O)3SiCH5CH2, Shin-Ethu Silicone) and g-methacryloxypropyl triethoxy silane (MPTES, (C2H5O)3SiC3H6OOCC(CH3)5CH2, Shin-Ethu Silicone), following the procedures proposed by Philipse and Vrij (35). Here, to ensure the mono-layer coverage on the silica surface, 72 ml of MPTES or VTES was added to 2400 ml of sol solution which was composed of [TEOS] 5 0.57 M, [NH3] 5 0.71 M, [H2O] 5 1.648 M in ethanol medium. Chemical adsorption of the silane coupling agent onto the silica surface was confirmed by elemental analysis, and the results for adsorbed elements were included in Table 1. The size distribution of the synthesized silica particles was observed by transmission electron microscopy (TEM). The particle suspensions were prepared in a solvent, tetrahydrofurfuryl alcohol (THFFA), which has the refractive index similar to that of the silica particle of ca. 1.45. Repeated evaporation and ultracentrifuge were applied to obtain the desired levels of purity and particle volume fraction. The particle density was determined using the Einstein equation in conjunction with the suspension viscosity data in a dilute concentration limit obtained from an Ubbelohde capillary. Finally, the z-potential of the coated silica particle was estimated with a zeta potential analyzer (Zeta Plus, Brookhaven), which measured directly the electrophoretic mobility of the particles. 2. Procedure and Apparatus for Analysis An ARES rheometer was used to measure the steady shear viscosity of the concentrated silica particle suspension. The shear rate was varied from 0.03 to 1000 s21 for the ARES steady sweep test at room temperature. The flow-induced dichroism of the suspension was investigated through two different optical alignments originally designed by Frattini and

FIG. 1. Schematic representation of two different optical alignments, I and II. Calibrating the various optical parameters from Alignment II, the flowinduced dichroism was measured from Alignment I.

400

LEE AND YANG

Fourier transformation on the signals detected by a photodiode. It has been shown that the optical train of Alignment I decouples effectively the dichroism and birefringence as well as their extinction and orientation angles without any mathematical manipulation (19, 20). To see this, suppose that the intensity of the polarizer and that detected by the photodiode of Alignment I are I 0 and I, respectively. Then, the intensity ratio I/I 0 can be expressed in terms of the harmonic oscillations for signal analysis. The decoupled amplitude functions I dc, I 1v, and I 2v of zeroth, first, and second harmonics in Alignment I, together with the calibration constants J 0 , J 1 , and J 2 , determine the orientation angle and the reduced dichroism via the Jones’ calculus. The results are

x 2 5 tan21

F

S

R 1v J 2~ A! z R 2v J 1~ A!

D

ÎS D S D G 2

d 0 5 tanh21 sgn$R 2v%

[1]

R 1v R2v 1 2J1~ A! 2J2~ A!

2

,

[2]

in which the ratios detected in the oscilloscope can be expressed as R 1v 5

I 1v 5 22 tanh d 0 sin 2 x 2J 1~ A! I dc FIG. 2. Transmission electron microphotograph of the mono-dispersed silica particles. The mono-dispersed particles were obtained from the composition of [TEOS] 5 0.57 M, [NH3] 5 0.71 M, [H2O] 5 1.648 M, and [MPTES] 5 0.097 M.

and R 2v 5

I 2v 5 22 tanh d 0 cos 2 x 2J 2~ A!. I dc

From the reduced dichroism of a given sample for the light signal of wavelength l, the flow-induced dichroism of the sample can be determined as Dn0 5

d0 l . 2p d

[3]

All the samples were loaded on a flow cell of the cone and plate geometry with a cone angle of 2° and an optical path length d of 0.52 mm. The cone and plate flow cell was made of quartz glass for the He–Ne laser beam to pass effectively. A uniform shear field was generated within the cone and plate geometry with the aid of a computer-controlled step-motor, and the shear rate was varied from 0.1 to 400 s21. The applied shear rate was determined from the rotational velocity of the plate divided by the cone angle. RESULTS AND DISCUSSION

1. Characterization of Particles and Suspensions In general, the sol– gel reaction of the alkoxide TEOS under a basic environment produces the spherical particles, while a

three-dimensional network structure is formed under an acidic environment. The morphology and the size distribution of the spherical silica particles depend largely on the concentration of ammonium hydroxide concentration, the types of alcohol, and the water content. For illustrative purposes, the TEM microphotograph of the synthesized mono-dispersed particles coated with silane coupling agent MPTES is reproduced in Fig. 2. The mono-dispersed particles were about 200.5 nm in radius and exhibited a very narrow size distribution with the standard deviation of 3.1 nm. Previous studies reported that these coating layers were very thin and usually less than 1 nm in thickness (24). Indeed, an almost negligible change in the particle size was observed in the present study during the coating process. The adsorbed layer of silane coupling agents was characterized by elemental analysis, and the results were included in Table 1. Distinct changes in both the contents of carbon and hydrogen and the hydrogen-to-carbon ratio were observed after treatment of the silane coupling agents. It can be noted that the carbon content was increased by the treatment with VTES or MPTES. The surface coverage of the adsorbed molecules was estimated from the increased amount of carbon. Although the number of the chemically adsorbed layers was not determined,

401

SILICA PARTICLE SUSPENSION

2. Rheological Behaviors

FIG. 3. Relative viscosity of a dilute silica suspension in THFFA solution as a function of the volume fraction.

the surface coverage of VTES was much larger than that of MPTES. It is also obvious that the adsorbed organic layer of the silane coupling agents could render the proper steric stabilization between the neighboring particles. However, the surface coverage alone is not sufficient to tell which silane agents would be more favorable for the stabilization of silica since the chain length of VTES is shorter than that of MPTES. To confirm the stability enhanced by the silane coupling agents, it is useful to measure the rheo-optical properties in addition to the conventional methods such as sedimentation ratio and turbidity tests. To determine the final particle volume fraction in the suspension, the density of silica particles should be measured somehow from the present experiment. One of the popular methods of determining the suspended particle density is the application of the Einstein equation for the suspension viscosity, which is strictly valid for a low-volume fraction. According to Einstein equation, the relative viscosity, i.e., the ratio of the suspension viscosity (h) to the continuous phase viscosity (hs) is linearly proportional to the particle volume fraction (f), and the proportionality constant is 2.5. In Fig. 3, the relative viscosity is plotted as a function of the particle volume fraction for the silica particles suspended in THFFA. Indeed, the bestfitted data with the theoretical slope of 2.5 was obtained when the particle density was 1.6 g/ml at room temperature. This value was very close to the previously reported results by Philipse and Vrij (35) and Bender and Wagner (23, 24). Finally, the concentrated suspensions with the volume fractions f 5 0.35 and f 5 0.45 of silica particles could be prepared in THFFA solvent with the measured density.

Significant deviations from Newtonian behaviors were induced by the combined contributions from the hydrodynamic and non-hydrodynamic inter-particle interaction and random Brownian motions under a uniform shear field. It can be simply expected that increments of the particle content decrease the inter-particle distance and consequently lead to considerable changes in the flow properties of the suspension. Thus, the rheological behaviors of the concentrated suspensions may be very complicated compared to those of other homogeneous fluids. In Figs. 4 and 5, the steady shear viscosity is plotted as a function of the shear rate for the suspensions with the particle volume fractions f 5 0.35 and f 5 0.45, respectively. The particles in the suspensions considered in Figs. 4 and 5 were treated through either VTES or MPTES coating. Also included in Figs. 4 and 5 is the shear viscosity versus the shear rate for the untreated particle suspensions with the same volume fractions. It can be easily seen from these figures that all three types of suspensions exhibited similar non-Newtonian behaviors of shear thinning, and that each suspension possessed a viscosity plateau at high shear rates independently of the surface modifications. The viscosity plateau or the limiting viscosity at high shear rates has been a subject related to the suspension rheology and has been discussed in the literature. For example, Krieger and Dougherty and Ball and Richmond proposed a simple expression for the limiting shear viscosity,

S

h f 5 12 hs fm

D

2@h#fm

,

[4]

FIG. 4. Steady shear viscosity as a function of the shear rate for three different suspensions with no treatment, with VTES treatment, and with MPTES treatment. The particle volume fraction was fixed at 0.35.

402

LEE AND YANG

FIG. 5. Steady shear viscosity as a function of the shear rate for three different suspensions, with no treatment, with VTES treatment, and with MPTES treatment. The particle volume fraction was fixed at 0.45.

in which f and fm denote the particle volume fraction and the maximum packing density, respectively, and [h] is the intrinsic viscosity of the suspension. In [4], the power, [h]fm, is known to be about 2 (6 – 8). In general, the random close packing density fm of a perfectly hard-sphere dispersion was reported as 0.63 when the shear rate was near zero while fm 5 0.71 was reported at infinitely high shear rates. As noted, the concentrated particle suspension prepared in the present study experienced a typical shear thinning and the viscosity plateau at high shear rates. However, the zero shear viscosity, which is another typical rheological property of a polymer melt or solution, was not measurable for the shear window in our ARES system. The maximum packing fraction calculated from [4] was 0.69, which was lower than the theoretical value of 0.71 for a hard-sphere suspension at infinite shear rates. The slight deviation from the hard-sphere value is due mainly to the existence of a small amount of electric charges on the silica particle surface, although each particle was very close to the hardsphere. Indeed, the z-potential of the silica particle was too small to be measured from the electrophoretic mobility in our Brookhaven instrument. However, as noted by Bender and Wagner, the weak electrostatic forces played on important role in increasing the high-shear limiting viscosity through repulsion between the neighboring particles (24). Thus, the small amount of residual charges at the particle surface was responsible for the lower value of the maximum volume fraction. It can be clearly seen from Fig. 5 that, although three different suspensions exhibited similar behaviors of shear thinning and limiting shear viscosity at high shear rates, a smooth shear thin-

ning was observed at a higher volume particle fraction, f 5 0.45, only in the MPTES-coated suspension. This is due to the fact that the coated layer of MPTES on the silica surface provided the effective steric hindrance which in turn stabilized the particle suspension. Meanwhile, the suspensions with no surface treatment were less stable than the surface-modified suspensions and likely to form the aggregated microstructure. Consequently, the untreated suspension displayed a considerably high shear viscosity, and was undergoing an abrupt shear thinning at very low shear rates. These phenomena were either reversible or irreversible and were observed frequently in heterogeneous systems like flocculated suspensions (14, 16). The present viscosity data were reproducible without appreciable deviation, which was indicative of the microstructural reversibility under a shear field. When the shear rate was low, the untreated particles tended to form a slightly aggregated or networked structure. Thus, the shear viscosity of the untreated particle suspension became higher than that of the stabilized suspension at low shear rates. However, the aggregated microstructures were broken up gradually as the shear rate (or equivalently the shear stress) was increased. Thus, the suspensions exhibited similar high shear limiting behaviors whether the suspensions were stabilized or not. The stabilized suspension with the coated VTES layer experienced a noticeable transition in the shear viscosity versus the shear rate as the particle volume fraction was changed. In particular, the silane coupling agent VTES became the less effective in the suspension stability for the higher particle volume fraction. As seen from Fig. 4, the stabilized suspensions with VTES and MPTES exhibited similar trends in the shear viscosity versus the shear rate for the particle volume fraction f 5 0.35. It can be also noted from Fig. 5, however, that the shear viscosity of the suspension coated with VTES approached that of the untreated bare particle suspension when the particle volume fraction increased up to f 5 0.45. Thus, the coupling agent VTES could not render effective steric stabilization, especially when the inter-particle distance became short. This deficiency was attributed to the short hydrocarbon chain of VTES compared with MPTES. 3. Flow-Induced Dichroism When the flow field is applied to the suspensions, the suspended particles are aligned to the flow direction and produce the flow-induced dichroism and birefringence. In general, the intensities of dichroism and birefringence for a spherical-particle suspension are very weak compared with those of anisotropic particle suspensions or concentrated polymer solutions. For a highly concentrated spherical-particle suspension, however, the dichroism and birefringence induced by flow can be measured under a sufficiently strong flow-field. In the present work, the intensity of the flow-induced birefringence of the silica suspension was too weak to be measured by the rheooptical alignments. Meanwhile, the flow-induced dichroism of the prepared suspensions could be measured in the shear field

SILICA PARTICLE SUSPENSION

403

FIG. 6. Flow-induced dichroisms as a function of the shear rate for two different volume fractions f 5 0.35 and f 5 0.45: (a) suspensions with no treatment, (b) suspensions with VTES treatment, and (c) suspensions with MPTES treatment.

of the cone and plate flow cell for the particle volume fractions f 5 0.35 and f 5 0.45. The flow-induced dichroism of the untreated suspension is illustrated as a function of the applied shear rate in Fig. 6a for the particle volume fractions f 5 0.35 and f 5 0.45. Also included in Figs. 6b and 6c were the flow-induced dichroisms of the stabilized suspensions with VTES and MPTES for the same set of parameters as in Fig. 6a. It can be readily noted from these figures, 6a– 6c, that the flow-induced dichroisms of the three different suspensions behaved quite differently one another as the shear rates increased gradually. Specifically, both of the untreated suspensions with f 5 0.35 and f 5 0.45 exhibited maximum peaks in the flow-induced dichroisms and underwent their sign changes at high shear rates of about 300 s21, as shown in Fig. 6a. In general, the unstable responses of the dichroism and

birefringence were observed for the heterogeneous systems including block copolymers, aggregated colloidal suspensions, and immiscible polymer blends together with suspensions or polymer solutions which exhibit shear thickening (14, 26, 31, 38 – 41). Furthermore, these heterogeneous systems give rise to the sign change in the dichroism and birefringence (31). This is because of the form effect arising from the formation of nonhomogeneity such as microdomains or aggregates (26, 29, 31, 38 – 41). Thus, the general features contained in Fig. 6a are indicative of the fact that the suspensions with no surface treatment are likely to form an aggregated structure due to the absence of steric repulsion between the particles. Further, the sign changes in the measured dichroisms at high shear rates imply that the flow stabilized the untreated suspension by causing the microstructural transition from an aggregated state

404

LEE AND YANG

to the individual particulate state aligned to the direction of shear field. Thus, the present rheo-optical behavior of the untreated suspension is consistent with the unstable shear viscosity behavior discussed previously. On the other hand, the VTES-coated suspensions exhibited quite different behaviors from those of the untreated suspension as seen in Fig. 6b. First, the sign change in the flowinduced dichroism, which occurred only at higher volume fraction f 5 0.45, was not detected at lower volume fraction f 5 0.35 for the VTES-coated suspension. This is clear evidence that the weakly structured aggregates were formed at higher volume fractions due to the deficiency in the steric stabilization of the VTES coating. Because the particle–particle distance became shorter for higher volume fractions, the coated layer of VTES at f 5 0.45 was not so effective as it was at f 5 0.35 in the steric stabilization. Second, the threshold shear rate at which the sign change occurred in the VTEScoated suspension of f 5 0.45 was about 10 s21. Thus, the shear-induced microstructural transition in the VTES-treated suspension occurred at relatively weak shear flow compared to that in the untreated suspension. This implies clearly that the aggregates formed in the VTES-treated suspension were not so structurally strong as those in the untreated suspension. As shown in Fig. 6c, both of the MPTES-coated suspensions with f 5 0.35 and f 5 0.45 were structurally stable and exhibited the homogeneous responses of flow-induced dichroisms. The flow-induced dichroism did not experience any sign change as the shear rate increased. Thus, the MPTES coating provided an effective stabilization by generating the proper steric repulsion between the particles, especially for higher particle volume fractions. The general features obtained from the rheo-optical study on the surface modification effects agree well with the rheometry results discussed earlier. The origin of the sign change in the flow-induced dichroism could be checked from the raw spectra detected in an oscilloscope, which are displayed in Figs. 7a–7c for the three types of silica suspension. It can be readily seen from these figures that at a shear rate as high as 290.7 s21, all types of suspension showed similar trends in the signals of R 2 v 5 I 2 v /I dc as a function of the elapsed time after the flow started. This is because either the treated or the untreated particles tend to fully align toward the flow direction under the sufficiently strong flow to break up the weakly aggregated structures. However, the aggregated structures of the untreated suspensions with either f 5 0.35 or f 5 0.45 were prolonged under the shear rate as low as 58.23 s21. Meanwhile, the VTES-coated suspension with only f 5 0.35 or both of the MPTES-coated suspensions with f 5 0.35 and f 5 0.45 displayed stable spectra at the low shear rate 58.23 s21, as noted in Figs. 7b and 7c. These structural behaviors of the silica suspensions can be illustrated schematically, as in Fig. 8. Under a weak shear rate, the hydrodynamic-interaction force is not as strong as the interaction force causing the aggregation, and thus particles behave as a form of aggregates. In a high shear regime,

however, the hydrodynamic-interaction force is strong enough to break up the particle aggregates and each particle may move as an isolated particle. Therefore, all the suspensions exhibited similar behaviors at high shear rates. Although the corresponding plot is not reproduced in this paper, the R 1 v 5 I 1 v /I dc observed through the oscilloscope showed the spectra opposite to R 2 v in sign. Since the sign change in dichroism causes both the sign changes in R 1 v and R 2 v at the same time, the signs of both signals are always opposite each other. This means that the orientation angle x2 was in the range from 245° to 0° and 2x2 remained in the fourth quadrant for the present flow geometry. Therefore, the sign change in the measured dichroism for the non-stabilized suspensions or the VTES-treated suspension of f 5 0.45 was caused by the form effect of phase separated or aggregated microstructure rather than by the jump of 2x2 to another quadrant. In addition, that the concentration fluctuations of the suspended particles switch from the vorticity to the flow direction with the strength of flow may be also responsible for the sign change in the dichroism (40, 41). 4. Correlation of the Stress Optical Rule As mentioned in the preceding section, the flow-induced dichroism and birefringence give a picture of the microstructure evolution in the particle suspension. In general, macroscopic rheological properties can be correlated to the microscopic dichroism and birefringence through the stress-optical rule (SOR) (19, 22–24, 26, 31). However, the breakdown of the SOR may be encountered in either the multi-component systems like block copolymers in which the form effects are significant or the state near or passing through a structural transition. In the present work, the untreated suspension with f 5 0.35 and f 5 0.45 or the VTES-coated suspension with higher volume fraction of f 5 0.45 showed the unstable behaviors owing to the formation of aggregates or microdomains. Thus, the form effects arising from the formation of micro-phase separation lead to the breakdown of the linear SOR for these non-stabilized suspensions. Therefore, the classical linear SOR can be applied only to the stabilized suspensions, such as a VTES-coated suspension with f 5 0.35 and MPTES-coated suspensions with f 5 0.35 and f 5 0.45. However, the acceptable constant stress-optical coefficient (C) for the stabilized suspension could not be obtained through a direct correlation between the shear stress and the flow-induced dichroism. To overcome the failure of the correlation of the SOR, Bender and Wagner (23, 24) proposed the modified SOR by considering separately the hydrodynamic and thermodynamic contributions as

h total 5 h thermo 1 h hydro 5

S D

1 Dn0 1 h hydro, C 12 g˙

[5]

in which C 12 is the stress optical coefficient relating the shear stress to the dichroism measured through the vorticity direc-

SILICA PARTICLE SUSPENSION

405

FIG. 7. Spectra of R2v 5 I2v/Idc as a function of the elapsed time at two different shear rates, 58.23 and 290.7 s21. The intensity of the R2v signal of each run measured at the oscilloscope was reproduced here on the same scale: (a) suspensions with no treatment, (b) suspensions with VTES treatment, and (c) suspensions with MPTES treatment.

tion. Thus, their modification was based on the theoretical considerations of the optical structure factor that only thermodynamic contributions could be correlated to the SOR. If the modified SOR is valid, the plot of total viscosity versus (Dn0/ g˙ ) will turn out a single line. In this case, the viscosity contributed from the purely hydrodynamic-interaction effect

can be calculated from the intersection with the total viscosity axis in the plot. In general, the flow dichroism and birefringence depend on the direction of light path relative to the flow. Suppose that the flow is in “1” direction with the velocity gradient in “2” direction in the typical Couette flow cell. Then, for the flow

406

LEE AND YANG

FIG. 8. Schematic representation of a non-stabilized suspension containing weakly aggregated structures. When the flow was weak, the micro-domains were formed by aggregation of nearby particles. In the strong flow regime, however, the micro-domain structures were broken up and the particles tended to align to the flow direction.

between two concentric cylinders, the light beam passes usually parallel to the vorticity direction (i.e., neutral “3” direction). In this case, the flow dichroism or birefringence measured through the 1,2-plane can be related directly to the shear stress (s12) by a linear SOR. Meanwhile, for the slit or coneand-plate geometry which has the light passed through the velocity gradient direction, the flow dichroism measured through the 1,3-plane cannot be applied directly to the SOR. To do so, the rheo-optical properties measured on the 1,3-plane should be related to those on the 1,2-plane as performed by Wales and Janeschitz-Kriegl (42, 43). Their rheo-optical studies for polymer melts in various flow configurations showed

that the flow birefringence on the 1,3-plane, ^n 11 2 n 33 & had the linear relationship to the square root of s12 and almost same value of ^n 11 2 n 22 &. In other words, the two normal stress differences of (s11 2 s22) and (s11 2 s33) were almost identical, and the normal stress difference (s11 2 s33) was proportional to the square root of s12. Thus, their results were consistent with a linear viscoelastic theory such as the Maxwell model in which the plot of (s11 2 s33) versus s12 gives rise to the slope of factor 2 in log–log scale as the Cole-Cole plot of G9 vs G0 (44, 45). Consequently, ^n 11 2 n 33 & is proportional to the square root of s12 and n 12 } =|^n 11 2 n 33 &| for a linear viscoelastic material. By making an analogy between flow-induced birefringence and dichroism, we can correlate the flow dichroism measured on the 1,3-plane to that on the 1,2-plane. Then the SOR proposed by Bender and Wagner can be expressed as

h total 5 h thermo 1 h hydro 5

1 C 13

SÎ

D

|^n 011 2 n 033&| 1 h hydro, g˙

[6]

in which C 13 is the stress optical coefficient which relates s12 to the dichroism measured through the 1,3-plane. In Fig. 9a, the total viscosity is plotted as a function of =|^n 011 2 n 033 &|/ g˙ for the VTES-coated suspension with f 5 0.35. Also shown in Fig. 9b are the same plots as in Fig. 9a for the MPTES-coated suspensions with f 5 0.35 and f 5 0.45. As expected, the linearity of this correlation [6] was confirmed successfully for the stabilized suspensions considered here, and the hydrodynamic-interaction contributions to the viscosity were measured from the intersections with the vertical axes of Figs. 9a and 9b.

FIG. 9. Total viscosity as a function of =|^n011 2 n033&|/g˙ . The intersection with the vertical axis was the viscosity contributed from the purely hydrodynamic interaction: (a) VTES-coated suspension of f 5 0.35 and (b) MPTES-coated suspensions of f 5 0.35 and f 5 0.45.

SILICA PARTICLE SUSPENSION

407

FIG. 10. Calculated stress-optical coefficients C13 as a function of the shear rate: (a) VTES-coated suspension of f 5 0.35, (b) MPTES-coated suspensions of f 5 0.35, and (c) MPTES-coated suspensions of f 5 0.45.

To show the validity of our extracted values of the hydrodynamic viscosity, we compared the values with the previous results. However, until now, only few reports have been available for separating two contributions through computer simulation (46) or well-defined theoretical study (23, 24). Bender and Wagner reported the value of ca. 0.05 Pas for hhydro (hr > 8.6) for f 5 0.52 and the value of near O (1021Pas) (hr > 20) for f 5 0.65, both of which were extracted before shear thickening (23, 24, 46). Meanwhile, the relative viscosity of the purely hydrodynamic contribution calculated from the “Stokesian dynamics” of Brady

and co-workers (46) was about 5 for f 5 0.45. In the present work, the estimated hydrodynamic viscosity was 0.0345 Pas (hr 5 5.56) for the VTES-treated suspension of f 5 0.35, 0.0384 Pas (hr 5 6.19) for the MPTES-treated suspension of f 5 0.35, and 0.0481 Pas (hr 5 7.75) for the MPTES-treated suspension of f 5 0.45. Hence, our results for the hydrodynamic contribution to the shear viscosity were in good agreement with the previously reported values. The stress-optical coefficient, C 13 , can then be expressed from considering the modified SOR [6] as follows:

408

LEE AND YANG

C 13 5

SÎ

|^n 011 2 n 033&| g˙

DS

D

1 . h total 2 h hydro

[7]

However, if we attempted to correlate directly the stress with the flow-induced dichroism without any modification, the stress-optical coefficient would be expressed as C 13 5

SÎ

D

|^n 011 2 n 033&| 1 . h total g˙

[8]

In Fig. 10a, the stress-optical coefficients given by [7] and [8] are illustrated as a function of the shear rate for the VTEScoated suspension with f 5 0.35. Likewise, the same plots are contained in Figs. 10b and 10c for the MPTES-coated suspensions with f 5 0.35 and f 5 0.45, respectively. As clearly seen from these figures, the modified SOR provided an accurate stress-optical coefficient for each stabilized suspension prepared in the present study. Without modification, a theoretically constant stress-optical coefficient would decrease monotonically as the shear rate increased. On the other hand, the stress-optical coefficient from the modified SOR sustained almost a constant value independent of the shear rate for the stabilized suspensions. Also noted from Figs. 10a–10c is the slight deviation from a constant value at high shear rates. This is due to the pronounced hydrodynamic-interaction contributions at strong flow regime to the total viscosity compared with thermodynamic contributions. The difference in the stress-optical coefficients of the VTESand MPTES-coated suspensions with the identical volume fraction f 5 0.35 was caused by the difference in the silica surface moieties. In particular, the refractive indices of the VTES and MPTES phases are 1.397 and 1.427, respectively, at room temperature, which are different from 1.45 of the solvent THFFA (37). For this reason, the VTES-coated suspension exhibited a little lower value of C than the MPTES-coated suspensions. Through the modified SOR, it was confirmed that the contribution from hydrodynamic interaction should be separated in the analysis of the stress-optical behaviors of a colloidal suspension, especially at a high shear region. By doing this, the more accurate SOR could be obtained at high shear rates. CONCLUSION

Mono-dispersed silica particle suspensions were prepared by the sol– gel method to examine the rheo-optical behaviors of the silica particle suspensions. The effects of surface modification on the suspension stability were also considered for different particle volume fractions f 5 0.35 and f 5 0.45. The conclusions from the present investigations are as follows: ● A flow-induced microstructure was possibly detected by the rheo-optical measurements. For the untreated bare particle suspension, the micro-phase domains or the particle aggregates

which were present at low shear rates were broken up as the shear rate increased. The microstructural transition was confirmed by both the flow-induced dichroism and the shear viscometry. ● Silane coupling agents such as VTES and MPTES were effective in enhancing the suspension stability. Surface modification with the silane coupling agent MPTES gave rise to the effective steric repulsion between the particles even at relatively high volume fractions. Consequently, the MPTEScoated suspensions showed smooth shear thinning and homogeneous rheo-optical behaviors. ● The VTES-coated suspension sustained its stability only at low volume fractions and exhibited unstable rheo-optical behaviors at higher volume fractions due to the diminishing stabilization effects of the coated VTES layer. At high shear rates, however, all the suspensions considered here possessed the limiting high-shear viscosities due to the predominant hydrodynamic interactions. ● The modified stress-optical rule, which reflects the thermodynamic contributions separately from the hydrodynamicinteraction contributions, provided an accurate correlation between the rheological properties and the optical anisotropy in the stabilized silica suspensions. ACKNOWLEDGMENT This work has been supported by the Korea Science and Engineering Foundation under Grant KOSEF 951-1105-1-011-2.

REFERENCES 1. Batchelor, G. K., J. Fluid Mech. 41, 245 (1970). 2. Batchelor, G. K., and Green, J. T., J. Fluid Mech. 56, 375 (1970). 3. Furth, R., and Cowper, A. D., “Investigations on the Theory of the Brownian Movement by A. Einstein.” Dover, New York, 1956. 4. Bossis, G., and Brady, J. F., J. Chem. Phys. 91, 1866 (1989). 5. Brady, J. F., J. Chem. Phys. 99, 567 (1993). 6. Krieger, I. M., and Dougherty, T. J., Trans. Soc. Rheol. 3, 137 (1959). 7. Krieger, I. M., Trans. Soc. Rheol. 7, 101 (1963). 8. Ball, R., and Richmond, P., J. Phys. Chem. Liquids 9, 99 (1980). 9. de Kruif, C. G., van Iersel, E. M. F., Vrij, A., and Russel, W. B., J. Chem. Phys. 83, 4717 (1985). 10. Russel, W. B., and Gast, A. P., J. Chem. Phys. 84, 1815 (1986). 11. van der Werff, J. C., and de Kruif, C. G., J. Rheol. 33, 421 (1989). 12. Metzner, A. B., and Whitlock, M., Trans. Soc. Rheol. 2, 239 (1958). 13. Hoffman, R. L., Trans. Soc. Rheol. 16, 155 (1972). 14. Toy, M. L., Scriven, L. E., Macosko, C. W., Nelson, Jr., N. K., and Olmsted, R. D., J. Rheol. 35, 887 (1991). 15. Boersma, W. H., Laven, J., and Stein, H. N., AIChE J. 36, 321 (1990). 16. Barnes, H. A., Hutton, J. F., and Walters, K., “An Introduction to Rheology.” Elsevier, New York, 1989. 17. Mellema, J., van der Werff, C. G., Blom, C., and de Kruif, C. G., Phys. Rev. A 39, 3696 (1989). 18. Kao, S. V., and Mason, S. G., Nature 253, 619 (1975). 19. Fuller, G. G., “Optical Rheometry of Complex Fluids.” Oxford Univ. Press, New York, 1995. 20. Frattini, P. L., and Fuller, G. G., J. Colloid Interface Sci. 100, 506 (1984). 21. Johnson, S. J., Frattini, P. L., and Fuller, G. G., J. Colloid Interface Sci. 104, 440 (1985).

SILICA PARTICLE SUSPENSION 22. Wagner, N. J., Fuller, G. G., and Russel, W. B., J. Chem. Phys. 89, 1580 (1988). 23. Bender, J., and Wagner, N. J., J. Rheol. 40, 899 (1996). 24. Bender, J. W., and Wagner, N. J., J. Colloid Interface Sci. 172, 171 (1995). 25. D’Haene, P., Mewis, J., and Fuller, G. G., J. Colloid Interface Sci. 156, 350 (1993). 26. Lee, J.-D., and Yang, S.-M., HWAHAK KONGHAK 35, 782 (1997). 27. KishBaugh, A. J., and McHugh, A. J., Rheol. Acta. 32, 9 (1993). 28. Pearson, D. S., Kiss, A. D., and Fetters, L. J., J. Rheol. 33, 517 (1989). 29. Hongladarom, K., and Burghardt, W. R., Macromolecules 26, 785 (1993). 30. Kim, W.-J., Yang, S.-M., and Kim, M., J. Colloid Interface Sci. 194, 108 (1997). 31. Macosko, C. W., “Rheology.” VCH, New York, 1994. 32. Stober, W., Fink, A., and Bohn, E., J. Colloid Interface Sci. 26, 62 (1968). 33. Bogush, G. H., Tracy, M. A., and Zukoski, C. G., J. Non-Cryst. Solids 104, 94 (1988). 34. van Helden, A. K., Jansen, J. W., and Vrij, A., J. Colloid Interface Sci. 81, 354 (1981).

409

35. Philipse, A. P., and Vrij, A., J. Colloid Interface Sci. 128, 121 (1989). 36. Tadros, Th. F., J. Colloid Interface Sci. 64, 36 (1978). 37. Korea Shin-Ethu Silicone Co., “Product Catalog of Silane Coupling Agents.” Korea, 1994. 38. Onuki, A., and Doi, M., J. Chem. Phys. 85, 1190 (1986). 39. Fredrickson, G. H., Macromolecules 20, 3017 (1987). 40. Yanase, H., Moldenaers, P., Mewis, J., Abetz, V., van Egmond, J., and Fuller, G. G., Rheol. Acta. 30, 89 (1991). 41. Moldenaers, P., Yanase, H., Mewis, J., Fuller, G. G., Lee, C.-S., and Magda, J. J., Rheol. Acta. 32, 1 (1993). 42. Wales, J. L. S., “The Application of Flow Birefringence to Rheological Studies of Polymer Melts.” Delft Univ. Press, Netherlands, 1976. 43. Janeschitz-Kriegl, H., “Polymer Melt Rheology and Flow Birefringence.” Springer-Verlag, Germany, 1983. 44. Han, C. D., and Jhon, M. S. J. Appl. Polym. Sci. 32, 3809 (1986). 45. Han, C. D., J. Appl. Polym. Sci. 35, 167 (1988). 46. Brady, J., in “Particulate Two-Phase Flow” (M. C. Roco, Ed.) Chap. 26, Butterworth-Heinemann, 1993.