- Email: [email protected]
Rheology and Structure of Flocculated Iron Oxide Suspensions
JOURNAL OF COLLOID AND INTERFACE SCIENCE ARTICLE NO.
180, 200–211 (1996)
0290
Rheology and Structure of Flocculated Iron Oxide Suspensions REINALDO C. NAVARRETE, 1 L. E. SCRIVEN,
AND
CHRISTOPHER W. MACOSKO
Center for Interfacial Engineering and Department of Chemical Engineering and Materials Science, University of Minnesota, Minneapolis, Minnesota 55455 Received July 13, 1995; accepted December 27, 1995
This paper focuses on the relation between structure and rheology of flocculated suspensions. Rheological measurements were taken both in shear and nearly extensional flows. The structures of magnetic iron oxide suspensions were visualized using cryogenic scanning electron microscopy ( Cryo-SEM ) and videoenhanced light microscopy ( VELM ) under shear, elongational, and rotational flows. Shear fields were generated by using a miniature Couette geometry, whereas elongational and rotational fields were generated by an innovative technique, which takes advantage of a surface tension gradient on the interface of an air bubble, and were visualized in real time under a light microscope. In Couette flow, the particle networks formed at high concentrations ( 9.4% by volume ) do not uniformly break up into smaller units of particles, as at lower concentrations, but rather break up at the center of the gap, segregating particles toward the walls. Real-time light microscopy observations of a spinning floc show that the deformation and breakup of flocs is a process that gradually occurs with time. In extensional flow, flocs elongate and break up, explaining the apparent extensional thinning behavior measured. Apparent extensional viscosities were one to two orders of magnitude larger than shear viscosities at the same applied stress. q 1996 Academic Press, Inc. Key Words: rheology; suspensions; colloids; flocculation; iron oxide; microscopy; extensional flows.
INTRODUCTION
The rheological behavior of colloidal suspensions is determined by the structures that particles form in suspension (1–10). Structures generally mean some sort of association of particles as extensive as the sample itself or nearly so, the association having something of the character of an elastic network. When there are long-range attractive forces and short-range repulsive forces, it is well known that the net force between two colloidal particles can vanish at separations determined by the interparticle forces (11). Departures from those separations are accompanied by elastic-like restoring forces, provided the departures are not too great. This 1 To whom correspondence should be addressed. Present Address: Schlumberger Dowell, Inc., P.O. Box 2710, Tulsa, OK 74101.
200
0021-9797/96 $18.00 Copyright q 1996 by Academic Press, Inc. All rights of reproduction in any form reserved.
AID
JCIS 4197
/
6g10$$$501
gives rise to ‘‘bonds’’ between particles; i.e., two particles are bonded to each other if there are forces that resist changes in their relative position. There can be different structural levels of bonding. At each of these levels, the strength of the bonds (i.e., the magnitude of the force necessary to separate two particles enough so that there is no resisting force to further changes in relative particle positions) between the structural units varies according to the length scale of the structure level. The degree of permanence of each structure level depends on the externally imposed disturbing forces and on the strength of the bonding at that level of structure. The relation between structure and rheology of flocculated suspensions can be manifested in different forms of rheological behavior. For example, elasticity implies the presence of a sample-spanning elastic structure, yield stress implies that the connectivity of the structure is extensively ruptured, shear thinning implies further disconnection as the rate of shear is raised, and time dependence implies that disconnection and reconnection proceed at detectable slow rates. There are a number of studies that propose structural models for flocculated suspensions. Michaels and Bolger (1) put forward a network model to interpret the yield stress of aqueous, flocculated kaolin suspensions and derived equations to predict the effects of kaolin concentration and shear rate. The basic flow units in the model are flocs, which are small associations of particles plus the water they enclose. These basic units are supposed to be the flocs that persist at the highest shear rate to which the system has been subjected. They have elastic character due to the forces acting between the particles that form the floc. At low shear rates, the flocs group into floc aggregates, and these in turn may interact to form a sample-spanning or effectively ‘‘infinite’’ network that is responsible for the elastic character of the suspension. When an external deformation of sufficiently large magnitude is applied to the network, it breaks up into floc aggregates or flocs depending on the magnitude of the applied deformation. In a series of papers, Firth and Hunter (4–6) advanced a somewhat similar model for coagulated sols which they called the ‘‘elastic floc model.’’ Van de Ven and Hunter (7)
05-13-96 20:41:08
coida
AP: Colloid
FLOCCULATED IRON OXIDE SUSPENSIONS
went a step further in modeling the structure of the flocs and introduced an additional structural level. They proposed that the flocs consisted of a number of smaller domains which they called flocculi. The flocculi were supposed to have been formed at the highest shear rate to which the system had ever been subjected, whereas the flocs and the floc aggregates were supposed to arise during the shear history after the flocculi themselves were formed. Following Hunter’s model, Kanai et al. (12) formulated a floc network model and used elastic percolation concepts to explain the scaling dependency of elasticity of iron oxide suspensions as a function of concentration. Non-zero elastic shear modulus appeared at low threshold particle volume fractions, indicating that flocs of iron oxide particles were very low density structures (2–5% by volume). Above the threshold concentration, the shear dynamic elastic modulus G * was found to follow a scaling law of the type G * Ç ( f f 0 f fc ) f , where f f is the floc volume fraction, f fc is the threshold floc volume fraction, and f is the scaling exponent. Buscall et al. (13, 14) studied the scaling relationship of the shear elastic modulus, the compressive strength and shear yield stress with the particle volume fraction for aggregated colloidal silica particle gels. The scaling exponent obtained suggested that the aggregation mechanism by which the particle network was formed was diffusion limited floc–floc aggregation, where the flocs were fractal in nature. Fractal geometry has been extensively used in the recent past to characterize aggregates of colloidal particles (15, 16). Uniformly sized microspheres interacting via longrange magnetic dipolar forces have also been reported to form clusters with a fractal dimension 1.16, i.e., clusters with pronounced chain like structures (17). Sonntag and Russel (18–20) in a series of papers investigated by small angle light scattering the effect of shear and extensional flows on the size and fractal dimension of polystyrene flocs. Under strong shear conditions they found that the size of the flocs decreased significantly under shear, in agreement with prior visual observations by Kao and Mason (21) of noncolloidal particle associations. The fractal dimension of the flocs under shear was found to be 2.48, which can be considered high. The implication is that smaller flocs, where the outer branches have been eliminated by hydrodynamic stresses, are closer to uniformly dense particles than those flocs formed under little or no shear. On the other hand, little variation in the floc size and structure was found under nearly extensional flows (flow through an abrupt contraction), in contrast to earlier observations by Kao and Mason. The effect of hydrodynamic stresses on particle associations in suspension has also been investigated by means of Stokesian dynamics simulations (22, 23) and by Brownian dynamics simulations (24). However, these simulations are computationally intensive, and therefore are restricted to small numbers of particles and simple flows. The present paper focuses on the relation between struc-
AID
JCIS 4197
/
6g10$$$502
05-13-96 20:41:08
201
ture and rheology of flocculated suspensions. Past studies indicate that the flocculated structures of magnetic particles under shear can give rise to unexpected microscopic structures and rheological behavior (25). The approach taken here was to measure the rheological behavior of a model flocculated suspension (magnetic iron oxide suspensions) and to visualize the structures under different flow fields, namely shear, elongational and rotational, by means of cryogenic scanning electron microscopy (Cryo-SEM) and videoenhanced light microscopy (VELM). Shear fields were generated using a miniature Couette geometry, whereas elongational and rotational fields were generated with an innovative technique which takes advantage of a surface tension gradient on the interface of an air bubble under a light microscope. MATERIALS AND METHODS
Experimental System The experimental suspension system consisted of needlelike particles of magnetic iron oxide suspended in mineral oil. The particle density was 4500 kg/m 3 . The average length of the particles was about 0.5 mm and the average length to diameter ratio was 5; transmission electron micrographs of similar particles can be found in Yang et al. (26). The liquid medium was heavy mineral oil from Sergent-Welch Co., with viscosity equal to about 0.1 Pars (257C) and density of 863 kg/m 3 . The iron oxide particles were treated with an alkoxysilane surfactant, which reacts with the hydroxyl sites on the surface of the particles to prevent particle aggregation by chemical bonding (27). A ball mill was employed for the preparation of the suspensions. Mill vessels were plastic bottles, 6 cm in diameter, and milling media were 9.5-mm steel balls. Powder was mixed with the mineral oil in a 40% by weight (11.3% by volume) concentration and then milled for 24 h. This suspension was then diluted to several concentrations. Rheological Measurements The shear viscosity of iron oxide suspensions was measured by means of constant stress or creep tests. Creep tests were used, as opposed to constant shear rate tests, because they allow the breakup of the structure of the suspension by following the shear deformation with time. A Rheometrics Stress Rheometer (RSR) 8600 (Rheometrics Scientific, Inc., Piscataway, NJ), fitted with a Couette geometry with the dimensions of a 13-mm bob diameter, 14-mm cup diameter, and 40-mm bob length, was used for creep measurements. A larger diameter bob (13.5 mm) was also used in some cases. A sequence of creep tests of increasing stress was applied to each sample. In between the creep tests, a period of 300 s was allowed for the sample to recover, which was not recorded. The duration of each creep test was 600 s. Before a sequence of creep tests was initiated, a unidirec-
coida
AP: Colloid
202
NAVARRETE, SCRIVEN, AND MACOSKO
FIG. 1. Low magnification SEM of 9.4% by volume iron oxide suspension in miniature Couette geometry. (1) Glass inner cylinder. (2) Frozen suspension in the gap. (3) Aluminum outer cylinder.
tional preshear was applied at a stress of 220 Pa for 900 s followed by a rest period of 1800 s at zero stress. All measurements were made at room temperature. An opposed nozzles indexer was used to measure the extensional viscosity of iron oxide suspensions (28–30). This instrument has two nozzles of the same dimensions placed opposing each other and submerged into a beaker with the sample. Each nozzle is connected to a syringe by means of an arm and tubing. One of the arms is fixed and the other is connected to a transducer which measures the torque exerted on the arm. The syringes can suck the sample from the beaker or return the sample to it. Ideally, the result in the first case is an extensional biaxial flow and in the last case an extensional uniaxial flow. In practice, however, these flows are not purely extensional due to the shear contributions originating from the regions close to the walls of the nozzles. If the flows are assumed to be ideally extensional, it is possible to calculate an apparent extensional viscosity from the volume flow rate of the sample in or out of the nozzles and the resulting torque on the moving arm. The apparent extensional viscosity calculated in this way should be taken as an index of extensional behavior rather than as a true extensional viscosity. All extensional measurements were taken with a RFX (Rheometrics Scientific, Inc., Piscataway, NJ).
AID
JCIS 4197
/
6g10$$$502
05-13-96 20:41:08
The opposed nozzles indexer is essentially a rate-controlled instrument; i.e., the extensional rate is set and the resulting tensile stress is measured. To achieve steady state, an initial period was allowed before taking the tensile stress measurement. This time was determined by monitoring the approach to the steady state of the torque on the transducer arm and varied according to the extension rate from 50 s for the lowest rate on the highest volume particle concentration, to 10 s for the highest rate on the lowest concentration tested. The actual torque measurement was made in 15 s. The range of extensional rates tested was determined by the nozzle diameter; i.e., large nozzle diameters are required to achieve small rates. The limit is a measurable torque on the transducer arm. Small nozzle diameters are required to achieve high rates. The limitations are steady state within the available volume capacity of the syringes and avoiding overloading the torque transducer. The nozzle diameters used varied between 1 to 4 mm. Following a suggestion of Schunk et al. (29), the nozzle separation distance was kept equal to the nozzle diameter to minimize shear contributions to the resulting apparent extensional viscosity. Visualization Techniques Two experimental techniques were used to visualize the particle structures of the iron oxide suspensions, Cryo-SEM and VELM. These techniques are described below:
coida
AP: Colloid
FLOCCULATED IRON OXIDE SUSPENSIONS
203
FIG. 2. Two forms of inducing a surface tension gradient at the bubble’s surface: (a) using a hot plate below the concave slide and (b) focus of light at the bubble’s surface.
Cryogenic scanning electron microscopy (Cryo-SEM). Freeze-fracture scanning electron microscopy, usually named cryogenic scanning electron microscopy (CryoSEM), was used to visualize the structure of iron oxide suspensions. This technique was developed by Bellare et al. (31, 32), Sheehan and Scriven (33), and Sheehan and Whalen-Shaw (34), and enables visualization of submicron particles in a liquid suspending medium. The steps involved are (i) freeze a sample of suspension, (ii) fracture the frozen sample, (iii) coat the fractured surface with a conducting
FIG. 4. Strain vs time curves of a creep test on a 9.4% iron oxide suspension with a gap of (a) 0.50 mm and (b) 0.25 mm at 10 Pa applied shear stress (r Å correlation coefficient).
FIG. 3. Shear creep viscosities of iron oxide suspensions of indicated particle volume fractions.
FIG. 5. Comparison of creep steady viscosities of a 9.4% by volume g-Fe2O3 suspension in two different gaps.
AID
JCIS 4197
/
6g10$$$502
05-13-96 20:41:08
metal, and (iv) observe the fractured, coated surface with an SEM. These steps are described in the references given above.
coida
AP: Colloid
204
NAVARRETE, SCRIVEN, AND MACOSKO
FIG. 6. Comparison of extensional vs shear viscosities of (a) 3.3% and (b) 1.9% suspensions.
The sample preparation procedure was modified with respect to those used in previous works to study the effect of shear on the microstructure of the iron oxide suspensions. A miniature Couette geometry was used to shear the suspensions. The suspension sample was loaded in the outer aluminum cylindrical cup, which consisted of two parts: a bottom cup and an upper tube of the same diameter that fitted together (35). The aluminum cup was then mounted on a Rheometrics Stress Rheometer (Rheometrics Scientific, Inc., Piscataway, NJ). There an inner glass rod was lowered in such a way that the suspension filled the annular space left between the glass rod and the aluminum cup. The diameter of the glass cylinder was 3.3 mm, leaving an annular gap of about 400 mm. A constant torque was applied to the glass rod, equivalent to a shear stress of 50 Pa for 1800 s. Next, the glass cylinder was stopped and secured to the outer cup by means of two lateral screws. Then the entire miniature
AID
JCIS 4197
/
6g10$$$502
05-13-96 20:41:08
Couette geometry was dismounted from the stress rheometer and plunged into liquid nitrogen. This was done in less than 60 s to reduce the relaxation of the structure induced by shear. After freezing, the sample was stored in liquid nitrogen. At the time of imaging, the sample was placed on a hinged fracture stage and transferred into a cryo-fracture chamber under vacuum (7 1 10 06 kPa) and at liquid nitrogen temperature ( 01757C). This chamber was connected to a metalcoating Hexland CT 1000A chamber. The two chambers were separated by a gate valve and the coating chamber was connected to a JEOL JSM 840 SEM by a gate valve. These attachments are well described by Sheehan and WhalenShaw (34). There the sample was fractured by separating the two parts of the outer cup and fracturing the inner glass rod and the frozen suspension in the annulus (see Fig. 1). Samples were then placed in the metal coating chamber
coida
AP: Colloid
FLOCCULATED IRON OXIDE SUSPENSIONS
205
FIG. 7. Effect of shear on a 9.4% by volume g-Fe2O3 suspension after an applied stress of 50 Pa. (1) Rotating glass cylinder. (2) Bright region adjacent to the wall. (3) Highly sheared region. (4) Transition region. (5) Broken down region of low particle concentration. (6) Region adjacent to stationary wall. (7) Stationary outer aluminum cylinder. Backscattered electron image.
and a film of conducting metal of controlled thickness was sputtered on the fractured surface to prevent electron charge build-up during imaging. The metals used were gold, platinum, and chromium; the metal or combination for each sample was found by trial-and-error. Thickness also varied from sample to sample. Typical thicknesses ˚. varied from 25 to 35 A The sample was then transferred to the cryo-chamber ( 01757C and 7 1 10 06 kPa) of a JEOL JSM 840 SEM equipped with an LaB6 filament for imaging. Short working distances, large objective apertures, small condenser apertures, and high accelerating voltages were used to enhance the resolution and contrast between the particles and the oil (35). The micrographs shown here are backscattered electron emission images. Video-enhanced light microscopy (VELM). Video-enhanced light microscopy (VELM) was used to observe the evolution of the structures of iron oxide suspensions in real time under different flow motions. Microflows were generated in the oil by means of a surface tension gradient-driven flow. A Nikon Optiphot-Pol light microscope (Frank Fryer Co., Carpenter Ville, IL) equipped with a black-and-white television camera and connected to a video recorder was used (35). The sample was placed between a glass slide with a spherical
AID
JCIS 4197
/
6g10$$$502
05-13-96 20:41:08
depression and a flat cover slide under a light microscope, as shown in Fig. 2. The depression provided enough depth and visual access to the three-dimensional structures in the suspension. Microflows were generated by entrapping an air bubble below the upper flat slide. A surface tension gradient was then induced on the oil/air interface of the bubble by an externally imposed temperature gradient. The temperature gradient was induced either by a hot plate below the concave bottom slide or by focusing the microscope’s source light on the bottom of the bubble’s surface using the condenser lens of the microscope. The interfacial tension gradient drove a flow that pulled the oil toward the interface. As a result a vortical flow was produced close to the corner formed between the bubble and the upper cover slide. The suspension was drawn into the vortex and subsequently flowed out into the bulk fluid, which was at rest. These observations were videotaped (35) and then played back. Pictures of still frames were then taken, which are shown here. RESULTS AND DISCUSSION
Results of Rheological Measurements Iron oxide suspensions exhibit strong yielding behavior, as is shown in Fig. 3, where the steady shear viscosity vs
coida
AP: Colloid
206
NAVARRETE, SCRIVEN, AND MACOSKO
FIG. 8. (a) Effect of shear on a 3.3% by volume g-Fe2O3 suspension after an applied stress of 50 Pa. Particle network evolves into elongated flocs. (1) Rotating glass cylinder. (2) Sheared suspension. (3) Stationary outer cylinder. Backscattered electron image. (b) Nonsheared 3.3% by volume iron oxide suspension in miniature Couette geometry, for comparison with sheared state. Numbers indicate same regions as in (a). Backscattered electron image.
AID
JCIS 4197
/
6g10$$4197
05-13-96 20:41:08
coida
AP: Colloid
FLOCCULATED IRON OXIDE SUSPENSIONS
207
FIG. 9. High magnification of deformed particle network in region (2) in Fig. 8a; 3.3% by volume iron oxide suspension after shear at 50 Pa. Backscattered electron image.
FIG. 10. Effect of shear on a 1.0% by volume iron oxide suspension after an applied stress of 50 Pa. Flocs become more elongated the closer they are to the rotating glass cylinder. (1) Rotating glass cylinder. (2) Sheared suspension. (3) Stationary outer cylinder. (4) Elongated floc.
AID
JCIS 4197
/
6g10$$4197
05-13-96 20:41:08
coida
AP: Colloid
208
NAVARRETE, SCRIVEN, AND MACOSKO
FIG. 11. Trapped bubble in a 1.0% by volume iron oxide suspension. (A) Bubble. (B) Floc spinning in a vortex (amplified in Fig. 12).
stress is plotted for different particle volume fractions. There is a drastic fall in viscosity within a certain stress range, which varies according to the concentration. The effect is especially dramatic at volume fractions of 9.4 and 6.0%, where the viscosity drops about 5 decades in less than 1 decade of stress. After yielding, iron oxide suspensions show shear-thinning behavior. Similar yielding behavior has been observed by Buscall et al. (37) for concentrated latex particles flocculated under a secondary minimum interparticle pair potential. The creep curves of the 9.4% suspension were studied more closely at 10 Pa, where the viscosity falls drastically. At 10 Pa the shape of the creep curve does not fit the traditional viscoelastic creep response, where the strain rate increases rapidly at the beginning due to elasticity and slows down afterward to a constant deformation rate, but rather shows a monotonically increasing deformation rate (see Fig. 4a). This behavior becomes much more drastic when the gap is cut in half (0.25 mm), as shown in Fig. 4b. This creep response suggests that the suspension is gradually yielding with time. A comparison between creep viscosity with the two different gaps 0.5 and 0.25 mm is shown in Fig. 5. The two curves coincide at low stresses (1–3 Pa), indicating that the zero shear rate viscosity plateau is real and not due to slip at the wall, and at stresses larger than 15Pa, where the particle network is broken down. The stress region between 4 and 15 Pa, where the particle network is being broken down shows discrepancies in the viscosities obtained by using different gaps; i.e., the viscosities from a 0.25-mm gap were consistently lower than those from a 0.5mm gap. It is also apparent that in the narrower gap the particle network begins to break down at lower stresses, namely, 4 Pa instead of 6 Pa. These results indicate that the length scales imposed by the wall boundaries become important once the particle network begins to break down.
AID
JCIS 4197
/
6g10$$$502
05-13-96 20:41:08
Extensional measurements were conducted with the opposed jets extensional indexer. Only two concentrations of iron oxide suspensions were tested: 3.3 and 1.0%. It was not possible to test any higher concentrations because of operational difficulties with the instrument; samples had difficulty flowing out of the narrow orifices of the syringes generating bubbles caused by leaks in the system. Iron oxide suspensions show apparent extensional thinning as shown in Figs 6a and 6b, where the apparent extensional viscosity of a 3.3 and 1.9% magnetic suspension is compared to their shear counterparts as a function of stress. The apparent extensional viscosity is at least two orders of magnitude larger than its shear counterpart for the 3.3% suspension and one order of magnitude larger for the 1.9% suspension. The ratios nominal of extensional over shear viscosities are much larger than the Trouton ratio of 2 (or 3, depending on the definition) for Newtonian fluids. The apparent extensional thinning behavior becomes more drastic with increasing particle concentration. Notably, the extensional viscosity vs stress curves are shifted about two orders of magnitude towards the right in stress with respect to their shear counterparts, implying that it requires a tensile stress 100 times larger to break the suspension to a similar degree under extension than that required under shear. These results do not follow the predictions of the theory of dilute, large aspect ratio, spheroidal particle suspension rheology (38), which is clearly not applicable to concentrated suspensions; instead these results suggest a completely different mechanism behind the apparent extensional thinning behavior. This behavior is opposite to that of dilute polymer solutions, i.e., extensional thickening by stretch of polymer coils under extensional flow, the stretch been elastic (39). These data indicate that iron oxide flocs break up with increasing extensional rates, supporting prior observations by Kao and Mason (21) of the breakup of noncolloidal aggregates under extensional flow. On the other hand, Kao and Mason concluded that the aggregates broke down faster under extension than under simple shear. This is contrary to the rheological data presented here, implying that the attractive colloidal interparticle forces and long range magnetic forces present between the iron oxide particles give iron oxide flocs added strength under extension as opposed to under shear. Results of Visualization Experiments Cryo-SEM was used to visualize the effect of shear on iron oxide suspensions in the miniature Couette geometry. A constant stress of 50 Pa was applied for 1800 s to iron oxide suspensions of three particle volume fractions, namely 9.4, 3.3, and 1.0%. As Fig. 3 indicates, this stress is well above the shear yield stress for these suspensions. A low magnification micrograph of the fracture plane of the miniature Couette fixture is shown in Fig. 1. The inner circle
coida
AP: Colloid
FLOCCULATED IRON OXIDE SUSPENSIONS
209
FIG. 12. Time sequence of a 1.0% iron oxide floc spinning in a vortex generated by a surface tension gradient on adjacent bubble surface. (A) Bubble’s interface. (B) Spinning floc.
corresponds to the glass rod, and the outer darker region corresponds to the aluminum cylinder. The dark annulus in between these two zones corresponds to a frozen 9.4% suspension. There is considerable debris left from the fracturing, mainly fragments of frozen suspension. A higher magnification micrograph of the gap reveals regions of different particle concentrations and structure, as shown in Fig. 7. Adjacent to the wall of the rotating glass rod there is a bright region where particles are more flocculated and their concentration appears lower (in some parts particles are completely absent). This region does not extend all the way around the glass rod perimeter, suggesting that this region was created by a misalignment of the rod during transfer of the miniature Couette fixture from the rheometer to the liquid nitrogen. Beyond this region (advancing in the
AID
JCIS 4197
/
6g10$$$503
05-13-96 20:41:08
radial direction from the inner cylinder to the outer cylinder) there is a sharp discontinuity to a region where the particle network is deformed in the shearing direction. Particles there appear to be more segregated due to the deformation. The thickness of this region is about 15 mm. Beyond this highly deformed region, there is a transition to a region where deformation is less apparent. At about the middle of the gap, beyond the previous region, there is a sharp transition to a region of low particle concentration. In this region the particle network has clearly been broken down into elongated flocs by the applied shear deformation. The thickness of this region is about 100 mm. Finally, there is a region of higher particle concentration close to the static outer cylinder, where the suspension seems unaffected by the applied deformation.
coida
AP: Colloid
210
NAVARRETE, SCRIVEN, AND MACOSKO
FIG. 13. Micrograph of a 0.5% by volume iron oxide suspension on a nearly extensional flow induced by the vortex. Arrow indicates the direction of the flow. (A) Reference floc. Flow was induced by focused light heating.
The suspension structure of a 3.3% by volume suspension under shear is shown in Fig. 8a. In this case none of the regions observed in the 9.4% suspension are present. Instead the particle structure is basically composed of elongated flocs throughout the gap. For comparison, a non-sheared 3.3% suspension is shown in Fig. 8b; there no elongation of the flocs is evident. A higher magnification micrograph of the elongated flocs in Fig. 8a shown in Fig. 9 reveals that the particles do not form elongated structures below 10 mm; i.e., the effect of the hydrodynamic forces generated by shear is not high enough to overcome the attractive colloidal interactions below this length scale. The last concentration, namely 1.0% by volume, is shown in Fig. 10. Here also elongated flocs can be observed. Flocs at this concentration are of the order of 10 mm which is smaller than those observed at higher concentrations. Below this size hydrodynamic forces at this shear stress are not high enough to overcome the attractive interparticle interactions. Cryo-SEM provided a means to obtain a snapshot of the structure of iron oxide suspensions during deformation. However, to understand the gradual yielding with time observed in creep experiments (Fig. 4) it was necessary to use the VELM to follow the evolution of the structures in real time. Figure 11 shows a low magnification micrograph of a trapped bubble in the geometry described in Fig. 2. There are flocs around the perimeter of the bubble’s interface, which are spinning in the vortex created by the surface tension gradient in the bubble’s surface. Figure 12 shows a sequence of high magnification micrographs of this region, where the evolution of a spinning floc is shown. This observation reveals a new dimension in the breakup of flocs under shear: flocs rotate around an axis perpendicular to the plane of shear. This dimension was missing from prior observations of floc or aggregate breakup (21). The floc in the vortex unravels, elongates and ultimately
AID
JCIS 4197
/
6g10$$$503
05-13-96 20:41:08
sections itself into smaller flocs. The structure of the floc appears very dense at first; however, as the floc unravels and elongates the structure appears more open and chainlike. It is clear from this sequence that the partition of a floc subjected to an imposed stress that exceeds the floc’s yield strength may not be quick, but rather a process that develops with time. In other words, there is a sequence of particle/ particle bonds of different strengths that break with time according to their relative strength and shifting stress concentration with exposure to the flow; the sequence ultimately leads to the separation of the floc into smaller parts. Figure 13 shows suspension being dragged into the vortex from its outer perimeter, where the suspension is mainly stationary. The flow in this zone is predominantly extensional; i.e., as the suspension converges into the vortex, there is a dramatic increase in velocity, which gives rise to a high gradient of the velocity in the direction opposed to the flow. Therefore, as flocs approach the vortex zone, they suffer a high extensional rate, which causes them to elongate and eventually section into smaller chainlike flocs. A high magnification micrograph confirms this assertion, as shown in Fig. 14, where flocs essentially look like chains with some branches and some denser flocs attached to them. These observations help explain the apparent extensional thinning behavior observed in the iron oxide suspensions. SUMMARY
Visual evidence has been presented that shows the effect of shear, extension and vorticity on the microstructure of iron oxide suspensions. These observations have been used to explain the rheological behavior of these suspensions under shear and extension, and they add new information on the microscopic details and dynamics of the breakup, which may prove valuable to the theoreticians attempting to model such phenomena. The following are the main conclusions: 1. Rheological measurements show that iron oxide suspensions exhibit both extensional and shear thinning. The apparent extensional viscosity measured is one to two orders of magnitude larger than the shear viscosity at the same applied stress. 2. Iron oxide flocs elongate and break up in an extensional flow, which explains the apparent extensional behavior exhibited by iron oxide suspensions. 3. The particle network of iron oxide suspensions at concentrations of 9.4% by volume does not uniformly break up into smaller units of particles in Couette flow, as happens at lower concentrations, but rather it breaks up in the center of the gap and segregates the particles toward the walls. 4. The breakup of the network structure of iron oxide particles in suspensions that results in yielding behavior of the suspension is a process that occurs gradually with time. This process was clearly demonstrated by the breakup sequence of spinning flocs.
coida
AP: Colloid
FLOCCULATED IRON OXIDE SUSPENSIONS
FIG. 14. High magnification of flocs of a 0.5% by volume iron oxide suspension flowing in a nearly extensional flow. Flow was induced by focused light heating.
5. Vorticity, which is present in shear flows, but not in extensional flows, plays an important role in the deformation and breakup of iron oxide flocs. This observation provides an explanation of the rheological measurements under extension which showed that it takes a tensile stress 100 times larger than a shear stress to break a suspension to a similar viscosity level. ACKNOWLEDGMENTS This research was supported by the Center for Interfacial Engineering at the University of Minnesota, which is sponsored by the National Science Foundation and industrial members. We thank the KAO Corporation for providing the particles used here. We are grateful to V. A. Le and J. G. Sheehan, who took the Cryo-SEM micrographs, and to G. G. Glasrud, who made the creep measurements. We thank Rheometrics Scientific, Inc. for the use of the RFX.
REFERENCES 1. Michaels, A. S., and Bolger, J. C., Ind. Eng. Chem. Fundam. 1(3), 153 (1962). 2. Papenhuizen, J. P. M., Rheol. Acta 11, 73 (1972). 3. Seto, J., J. Soc. Rheol. Jpn. 5(4), 156 (1977). 4. Firth, B. A., and Hunter, R. J., J. Colloid. Interface Sci. 57(2), 248 (1976). 5. Firth, B. A., J. Colloid. Interface Sci. 57(2), 257 (1976). 6. Firth, B. A., and Hunter, R. J., J. Colloid. Interface Sci. 57(2), 266 (1976). 7. Van de Ven, T. G. M., and Hunter, R. J., Rheol. Acta 16, 534 (1977). 8. Kamphuis, H., Jongschaap, R. J. J., and Bouter, P., J. Colloid. Interface Sci. 98(2), 459 (1984). 9. Wildemuth, C. R., and Williams, M. C., Rheol. Acta 24, 75 (1985). 10. Schoukens, G., and Mewis, J., J. Rheol. 22, 381 (1978).
AID
JCIS 4197
/
6g10$$$503
05-13-96 20:41:08
211
11. Russell, W. B., J. Rheol., 24, 287 (1980). 12. Kanai, H., Navarrete, R. C., Macosko, C. W., and Scriven, L. E., Rheol. Acta 31, 333 (1992). 13. Buscall, R., McGowan, I. J., Mills, P. D. A., Stewart, R. F., Sutton, D., White, L. R., and Yates, G. E., J. Non-Newtonian Fluid Mech., 24, 183 (1987). 14. Buscall, R., Mills, P. D. A., Goodwin, J. W., and Lawson, D. W., J. Chem. Soc. Faraday Trans. 1 84, 4249 (1988). 15. Weitz, D. A., and Huang, J. S., in ‘‘Kinetics of Aggregation and Gelation’’ (F. Family and D. P. Landau, Eds.), p. 19. Elsevier, Amsterdam, 1984. 16. Schaefer, D. W., Martin, J. E., Wiltzius, P., and Cannell, D. S., Phys. Rev. Lett. 52, 2371 (1984). 17. Helgesen, G., Skjeltorp, A. T., Mors, P. M., R. Botet, R., and Jullien, R., Phys. Rev. Lett. 61, 1736 (1988). 18. Sonntag, R. C., and Russel, W. B., J. Colloid Interface Sci. 113, 399 (1986). 19. Sonntag, R. C., and Russel, W. B., J. Colloid Interface Sci. 115, 378 (1987). 20. Sonntag, R. C., and Russel, W. B., J. Colloid Interface Sci. 115, 390 (1987). 21. Kao, S. V., and Mason, S. G., Nature (London) 253, 619 (1975). 22. Bossis, G., and Brady, J. F., J. Chem. Phys. 80, 5141 (1984). 23. Brady, J. F., Phillips, R. J., Lester, J. C., and Bossis, G., J. Fluid Mech. 195, 257 (1988). 24. Ansell, G. C., Dickinson, E., and Ludvigsen, M., J. Chem. Soc. Faraday Trans. 2 80, 1269 (1985). 25. Toy, M. L., Scriven, L. E., Macosko, C. W., Nelson, N. K., and Olmsted, R. D., J. Rheol. 35(5), 887 (1991). 26. Yang, M. C., Scriven, L. E., and Macosko, C. W., J. Rheol. 30(5), 1015 (1986). 27. Utsugi, H., Endo, A., Suzuki, N., and Ono, K., J. Jpn. Soc. Powder and Powder Metallurgy 32, 100 (1985). 28. Fuller, G. G., Cathey, C. A., Hubbard, B. and Zebrowski, B. E., J. Rheol., 31, 235 (1987). 29. Schunk, R., De Santos, J. M., and Scriven, L. E., J. Rheol. 34, 387 (1990). 30. Cai, J. J., Souza Mendes, P. R., Macosko, C. W., Scriven, L. E., and Secor, R. B., in ‘‘Theoretical and Applied Rheology’’ (P. Moldenaers and R. Keunings, Eds.). Proc. XIth Int. Congr. on Rheol., Vol. 2, p. 1012. Elsevier, Brussels, 1992. 31. Bellare, J., Davis, H. T., Scriven, L. E., and Talmon, Y., in ‘‘Proc. XIth Int. Cong. on Electron Microscopy,’’ 1986 Vol. 1, p. 367. 32. Bellare, J., Sheehan, J. G., Davis, H. T., and Scriven, L. E., Mater. Res. Soc. Symp. Proc. 115, 75 (1988). 33. Sheehan, J. G., and Scriven, L. E., in ‘‘Proc. of the 46th Annual Meeting of the Electron Microscopy Society of America, Elec. Micros. Soc. Am., Woods Hole, MA, 1988,’’ p. 102. 34. Sheehan, J. G., and Whalen-Shaw, M., Tappi J. 23(5), 171 (1990). 35. Navarrete, R. C., Ph. D. Thesis, Univ. of Minnesota, Minneapolis, 1991. 36. Navarrete, R. C., Le, V. A., Glasrud, G. G., Macosko, C. W., and Scriven, L. E., in ‘‘Theoretical and Applied Rheology’’ (P. Moldenaers and R. Keunings, Eds.). Proc. XIth Int. Congr. on Rheol., Vol. 2, p. 625. Elsevier, Brussels, Belgium, 1992. 37. Buscall, R., McGowan, I. J., and Mumme-Young, C. A., Faraday Discuss. Chem. Soc., 90, 115 (1990). 38. Brenner, H., in ‘‘Progress in Heat and Mass Transfer’’ (W. R. Schowalter et al., Eds.), Vol. 5, p. 5. Pergamon, Oxford, 1972. 39. Schunk, R., and Scriven, L. E., J. Rheol. 34(7), 1085 (1990).
coida
AP: Colloid
180, 200–211 (1996)
0290
Rheology and Structure of Flocculated Iron Oxide Suspensions REINALDO C. NAVARRETE, 1 L. E. SCRIVEN,
AND
CHRISTOPHER W. MACOSKO
Center for Interfacial Engineering and Department of Chemical Engineering and Materials Science, University of Minnesota, Minneapolis, Minnesota 55455 Received July 13, 1995; accepted December 27, 1995
This paper focuses on the relation between structure and rheology of flocculated suspensions. Rheological measurements were taken both in shear and nearly extensional flows. The structures of magnetic iron oxide suspensions were visualized using cryogenic scanning electron microscopy ( Cryo-SEM ) and videoenhanced light microscopy ( VELM ) under shear, elongational, and rotational flows. Shear fields were generated by using a miniature Couette geometry, whereas elongational and rotational fields were generated by an innovative technique, which takes advantage of a surface tension gradient on the interface of an air bubble, and were visualized in real time under a light microscope. In Couette flow, the particle networks formed at high concentrations ( 9.4% by volume ) do not uniformly break up into smaller units of particles, as at lower concentrations, but rather break up at the center of the gap, segregating particles toward the walls. Real-time light microscopy observations of a spinning floc show that the deformation and breakup of flocs is a process that gradually occurs with time. In extensional flow, flocs elongate and break up, explaining the apparent extensional thinning behavior measured. Apparent extensional viscosities were one to two orders of magnitude larger than shear viscosities at the same applied stress. q 1996 Academic Press, Inc. Key Words: rheology; suspensions; colloids; flocculation; iron oxide; microscopy; extensional flows.
INTRODUCTION
The rheological behavior of colloidal suspensions is determined by the structures that particles form in suspension (1–10). Structures generally mean some sort of association of particles as extensive as the sample itself or nearly so, the association having something of the character of an elastic network. When there are long-range attractive forces and short-range repulsive forces, it is well known that the net force between two colloidal particles can vanish at separations determined by the interparticle forces (11). Departures from those separations are accompanied by elastic-like restoring forces, provided the departures are not too great. This 1 To whom correspondence should be addressed. Present Address: Schlumberger Dowell, Inc., P.O. Box 2710, Tulsa, OK 74101.
200
0021-9797/96 $18.00 Copyright q 1996 by Academic Press, Inc. All rights of reproduction in any form reserved.
AID
JCIS 4197
/
6g10$$$501
gives rise to ‘‘bonds’’ between particles; i.e., two particles are bonded to each other if there are forces that resist changes in their relative position. There can be different structural levels of bonding. At each of these levels, the strength of the bonds (i.e., the magnitude of the force necessary to separate two particles enough so that there is no resisting force to further changes in relative particle positions) between the structural units varies according to the length scale of the structure level. The degree of permanence of each structure level depends on the externally imposed disturbing forces and on the strength of the bonding at that level of structure. The relation between structure and rheology of flocculated suspensions can be manifested in different forms of rheological behavior. For example, elasticity implies the presence of a sample-spanning elastic structure, yield stress implies that the connectivity of the structure is extensively ruptured, shear thinning implies further disconnection as the rate of shear is raised, and time dependence implies that disconnection and reconnection proceed at detectable slow rates. There are a number of studies that propose structural models for flocculated suspensions. Michaels and Bolger (1) put forward a network model to interpret the yield stress of aqueous, flocculated kaolin suspensions and derived equations to predict the effects of kaolin concentration and shear rate. The basic flow units in the model are flocs, which are small associations of particles plus the water they enclose. These basic units are supposed to be the flocs that persist at the highest shear rate to which the system has been subjected. They have elastic character due to the forces acting between the particles that form the floc. At low shear rates, the flocs group into floc aggregates, and these in turn may interact to form a sample-spanning or effectively ‘‘infinite’’ network that is responsible for the elastic character of the suspension. When an external deformation of sufficiently large magnitude is applied to the network, it breaks up into floc aggregates or flocs depending on the magnitude of the applied deformation. In a series of papers, Firth and Hunter (4–6) advanced a somewhat similar model for coagulated sols which they called the ‘‘elastic floc model.’’ Van de Ven and Hunter (7)
05-13-96 20:41:08
coida
AP: Colloid
FLOCCULATED IRON OXIDE SUSPENSIONS
went a step further in modeling the structure of the flocs and introduced an additional structural level. They proposed that the flocs consisted of a number of smaller domains which they called flocculi. The flocculi were supposed to have been formed at the highest shear rate to which the system had ever been subjected, whereas the flocs and the floc aggregates were supposed to arise during the shear history after the flocculi themselves were formed. Following Hunter’s model, Kanai et al. (12) formulated a floc network model and used elastic percolation concepts to explain the scaling dependency of elasticity of iron oxide suspensions as a function of concentration. Non-zero elastic shear modulus appeared at low threshold particle volume fractions, indicating that flocs of iron oxide particles were very low density structures (2–5% by volume). Above the threshold concentration, the shear dynamic elastic modulus G * was found to follow a scaling law of the type G * Ç ( f f 0 f fc ) f , where f f is the floc volume fraction, f fc is the threshold floc volume fraction, and f is the scaling exponent. Buscall et al. (13, 14) studied the scaling relationship of the shear elastic modulus, the compressive strength and shear yield stress with the particle volume fraction for aggregated colloidal silica particle gels. The scaling exponent obtained suggested that the aggregation mechanism by which the particle network was formed was diffusion limited floc–floc aggregation, where the flocs were fractal in nature. Fractal geometry has been extensively used in the recent past to characterize aggregates of colloidal particles (15, 16). Uniformly sized microspheres interacting via longrange magnetic dipolar forces have also been reported to form clusters with a fractal dimension 1.16, i.e., clusters with pronounced chain like structures (17). Sonntag and Russel (18–20) in a series of papers investigated by small angle light scattering the effect of shear and extensional flows on the size and fractal dimension of polystyrene flocs. Under strong shear conditions they found that the size of the flocs decreased significantly under shear, in agreement with prior visual observations by Kao and Mason (21) of noncolloidal particle associations. The fractal dimension of the flocs under shear was found to be 2.48, which can be considered high. The implication is that smaller flocs, where the outer branches have been eliminated by hydrodynamic stresses, are closer to uniformly dense particles than those flocs formed under little or no shear. On the other hand, little variation in the floc size and structure was found under nearly extensional flows (flow through an abrupt contraction), in contrast to earlier observations by Kao and Mason. The effect of hydrodynamic stresses on particle associations in suspension has also been investigated by means of Stokesian dynamics simulations (22, 23) and by Brownian dynamics simulations (24). However, these simulations are computationally intensive, and therefore are restricted to small numbers of particles and simple flows. The present paper focuses on the relation between struc-
AID
JCIS 4197
/
6g10$$$502
05-13-96 20:41:08
201
ture and rheology of flocculated suspensions. Past studies indicate that the flocculated structures of magnetic particles under shear can give rise to unexpected microscopic structures and rheological behavior (25). The approach taken here was to measure the rheological behavior of a model flocculated suspension (magnetic iron oxide suspensions) and to visualize the structures under different flow fields, namely shear, elongational and rotational, by means of cryogenic scanning electron microscopy (Cryo-SEM) and videoenhanced light microscopy (VELM). Shear fields were generated using a miniature Couette geometry, whereas elongational and rotational fields were generated with an innovative technique which takes advantage of a surface tension gradient on the interface of an air bubble under a light microscope. MATERIALS AND METHODS
Experimental System The experimental suspension system consisted of needlelike particles of magnetic iron oxide suspended in mineral oil. The particle density was 4500 kg/m 3 . The average length of the particles was about 0.5 mm and the average length to diameter ratio was 5; transmission electron micrographs of similar particles can be found in Yang et al. (26). The liquid medium was heavy mineral oil from Sergent-Welch Co., with viscosity equal to about 0.1 Pars (257C) and density of 863 kg/m 3 . The iron oxide particles were treated with an alkoxysilane surfactant, which reacts with the hydroxyl sites on the surface of the particles to prevent particle aggregation by chemical bonding (27). A ball mill was employed for the preparation of the suspensions. Mill vessels were plastic bottles, 6 cm in diameter, and milling media were 9.5-mm steel balls. Powder was mixed with the mineral oil in a 40% by weight (11.3% by volume) concentration and then milled for 24 h. This suspension was then diluted to several concentrations. Rheological Measurements The shear viscosity of iron oxide suspensions was measured by means of constant stress or creep tests. Creep tests were used, as opposed to constant shear rate tests, because they allow the breakup of the structure of the suspension by following the shear deformation with time. A Rheometrics Stress Rheometer (RSR) 8600 (Rheometrics Scientific, Inc., Piscataway, NJ), fitted with a Couette geometry with the dimensions of a 13-mm bob diameter, 14-mm cup diameter, and 40-mm bob length, was used for creep measurements. A larger diameter bob (13.5 mm) was also used in some cases. A sequence of creep tests of increasing stress was applied to each sample. In between the creep tests, a period of 300 s was allowed for the sample to recover, which was not recorded. The duration of each creep test was 600 s. Before a sequence of creep tests was initiated, a unidirec-
coida
AP: Colloid
202
NAVARRETE, SCRIVEN, AND MACOSKO
FIG. 1. Low magnification SEM of 9.4% by volume iron oxide suspension in miniature Couette geometry. (1) Glass inner cylinder. (2) Frozen suspension in the gap. (3) Aluminum outer cylinder.
tional preshear was applied at a stress of 220 Pa for 900 s followed by a rest period of 1800 s at zero stress. All measurements were made at room temperature. An opposed nozzles indexer was used to measure the extensional viscosity of iron oxide suspensions (28–30). This instrument has two nozzles of the same dimensions placed opposing each other and submerged into a beaker with the sample. Each nozzle is connected to a syringe by means of an arm and tubing. One of the arms is fixed and the other is connected to a transducer which measures the torque exerted on the arm. The syringes can suck the sample from the beaker or return the sample to it. Ideally, the result in the first case is an extensional biaxial flow and in the last case an extensional uniaxial flow. In practice, however, these flows are not purely extensional due to the shear contributions originating from the regions close to the walls of the nozzles. If the flows are assumed to be ideally extensional, it is possible to calculate an apparent extensional viscosity from the volume flow rate of the sample in or out of the nozzles and the resulting torque on the moving arm. The apparent extensional viscosity calculated in this way should be taken as an index of extensional behavior rather than as a true extensional viscosity. All extensional measurements were taken with a RFX (Rheometrics Scientific, Inc., Piscataway, NJ).
AID
JCIS 4197
/
6g10$$$502
05-13-96 20:41:08
The opposed nozzles indexer is essentially a rate-controlled instrument; i.e., the extensional rate is set and the resulting tensile stress is measured. To achieve steady state, an initial period was allowed before taking the tensile stress measurement. This time was determined by monitoring the approach to the steady state of the torque on the transducer arm and varied according to the extension rate from 50 s for the lowest rate on the highest volume particle concentration, to 10 s for the highest rate on the lowest concentration tested. The actual torque measurement was made in 15 s. The range of extensional rates tested was determined by the nozzle diameter; i.e., large nozzle diameters are required to achieve small rates. The limit is a measurable torque on the transducer arm. Small nozzle diameters are required to achieve high rates. The limitations are steady state within the available volume capacity of the syringes and avoiding overloading the torque transducer. The nozzle diameters used varied between 1 to 4 mm. Following a suggestion of Schunk et al. (29), the nozzle separation distance was kept equal to the nozzle diameter to minimize shear contributions to the resulting apparent extensional viscosity. Visualization Techniques Two experimental techniques were used to visualize the particle structures of the iron oxide suspensions, Cryo-SEM and VELM. These techniques are described below:
coida
AP: Colloid
FLOCCULATED IRON OXIDE SUSPENSIONS
203
FIG. 2. Two forms of inducing a surface tension gradient at the bubble’s surface: (a) using a hot plate below the concave slide and (b) focus of light at the bubble’s surface.
Cryogenic scanning electron microscopy (Cryo-SEM). Freeze-fracture scanning electron microscopy, usually named cryogenic scanning electron microscopy (CryoSEM), was used to visualize the structure of iron oxide suspensions. This technique was developed by Bellare et al. (31, 32), Sheehan and Scriven (33), and Sheehan and Whalen-Shaw (34), and enables visualization of submicron particles in a liquid suspending medium. The steps involved are (i) freeze a sample of suspension, (ii) fracture the frozen sample, (iii) coat the fractured surface with a conducting
FIG. 4. Strain vs time curves of a creep test on a 9.4% iron oxide suspension with a gap of (a) 0.50 mm and (b) 0.25 mm at 10 Pa applied shear stress (r Å correlation coefficient).
FIG. 3. Shear creep viscosities of iron oxide suspensions of indicated particle volume fractions.
FIG. 5. Comparison of creep steady viscosities of a 9.4% by volume g-Fe2O3 suspension in two different gaps.
AID
JCIS 4197
/
6g10$$$502
05-13-96 20:41:08
metal, and (iv) observe the fractured, coated surface with an SEM. These steps are described in the references given above.
coida
AP: Colloid
204
NAVARRETE, SCRIVEN, AND MACOSKO
FIG. 6. Comparison of extensional vs shear viscosities of (a) 3.3% and (b) 1.9% suspensions.
The sample preparation procedure was modified with respect to those used in previous works to study the effect of shear on the microstructure of the iron oxide suspensions. A miniature Couette geometry was used to shear the suspensions. The suspension sample was loaded in the outer aluminum cylindrical cup, which consisted of two parts: a bottom cup and an upper tube of the same diameter that fitted together (35). The aluminum cup was then mounted on a Rheometrics Stress Rheometer (Rheometrics Scientific, Inc., Piscataway, NJ). There an inner glass rod was lowered in such a way that the suspension filled the annular space left between the glass rod and the aluminum cup. The diameter of the glass cylinder was 3.3 mm, leaving an annular gap of about 400 mm. A constant torque was applied to the glass rod, equivalent to a shear stress of 50 Pa for 1800 s. Next, the glass cylinder was stopped and secured to the outer cup by means of two lateral screws. Then the entire miniature
AID
JCIS 4197
/
6g10$$$502
05-13-96 20:41:08
Couette geometry was dismounted from the stress rheometer and plunged into liquid nitrogen. This was done in less than 60 s to reduce the relaxation of the structure induced by shear. After freezing, the sample was stored in liquid nitrogen. At the time of imaging, the sample was placed on a hinged fracture stage and transferred into a cryo-fracture chamber under vacuum (7 1 10 06 kPa) and at liquid nitrogen temperature ( 01757C). This chamber was connected to a metalcoating Hexland CT 1000A chamber. The two chambers were separated by a gate valve and the coating chamber was connected to a JEOL JSM 840 SEM by a gate valve. These attachments are well described by Sheehan and WhalenShaw (34). There the sample was fractured by separating the two parts of the outer cup and fracturing the inner glass rod and the frozen suspension in the annulus (see Fig. 1). Samples were then placed in the metal coating chamber
coida
AP: Colloid
FLOCCULATED IRON OXIDE SUSPENSIONS
205
FIG. 7. Effect of shear on a 9.4% by volume g-Fe2O3 suspension after an applied stress of 50 Pa. (1) Rotating glass cylinder. (2) Bright region adjacent to the wall. (3) Highly sheared region. (4) Transition region. (5) Broken down region of low particle concentration. (6) Region adjacent to stationary wall. (7) Stationary outer aluminum cylinder. Backscattered electron image.
and a film of conducting metal of controlled thickness was sputtered on the fractured surface to prevent electron charge build-up during imaging. The metals used were gold, platinum, and chromium; the metal or combination for each sample was found by trial-and-error. Thickness also varied from sample to sample. Typical thicknesses ˚. varied from 25 to 35 A The sample was then transferred to the cryo-chamber ( 01757C and 7 1 10 06 kPa) of a JEOL JSM 840 SEM equipped with an LaB6 filament for imaging. Short working distances, large objective apertures, small condenser apertures, and high accelerating voltages were used to enhance the resolution and contrast between the particles and the oil (35). The micrographs shown here are backscattered electron emission images. Video-enhanced light microscopy (VELM). Video-enhanced light microscopy (VELM) was used to observe the evolution of the structures of iron oxide suspensions in real time under different flow motions. Microflows were generated in the oil by means of a surface tension gradient-driven flow. A Nikon Optiphot-Pol light microscope (Frank Fryer Co., Carpenter Ville, IL) equipped with a black-and-white television camera and connected to a video recorder was used (35). The sample was placed between a glass slide with a spherical
AID
JCIS 4197
/
6g10$$$502
05-13-96 20:41:08
depression and a flat cover slide under a light microscope, as shown in Fig. 2. The depression provided enough depth and visual access to the three-dimensional structures in the suspension. Microflows were generated by entrapping an air bubble below the upper flat slide. A surface tension gradient was then induced on the oil/air interface of the bubble by an externally imposed temperature gradient. The temperature gradient was induced either by a hot plate below the concave bottom slide or by focusing the microscope’s source light on the bottom of the bubble’s surface using the condenser lens of the microscope. The interfacial tension gradient drove a flow that pulled the oil toward the interface. As a result a vortical flow was produced close to the corner formed between the bubble and the upper cover slide. The suspension was drawn into the vortex and subsequently flowed out into the bulk fluid, which was at rest. These observations were videotaped (35) and then played back. Pictures of still frames were then taken, which are shown here. RESULTS AND DISCUSSION
Results of Rheological Measurements Iron oxide suspensions exhibit strong yielding behavior, as is shown in Fig. 3, where the steady shear viscosity vs
coida
AP: Colloid
206
NAVARRETE, SCRIVEN, AND MACOSKO
FIG. 8. (a) Effect of shear on a 3.3% by volume g-Fe2O3 suspension after an applied stress of 50 Pa. Particle network evolves into elongated flocs. (1) Rotating glass cylinder. (2) Sheared suspension. (3) Stationary outer cylinder. Backscattered electron image. (b) Nonsheared 3.3% by volume iron oxide suspension in miniature Couette geometry, for comparison with sheared state. Numbers indicate same regions as in (a). Backscattered electron image.
AID
JCIS 4197
/
6g10$$4197
05-13-96 20:41:08
coida
AP: Colloid
FLOCCULATED IRON OXIDE SUSPENSIONS
207
FIG. 9. High magnification of deformed particle network in region (2) in Fig. 8a; 3.3% by volume iron oxide suspension after shear at 50 Pa. Backscattered electron image.
FIG. 10. Effect of shear on a 1.0% by volume iron oxide suspension after an applied stress of 50 Pa. Flocs become more elongated the closer they are to the rotating glass cylinder. (1) Rotating glass cylinder. (2) Sheared suspension. (3) Stationary outer cylinder. (4) Elongated floc.
AID
JCIS 4197
/
6g10$$4197
05-13-96 20:41:08
coida
AP: Colloid
208
NAVARRETE, SCRIVEN, AND MACOSKO
FIG. 11. Trapped bubble in a 1.0% by volume iron oxide suspension. (A) Bubble. (B) Floc spinning in a vortex (amplified in Fig. 12).
stress is plotted for different particle volume fractions. There is a drastic fall in viscosity within a certain stress range, which varies according to the concentration. The effect is especially dramatic at volume fractions of 9.4 and 6.0%, where the viscosity drops about 5 decades in less than 1 decade of stress. After yielding, iron oxide suspensions show shear-thinning behavior. Similar yielding behavior has been observed by Buscall et al. (37) for concentrated latex particles flocculated under a secondary minimum interparticle pair potential. The creep curves of the 9.4% suspension were studied more closely at 10 Pa, where the viscosity falls drastically. At 10 Pa the shape of the creep curve does not fit the traditional viscoelastic creep response, where the strain rate increases rapidly at the beginning due to elasticity and slows down afterward to a constant deformation rate, but rather shows a monotonically increasing deformation rate (see Fig. 4a). This behavior becomes much more drastic when the gap is cut in half (0.25 mm), as shown in Fig. 4b. This creep response suggests that the suspension is gradually yielding with time. A comparison between creep viscosity with the two different gaps 0.5 and 0.25 mm is shown in Fig. 5. The two curves coincide at low stresses (1–3 Pa), indicating that the zero shear rate viscosity plateau is real and not due to slip at the wall, and at stresses larger than 15Pa, where the particle network is broken down. The stress region between 4 and 15 Pa, where the particle network is being broken down shows discrepancies in the viscosities obtained by using different gaps; i.e., the viscosities from a 0.25-mm gap were consistently lower than those from a 0.5mm gap. It is also apparent that in the narrower gap the particle network begins to break down at lower stresses, namely, 4 Pa instead of 6 Pa. These results indicate that the length scales imposed by the wall boundaries become important once the particle network begins to break down.
AID
JCIS 4197
/
6g10$$$502
05-13-96 20:41:08
Extensional measurements were conducted with the opposed jets extensional indexer. Only two concentrations of iron oxide suspensions were tested: 3.3 and 1.0%. It was not possible to test any higher concentrations because of operational difficulties with the instrument; samples had difficulty flowing out of the narrow orifices of the syringes generating bubbles caused by leaks in the system. Iron oxide suspensions show apparent extensional thinning as shown in Figs 6a and 6b, where the apparent extensional viscosity of a 3.3 and 1.9% magnetic suspension is compared to their shear counterparts as a function of stress. The apparent extensional viscosity is at least two orders of magnitude larger than its shear counterpart for the 3.3% suspension and one order of magnitude larger for the 1.9% suspension. The ratios nominal of extensional over shear viscosities are much larger than the Trouton ratio of 2 (or 3, depending on the definition) for Newtonian fluids. The apparent extensional thinning behavior becomes more drastic with increasing particle concentration. Notably, the extensional viscosity vs stress curves are shifted about two orders of magnitude towards the right in stress with respect to their shear counterparts, implying that it requires a tensile stress 100 times larger to break the suspension to a similar degree under extension than that required under shear. These results do not follow the predictions of the theory of dilute, large aspect ratio, spheroidal particle suspension rheology (38), which is clearly not applicable to concentrated suspensions; instead these results suggest a completely different mechanism behind the apparent extensional thinning behavior. This behavior is opposite to that of dilute polymer solutions, i.e., extensional thickening by stretch of polymer coils under extensional flow, the stretch been elastic (39). These data indicate that iron oxide flocs break up with increasing extensional rates, supporting prior observations by Kao and Mason (21) of the breakup of noncolloidal aggregates under extensional flow. On the other hand, Kao and Mason concluded that the aggregates broke down faster under extension than under simple shear. This is contrary to the rheological data presented here, implying that the attractive colloidal interparticle forces and long range magnetic forces present between the iron oxide particles give iron oxide flocs added strength under extension as opposed to under shear. Results of Visualization Experiments Cryo-SEM was used to visualize the effect of shear on iron oxide suspensions in the miniature Couette geometry. A constant stress of 50 Pa was applied for 1800 s to iron oxide suspensions of three particle volume fractions, namely 9.4, 3.3, and 1.0%. As Fig. 3 indicates, this stress is well above the shear yield stress for these suspensions. A low magnification micrograph of the fracture plane of the miniature Couette fixture is shown in Fig. 1. The inner circle
coida
AP: Colloid
FLOCCULATED IRON OXIDE SUSPENSIONS
209
FIG. 12. Time sequence of a 1.0% iron oxide floc spinning in a vortex generated by a surface tension gradient on adjacent bubble surface. (A) Bubble’s interface. (B) Spinning floc.
corresponds to the glass rod, and the outer darker region corresponds to the aluminum cylinder. The dark annulus in between these two zones corresponds to a frozen 9.4% suspension. There is considerable debris left from the fracturing, mainly fragments of frozen suspension. A higher magnification micrograph of the gap reveals regions of different particle concentrations and structure, as shown in Fig. 7. Adjacent to the wall of the rotating glass rod there is a bright region where particles are more flocculated and their concentration appears lower (in some parts particles are completely absent). This region does not extend all the way around the glass rod perimeter, suggesting that this region was created by a misalignment of the rod during transfer of the miniature Couette fixture from the rheometer to the liquid nitrogen. Beyond this region (advancing in the
AID
JCIS 4197
/
6g10$$$503
05-13-96 20:41:08
radial direction from the inner cylinder to the outer cylinder) there is a sharp discontinuity to a region where the particle network is deformed in the shearing direction. Particles there appear to be more segregated due to the deformation. The thickness of this region is about 15 mm. Beyond this highly deformed region, there is a transition to a region where deformation is less apparent. At about the middle of the gap, beyond the previous region, there is a sharp transition to a region of low particle concentration. In this region the particle network has clearly been broken down into elongated flocs by the applied shear deformation. The thickness of this region is about 100 mm. Finally, there is a region of higher particle concentration close to the static outer cylinder, where the suspension seems unaffected by the applied deformation.
coida
AP: Colloid
210
NAVARRETE, SCRIVEN, AND MACOSKO
FIG. 13. Micrograph of a 0.5% by volume iron oxide suspension on a nearly extensional flow induced by the vortex. Arrow indicates the direction of the flow. (A) Reference floc. Flow was induced by focused light heating.
The suspension structure of a 3.3% by volume suspension under shear is shown in Fig. 8a. In this case none of the regions observed in the 9.4% suspension are present. Instead the particle structure is basically composed of elongated flocs throughout the gap. For comparison, a non-sheared 3.3% suspension is shown in Fig. 8b; there no elongation of the flocs is evident. A higher magnification micrograph of the elongated flocs in Fig. 8a shown in Fig. 9 reveals that the particles do not form elongated structures below 10 mm; i.e., the effect of the hydrodynamic forces generated by shear is not high enough to overcome the attractive colloidal interactions below this length scale. The last concentration, namely 1.0% by volume, is shown in Fig. 10. Here also elongated flocs can be observed. Flocs at this concentration are of the order of 10 mm which is smaller than those observed at higher concentrations. Below this size hydrodynamic forces at this shear stress are not high enough to overcome the attractive interparticle interactions. Cryo-SEM provided a means to obtain a snapshot of the structure of iron oxide suspensions during deformation. However, to understand the gradual yielding with time observed in creep experiments (Fig. 4) it was necessary to use the VELM to follow the evolution of the structures in real time. Figure 11 shows a low magnification micrograph of a trapped bubble in the geometry described in Fig. 2. There are flocs around the perimeter of the bubble’s interface, which are spinning in the vortex created by the surface tension gradient in the bubble’s surface. Figure 12 shows a sequence of high magnification micrographs of this region, where the evolution of a spinning floc is shown. This observation reveals a new dimension in the breakup of flocs under shear: flocs rotate around an axis perpendicular to the plane of shear. This dimension was missing from prior observations of floc or aggregate breakup (21). The floc in the vortex unravels, elongates and ultimately
AID
JCIS 4197
/
6g10$$$503
05-13-96 20:41:08
sections itself into smaller flocs. The structure of the floc appears very dense at first; however, as the floc unravels and elongates the structure appears more open and chainlike. It is clear from this sequence that the partition of a floc subjected to an imposed stress that exceeds the floc’s yield strength may not be quick, but rather a process that develops with time. In other words, there is a sequence of particle/ particle bonds of different strengths that break with time according to their relative strength and shifting stress concentration with exposure to the flow; the sequence ultimately leads to the separation of the floc into smaller parts. Figure 13 shows suspension being dragged into the vortex from its outer perimeter, where the suspension is mainly stationary. The flow in this zone is predominantly extensional; i.e., as the suspension converges into the vortex, there is a dramatic increase in velocity, which gives rise to a high gradient of the velocity in the direction opposed to the flow. Therefore, as flocs approach the vortex zone, they suffer a high extensional rate, which causes them to elongate and eventually section into smaller chainlike flocs. A high magnification micrograph confirms this assertion, as shown in Fig. 14, where flocs essentially look like chains with some branches and some denser flocs attached to them. These observations help explain the apparent extensional thinning behavior observed in the iron oxide suspensions. SUMMARY
Visual evidence has been presented that shows the effect of shear, extension and vorticity on the microstructure of iron oxide suspensions. These observations have been used to explain the rheological behavior of these suspensions under shear and extension, and they add new information on the microscopic details and dynamics of the breakup, which may prove valuable to the theoreticians attempting to model such phenomena. The following are the main conclusions: 1. Rheological measurements show that iron oxide suspensions exhibit both extensional and shear thinning. The apparent extensional viscosity measured is one to two orders of magnitude larger than the shear viscosity at the same applied stress. 2. Iron oxide flocs elongate and break up in an extensional flow, which explains the apparent extensional behavior exhibited by iron oxide suspensions. 3. The particle network of iron oxide suspensions at concentrations of 9.4% by volume does not uniformly break up into smaller units of particles in Couette flow, as happens at lower concentrations, but rather it breaks up in the center of the gap and segregates the particles toward the walls. 4. The breakup of the network structure of iron oxide particles in suspensions that results in yielding behavior of the suspension is a process that occurs gradually with time. This process was clearly demonstrated by the breakup sequence of spinning flocs.
coida
AP: Colloid
FLOCCULATED IRON OXIDE SUSPENSIONS
FIG. 14. High magnification of flocs of a 0.5% by volume iron oxide suspension flowing in a nearly extensional flow. Flow was induced by focused light heating.
5. Vorticity, which is present in shear flows, but not in extensional flows, plays an important role in the deformation and breakup of iron oxide flocs. This observation provides an explanation of the rheological measurements under extension which showed that it takes a tensile stress 100 times larger than a shear stress to break a suspension to a similar viscosity level. ACKNOWLEDGMENTS This research was supported by the Center for Interfacial Engineering at the University of Minnesota, which is sponsored by the National Science Foundation and industrial members. We thank the KAO Corporation for providing the particles used here. We are grateful to V. A. Le and J. G. Sheehan, who took the Cryo-SEM micrographs, and to G. G. Glasrud, who made the creep measurements. We thank Rheometrics Scientific, Inc. for the use of the RFX.
REFERENCES 1. Michaels, A. S., and Bolger, J. C., Ind. Eng. Chem. Fundam. 1(3), 153 (1962). 2. Papenhuizen, J. P. M., Rheol. Acta 11, 73 (1972). 3. Seto, J., J. Soc. Rheol. Jpn. 5(4), 156 (1977). 4. Firth, B. A., and Hunter, R. J., J. Colloid. Interface Sci. 57(2), 248 (1976). 5. Firth, B. A., J. Colloid. Interface Sci. 57(2), 257 (1976). 6. Firth, B. A., and Hunter, R. J., J. Colloid. Interface Sci. 57(2), 266 (1976). 7. Van de Ven, T. G. M., and Hunter, R. J., Rheol. Acta 16, 534 (1977). 8. Kamphuis, H., Jongschaap, R. J. J., and Bouter, P., J. Colloid. Interface Sci. 98(2), 459 (1984). 9. Wildemuth, C. R., and Williams, M. C., Rheol. Acta 24, 75 (1985). 10. Schoukens, G., and Mewis, J., J. Rheol. 22, 381 (1978).
AID
JCIS 4197
/
6g10$$$503
05-13-96 20:41:08
211
11. Russell, W. B., J. Rheol., 24, 287 (1980). 12. Kanai, H., Navarrete, R. C., Macosko, C. W., and Scriven, L. E., Rheol. Acta 31, 333 (1992). 13. Buscall, R., McGowan, I. J., Mills, P. D. A., Stewart, R. F., Sutton, D., White, L. R., and Yates, G. E., J. Non-Newtonian Fluid Mech., 24, 183 (1987). 14. Buscall, R., Mills, P. D. A., Goodwin, J. W., and Lawson, D. W., J. Chem. Soc. Faraday Trans. 1 84, 4249 (1988). 15. Weitz, D. A., and Huang, J. S., in ‘‘Kinetics of Aggregation and Gelation’’ (F. Family and D. P. Landau, Eds.), p. 19. Elsevier, Amsterdam, 1984. 16. Schaefer, D. W., Martin, J. E., Wiltzius, P., and Cannell, D. S., Phys. Rev. Lett. 52, 2371 (1984). 17. Helgesen, G., Skjeltorp, A. T., Mors, P. M., R. Botet, R., and Jullien, R., Phys. Rev. Lett. 61, 1736 (1988). 18. Sonntag, R. C., and Russel, W. B., J. Colloid Interface Sci. 113, 399 (1986). 19. Sonntag, R. C., and Russel, W. B., J. Colloid Interface Sci. 115, 378 (1987). 20. Sonntag, R. C., and Russel, W. B., J. Colloid Interface Sci. 115, 390 (1987). 21. Kao, S. V., and Mason, S. G., Nature (London) 253, 619 (1975). 22. Bossis, G., and Brady, J. F., J. Chem. Phys. 80, 5141 (1984). 23. Brady, J. F., Phillips, R. J., Lester, J. C., and Bossis, G., J. Fluid Mech. 195, 257 (1988). 24. Ansell, G. C., Dickinson, E., and Ludvigsen, M., J. Chem. Soc. Faraday Trans. 2 80, 1269 (1985). 25. Toy, M. L., Scriven, L. E., Macosko, C. W., Nelson, N. K., and Olmsted, R. D., J. Rheol. 35(5), 887 (1991). 26. Yang, M. C., Scriven, L. E., and Macosko, C. W., J. Rheol. 30(5), 1015 (1986). 27. Utsugi, H., Endo, A., Suzuki, N., and Ono, K., J. Jpn. Soc. Powder and Powder Metallurgy 32, 100 (1985). 28. Fuller, G. G., Cathey, C. A., Hubbard, B. and Zebrowski, B. E., J. Rheol., 31, 235 (1987). 29. Schunk, R., De Santos, J. M., and Scriven, L. E., J. Rheol. 34, 387 (1990). 30. Cai, J. J., Souza Mendes, P. R., Macosko, C. W., Scriven, L. E., and Secor, R. B., in ‘‘Theoretical and Applied Rheology’’ (P. Moldenaers and R. Keunings, Eds.). Proc. XIth Int. Congr. on Rheol., Vol. 2, p. 1012. Elsevier, Brussels, 1992. 31. Bellare, J., Davis, H. T., Scriven, L. E., and Talmon, Y., in ‘‘Proc. XIth Int. Cong. on Electron Microscopy,’’ 1986 Vol. 1, p. 367. 32. Bellare, J., Sheehan, J. G., Davis, H. T., and Scriven, L. E., Mater. Res. Soc. Symp. Proc. 115, 75 (1988). 33. Sheehan, J. G., and Scriven, L. E., in ‘‘Proc. of the 46th Annual Meeting of the Electron Microscopy Society of America, Elec. Micros. Soc. Am., Woods Hole, MA, 1988,’’ p. 102. 34. Sheehan, J. G., and Whalen-Shaw, M., Tappi J. 23(5), 171 (1990). 35. Navarrete, R. C., Ph. D. Thesis, Univ. of Minnesota, Minneapolis, 1991. 36. Navarrete, R. C., Le, V. A., Glasrud, G. G., Macosko, C. W., and Scriven, L. E., in ‘‘Theoretical and Applied Rheology’’ (P. Moldenaers and R. Keunings, Eds.). Proc. XIth Int. Congr. on Rheol., Vol. 2, p. 625. Elsevier, Brussels, Belgium, 1992. 37. Buscall, R., McGowan, I. J., and Mumme-Young, C. A., Faraday Discuss. Chem. Soc., 90, 115 (1990). 38. Brenner, H., in ‘‘Progress in Heat and Mass Transfer’’ (W. R. Schowalter et al., Eds.), Vol. 5, p. 5. Pergamon, Oxford, 1972. 39. Schunk, R., and Scriven, L. E., J. Rheol. 34(7), 1085 (1990).
coida
AP: Colloid