8460
Langmuir 2004, 20, 8460-8467
Flow Electrification in Nonaqueous Colloidal Suspensions, Studied with Video Microscopy V. A. Tolpekin, D. van den Ende, M. H. G. Duits,* and J. Mellema Physics of Complex Fluids group, Associated with the J.M. Burgerscentrum for Fluid Mechanics and Institute of Mechanics, Processes and Control - Twente (IMPACT), Faculty of Science and Technology, University of Twente, P.O. Box 217, Enschede 7500 AE, The Netherlands Received February 9, 2004. In Final Form: July 8, 2004 Flow electrification in nonaqueous suspensions has been scarcely reported in the literature but can significantly affect colloidal stability and (phase) behavior, perhaps even without being recognized. We have observed it in shear flow experiments on concentrated binary suspensions of hydrophobized silica particles in chloroform. In this low-polarity solvent, electrical charges on the large-particles’ surfaces manifest themselves via long-ranged forces, because hardly any screening can take place through counterions. By shearing the suspension for a prolonged time, we could demonstrate that the effective interactions between the large particles change from weakly attractive (due to the small particles) to strongly repulsive (due to acquired Coulomb interactions). One of the conditions required for flow electrification was the presence of a glass surface in the shear cell. A spectacular manifestation of the phenomenon was observed with confocal video microscopy. First, the formation of large-particle aggregates was seen, while subsequently (over a long shearing time) the aggregates disintegrated into small entities, mostly primary particles. The spatial distribution of these entities in the quiescent state after stopping the flow showed evidence for acquired long-range repulsion. The occurrence of flow electrification was further corroborated by control experiments, where no flow was imposed, antistatic agent was added, or the glass bottom was coated with a conducting (indium tin oxide, ITO) layer: here, the aggregates kept growing until they became very large. To further diagnose the phenomenon, we have also done experiments in which an external electric field was applied (via the ITO layer) to an aggregated suspension. When the lower electrode was given the lowest potential, the aggregates were found to move away from the bottom and disintegrate. The qualitative similarity hereof with the flow electrification experiment suggests that in the latter, the glass acquired negative charges. After prolonged application of an external electric field, we observed segregation into regions enriched in large particles and regions completely depleted of them. In the quiescent fluid these regions exist as isolated units, but in shear flow they merge into bands, a behavior which resembles shear banding.
1. Introduction Flow electrification is the process of accumulation of electrostatic charges on solid surfaces, which are in contact with flowing liquids.1,2 In principle, any interface of a flowing liquid with a solid or another liquid might accumulate (frictional) charges. Especially in low conductivity media such charges can manifest themselves, because here the subsequent discharging can take a very long time. In ref 3 the typical time scale is estimated as τ ) D0/λ, with D being the dielectric constant, 0 being the permittivity of a vacuum, and λ being the specific conductivity of the solvent. If this time is much longer than the characteristic time of the flow (e.g., γ-1 with γ being the shear rate), then significant charges might be generated that affect the behavior of such systems. For example, in the petroleum industry the accumulation of frictional charges has led to numerous explosions.4 Prevention of this phenomenon can be effected by adding an “antistatic” agent - a dissociating substance to provide an electric screening of the interactions or to increase the solvent conductivity and so enable discharging of the surfaces by means of a leaking current. Less dramatic effects can also take place at the interface of a colloidal particle and a flowing liquid. Flow electri* Corresponding author. (1) Touchard, G. J. Electrost. 2001, 51, 440-447. (2) Touchard, G. G.; Patzek, T. W.; Radke, C. J. IEEE Trans. Ind. Appl. 1996, 32, 1051-1057 (3) Morrison, I. D. Colloids Surf., A 1993, 71 (1), 1-37.
fication can in principle take place during the handling of colloidal dispersions (e.g., widely used stirring) leading to a mechanical history dependence of the particle surface charge. For nonaqueous colloidal systems, flow electrification can, thus, modify the particle interactions and, via this, the phase diagram and settling behavior. At present, the understanding of the charge formation mechanism(s) in nonaqueous suspensions by the scientific community is still incomplete. The charge formation seems to be related either to preferential adsorption of ions or to dissociation of some groups on the particle’s surface.3 In a quiescent fluid, the counterion cloud can stay close to the particle surface to neutralize the charge. However, macroscopic flow of the fluid around particles can take away this diffusive part of the charge. Another important issue for flow electrification is the nature of the walls of the fluid container. For example, in the case of a conducting earthened container diffusive charges can be removed from the system, whereas a container made from an insulating material may accumulate a part of the diffusive charges. Also, charged particles coming into contact with a conducting wall can sometimes lose their charge. The goal of the present study is to demonstrate that flow electrification occurs in suspensions of hydrophobized silica particles in a solvent with a low-dielectric constant (chloroform). The occurrence of flow electrification in (nonaqueous) suspensions is not always so easily recognized. Yet it may have large effects, especially at low particle concentrations.3 One of the reasons for this
10.1021/la0496587 CCC: $27.50 © 2004 American Chemical Society Published on Web 09/02/2004
Flow Electrification in Colloidal Suspensions
paradoxical situation is the scarceness of diagnostic tools for charges in nonaqueous suspensions. In this paper, we have used the microscopic observation of colloidal structures formed in flow as a tool. Especially the structure of weakly aggregating suspensions can be very sensitive to flow electrification. For that purpose we have used a confocal scanning laser microscope (CSLM) to observe the flow-induced structural changes of such dispersions in a shear cell. In an attempt to further diagnose the phenomenon, also CSLM experiments in an externally applied electric field have been carried out. This paper is further organized as follows: in section 2 the preparation of fluids and the instrumental setup will be described. In section 3 the results of experimental observations will be presented, along with interpretation. In section 4, a reflection will be given on possible mechanisms, while in section 5 conclusions will be drawn. 2. Experiments 2.1. Synthesis and Characterization of Silica Particles. Different ways exist to make weak aggregates from colloidal silica particles. We have chosen to use (substantially) smaller silica particles as a “flocculating agent” for larger silica particles. By coating the silica spheres with (end-grafted) octadecyl chains and using (near-refractive index matching) chloroform as the solvent, it is achieved that all spheres show negligible attractions and short-ranged repulsions between each other. As a result of these direct interactions, pairs of large particles (from now on termed LPs) can, however, experience an “effective attraction” (see also section 3.1). Both kinds of colloidal silica particles were synthesized inhouse. The method of Sto¨ber et al.5 was used to synthesize primary silica cores from tetraethyl orthosilicate (TEOS) in the presence of ammonia, water, and ethanol. To make the LPs visible with the fluorescence confocal microscope, a relatively small amount of fluorescein isothiocyanate (FITC) dye was added to the TEOS in the first preparation step as described by Verhaegh and Van Blaaderen.6 Primary particles were grown out to their target radius via seeded growth. The final particle radii amounted to 25 nm for the small particles (from now on termed SPs) and 460 nm for the LPs. These silica cores were coated with a dense layer of octadecyl chains via the method described by van Helden et al.7,8 This surface coating renders the particles hydrophobic and provides a steric stabilization against flocculation. We remark here that not all surface silanol groups take part in the surface modification, as follows from a comparison of typical grafting densities7,8 and analysis of bare silica surfaces.9 Sedimentation (using an ultracentrifuge for the SPs, at 10.000 g and with stainless steel bottles) and resuspension in pure solvent was applied to remove unreacted octadecyl chains and to transfer the particles. The SPs were initially transferred to pure cyclohexane to facilitate their sedimentation and subsequently dried and redispersed into chloroform. The LPs were directly transferred to chloroform.17 The chloroform (Merck) was of p.a. quality, stabilized with ethanol (0.6-1.0%). The water content in chloroform was less than 0.01%, according to the manufacturer’s specifications. Particle size and shape (distributions) were measured with transmission electron microscopy. The (460 nm) LPs showed a good sphericity and a polydispersity of 8 ( 2% (defined as the relative standard deviation of the Gaussian function which best fit the radius distribution function). The (25 nm) SPs showed some irregularities in the shape but were still fairly spherical. Here, the polydispersity amounted to 11 ( 2%. (4) Klinkenberg, A.; van der Minne, J. L. Electrostatics In The Petroleum Industry; Elsevier: Amsterdam, 1958; pp 1-191. (5) Sto¨ber, W.; Fink, A.; Bohn, E. J. Colloid Interface Sci. 1968, 26, 62. (6) Verhaegh, N. A. M.; Van Blaaderen, A. Langmuir 1994, 10, 14271438. (7) Van Helden, A. K. J. Colloid Interface Sci. 1980, 76, 418. (8) Van Helden, A. K. J. Colloid Interface Sci. 1981, 81, 354. (9) Iler, R. K. The chemistry of silica; John Wiley & Sons: New York, 1979.
Langmuir, Vol. 20, No. 20, 2004 8461 2.2. Preparation of the Mixtures. Mixtures of SPs and LPs in chloroform were prepared from stock dispersions of the separate particle systems. For each experiment, 1.0 mL of mixture was prepared in a 2-mL capped vial by first adding the SP stock, then the chloroform, and finally the LP stock. Mixtures were freshly prepared, that is, (typically 10) minutes before the start of the experiment. Homogenization was achieved by shaking the vial shortly and gently with the hand. The yellow color of the LPs was helpful in verifying that indeed a homogeneous mixture was obtained. In selected experiments, also an antistatic agent was added to prevent the buildup of static charges within the fluid. We used calcium 2-ethylhexanoate (ABCR), added as 10 µL of a 0.25% (by wt) solution in chloroform. Most of the experiments described in this paper were done at a standard composition (if not specified otherwise): 35 vol % for the SPs and 0.5 vol % for the LPs. 2.3. CSLM. An UltraView CSLM System (Perkin-Elmer) was used for imaging the fluorescent LPs. This system comprises (among others) a Nikon Eclipse TE200 inverted microscope, supplied with a 100×, N.A. 1.30 immersion oil objective. Using a high-speed Z-axis controller, the position of the objective (and with it, the focus in the sample) can be set reproducibly and with submicrometer resolution. Excitation of the FITC fluorescence was done with a 488-nm laser line. Optical sections (i.e., x,y images) were built up by scanning the laser line through a set of two coupled, spatially separated Nipkow disks (one with pinholes and one with microlenses), rotating at high speed. Images were recorded with a Hamamatsu IEEE 1394 C474295-12ERG camera having 1344 × 1024 (square) pixels. Pixel binning ×2 allowed scanning rates up to 65 ms per image, at an effective pixel size of (2 × 0.0645)2 µm2. Images from the camera were directly stored on a PC hard disk. 2.4. Shear Cell. A home-built shear cell with a plate-plate geometry was mounted on top of the inverted microscope of the CSLM. The bottom plate of the shear cell was immobile and made of 0.17-mm thin glass to allow the observations. The glass slide itself was bigger (55 mm) than the sample compartment of the cell (30 mm). By rotating the upper plate, a shear flow is set up in the fluid. The shear rate as defined from the cell geometry and the rotational speed ranged from 0 to 6.3 s-1 (for radial positions between 0 and 9 mm, respectively). Observations were done at 3.5 s-1 unless specified otherwise. The width between the two plates was measured with a micrometer and was set to 300 µm ((25 µm) for all experiments. Only the lowest 80 micrometers of the fluid can be accessed for CSLM observations. Still, any aggregates will manifest themselves at the bottom of the cell, because the particles have a significantly higher density than the continuous phase. To prevent evaporation of the chloroform, a vapor lock filled with glycerol was used. In part of the experiments, modified bottom plates were used to allow setting up an electric field inside the shear cell. For this purpose, glass slides were covered with a thin layer of indium tin oxide (ITO; In/Sn was 90:10 by weight). By depositing approximately 300 nm of this material, electrical conductivity was achieved while keeping the glass transparent enough for the CSLM observations. The coated area did not cover the complete glass surface, but it did cover the sample compartment area and in one radial direction it was made large enough to allow electrical contact with an external voltage supply. Insulation of the glass/ITO electrode from the other parts of the shear cell (all stainless steel or aluminum) was achieved as illustrated in Figure 1 (note that the glass slide is inserted with the ITOcovered side on top). Special care was taken not to break the thin glass slide while ensuring a good electric contact. The upper plate of the shear cell was connected to the other contact of the voltage supply and earthened. The voltage supply could establish an electric current up to 1 A. Because of this we secured the setup with a current-limiting resistor. Prior to the experiments we checked electrical connections and the insulation of the two plates of the shear cell by measuring the resistance between different parts of the electric chain. Given the dimensions of the shear cell and the fact that all parts of the shear cell except the bottom plate were in electric contact with the upper plate, the electric field obtained can be regarded as oriented perpendicular to the shear cell plates, except for the very small region close to the edge of the sample
8462
Langmuir, Vol. 20, No. 20, 2004
Tolpekin et al.
Figure 1. Schematic illustration (not to scale) of the shear cell, including electric connections to a voltage supply: 1, upper plate; 2, glass slide; 3, CSLM objective; 4, shear cell base; 5, vapor lock; 6, vapor lock liquid; 7, poly(tetrafluoroethylene) (PTFE) rings; 8, electric contact with the glass slide; 9, PTFE stopper; 10, rubber O-rings; 11, fixing screw; and 12, insulating PTFE pipe. Fluid samples are inserted between plates 1 and 2. compartment. Because all observations were made within 9 mm from the center of the (15-mm radius) compartment, the electric field can be regarded to be homogeneous in the observed part of the cell. A voltmeter was used to measure the precise potential difference between the electrodes. 2.5. Image Acquisition. Complete three-dimensional information on the fluorescent particle configuration (positions) at rest and under shear was obtained by performing XY scans at different Z positions in the specimen. Each three-dimensional scan consists of (typically) 60 XY scans (images) spaced by 1 µm in the Z direction. Each image, in turn, was a two-dimensional 87 × 66 µm2 CSLM image captured with an exposure time of 65 ms. The time interval between successive three-dimensional scans was gradually changed during the experiment. In the beginning of the experiment, where the quickest changes were expected, it was set to 0. Later the delay was increased up to 120 s in some cases, thus, optimizing the use of disk space for image storage and minimizing particle photobleaching. 2.6. Image Analysis. The analysis of images was done with in-house-developed software written in ALI programming language incorporated in the OPTIMAS 6.5 software package. Two different kinds of analysis were used: identification and coordinate measurements of individual particles, as described in ref 10, and measurements of aggregate sizes in flow, which was described in ref 11.
3. Results 3.1. Interparticle Repulsion Created by Flow. (Flow Restabilization). In the absence of flow, our binary suspensions show evidence for (effective) attractive interactions between the LPs. The origin of this attraction lies in the entropy of the ensemble of SPs and LPs (see, for example, ref 12). The strength of this attraction grows with increasing SP concentration and leads to aggregate formation (and subsequently phase separation) above a (10) Tolpekin, V. A.; Duits, M. H. G.; van den Ende, D.; Mellema, J. Langmuir 2003, 19, 4127-4137. (11) Tolpekin, V. A.; Duits, M. H. G.; van den Ende, D.; Mellema, J. Langmuir 2004, 20, 2614-2627. (12) Dijkstra, M.; van Roij, R.; Evans, E. Phys. Rev. Lett. 1998, 81, 2268-2271.
certain concentration. We have not attempted to measure the precise threshold concentration, but for SP volume fractions of 30% and higher, the formation of LP aggregates was clearly manifested via the sedimentation behavior. This was discussed in more detail in a previous paper.10 Sedimented aggregates accumulated at the bottom of the cell, forming a dense network, which did not change noticeably over time, other than growing higher due to the arrival of new aggregates. In shear flow, remarkable observations were made with the CSLM. Initially, the LPs formed aggregates which subsequently grew over time. This growth stage lasted typically one or a few hours. At much longer time scales, typically after shearing the fluid overnight, the aggregates had disappeared almost completely, that is, only single particles and small aggregates (a few particle diameters in size) were observed (see Figure 2). By inspecting the shear cell at different radial positions, it was ascertained that the large aggregates could not have disappeared by migrating laterally. To quantify these changes, we have analyzed images as a function of time, for the volume-averaged aggregate size, as explained in more detail in ref 11. Measurements were done close to the bottom of the cell to detect also the biggest aggregates. The evolution of the average aggregate diameter in a typical experiment is given in Figure 3. Stopped Flow Experiment. To further diagnose the changes caused by the shear flow, the flow was stopped after 24 h, after which the three-dimensional structure of the fluid was monitored over time. The decision to stop the flow at this time was based on visual CSLM observations, made in the preceding 4 h (i.e., 20-24 h after the experiment was started). In this time, no significant changes in fluid structure and concentration were observable. This is also evidenced by Figure 3, which shows that the average size remained small. Remarkably enough, after stopping the flow the fluid kept on consisting of predominantly single particles, at
Flow Electrification in Colloidal Suspensions
Figure 2. Typical pictures of our dispersions in an early aggregation stage (a) and after 20 h of shearing at 3.5 s-1 (b). The velocity direction is vertical, while the vorticity direction is horizontal. Numbers 1, 2, and 3 correspond to different heights of observation relative to the bottom plate: 5, 10, and 20 µm, respectively. Dimensions of all images are 87 × 66 µm. The elongations seen in pictures b reflect that the fluid velocities are higher than in pictures a. Note that in the presence of large aggregates (as in pictures a), the velocity profile near the bottom becomes nonlinear: as a result of the high concentration of particulate material, the viscosity becomes high (or in microscopic terms, particle obstructions counteract relative motions). For this reason the aggregates in the pictures a have almost no velocity. Redistribution of the particulate material over the volume (as in pictures b) causes the viscosity gradients (as in pictures a) to disappear, thus, restoring a constant velocity gradient and noticeable velocities near the bottom.
Figure 3. Volume-weighted average (particle or aggregate) diameter versus the duration of the shear flow. The procedure for obtaining 〈D〉v can be found in ref 11. For the present fluid, the small number of large aggregates observed per time caused a noticeable statistical error in the measured 〈D〉v. The solid lines are exponential fits, made to characterize the time scales of aggregation (6 min) and breakup (6 h). Data indicated with + are not included in the fit: at this stage, the average size was difficult to measure, because also aggregates bigger than our 87 × 66 µm observation window were present.
least for the next 20 h. This suggests that even after this long resting time, there was enough repulsion left between the LPs to prevent them from aggregating again. The time evolution of the LP concentration height profile is shown in Figure 4. The concentration profile obtained directly after stopping the flow can be regarded as a “snapshot” of the steady-state height distribution created by the shear flow. The LP concentration shows a peak for z < 10 µm, after which it appears to become constant (within the inaccuracy of the measurement) for 10 < z < 80 µm. However, assuming mass conservation for the LP (i.e., the total number of particles in a gap-spanning volume element divided by its volume must equal the initial number density), it is implied that there must be a substantial increase in particle concentration for 80 < z < 300 µm. This is also illustrated by the LP concentration
Langmuir, Vol. 20, No. 20, 2004 8463
profiles recorded 3, 10, and 20 h after stopping the flow, each of which on extrapolation to z ) 300 µm and integration over z results in an about three times larger number of LPs. Given the tendency of the LPs to settle in normal gravity, it is clear that an upward force must be acting on the LPs after a long flow experiment. Stopping the flow, this upward force seems to be gradually reduced, as indicated by the growth of the “concentration versus height” profiles over time. Estimating the LP sedimentation velocity with vsed0 ) 2∆F(φSP)ga2/9η(φSP) (the equation for an isolated LP in an SP-containing continuum fluid and with no other forces than gravity, with a being the particle radius, ∆F being the excess mass density, and η being the viscosity),10 a velocity of 0.02 µm/s is obtained, implying that 4 h would be needed for a single LP to move from the top to the bottom plate under normal gravity. In our system, gravity settling still goes on after 10 h, indicating once again the presence of additional (upward) forces in the vertical direction. Also the sedimentation length z0 ) 3kT/4πa3∆F(φSP)g (i.e., the decay length of the concentration vs height profile) near the bottom of the cell provides information. For single LPs in the absence of forces other than gravity, it should amount 6 µm.10 Fitting the initial decay length (i.e., the constant-slope part in the right-hand plot), we find values ranging between 3 and 5 µm. This suggests that at least the LPs near the bottom of the cell do not feel significant vertical forces other than gravity anymore. Similar experiments were performed for different particle mixture compositions (see Table 1), each time with qualitatively the same results as in the aforementioned experiment, which from now on will be designated as the reference experiment. A quantitative difference between the various experiments was the maximum size that the aggregates could reach and, in relation with that, the duration of the breakup stage. For an SP concentration of 30 vol %, the aggregates initially grew much bigger than in the reference experiment, and it took about 48 h to achieve complete aggregate disintegration. In contrast, for an SP volume fraction of 40% almost no aggregate growth was observed at all. However, the flow-induced repulsions still manifested themselves via a LP redistribution after initial sedimentation. Shear Rate Dependence. Considering that as long as the fluid remains homogeneous, the shear rate is proportional to the radial position in our setup; we have attempted to study the dependence of the flow electrification on the (shear) flow rate. The formation of aggregates was found to take place more quickly near the periphery of the cell, which is in line with the Peclet number being >1 (see ref 10): the shear rate then dictates the encounter frequency between aggregating entities. After a shearing time of 1 h, a distinction could be made between giant aggregates (in cases larger than the 87 × 66 µm observation window) near the center and smaller aggregates near the periphery. This is qualitatively in line with shearinduced breakup; no other mechanisms are needed to explain it. After shearing overnight, the aggregates were found to be broken up almost everywhere in the cell. Experiments in Other Flow Fields. Also some other experiments in less well-defined flows were performed. A number of SP + LP suspensions were put in cuvettes, which were subsequently tumbled end-over for typically 24 h. In this experiment a volume of entrapped air oscillates between the bottom and the top of the cuvette for each rotation. This creates a combination of shear and extensional flows. The state of the aggregation was checked
8464
Langmuir, Vol. 20, No. 20, 2004
Tolpekin et al.
Figure 4. LP concentration versus height, after stopping the long shear flow experiment of Figure 3. N is the number of particles per Z slice. Each (transient) profile was recorded within 15 s. Time lapsed after stopping the flow: 0 h (b), 3 h (9), 10 h (2), and 20 h (+). Table 1. Sample Compositions (Volume Fractions) for the Shear Flow Experiments φSP (% v/v) φLP (% v/v) a
30 2.0
32 1.0
35a 0.5a
37 0.3
37 0.5
37 1.0
40 0.3
Reference experiment.
periodically, by shortly removing the cuvette from the tumbling wheel and inspecting the contents with the CSLM. The aggregate size was seen to go through a maximum, that is, qualitatively similar to Figure 3. In addition, also some LP + SP suspensions were sonicated and subsequently inspected (after transferring them gently into cuvettes). Sonication is obviously not truly a flow, but it does involve (forced) rapid displacements of particles relative to the surrounding liquid. In these experiments a segregation into two fluid microphases was observed, one fluorescent (LP containing) and one “dark” (SP only). The LPs turned out to be disaggregated. 3.2. Addition of an Antistatic. In the presence of a small amount of calcium-2-ethylhexanoate (0.1 mmol of solid added per liter, corresponding to a calculated Debye screening length κ-1 of 7 nm), no flow-induced disintegration of aggregates was observed anymore. Aggregates kept on growing and were found as large sedimented structures at the bottom of the flow cell. The kinetics of aggregation in the presence of the antistatic has been extensively described in a previous paper,10 where a good agreement was found between experimental data and a model, which assumed a binary hard sphere system. In other words, no significant long-range repulsions were needed to describe the aggregation kinetics in the presence of the antistatic. The antistatic was also added to a mixture of SPs and LPs in an experiment without flow to study its influence on the sedimentation behavior. Macroscopic observations of the sedimentation velocity revealed no significant difference in the presence or absence of the salt. This result is in line with the previous observations: in experiments involving just a short shaking, followed by a long resting time, there is hardly any build up of flow-induced repulsions, while the time to “erase the flow history” is long. Considering the results obtained in the presence and in the absence of the antistatic, we conclude that the flow restabilization as described in section 3.1 is due to flowinduced electrical charging of the LPs. This causes a repulsion between LPs that is strong enough to break up aggregates and redistribute sedimented LP material over the volume of the cell. Viscous resuspension (i.e., a vertical redistribution of suspended matter as a consequence of hydrodynamic forces) is ruled out as a cause, because this should not depend on the presence of the antistatic. All further experiments to be described were done in the absence of the antistatic, to allow a deeper investigation of the flow electrification phenomenon.
3.3. Coating the Glass Bottom with a Conducting Layer. A number of experiments were performed in which the glass bottom of the shear cell was coated with a thin layer of ITO to provide electrical conductivity. All fluid samples had the same composition as those in the reference experiment. The first experiment was to shortcut the two plates electrically, to ensure that no electric field could be present in the cell. Subjecting the sample to a shear flow in this condition, the LPs were seen to aggregate, and subsequently the aggregates were found to sediment to the bottom of the cell (see Figure 5a1). No aggregate breakup occurred, even for shearing times longer than 24 h. This finding sheds additional light on the flow electrification: apparently it does not occur in the absence of an electric field or if a leaking current is facilitated. This suggests that in the absence of the ITO coating, charge accumulates on the glass bottom plate, which in turn sets up an electric field in the cell. Also it is once more indicated that the restabilization and LP redistribution in our fluid is not due to shear-induced resuspension, because the ITO does not make a mechanical difference. 3.4. Application of External Electric Fields. Anticipating that the situation created in the flow electrification experiment (of section 3.1) could be mimicked by applying an external potential difference over the top and bottom plates, we performed experiments with and without shear flow and for different polarities. 3.4.1. Electric Field, No Shear Flow. External electric fields were applied to samples in a standard quiescent state, reached by shearing the fluid for 1 h at zero electric field (to effect aggregation) and then stopping the flow. Hereafter, several experiments were done: for different polarities and magnitudes of the electric field and also more complicated experiments in which the polarity was reversed after some time. The steady state of our suspension in electric fields was found to be independent of the previous mechanical and electric field history. Also the results did not depend too sensitively on the magnitude of the applied voltage: 5 or 10 V (corresponding to a calculated field of 3 × 104 V/m in the latter case). Choosing the polarity “plus” or “minus” or switching the polarity, however, gave dramatically different results, as will be discussed below. Lower Plate as a Negative Electrode. Before switching on the electric field, the LPs were allowed to form aggregates in a shear flow and sediment to the bottom of the cell. A typical image of the thus obtained fluid structure is shown in Figure 5a1. Application of a 10-V potential difference (on the voltage supply output) caused an almost complete disaggregation of the LPs. After 1 min only a few doublets were left; the rest of the aggregates had disintegrated into single particles (see Figure 5b1) and redistributed over space.
Flow Electrification in Colloidal Suspensions
Langmuir, Vol. 20, No. 20, 2004 8465
Figure 5. Typical images of aggregates taken (a1) after preshearing in the absence of an electric field, (b1) in an electric field with a negative electrode at the bottom, and (c1) in an electric field with positive electrode at the bottom. Note that Figure a1,b1 corresponds to the same sample, before and after switching the electric field. Dimensions of images a1 and b1 are 87 × 66 µm; image c1 has dimensions 22 × 17 µm. Graphs below the images correspond to particle distributions with respect to height (a2, b2, c2) and pair correlation functions g(r) measured in plane (a3, b3, c3). The images were taken respectively at 8, 12, and 15 µm above the glass, corresponding approximately to the locations of maximum particle concentrations. Note that there is a clear long-range order without periodicity for the aggregated state (a3), very little correlation in the disaggregated state (b3), and periodic order with a crystal-like structure in case c3. Also note that all samples show a nonhomogeneous distribution of LP with height; for c2 a depletion zone near the bottom is observed (as well as artificial “peaks” between 10 and 30 µm which are due to an increased inaccuracy in the measurement of high concentrations).
Another noticeable observation was that a fraction of the LPs was expelled from the first 10 µm next to the bottom plate, ending up in the adjacent 40 µm (see Figure 5a2,b2). Inspecting the 50-80-µm range (not clearly visible in the figures) revealed no significant differences between the two situations. While Figure 5a2-c2 shows concentrations obtained by averaging over the whole xy image, it is also interesting to look at the spatial correlations within the pictures. Two length scales should be considered here. The first is that of the radial distribution function g(r), defined at the particle level as the normalized probability of encountering another particle at an in-plane distance r: when parts a3 and b3 of Figure 5 are compared, it is evident that the fraction of nearest-neighbor contacts has diminished considerably. Also correlations at larger length scales should be considered. As is visible in Figure 5b1 (and more clearly in Figure 6c,d, which are also representative for the presently considered experiment), a segregation into LPrich and LP-poor phases had taken place. The LP concentration gradients at the interface of the segregated domains were the sharpest near z ) 0 and gradually decreased with height. At z ) 50 µm, the fluid was homogeneous again. Over time, the gradients gradually became sharper, resulting in very clearly separated regions near the bottom of the cell (as in Figure 6d).
After switching off the external electric field, the LP distribution in space turned from segregated to homogeneous, while also LP aggregation was seen to occur again. Studying the kinetics of aggregation after switching off the electric field, we found that both under quiescent conditions and in shear flow the characteristic aggregation times corresponded well to the results of experiments in which no electric field was ever applied. This indicates that our fluids have only a short-time memory (minutes) for externally applied electric fields. Lower Plate as a Positive Electrode. In this case an entirely different response was obtained in the lower part of the cell. All LP aggregates and single LPs were now found to assemble into a single, relatively compact structure, which extended over the full xy image range (see Figure 5c1). Remarkably enough, also in this case a LP-depletion zone near the bottom plate was observed (see Figure 5c2). The compact assembly extended up to z ) 35 µm, while at larger z, there were still loose LPs and very small LP aggregates. The particle radial distribution function shown in Figure 5c3 shows an ordered arrangement of the particles within the assembly. Switching off the electric field revealed no significant changes on the time scale of minutes. However, when switching to opposite polarity, the large compact LP agglomerate was destroyed within 1 min, resulting in a
8466
Langmuir, Vol. 20, No. 20, 2004
Tolpekin et al.
Figure 6. “Shear banding” effect (a), snapshot of bands directly after stopping the flow (b), evolution of the “band” structure after some time in a quiescent state (c and d). Dimensions of all images are 87 × 66 µm. The velocity direction is vertical, while the vorticity direction is horizontal.
very similar LP space distribution as previously found with a negative lower electrode: apparently the latter condition leads to erasure of the previous mechanical and electric field history. 3.4.2. Electric Field and Shear Flow. We also applied external electric fields to suspensions in flow. Also in these experiments, suspensions were sheared for 1 h before the voltage supply was switched on. Regardless of the field orientation, marked space inhomogeneities were now observed: “bands” enriched in LPs, separated by dark “bands” (see Figure 6a). This behavior bears at least a close resemblance to the so-called “shear banding” phenomenon, which has been receiving a lot of interest since the 1990s (see refs 13-15 and references therein). Inspecting all accessible radial positions (0 < r < 9 mm) in the shear cell, the bands were found everywhere. Differences found for opposed field orientations were as follows: (1) When the lower electrode was given a negative potential, similar LP concentration profiles were found as for the corresponding experiments in the quiescent state. For opposed polarity, the concentration z profile was rather similar, except that the z range 0-10 µm was depleted of LPs. The transition to z heights where bands were seen was a sharp one. (2) When the lower electrode was given a negative potential, all LPs occurred as individual entities. For opposed polarity, a certain amount of (not very large) aggregates was formed within the bands. Like in the experiments without flow, no history dependence of the final state on the order of switching the shear flow or electric field was observed. After stopping the flow, the structure of the bands slowly evolved to almost (13) Dhont, J. K. G.; Lettinga, M. P.; Dogic, Z.; Lenstra, T. A. J.; Wang, H.; Rathgeber, S.; Carletto, P.; Willner, L.; Frielinghaus, H.; Lindner, P. Faraday Discuss. 2003, 123, 157-172. (14) Dhont, J. K. G. Phys. Rev. E 1999, 60, 4534-4544. (15) Fielding, S. M.; Olmsted, P. D. Phys. Rev. Lett. 2003, 90, No. 224501.
spherical segregation regions as shown in Figure 6c,d. If the flow was applied again in a situation as in Figure 6d, the reverse process was observed, thus, reflecting that the separation of the LPs in bands is an effect of the flowinduced elongation or merging of the LP-containing regions. 4. Reflection 4.1. On the Effects of an Electric Field. One may wonder what happens on applying an external electric field to our flow-electrification suspensions. At least two processes can in principle play a role. First of all, particles that carry initial charge will respond to the electric field via their electrophoretic mobilities. The question is how much surface charge was present before the electric field was applied. The observed aggregation speed in the preceding hour suggests that at least a significant part of the LP was not noticeably charged. However, a strong electric field (as applied) could still have a strong impact on the present charges. Also, it could activate “latent charges” on the particles, by removing a counterion. Second, there may be an indirect effect of electric-field induced motion of LPs through the suspension. Our online observations showed that these motions are fast, which makes it conceivable that additional frictional charges may be created on the LP surfaces. Because such a “secondary charge accumulation” would also enhance the electrophoretic mobility, even a self-accelerating process would be possible. Another question is to what extent our externally applied electric field resembles the situation created in the flow electrification experiment with the insulating glass bottom. First of all, the disaggregation observed with the lower plate as a negative electrode bears a striking resemblance with the flow electrification experiment. It is also in line with the expectation that the insulating
Flow Electrification in Colloidal Suspensions
glass bottom should carry a negative charge. A closer correspondence between the two experiments was not to be expected in our opinion. The typical time scales for the disaggregation are entirely different, but this might be expected in view of the long time needed to build up (and relax) charges on an insulating surface in contact with an insulating liquid. Also the strength of the electric field created by the charges on the insulating glass may be of an entirely different magnitude. A last remark in this respect concerns the homogeneity of the electric field. Because our fluid contains charged species which respond to an electric field via a spatial redistribution, it is beyond doubt that the electric field will be modulated by the particles and should, hence, be considered as a local variable. For our system it is not possible to further address this issue, because the local field strength cannot be measured, while the spatial charge distributions (for the glass, the LP and possibly also the SP) are also unknown. In this light, the observed “shear banding” may involve a complex spatial dependence of the charge density and the electric field. 4.2. On Possible Mechanism(s) for Flow Electrification. In this section we want to present what we think of as the simplest picture that might explain our key observation: the flow electrification. This “explanation” will be inherently inconclusive, because the distribution of electrical charges over space and over the constituents of our system cannot be assessed. Nevertheless, we feel that some considerations could guide further thinking. Because silica and glass are known to be “acidic” surfaces,16 the most straightforward assumption would be that the glass and the LPs (and possibly also the SPs) act as proton donors and that the chloroform solvent acts as an acceptor. We recall here that even silica particles, which are stearylated, still have ionizable silanol groups available and that indications for a negative charge on the glass and the LP were obtained from our experiments. A velocity difference between an ionizable surface and its solvent environment is needed to separate the protons from their donors. Imposed flow can facilitate this, by supplying the required energy and by separating the charged species in a sufficiently short time to make recombination an unlikely event. For aggregates in flow, velocity differences between particle monomers and the solvent environment are the largest in the outer shells, which explains why our aggregates appeared to be “peeled down” over time. As a result of the accumulation of negative charges on the glass bottom, an electric field is set up in the cell. Depending on its strength, this electric field may induce additional charge separation at the particle surfaces, via its electrokinetic effect on particles that already carry some charge. When the glass is coated with ITO, charge accumulation on the bottom is prevented and no electric field is set up across the suspension. In the bulk fluid, the creation of charge on LP surfaces will still take place, but this is no longer amplified by an electric field. In the presence of 0.1 mM calcium 2-ethylhexanoate, charge separation is modulated by the availability of Ca2+ ions and 2-ethylhexanoate anions. Ca2+ could replace protons at the glass/silica surfaces, thus, diminishing the (16) Van der Hoeven, P. H. C.; Lyklema, J. Adv. Colloid Interface Sci. 1992, 42, 205-277. (17) It is interesting to note that Verhaegh and van Blaaderen6 found at this stage for their stearylated silica particles in chloroform a reduced sedimentation rate and the formation of colloidal crystals, which were attributed to electrical charge on the particles’ surfaces. For our particles we did not observe such behavior during the particle synthesis. However, exposing the particles to prolonged flow did have a strong effect on our particles (see text).
Langmuir, Vol. 20, No. 20, 2004 8467
accumulation of negative charge. In addition, assuming a full dissociation of the salt, an electrostatic screening with a double-layer thickness of 7 nm will be provided. Flow-induced charges may also be present on the SPs. A particular aspect to consider for this study is that the concentration of SPs has been rather high. As a consequence, long-range repulsions between the SPs are not to be expected, even if they would carry charge. The reason for this is that the interparticle distance for SPs is much smaller than the typical distance over which the overall electrostatic potential experienced by a SP decays.3 Therefore, the disintegration of the LP aggregates has to involve at least a change in the charge state of the LP surfaces themselves. 5. Conclusion We have presented evidence for the occurrence of flow electrification on hydrophobized silica particles in chloroform. The effect of the charges on the particles is strong enough to turn the phase behavior from unstable (depletion flocculation) to a highly stable one and to affect the spatial distribution of the particles in the system. Necessary conditions for the frictional charge development are a dielectric (glass) bottom of the fluid container and sufficiently long flow history of the fluid. The initial aggregate growth process is described by the same time scale as for noncharged (or slightly charged) particles. Because the time scale is very sensitive to the repulsive part of the interaction potential (see ref 10), the amount of charge on the particles in that part of the experiment can be supposed to be negligibly small. However, after a long shearing time, the charge on the particles grows big enough to change the system behavior drastically. The time needed to reduce the size of aggregates to the level of single particles was found to be dependent on the maximum size of aggregates, attained during the initial growth stage. This suggests that the LPs on the outer shell of the aggregates are charged first and then leave the aggregate. We have also demonstrated that a similar (though not entirely the same) disintegration of aggregates occurs when an external electric field is applied, with the lower potential applied to the bottom plate. This could indicate that in the flow experiments with a nonconducting glass surface, negative charge is collected on this surface, which subsequently creates an electric field in the cell. Uncovering the exact mechanism for the flow electrification remains difficult. Proton transfer between the “acidic”16 silanol groups on the glass and particles’ surfaces and the chloroform solvent molecules could result in negatively charged glass and particles and positive counterions. This charge separation appears to be facilitated both by flow and by electric fields. Perhaps the most important for practical situations is the implication of our findings that, without precautions such as adding an appropriate antistatic or using sufficiently earthened containers, suspensions of particles with ionizable surfaces may show a “memory” for their electric or mechanical history. Acknowledgment. We thank Matthijn Dekkers from Inorganic Material Science Group of University of Twente for covering glass slides with ITO and Martijn Kamps for exploratory experiments. LA0496587
