摘要:

Andrew P. Michelmore* and Robert A. Hayes† Ian Wark Research Institute, University of South Australia, Mawson Lakes, SA 5095, Australia. E-mail: micap001@students.unisa.edu.au Received 15th March 2000, Accepted 9th May 2000, Published 16th May 2000 Two models have recently been proposed to explain the electrokinetic properties of surfaces with electrostatically deposited particles (R. A. Hayes, Colloids Surf., A, 1999, 146, 89, and M. Zembala and Z. Adamczyk, Langmuir, 2000, 16, 1593). However, supporting experimental data is sparse. In this study, the effects of particle size, particle coverage and ionic strength on the electrokinetic behaviour of positively charged surfaces with negatively charged particles deposited were investigated by streaming current measurements.Results confirm that deposition of oppositely charged particles decrease the magnitude of the zeta potential of the underlying surface. Reversal of the zeta potential was observed at all particle sizes with larger particles reversing the potential at lower coverages. The results presented are fitted to the respective models, and the relative merits of each are discussed. The effect of deposition of negatively charged particles on the electrokinetic behaviour of oppositely charged surfaces this substrate, silica particles (0.25–1 P GLDPHWHU ZHUH adsorbed, the process being governed by electrostatic attraction between the positive substrate and negative particles. Introduction Electrostatic adsorption of particles occurs in a wide range of industries.The adsorption of unwanted "slime" particles onto the surface of valuable minerals, for example, can drastically alter the zeta potential of mineral slurries. It has been shown previously1,2 that the yield stress of suspensions varies linearly with the square of the zeta potential. Electrostatically deposited slime particles can thus flocculate mineral slurries, raising the yield stress and necessitating a large energy input to induce the required flow rate. The kinetics of particle adsorption have been investigated extensively using the stagnant point flow apparatus3–5 and other techniques.6–8 The effect that the adsorbed particles have on the electrokinetic properties of the underlying surface is surprisingly poorly understood, yet it can have a dramatic effect on the properties of many systems. While theoretical solutions exist for single particles adsorbed on a surface,9–11 the solution for colloidal particles adsorbed on a macroscopic surface has not yet been determined.An increased understanding of the way in which adsorbed particles alter the electrical properties of the macroscopic surface would therefore be of enormous benefit, both practically and fundamentally. The purpose of this study was to provide quantitative experimental evidence using a model material system, and to assess two recently proposed electrokinetic models.12,13 The electrokinetic properties of glass microscope slides modified by adsorption of di(aminopropyl) trimethoxysilane were measured by the streaming current technique.The adsorption of the silane renders the glass surface positively charged at pH values less than 8. A pH of 5.6 ± 0.1 was used for all experiments reported here. Onto DOI: 10.1039/b001277g PhysChemComm, 2000, 6 Experimental Apparatus An Anton–Paar electrokinetic analyser was used in all streaming current experiments. The streaming current apparatus is schematically represented in Fig. 1. The test substrates are clamped parallel to each other with a channel between them created by a Teflon spacer. By means of a pump, a pressure drop is created across the channel between the two surfaces, inducing a liquid flow. The movement of the liquid convects ions in the mobile part of the electrical double layer of the surfaces in the direction of the flow.Platinum black electrodes are placed in the vicinity of the ends of the channel. A wire is connected to the platinum electrodes to short circuit the cell, allowing current to flow continuously around the cell. The connecting wire must be of insignificant resistance compared to that of the liquid. As the liquid flow convects ions in the direction of flow, the so-called streaming current (Is) is created. Due to the short circuit, there is no build up of charge at either end of the channel under ideal conditions. The streaming current is typically of the order of 50 nA. To enable the current to be measured, the electrodes were connected to a Keithley 427 current amplifier.The gain was typically set to 107 V A–1. The resulting signal was then accumulated on a chart recorder. The zeta potential ( z) of the surfaces is determined from the slope of the streaming current–pressure drop relationship in eqn. (1).hL s cell ) (1) dI dP D 0 cell eA Fig. 1 Above, a schematic of the streaming current apparatus showing pressure controller, ammeter and platinum black electrodes. Substrates are shown in blue. Below, a schematic of the effect of polarisation on the measured streaming current with time. (2a) (2b) z= ( where is the viscosity of the fluid, Lcell is the length of the cell (76 mm), D is the dielectric constant of the medium, 0 is the permittivity of a vacuum, and Acell is the cross sectional area of the channel between the surfaces (0.3 mm × 10 mm).Typically, the streaming current was measured at four pressure drops, up to 400 mbar, in both directions of flow. The resulting plots of streaming current versus pressure drop all gave linear plots with r2 > 0.98. The streaming current induced by the liquid flow is ionic in nature. However, current flow through the short circuit wire must be electronic. For the current to flow continuously around the short circuit shown in Fig. 1, a charge transfer reaction must occur at the interface between the liquid and the electrodes.14 The charge transfer mechanism is via reduction–oxidation reactions.15 2H2O + 2e– � H2 + 2OH– 2H2O � O2 + 4H+ + 4e- In an ideal system, the reduction–oxidation reactions at the electrode will be fast enough that no charge will accumulate at the electrodes.However, if the flow of ions to the electrodes is not balanced by electronic flow through the short circuit wire, then charge will accumulate resulting in a non-steady state where the streaming current will decay with time, as shown in Fig. 1. This phenomenon is called polarization. After the liquid flow is stopped, the polarization results in a reversal of the current which returns the system to its initial state. The rate of redox reactions is the rate-limiting step. Thus, as the flow of ions due to liquid flow increases, polarization becomes increasingly important. Consequently, the rate of polarization increases with both pressure drop and ionic strength.The true streaming current is the peak value which occurs before polarization becomes important. Under ideal conditions, the pressure drop (and streaming current) across the cell is created instantaneously. In practical applications though, the time taken to apply and control a set pressure drop is of the order of seconds. If the ionic strength is too high, then polarization can be of importance before the set pressure drop is achieved and maintained. As a result, streaming current measurements can be difficult at high ionic strength (>0.001 M), often resulting in an S-shaped plot of pressure drop versus measured streaming current. The ionic strength for this study was therefore limited to a maximum value of 0.001 M KCl to avoid this potential problem.Materials The substrates used in all experiments were glass microscope slides treated with di(aminopropyl) trimethoxysilane. All slides were cleaned prior to treatment using the following procedure. The glass was immersed in a 2% Extran (detergent) solution and sonicated for 10 min to remove any particles on the glass surface. The samples were then rinsed extensively in distilled water, then ethanol and finally in heptane before being dried under a stream of filtered nitrogen gas (0.05 mm). The slides were then exposed to plasma (0.25 mbar O2, 300 W) for 15 min. The glass surfaces were tplaced in a 0.15 wt.% di(aminopropyl) trimethoxysilane solution in 50 wt.% isopropanol, 50 wt.% high purity water.The substrates were immersed in the solution for 15 min before being thoroughly rinsed in isopropanol, dried under nitrogen and then temporarily stored in aluminium foil prior to testing. Monospher 1000, 500 and 250 silica particles were used as the adsorbate in particle adsorption experiments. The particle sizes are nominally 1, 0.5 and 0.25 mm, respectively. The samples were white powders, and were used as provided from Merck without further purification. 200 ppm dispersions were prepared in cleaned one-litre HDPE bottles with high purity water at pH 5.6 at 0.001 M KCl. The dispersions were sonicated initially for 10 min to break up large aggregates, then were left on rollers for at least 72 h to allow the silica surface to hydrate and disperse.Observation by optical microscopy revealed that the Monospher 1000 and 500 particles were fully dispersed after 72 h, however the Monospher 250 particles took at least one week to disperse. Samples for streaming current experiments were prepared by placing two modified glass substrates into a petri dish with a 200 ppm dispersion of particles to allow particles to deposit onto the surface. The ionic strength for this stage was kept constant at 0.001 M KCl throughout. The particle coverage was controlled by altering the time allowed for adsorption.5 For the Monospher 1000 and 500 particles, after a short time the substrates were removed and rinsed thoroughly with high-purity water and placed under an optical microscope for image analysis purposes.The samples were then placed in the Anton–Paar apparatus and the streaming current was measured at pH 5.6 at different ionic strengths starting at low ionic strength. All electrolyte solutions were made up from reagent-grade KCl. The slides were kept in contact with electrolyte throughout to eliminate surface aggregation due to capillary forces. Thesubstrates were then placed in the dispersion again to allow further deposition to occur. Monospher 250 particles were too small to observe optically while in contact with water, so different substrates were used each time and the surface was dried prior to image analysis. The surface coverages were determined by averaging three images. Fig. 2 shows an example of Monospher 1000 particles adsorbed on a modified glass surface.Fig. 2 Optical microscope image of Monospher 1000 particles adsorbed on modified glass surface (coverage = 0.32) after deposition from 200 ppm dispersion. Ionic strength = 0.001 M KCl, deposition time = 1 h. Modelling Recently an empirical model has been proposed by Hayes to empirically describe the observed electrokinetic behaviour of surfaces with deposited particles of size 0.03– 0.2 mm diameter.12 This attempts to reconcile the observed decrease in the magnitude of the zeta potential of the surface with adsorbed particles by predicting a shift of the plane of shear due to the deposited particles. The model assumes the shear plane is displaced from the surface in the particle-free areas by an effective distance (deff), as shown in Fig.3. The predicted zeta potential of the macroscopic surface is given in eqn. (3). Fig. 3 Physical representation of the shear plane being displaced from the underlying surface due to the deposition of particles. exp( ) ) 1 ( = - k z q e s p p p d ff + p z deff C q 0p - q zp s q+ zzs s = s q dp where z is the zeta potential, q is the particle surface coverage, k is the inverse of the Debye length, d is the particle diameter and the subscripts p and s denote particle and surface contributions respectively. The parameter s is a free fitting parameter. deff is an average quantity and no local information about the position of the surface of shear is provided by the model. A theoretical model has been proposed by Zembala and Adamczyk, based on the flow properties of the electrolyte across a surface with deposited particles, and the electrical properties of the particles and underlying surface.13 This model predicts that the observed electrokinetic behaviour is due to damping of the flow of electrolyte near the surface due to the presence of the particles.For this model the zeta potential of the macroscopic surface can be found using eqn. (4). = e-C0 s z z 0 and C 0 are parameters relating to the flow p where C properties at the surface and over the particles respectively. Results and discussion Fig. 4 shows the normalised zeta potential of di(aminopropyl) trimethoxysilane treated surfaces with adsorbed silica particles of different sizes as a function of the surface coverage. The solid lines are fits using the Zembala and Adamczyk model and the dotted lines are fits using the Hayes model, determined by non-linear least squares analysis.The fitting parameters were C 0 and C 0 for the Zembala and Adamczyk model, and s for the Hayes model. The models show good agreement for coverages up to 0.3. The zeta potential of the bare surface at pH 5.6 is 63 and 87 mV at KCl concentrations of 0.001 and 0.0001 M, respectively. Monospher particles are negatively charged at pH 5.6 (zeta potentials were measured independently and are reproduced in Table 1) and so lower the potential of the surface as expected. A sharp initial decrease in the potential is observed, followed by a plateau of the potential at high coverages.Reversal of the potential at sufficiently high silica particle coverages is observed in all cases. At high coverages, the zeta potential of the macroscopic surface is significantly less than that of the silica particles. This may indicate either that the surface still contributes to the total measured potential, or the particles contribute only to the area-weighted component of their potential.12 (3) (4) p sFig. 4 Normalised zeta potential of modified glass surface with adsorbed Monospher particles at I = 0.001 M ( ) and 0.0001 M KCl ( ). Particle sizes are Monospher 1000 (black) Monospher 500 (blue) and Monospher 250 (red). Dotted lines are fits to the Hayes model, solid lines are fits to the Zembala and Adamczyk model.Table 1 Zeta potential of Monospher particles measured using a Malvern Zetasizer II Zeta potential/mV Zeta potential/mV Particle size/nm I = 0.0001 M I = 0.001 M –43 –41 250 –37 –28 500 –52 –49 1000 Similar to the data of Zembala and Adamczyk, all plots show a common intersection close to the point of zero potential for the two ionic strengths used. Vincent et al. also observed this phenomenon.16 In this case the electrophoretic mobility of 3.2 P QHJDWLYHO\ FKDUJHG polystyrene latex particles was measured as smaller (0.2 P SRVLWLYHO\ FKDUJHG ODWH[ SDUWLFOHVZHUH DGVRUEHGRQWR the surface. At all ionic strengths (10–5–10–2 M NaCl) reversal of the potential occurred at around 10% coverage.An initial sharp increase in potential was shown, with a plateau value evident at higher coverages. Interestingly, the result is similar to that found for ionic surfactant adsorption.17 The common i.e.p. for ionic surfactant adsorption is attributed to neutralisation of the surface charge. Due to their macroscopic size this result is not necessarily expected for particle adsorption. Fig. 5 shows the dependence of the fitting parameter s for the Hayes model on the particle size and ionic strength. There is a clear difference between the s values obtained at the different ionic strengths, and there appears to be a slight dependence on the particle size. Higher s values at lower ionic strength indicate that the shear plane is displaced from the surface to a greater extent.The physical significance of the s parameter is unclear at this stage, however the linear trends observed indicate that the empirical model, despite its simplicity, is capable of adequately describing the electrokinetic behaviour of surfaces with adsorbed particles, despite the fact that the s parameter is not explicitly determinable. Fig. 5 s value determined from eqn. (3) as a function of particle size. s p p s p 0 s The initial slope of the normalised zeta potential with coverage shows a clear dependence on the particle size, with larger particle reversing the zeta potential at lower coverages. This manifests itself in the fitting parameters 0 and C 0 C used in the Zembala and Adamczyk model, as the particle size does not appear explicitly in the model.As shown in Table 2, C 0 varies little with particle size, however C 0 increases with particle size. This result is not observed by Zembala and Adamczyk as their data show a universal relationship, independent of both ionic strength and particle size. According to their analysis, C and C 0 are constants (10.21 and 6.51, respectively), independent of particle size and ionic strength. The values determined here are of the same order of magnitude. Table 2 Fitting parameters for the Zembala and Adamczyk model Particle size/nm Cp0 Cs0 2.20 1.98 250 1.29 6.55 500 1.52 15.6 1000 Intuitively, the flow of electrolyte over the particles should not be affected by the particle size, however the damping of flow over the surface should depend on the particle dimension.A similar result has been shown by Warszynski18 for particle adsorption, where an "electrohydrodynamic shadow" is observed downstream of adsorbed particles. Due to the flow properties, this region is unavailable for particles to adsorb to. The size of this region is dependent on the particle diameter. Conclusions It has been shown that electrostatically adsorbed particles affect the electrokinetic behaviour of the underlying surface, decreasing the magnitude of the zeta potential at low surface coverages, and completely reversing the potential at higher coverages. There is a strong dependence on the particle size.Apart from illumination of the basic physics involved, the empirical model developed by Hayes can be used to reconcile measured electrokinetic behaviour with that of the constituent surfaces, for particles in the size range 0.03–1 P 7KHPRUHULJRURXVPRGHORIZembala and p 0 s Adamczyk fits the experimental data, however the dependence of the C and C 0 terms on the particle size remains unclear.Acknowledgements The authors would like to warmly thank both Dr Gerald Belder and Dr Marcel Böhmer from Philips Research Laboratories, Eindhoven, The Netherlands, for their generously offered assistance regarding streaming current work. Financial assistance from AMIRA and DIST is also acknowledged. References 1 W. B. Russel, J. Rheol., 1980, 24, 287. 2 B. Firth and R. J. Hunter, J. Colloid Interface Sci., 1976, 57, 266. 3 Z. Adamczyk, B. Siwek, M. Zembala and P. Belouschek, Adv. Colloid Interface Sci., 1994, 48, 151. 4 T. Dabros and T. G. M. van de Ven, Colloid Polym. Sci., 1983, 261, 694. 5 R. A. Hayes, M. R. Böhmer and L. G. J. Fokkink, Langmuir, 1999, 15, 2865. 6 M. Elimelech, J. Colloid Interface Sci., 1991, 146, 337. 7 M. Elimelech, J. Colloid Interface Sci., 1994, 164, 190. 8 F. Vasak, B. D. Bowen, C. Y. Chen, F. Kastanek and N. Epstein, Can. J. Chem. Eng., 1995, 73, 785. 9 P. Warszynski and Z. Adamczyk, J. Colloid Interface Sci., 1997, 187, 283. 10 J. Stahlberg, U. Appelgren and B. Jonsson, J. Colloid Interface Sci., 1995, 176, 397. 11 H. Ohshima, J. Colloid Interface Sci., 1998, 198, 42. 12 R. A. Hayes, Colloids Surf., A, 1999, 146, 89. 13 M. Zembala and Z. Adamczyk, Langmuir, 2000, 16, 1593. 14 A. G. van der Put, Ph.D Thesis, University of Wageningen, 1980. 15 CRC Handbook of Chemistry and Physics, 74th edn., CRC Press Inc., Boca Raton, FL, 1993. 16 B. Vincent, C. A. Young and T. F. Tadros, J. Chem. Soc., Faraday Trans., 1980, 76, 665. 17 A. de Keizer, M. R. Böhmer, T. Mehrian and L. K. Koopal, Colloids Surf., 1990, 51, 339. 18 P. Warszynski, Adv. Colloid Interface Sci., 2000, 84, 47. Footnote † Current address: Philips Research, Prof. Holstlaan 4, 5656AA, Eindhoven, Netherlands. PhysChemComm © The Royal Society of Chemistry 2000

ISSN:1460-2733    

DOI:10.1039/b001277g

出版商:RSC    

年代:2000

数据来源: RSC