Colloids and Surfaces A: Physicochemical and Engineering Aspects

Transcription

1 Colloids and Surfaces A: Physicochem. Eng. Aspects 364 (2010) 7 15 Contents lists available at ScienceDirect Colloids and Surfaces A: Physicochemical and Engineering Aspects journal homepage: Permeation of a cationic polyelectrolyte into mesoporous silica. Part 2. Effects of time and pore size on streaming potential Martin A. Hubbe a,, Ning Wu a, Orlando J. Rojas a,b, Sunkyu Park a a North Carolina State University, Department of Forest Biomaterials, Campus Box 8005, Raleigh, NC , USA b Helsinki University of Technology, Department of Forest Products Technology, Faculty of Chemistry and Materials Sciences, P.O. Box 3320, FIN TKK, Espoo, Finland article info abstract Article history: Received 17 December 2009 Received in revised form 5 April 2010 Accepted 7 May 2010 Available online 15 May 2010 Keywords: Permeation Polyelectrolyte Silica gel Pore size Time Streaming potential Desorption Salt concentration Molecular mass Streaming potential tests were carried out to determine effects of time and pore size in the adsorption and desorption from aqueous suspensions of cationic polyelectrolytes on silica gel particles. Results in Part 1 of this series showed that the adsorption of cationic polyelectrolytes exposed to mesoporous silica gels can be highly dependent on ph, the polyelectrolyte s molecular mass, and the solution s electrical conductivity. Also, the observed changes in streaming potential indicated that the adsorption tended to be relatively slow and incomplete under the conditions of analysis. The present results indicate that the rate of change of streaming potential is proportional to the logarithm of exposure time. The related changes in adsorbed amounts of polyelectrolyte were below the detection limits of typical polyelectrolyte titration procedures. Contrasting charge behaviors were observed on the exterior vs. interior surfaces of silica gel particles as a function of pore size, electrical conductivity, and polyelectrolyte molecular mass. Increasing ionic strength tended to enhance the effect of adsorption of high-mass cationic polymers on the outer surfaces, but produced only a relatively small effect on streaming potential related to their permeation into silica gel (nominal pore sizes of 6 nm or 30 nm). Adsorption of very-low-mass cationic polymer onto the outer surfaces and inside the 6 nm pore size silica gel appeared to be maximized at an intermediate salt level. Finally, electrokinetic tests were used for the first time in a protocol designed to provide evidence of polyelectrolyte desorption from the interiors of mesoporous materials Elsevier B.V. All rights reserved. 1. Introduction The interactions of cationic polyelectrolytes with porous materials have current and potential applications in a wide range of fields, including paper manufacture, pollution abatement, textiles, and the preparation of nanomaterials. Many unit operations in these applications depend on the ability and rate for the polyelectrolytes to diffuse from solution into the sub-micron interior spaces of the adsorbate material. For instance, in papermaking applications it is often preferred that polyelectrolytes remain on the outer surfaces of cellulosic fibers, where they can participate in such useful functions as retention of fine particles during the formation of a sheet. Though those polyelectrolyte molecules that are able to permeate into the cell wall of the fibers, before the fiber web is consolidated, have little influence on retention and inter-fiber bonding, other effects are expected, such as reinforcement of fiber wall strength and dimensional stability under different relative humidities. Part 1 of this series of articles considered electrokinetic tests to demonstrate how the adsorption of linear, highly cationic Corresponding author. Tel.: address: hubbe@ncsu.edu (M.A. Hubbe). poly-(diallyldimethylammonium chloride) (poly-dadmac) was affected by molecular mass, electrical conductivity, and ph [1]. By measuring the streaming potential of silica gel particles held within a packed bed and by comparing the results in the presence or absence of salt it has been possible to gain information concerning the location (outside or inside the porous network) of the adsorbed cationic polymers [2 4]. The present set of experiments was carried out to investigate the effects of time and pore size, together with molecular mass of the cationic polymer used in pretreatment, and the concentration of background electrolyte, on the streaming potential of silica gel particles. As summarized in a recent review article [5], it is reasonable to expect that considerable time is required for a polyelectrolyte to reach adsorption equilibrium in porous material, especially if the molecular mass is relatively high and the adsorbate material is mesoporous, i.e., pore size of ca m [6]. This pore size range generally coincides with the range associated with the characteristic dimensions of cell walls of cellulosic fibers after various preparation procedures [3 4,7 12]. In various papermaking applications it can be important for polymeric additives to remain on the outer surfaces of cellulosic fibers for long enough so that they can efficiently affect flocculation behaviors, as well as for junctions of hydrogen-bonds to be formed between adjacent fibers [13 15].On /$ see front matter 2010 Elsevier B.V. All rights reserved. doi: /j.colsurfa

2 8 M.A. Hubbe et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 364 (2010) 7 15 the other hand, some applications, such as cell-wall filling, demand effective penetration of nanomaterials in the interior of porous systems [16,17]. Key questions to be considered in this article are: (1) Can streaming potential be used as a highly sensitive method to evaluate rates of permeation of a cationic linear polyelectrolyte into the interior spaces of a negatively charged mesoporous material from aqueous solution? (2) What is the mathematic form of the time dependencies of adsorption and desorption of the cationic polymer? and (3) How do changes in pore size, in combination with ionic strength and polymer molecular mass, affect the streaming potential as a consequence of permeation and adsorption of the cationic polymers on the silica gel? Based on transport mechanisms, including diffusion, it is expected that the changes in streaming potential associated with adsorption or desorption of polyelectrolyte from a mesoporous material should be related to the extent of adsorption. However, to our knowledge, no model is available to permit calculation of streaming potential for an arbitrary distribution of polyelectrolyte adsorbed within a mesopore structure. Therefore, as a first approximation, a direct relation between the change in streaming potential and the corresponding adsorbed amounts will be assumed within suitably narrow intervals of potential. Under such an assumption, different mathematical relations can be drawn, depending on whether the transport is governed by simple diffusion or other mechanisms. For example, an expected dependency of adsorbed amount on the square-root time of exposure of a solution to a porous material can follow from Fick s second law, which can be expressed in integrated from as [18,19], C(x, t) = S[(Dt) 1/2 ]exp [ ] x 2 4Dt where S is the amount of diffusing substance added at the initial time, D is the diffusion rate constant, t is the elapsed time, and x is the distance of the sampling point from the surface. As an approximation, Eq. (1) can be linearized as follows: ] x C(x, t) C o [1 (Dt) 1/2 where C o is the initial concentration at the sampling location. Thus, a diffusion-controlled mechanism implies that the adsorbed amount is approximately proportional the square-root of elapsed time following a change in bulk solution concentration. Alternatively, from diffusion experiments with fluorescently tagged dextran molecules in porous silica, similar to those used in the present work [20], a logarithmic dependency of adsorbed amount within a mesoporous substrate with elapsed time was found. Specifically, the time-dependent concentration at a specified location within the porous material was described by an equation (1) (2) of the form: C(x, t) = C o 1 B ln [ t t o ] where B is a constant and t o is the starting time of observation. Landsberg [21] attributed such a relationship to a time-dependent chemical interaction with the substrate, in addition to diffusion. Thus, if this mechanism governs the interaction of cationic polyelectrolytes adsorbing from aqueous solution within an acidic, mesoporous material, the attraction between the polyelectrolyte and the substrate would be expected to yield a dependency of the adsorbed amount as that shown in Eq. (3). Interestingly, a similar dependency of dissolved polymer transport distance on the logarithm of exposure time was found in simulation work by Wolterink et al. [22], who considered the reptation of non-interacting polymers in solution through a membrane. By performing direct measurements of changes in streaming potential, our experimental work was aimed to elucidate the effects of pore size and polymer molecular mass on the dynamics and adsorption/desorption behaviors in suspensions of mesoporous silica gel particles. 2. Experimental 2.1. Materials Some materials used in this work were described in Part 1 of this series of articles [1]. Water used in the experiments was deionized with a Pureflow system. The default buffer solution was 1000 S/cm conductivity sodium sulfate solution that also contained 10 4 M NaHCO 3. An alternative salt-free buffer solution had the same NaHCO 3 concentration, but no Na 2 SO 4. Inorganic chemicals were of reagent grade. The cationic polyelectrolytes were linear poly-(diallyldimethylammonium chloride) (poly-dadmac) from Aldrich, catalogue numbers 52,237-6 (very-low-mass) and 40,903-0 (high-mass). The range of nominal molecular masses of the polyelectrolytes are given as 5 20 kda and kda, respectively. Polyvinylsulfate potassium salt (PVSK) used in polymer titrations was from Nalco Chemical Co., having product code 460-S5434. Three types of mesoporous silica gels, having different nominal pore sizes, were used in the experiments. The default substrate was catalogue S745-1 from Fisher Scientific, also known as Davisil Silica Gel 150, labeled as S 15 in this article. It was reported to have a nominal pore size of 15 nm and a mesh size range of Tests also were carried out with silica gel having a smaller nominal pore size of 6 nm (Fisher S735-1), also known as Davisil Silica Gel 60, labeled as S 15 and a larger nominal pore size of 30 nm (Fisher S814-1), also known as Davisil Silica Gel 300, labeled as S 30 (see Table 1). (3) Table 1 Characteristics of silica gel particles. Silica type Pore size (nm) Pore volume (cm 3 /g) Particle size ( m and mesh sizes) Nominal Surface area (m 2 /g) Experimental surface area (m 2 /g) * S 6 Davisil, Grade mesh m S 15 Davisil, Grade mesh m S 30 Davisil Sorbent, Mesh Grade 653 XWP 50 m * BET surface area was determined by 3-point analysis in a HORIBA SA-9601-MP surface area analyzer. The samples were dried at 150 C under nitrogen atmosphere for 2 h before surface area was tested.

3 M.A. Hubbe et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 364 (2010) Table 2 Levels of independent variables and default conditions. Factors Level Poly-DADMAC dosage, % on dry 0.1, 1, 3, 15, 30 mass of silica gel Silica gel pore size, nm 6, 15, 30 Time of equilibration of silica gel 0.5, 1, 2, 4, 12, 24 polyelectrolyte solution, h Electrical conductivity, S/cm 10, 100, 1000, 10,000 Values in bold indicate the default condition Streaming potential tests to observe effects of adsorption Modified streaming potential procedures were used to observe certain electrokinetic effects as described in Part 1 [1]. In that research we obtained preliminary results for the effect of poly- DADMAC concentration, molecular mass of poly-dadmac, ph, and conductivity. In the present work we designed our experimental conditions to investigate factors affecting polyelectrolyte penetration into mesoporous silica gel particles with a focus on the effects of pore size and time, as shown in Table 2. When considering each of the factors, default variable levels (marked in bold font) were used. The following procedure was used to compare the effects of different conditions on the degree to which adsorbed poly-dadmac shifted the streaming potential: Silica gel having a specified mean pore size was added to one liter of the specified polyelectrolyte or salt-free buffer solution and equilibrated during the indicated time. Poly-DADMAC treatment was based on the dry mass percentage relative to silica gel. The default equilibration time of silica gel particles in poly-dadmac solutions was approximately 24 h. Batches of six mixtures were stirred simultaneously during each equilibration period, using a shear rate large enough to keep the solid particles suspended (approximately 100 rpm, see details in Part 1 [1]). Then the solids were collected as a packed bed in the streaming potential device. When considering the time variable, the silica gel held in the packed bed was rinsed once by flowing fresh default buffer solution. When considering conductivity variable, the silica gel trapped in the packed bed was rinsed by flowing buffer solution corresponding to the conductivity of testing, rather than the equilibrium condition. The streaming potential was measured by continuously recording the potential between a pair of electrodes placed adjacent to each of two screens used to hold in place the packed bed Streaming potential tests to observe effects of desorption Modified streaming potential analyses were also used to monitor effects related to desorption of high-mass poly-dadamac. The following procedure was used to elucidate the effect of time on the degree to which apparent desorption of poly-dadmac shifted the streaming potential: The selected dry silica gel (Fisher S735-1) was added to poly-dadmac aqueous solution (15% on dry mass of silica gel). The mixture was shaken within a given time period (between 1 h and 24 h) in a thermostated shaker. Then the solids were collected as a packed bed for experiments with the streaming potential jar and rinsed once by flowing fresh default buffer solution. The procedure used for evaluation of streaming potentials was the same as that described in Part 1 of this series [1]. Briefly stated, streaming potential measurements require the evaluation of the difference in electrical potential between suitable electrode probes adjacent to the screens at either end of a packed bed or mat of a porous material, through which liquid is flowing at a determined pressure, which was 207 kpa in the present work. The method used in the present work included a second measurement at zero applied pressure as a means of calibration [23]. Each measurement cycle was completed by applying a slight negative pressure to allow ample time for gently returning the filtrate to the reservoir of the streaming potential device. A complete cycle, involving pressure application (8 16 s), then zero applied pressure (20 s), then gentle vacuum (150 s) required about 180 s. The streaming potential tests were started promptly after preparing the packed bed, using polyelectrolyte-free buffer solution. It should be noted that application of pressure during a streaming potential test tends to induce flow both around and through mesoporous particles. Because the channels associated with flow around the particles are much larger, it follows that most of the volumetric flow will be associated with such channeling. However, electrokinetic effects depend not on volumetric flow, but on the gradient of velocity very near to a surface, which is proportional to the applied pressure Desorption evaluated by streaming current A polyelectrolyte titration method was used in order to evaluate whether poly-dadmac desorbed from silica gel within a given time period (between ca. 1 min and 48 h) following replacement of the polyelectrolyte solution with polyelectrolyte-free buffer solution. As before, each experiment started with equilibration of the silica gel with a poly-dadmac solution. Two procedures were used. In the first procedure 30 ml poly-dadmac solution of 1% concentration base on silica solids were prepared in 40 ml vials. To this mixture 0.3 g of dry silica gel was added. The mixtures, confined within screw-capped cylindrical glass vial (40 ml volume), were kept in suspension with a holder rotating the vials end over end at 18 rpm. After approximately 24 h of equilibration, the following steps were applied in sequence: (a) the silica gel particles were allowed to settle for 60 s, (b) the vial openings were covered with a 500 mesh screen, inverting the vial, and applying vacuum to the aqueous solution, so that the silica gel was retained on the 500- mesh screen and kept within the vial, (c) the vials were carefully filled with 25 ml default buffer by pipette, (d) the mixture was rotated for a selected period of time: 60 s, 4 h, 24 h, and 48 h, (e) the silica gel particles were allowed to settle, and (f) the concentration of cationic polyelectrolytes in the supernatant solution was determined by titration with PVSK. An aliquot of 5 ml of supernatant solution was titrated with N PVSK. A second set of tests was carried out with a larger batch size and a higher ratio of silica gel to poly-dadmac solution. This set of tests employed 120 ml aliquots of 1% poly-dadmac solution, to which 30 g of dry silica gel was added. The mixtures, confined within screw-capped glass bottle (120 ml volume), were kept in suspension with a holder rotating at 18 rpm. (a) After 24 h, the silica gel particles were allowed to settle for 60 s, (b) the bottle openings were covered with a 500 mesh screen, inverting the bottle, and applying vacuum to the aqueous solution, so that the silica gel was retained on the 500 mesh screen and kept within the bottle, (c) the bottle was carefully filled with 80 ml default buffer by syringe, (d) the mixture was rotated for a selected period of time, and (e) samples of cationic polyelectrolytes in the supernatant solution were periodically taken out from the bottle at 60 s, 15 min, 4 h, 24 h, and 48 h, (f) 0.45 m nylon filters were used to further exclude any silica gel fragments from the supernatant poly-dadmac solutions, and (g) the concentration of cationic polyelectrolytes in the supernatant solution was determined by titration with PVSK. An aliquot of 5 ml of supernatant solution was titrated with 10 to 1000 N PVSK. The endpoint of the polyelectrolyte titration was determined by means of a Mütek PDC-03pH streaming current detector. It is worth noting that the initial poly-dadmac concentration of samples was always high enough to provide a positive signal of the streaming current detector at the start of the titration. Thus, the amounts of poly-dadmac adsorbed onto and into silica gel later desorb from

4 10 M.A. Hubbe et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 364 (2010) 7 15 Fig. 1. Effect of equilibration time on the streaming potential of silica gel particles measured in fresh default buffer. silica gel under specified conditions were quantified by means of polyelectrolyte titrations [24,25]. 3. Results and discussion 3.1. Equilibration time and polymer permeation As shown in Fig. 1, equilibration of the default silica gel sample (pore size 15 nm) with a dilute solution of high-mass poly-dadmac produced a significant change in the streaming potential. Since both the equilibration and the SP measurement were done at an electrical conductivity of 1000 S/cm (default buffer), the observed change in streaming potential must be attributed to increasing adsorption of the cationic polymer within the fine pores of the silica gel with increasing time of equilibration. As shown in previous work [2 4], when streaming potential tests of this class of silica gel are carried out in solution having an electrical conductivity of 1000 m, changes in streaming potential resulting from exposure to polyelectrolyte solution mainly indicate changes in the charged nature of the interior surfaces. The fact that even after about 4 days the signal was still negative implies that coverage of the internal surface area was far from complete. Had all of the surfaces become covered with an adsorbed layer of polyelectrolytes, a positive value of streaming potential would have been observed. The logarithmic scale used for the time variable in Fig. 1 helps to emphasize the fact that adsorption still continued to some extent even after 24 h of equilibration. The rather slow approach to equilibrium is consistent with predictions of slow diffusion through confined zones in porous media [5,26 28]. Because each streaming potential measurement involves a comparative test at zero applied pressure, instrumental drift can be ruled out as an explanation for the observed changes in streaming potential vs. time in the presence of polymer-free solution Desorption experiments Unlike the case considered in Fig. 1, streaming potential tests shown in Fig. 2A were carried out with fresh buffer solution, making it possible to probe polyelectrolyte desorption. As such, changes in the resulting streaming potential indicated in Fig. 2A suggest that poly-dadmac desorbs from the internal surfaces of the silica gel following suspension in polyelectrolyte-free solution. It is also possible that some macromolecules diffuse deeper into the mesoporous structure, though based on related reports [2 4] it can be argued that the relative contribution of such movement to the observed streaming potential is small. Fig. 2A also shows results from a series of streaming potential measurements carried out with default buffer solution after the silica gel had been equili- Fig. 2. (A) Fall-off of streaming potential with time in fresh default buffer during streaming potential analyses. Note that the streaming potential increased with increasing time of equilibration time with poly-dadmac, before the solids were exposed to polymer-free solution. (B) Streaming potential as a function of equilibration time. The streaming potential value was obtained by extrapolating the data in this figure to time equal to zero, using the regression results. brated for different lengths of time with high-mass poly-dadmac solution (15% treatment level based on SiO 2 solids). The horizontal axis corresponds to the cumulative time during which a continuous series of streaming potential measurements was conducted with polyelectrolyte-free buffer solution. The negative slopes of the linear regression lines through the data were interpreted as evidence of progressive desorption of the cationic polymer from the silica surfaces. Linear regression was used for simplicity and to avoid excessive emphasis on data collected during the earliest cycles of testing (the potential baseline is typically more stable during later cycles). By extrapolating the results in Fig. 2A to zero seconds, one can obtain an estimate of the streaming potentials at the moment of replacement of the supernatant solution with fresh buffer. The extrapolated potentials are presented in Fig. 2B as a function of (the logarithm of) time following the replacement of the supernatant solution with buffer solution. As shown, a linear regression of streaming potential vs. the log of time according to Eq. (4) was found suitable: SP = 1.225log ( te s ) (4) where SP is the streaming potential measured with a pressure difference of 207 kpa (see Section 2), and t e is the equilibration time, during which the silica gel was exposed to the default solution of high-mass poly-dadmac before the testing. This semi-logarithmic

5 M.A. Hubbe et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 364 (2010) relationship implies that the rate of change of streaming potential decreased with the passage of time. Also, the relation in Eq. (4) indicates a departure from a simply diffusion-limited transport process. As was noted in Section 1, if one assumes that relatively small changes in streaming potential are linearly related to changes in adsorbed amount, then it follows that the observed dependency of streaming potential on the logarithm of time may be attributable to either (a) a rate that is at least partly governed by the activation energy for release of the polyelectrolyte from the surfaces [20,21], or (b) a rate that is governed by a process such as reptation of polyelectrolytes through passages that are small enough to restrict the conformational changes [22]. Cule and Hwa [27] showed that heterogeneities in the pore structure can be expected to further extend the time required for a polyelectrolyte to reptate through mesoporous material. The observed logarithmic dependency on time also agrees with the confocal microscopic observations of Horvath et al. [29] who studied factors affecting permeation of a cationic polyelectrolyte into cellulose. Alternative explanations to account for the observed time dependency also were considered. For instance, time effects such as those shown in Fig. 2B might be attributed to reconformation of the polyelectrolyte molecules. Finite time is required for an adsorbing polyelectrolyte molecule to change its equilibrium conformation as it comes from the bulk of solution and interacts with a charged solid surface, and such time effects would not necessarily depend on the polyelectrolyte being able to enter into the pores. Recent work by Enarsson and Wågberg [30] considered the adsorption of several well-characterized cationic polyelectrolytes onto silica, using reflectometry and a quartz crystal microbalance. Even though the polyelectrolytes under consideration had substantially higher molecular mass than in the present study, it was found that equilibration of the adsorbed conformation on the non-porous substrate was achieved within a time far shorter than seconds. In related work Theodoly et al. [31] found that streaming potential signals reached equilibrium values in less than 60 s when cationic polyelectrolytes were adsorbing onto a non-porous substrate. More sluggish changes in the adsorbed conformation of a cationic polyelectrolyte were found by Notely et al. [32], but that study involved weak polyelectrolytes, and the authors explained their results by a mechanism that was specific to such situations. Relatively slow approach to equilibrium effects due to polyelectrolyte adsorption, in agreement with the present results, have been reported in the case of porous substrates. For instance, Hostetler and Swanson [33] reported that 36 h was needed to achieve equilibrium adsorption of polyethyleneimine (PEI) onto silica gel, over a range of molecular mass and pore size. Petlicki and van de Ven carried out related tests with pulp fibers, which have nano-size porosity, and observed continuing changes in adsorbed amounts when the time of exposure to PEI was extended beyond a minute. Wågberg et al. obtained similar results when evaluating the kinetics of adsorption of acrylamidetype polymers on cellulosic fibers [34] Desorption for different concentrations and molecular masses Fig. 3 compares the decay of streaming potential following different levels of poly-dadmac treatment. In the case of verylow-mass poly-dadmac, Fig. 3A shows that at the highest level of poly-dadmac treatment (3% based on silica) there was a steady, positive value of streaming potential. At the lowest level of treatment there was evidence of desorption, since the streaming potential was reduced from 2mVtoca. 7 mv with an increase of rinsing time or number of cycles. For reference, a streaming potential of 13 mv was observed for the default silica gel with the same buffer solution and no polymer treatment. A tentative explanation for the results in Fig. 3A follows from preceding studies in which it was proposed [2] and demonstrated Fig. 3. (A) Fall-off of streaming potential in fresh default buffer during streaming potential analyses after equilibration of the silica gel with different dosages of verylow-mass poly-dadmac (based on SiO 2 dry mass for about 20 h). (B) Fall-off of streaming potential with time when in the fresh default buffer during streaming potential analyses after equilibration of the silica gel with different dosages of highmass poly-dadmac (based on SiO 2 dry mass) for about 20 h. [1] that the positive values of streaming potential observed under similar conditions were due to very-low-mass oligomers present in the poly-dadmac samples. It was demonstrated that very-lowmass cationic oligomers were able to permeate into the silica gel to an extent that was sufficient to reverse the streaming potential to a positive value, even in the presence of background electrolyte (1000 S/cm conductivity). The results in Fig. 3A imply that after such permeation, the rate of desorption of cationic oligomers from the pore structure must have been slow relative to the time of observation. The fact that a significant decrease in streaming potential with time was observed in the case where the silica gel had been exposed to the lowest concentration of cationic polymer is consistent with relatively little permeation into the pores of the silica gel. Due to a lower concentration used during the equilibration step, there was less oligomeric material capable of permeating to a significant extent into the 15-nm pores of the silica gel. Likewise, the greater stability of streaming potential in the cases where the equilibration step involved a higher concentration is attributed to greater permeation during the equilibration period so that the surfaces of pores still remained substantially coated with polyelectrolyte throughout the observed period allowed for desorption. A non-linear effect of time of exposure to fresh buffer solution is shown at low polymer concentrations (lower curves in Fig. 3A). The time trends are similar to those obtained by Tanaka and Ödberg when an initial population of labeled cationic polymers was displaced from cellulosic fibers by a second addition of unlabeled cationic polymers [35].

6 12 M.A. Hubbe et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 364 (2010) 7 15 A unique effect was observed during the collection of the data in Fig. 3A; the streaming potential returned to slightly more positive or less negative values each time that the system was allowed to rest for extended periods between rinsing cycles. These shifts can be seen in Fig. 3A corresponding to cycles number 5, 3, and 26 for exposure levels of 3%, 1%, and 0.1%, respectively. The cause of this effect is not known. It is possible that flow causes the adsorbed polyelectrolytes to adopt a less extended conformation, and a rest period between applications of pressure or vacuum allows the adsorbed polymer to revert to a more extended configuration. Another possibility is that some time between cycles may be required in order for the packing density of the bed with adsorbed polyelectrolyte to relax towards its unperturbed state (the same effect is evident in Fig. 3B for the 10% level of exposure with high-mass poly-dadmac, see cycle 12). Fig. 3B shows corresponding results for high-mass poly- DADMAC. In this case there was clear evidence of desorption of poly-dadmac with rinsing, as judged from the increasingly negative streaming potential for all levels of polymer concentration after equilibration (adsorption) with the silica gel. Based on the higher initial streaming potential value at corresponding poly-dadmac concentration in Fig. 3B compared to Fig. 3A, it can be expected that large molecules are less able to reptate into the entrances of the internal pores of the silica gel during the equilibration period, and that significant time is required in order to reverse this process, when the supporting medium is replaced by fresh default buffer [36]. In addition, experiments with gel permeation chromatography indicated a lesser proportion of very-low-mass oligomers in the high-mass sample of poly-dadmac (data not shown). Two mechanisms are likely to account for the general shape of the curves in Fig. 3B. First, it is likely that some of the greater persistence of relatively high values of streaming potential following the highest level of treatment was due to the same effects mentioned in the context of Fig. 3A. The lesser proportion of verylow-mass oligomers in the high-mass polymer sample explains why no streaming potential values above zero were observed (i.e. much less material having the capability of permeating into the small pores was available). The observed decay of streaming potential values with time is consistent with desorption of polyelectrolyte molecules from the outer surfaces of the silica gel. The reason to expect significant desorption even in the case of a veryhigh-mass cationic polyelectrolyte is due to the expected crowded condition of the adsorbed layer. Because the amount of poly- DADMAC used in the experiments was about 100 times greater than the adsorption capacity of the silica gel samples [2], strongly adsorbing polyelectrolyte will compete for adsorption sites. Upon resuspension of the silica gel in polyelectrolyte-free medium, the adsorbed macromolecules can be expected to gradually adopt flatter adsorbed conformation, a process that tends to evict some of the macromolecules from the surface [37,38]. Past studies have suggested that over time the higher-mass macromolecules tend to out-compete lower-mass macromolecules, and the expected result is (a) a lower net adsorbed amount and (b) less extension of loops and tails of polyelectrolyte outward from the surface. Both of these effects can be expected to account for a drift in streaming potential signals to more negative values, as shown in Fig. 5 to be discussed later Streaming current titrations and desorbed amounts Fig. 4 shows results of tests that were designed to detect possible desorption of very-low-mass poly-dadmac from silica gel particles, following their treatment, removal of most of the suspending medium, and then their prompt resuspension in polyelectrolytefree buffer solution. As shown, there was no significant change over time in the amount of poly-dadmac in the solution. Thus, stream- Fig. 4. Desorption amount of very-low-mass poly-dadmac from SiO 2 with 15 nm pore size detected by streaming current titration of the supernatant solutions. Diamonds represent results of tests using the first protocol described in Section 2, whereas circles correspond to the second protocol. ing current (SC) titration was not able to detect the trends that were observed by the streaming potential (SP) tests. The regression results showed, with 95% confidence, that the desorbed amount of poly-dadmac was no higher than 0.31 mg/g of SiO 2 for the tests in the small vials and was no higher than 0.01 mg/g of SiO 2 for the follow-up set of tests carried out with the larger volume batches. Based on these results, it is possible that some net desorption occurred upon resuspension in polyelectrolyte-free buffer, but the amount was below the detection limit of the streaming current titrations. The fact that changes were observed in streaming potential, even though it was not possible to detect changes in the net amount of adsorbed polyelectrolyte is consistent with the proposal that the streaming potential measurements were mainly responding to changes in the amounts of oligomers that had permeated into mesopore spaces [2]. Because of the high ratio of high-mass to verylow-mass moieties in the solution, relative changes in the amount of oligomer in the bulk solution would have relatively little effect on the concentration of titratable polyelectrolyte in the bulk of solution. Another possible way to explain the difference in findings of the streaming potential tests and the adsorption tests is based on differences in the flow regime during the performance of the two kinds of experiment. In the SC tests, mixing of silica gel and poly-dadmac was achieved by rotating the sample containers, and the concentrations of the residual poly-dadmac in solution were determined in the absence of solids. By contrast, in the SP tests, part of the flow induced by pressure went though silica gel particles, giving rise to the streaming potential signals. It is proposed that pressure-induced flow might be the reason for poly-dadmac to have desorbed from the silica gel particles Effects of pore size, molecular mass, and salt content Fig. 5 shows that whereas the high-mass poly-dadmac had a greater effect on streaming potential when the measurements were in the absence of salt (Fig. 5A), the very-low-mass poly-dadmac had a greater effect when salt was present during the streaming potential measurements. These tests were conducted with silica gel having a characteristic pore size twice as large a those reported so far in this work (including in Part 1 of this series [1]). It is reasonable to expect that larger pores (30 nm, compared to a default value of 15 nm) would permit more rapid and complete permeation, though the results also could be expected to depend on

7 M.A. Hubbe et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 364 (2010) Fig. 5. (A) Effect of electrical conductivity during equilibration with poly-dadmac of very-low- vs. high-mass on the streaming potential of the outer surfaces, later measured with fresh salt-free buffer solution on silica gel having 30 nm a pore size. (B) Effect of electrical conductivity during equilibration with poly-dadmac of verylow- vs. high-mass on the streaming potential (with the major contribution to effect coming from pre-surfaces) the later measured with fresh 1000 S/cm buffer solution on silica gel having 30 nm a pore size. molecular mass and other factors. The conductivity values plotted in Fig. 5 (and also Fig. 6) correspond to the conditions used during overnight equilibration of the cationic polyelectrolyte and the silica gel. The electrolyte conditions during the cyclic streaming potential runs are specified in the legends and captions. Following the strategy developed in previous work [2,3], tests were conducted under highly contrasting concentration of background electrolyte in order to distinguish between effects due to cationic polymer adsorbed either on the outer surface of the silica gel (mainly observed by streaming potential tests in the absence of electrolyte) vs. effects due to permeation of cationic polymer into the mesopores (mainly observed by streaming potential tests carried out in the presence of salt). As shown in Fig. 5A, streaming potential measurements carried out in the absence of salt (with just 10 4 M NaHCO 3 to buffer the ph) revealed evidence of increasing poly-dadmac adsorption on the outer surface of the silica gel particles with increasing background salt during equilibration. Increasing ionic strength resulted in a more compact conformation of polyelectrolytes, allowing more molecules to pack into a given area of surface. The trend was especially evident in the case of the high-mass cationic polymer. These results are consistent with an expectation of very highaffinity adsorption of the large cationic macromolecules; thus it is reasonable to expect suitably strong adsorption even at the very Fig. 6. (A) Results for salt-free system with smaller pores: effect of electrical conductivity during equilibration with poly-dadmac of very-low- vs. high-mass on the streaming potential of the outer surfaces, later measured with fresh salt-free buffer solution on silica gel having 6 nm a pore size. (B) Same system as the previous (6 nm pore size), but with streaming potential evaluation in the presence of 1000 S/cm buffer solution. (C) Same system as the previous (6 nm pore size), but with streaming potential evaluation in the presence of 8000 S/cm buffer solution. high conductivity level of 10,000 S/cm. High-affinity adsorption is attributable to the large change in free energy when a multiplicity of the polyelectrolyte s ionic groups replace counter-ions at a substrate surface during the adsorption event. In addition, increased salt content favors polyelectrolyte to adopt a more condensed, coiled conformation in solution, allowing more of it to adsorb on a given amount of accessible surface area. The situation

8 14 M.A. Hubbe et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 364 (2010) 7 15 is expected to be somewhat different in the case of the very-lowmass poly-dadmac, since the free energy change associated with adsorption of one such oligomer is much lower. As the salt concentration is increased, a balance between the effects of weakening of the adsorption affinity and increased packing ability is reached. This balance between a more condensed macromolecular conformation and decreased adsorption affinity with increasing salt may explain why the effect of very-low-mass poly-dadmac on streaming potential appeared to go through a maximum with respect to salt content of the solution during equilibration. As shown in Fig. 5B, streaming potential measurements carried out with polymer-free default buffer solution (with sodium sulfate added to raise the electrical conductivity to 1000 S/cm) revealed a strong dependency of the resulting streaming potential on molecular mass of the poly-dadmac used in the equilibration step. As described earlier [2 4], results of streaming potential tests of mesoporous material carried out in the presence of salt can be interpreted as mainly revealing the charged condition of internal surfaces i.e. the surfaces of the pores. Thus, the results in Fig. 5B indicate a substantial effect of salt on the permeation of the verylow-mass cationic polyelectrolyte. No such effect was observed in the case of the high-mass cationic polymer. This difference can be attributed to space hindrance in the permeation of a very large polymer into small pores even in cases where electrostatic effects have been weakened by the presence of salts. Further evidence in support of this explanation is the fact that Fig. 5B shows negative values of streaming potential, whereas Fig. 5A shows positive values for samples that were equilibrated under identical conditions. The difference is attributed to an inability of the high-mass polyelectrolyte to reach the interior surfaces of the mesoporous material, allowing those surfaces to remain negative in charge, a condition that is observable only when the streaming potential tests are carried out in the presence of salt [2 4]. Results shown in Fig. 6A C correspond to tests carried out with silica gel having nominal pore sizes of 6 nm, about half the size of the default silica gel (noting that some corresponding results for the default silica gel already were given in Part 1 [1]). Fig. 6A, which involves streaming potential tests in the absence of salt, shows some of the same trends as in Fig. 5A (for which the treatment and testing conditions were the same, but the pore size was 30 nm). Thus, the effect of the high-mass poly-dadmac on increasing the streaming potential became more evident at higher salt concentrations during the equilibration phase. By contrast, the effect of the very-low-mass poly-dadmac achieved a maximum when the equilibration took place at the intermediate electrical conductivity value of 1000 S/cm. The explanations for these two observations presumably are just the same as those that were given in the discussion of Fig. 5A, and the main findings are confirmed. Closer comparison of the results in Figs. 5A and 6A shows that in the case of the high-mass poly-dadmac approximately twice as high values of streaming potential were achieved throughout the range of electrical conductivities tested when the polyelectrolyte was adsorbing onto the substrate having the larger pore size. One possible explanation is the expected slow interaction of a highmass polyelectrolyte in the entrance of a suitably-sized mesopore network. If the pores are large enough, one can consider a mechanism in which the ends of a high-mass polymer can function like the roots of a plant; reptation of the polymer ends even a short distance into the mesopore structure can be expected to achieve harder-toreverse adsorption [36]. Though such a mechanism could account for the results shown in Fig. 5A (with the 30 nm pores), the 6 nm pores (Fig. 6A) can be expected to be too small for the high-mass poly-dadmac to have an appreciable tendency to interact with the pores. The experiments represented in Fig. 6B were essentially the same as those in Fig. 5B, except that there was a very large difference in pore size. Even though earlier work had shown that a conductivity of 1000 S/cm was enough to suppress the double layer effects in the mesopore structure, the same assumption would not necessarily be true in the case of smaller pores. Electrokinetic effects are suppressed in cases where the pores are too small to accommodate the thickness of the double layers. So the large contrast between the results in Figs. 5B and 6B maybedue to the fact that the 1000 S/cm conductivity level was not sufficiently high in order for the 6 nm mesopore surfaces to make a dominant contribution to the observed streaming potential. Thus, the positive values of streaming potential shown Fig. 6B can be tentatively attributed to adsorption of poly-dadmac on the external surfaces. To help evaluate this hypothesis, Table 3 compares the factors by which electrokinetic effects are suppressed [39], when comparing the three types of silica gel and the levels of salt concentration used in the present work. The Debye Hückel lengths 1 shown in Table 3 were calculated from the formula [40]: = {[ ] 1/2 4e 2 N A z 2 (1000εkT) i i} M (5) where e is the electron charge, N A is Avagadro s number, ε is the dielectric constant of water, k is Boltzman s constant, T is absolute temperature, z is the valence of the ion of type i, and M i is the molar concentration of the same ion. Based on the values shown in Table 3 it can be expected that a sodium sulfate level sufficient to achieve about 8000 S/cm conductivity is enough to allow about 80% of the electrokinetic effects originating within the mesopore structure to be expressed. Therefore, to confirm the explanation just given in the context of Fig. 6B, similar tests were carried out at a higher level of sodium sulfate addition, sufficient to give an electrical conductivity of 8000 S/cm. As shown in Fig. 6C, this high level of salt during the analyses yielded more negative values of streaming potential, thus supporting the hypothesis. The longer error bars in Fig. 6C are attributed to the decreasing resolution of streaming potential measurements at very high conductivity. However, it is notable that the trends shown in Fig. 6C generally match those in Fig. 6A, suggesting that permeation as well as adsorption of very-low-mass poly-dadmac went through a maximum with respect to conductivity of the equilibra- Table 3 Suppression factors giving relative magnitude of electrokinetic effects within pores, compared to outer exposed surfaces, for various electrolyte conditions and pore sizes. a. Conductivity ( S/cm) Na 2SO 4 conc. (mm) Debye Hückel length (nm) Suppression factor Pore size (nm) , a Based on sodium sulfate as main electrolyte; conditions match those used in this study.

9 M.A. Hubbe et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 364 (2010) tion solution, whereas permeation and adsorption of the high-mass poly-dadmac increased monotonically with increasing conductivity within the range considered. Due to the intriguing nature of the observations reported here, a logical next step to consider in the investigation would consist of evaluation of polyelectrolyte adsorption isotherms. Results of such work will be considered in Part 3 of this series. 4. Conclusions 1. Evidence of progressive, though limited permeation of highcharge linear cationic polymer was obtained by measuring streaming potential of polymer-treated silica gel particles in the presence of polymer-free buffer solution, following h of equilibration with a poly-dadmac solution. Changes in streaming potential observed in the presence of salt suggested that oligomers in the poly-dadmac solutions permeated into the silica gel during the equilibration period to a much greater extent than higher-mass macromolecules. 2. The change in streaming potential during adsorption and permeation was proportional to the logarithm of time of exposure to polyelectrolyte solution. 3. Evidence consistent with desorption of poly-dadmac from silica gel was monitored by recording the changes in streaming potential during series of automated cycles of pressure (high, ambient, and negative pressures). Desorption effects were especially apparent when high-mass cationic polymer was used during the equilibration part of the experiment. 4. The sensitivity of the streaming potential measurements, with respect to certain effects related to polyelectrolyte permeation and adsorption, was judged to be high relative to the sensitivity of adsorption tests based on streaming current determinations of polyelectrolyte concentrations in the solution phase. The amount of desorbed polymer was below the level of detection of streaming current titration measurements. 5. Tests with different nominal pore size samples of silica gel carried out with different levels of background electrolyte were consistent with only limited permeation of either the high-mass or the very-low-mass cationic polymer. Streaming potential tests carried out at very-low electrolyte levels gave evidence, however, that adsorption of the high-mass cationic polymer on the outer surfaces of the silica gel increased with increasing salt throughout the range of experimentation. Permeation of verylow-mass poly-dadmac into silica gel having a small pore size was favored by an intermediate level of electrical conductivity. Acknowledgments The authors wish to acknowledge the support of the graduate student Ning Wu by the Petroleum Research Fund, and the administration of the fund by the National Science Foundation (grant no AC5). References [1] N. Wu, M.A. Hubbe, O.J. Rojas, S. Park, Permeation of a cationic polyelectrolyte into meso-porous silica. Part 1. Factors affecting changes in streaming potential, Colloids Surf. A. [2] M.A. Hubbe, O.J. Rojas, S.Y. Lee, S. Park, Distinctive electrokinetic behavior of nanoporous silica particles treated with cationic polyelectrolyte, Colloids Surf. A 292 (2007) 271. [3] M.A. Hubbe, O.J. Rojas, L.A. Lucia, T.M. Jung, Consequences of the nanoporosity of cellulosic fibers on their streaming potential and their interactions with cationic polyelectrolytes, Cellulose 14 (2007) 655. [4] M.A. Hubbe, Sensing the electrokinetic potential of cellulosic fiber surfaces, BioResources 1 (2006) 116. [5] N. Wu, M.A. Hubbe, O.J. Rojas, S. Park, Permeation of polyelectrolytes and other solutes into the pore spaces of water-swollen cellulose: a review, BioResources 4 (2009) [6] J. Rouquerol, D. Avnir, C.W. Fairbridge, D.H. Everett, J.H. Haynes, N. Pernicone, J.D.F. Ramsay, K.S.W. Sing, K.K. Unger, Recommendations for the characterization of porous solids, Pure Appl. Chem. 66 (1994) [7] J.E. Stone, A.M. Scallan, A structural model of the cell wall of water-swollen wood pulp fibers based on their accessibility to macromolecules, Cellulose Chem. Technol. 2 (1968) 343. [8] T.-Q. Li, U. Henriksson, L. Ödberg, Determination of pore sizes in wood cellulose fibers by 2 H and 1 H NMR, Nordic Pulp Paper Res. J. 8 (1993) 326. [9] B. Alince, T.G.M. van de Ven, Porosity of swollen pulp fibers evaluated by polymer adsorption, in: C.F. Baker (Ed.), The Fundamentals of Papermaking Materials, Pira Int l., 1997, p [10] J. Berthold, L. Salmén, Effects of mechanical and chemical treatments on the pore-size distribution in wood pulps examined by inverse size-exclusion chromatography, J. Pulp Paper Sci. 23 (1997) J245. [11] B. Alince, Porosity of swollen pulp fibers revisited, Nordic Pulp Paper Res. J. 17 (2002) 71. [12] B. Andreasson, J. Forsström, L. Wågberg, The porous structure of pulp fibres with different yields and its influence on paper strength, Cellulose 10 (2003) 111. [13] J.L. Koethe, W.E. Scott, Polyelectrolyte interactions with papermaking fibers the mechanism of surface charge decay, Tappi J. 76 (6) (1993) 123. [14] E. Gruber, Großmann, W. Schempp, Interactions of synthetic cationic polymers with fibers and fillers; influence on adsorption, Wochenbl. Papierfabr. 124 (1) (1996) 4. [15] C.E. Farley, Factors influencing the rate of charge decay, Tappi J. 80 (10) (1997) [16] G.G. Allan, A.R. Negri, P. Ritzenthaler, The microporosity of pulp the properties of paper made from pulp fibers internally filled with calcium carbonate, Tappi J. 75 (3) (1992) 239. [17] J.H. Klungness, A. Ahmed, N. Ross-Sutherland, S. AbuBakr, Lightweight, highopacity paper by fiber loading: filler comparison, Nordic Pulp Paper Res. J. 15 (2000) 345. [18] W.F. Smith, Foundations of Materials Science and Engineering, 3rd edition, McGraw-Hill, [19] B. Koumanova, P. Peeva, S.J. Allen, Variation of intraparticle diffusion parameter during adsorption of p-chlorophenol onto activated carbon made from apricot stones, J. Chem. Technol. Biotechnol. 78 (2003) 582. [20] J.B.S. Ng, P. Kamali-Zare, M. Sörensen, H. Brismar, N. Hedin, L. Bergström, Intraparticle transport and release of dextran in silica spheres with cylindrical mesopores, Langmuir 26 (2010) 466. [21] P.T. Landsberg, On the logarithmic rate law in chemisorption and oxidation, J. Chem. Phys. 23 (1955) [22] J.K. Wolterink, G.T. Barkema, D. Panja, Passage times for unbiased polymer translocation through a narrow pore, Phys. Rev. Lett. 96 (2006) [23] F. Wang, M.A. Hubbe, Development and evaluation of an automated streaming potential measurement device, Colloids Surf. A 194 (2001) 221. [24] J. Chen, J.A. Heitmann, M.A. Hubbe, Dependency of polyelectrolyte complex stoichiometry on the order of addition. 1. Effect of salt concentration during streaming current titrations with strong poly-acid and poly-base, Colloids Surf. A 223 (2003) 215. [25] M.A. Hubbe, J. Chen, Charge-related measurements a reappraisal. Part 1: streaming current, Paper Technol. 45 (2004) 17. [26] I. Teraoka, K.H. Langley, F.E. Karasz, Reptation dynamics of semirigid polymer in porous media, Macromolecules 25 (1992) [27] D. Cule, T. Hwa, Polymer reptation in disordered media, Phys. Rev. Let. 80 (1998) [28] C. Wang, M.B. Luo, Effect of interchain interactions on the translocation of polymer chains through small holes, J. Appl. Polymer Sci. 103 (2007) [29] A.T. Horvath, A.E. Horvath, T. Lindström, L. Wågberg, Diffusion of cationic polyelectrolytes into cellulose fibers, Langmuir 24 (2008) [30] L.E. Enarsson, L. Wågberg, Langmuir 24 (2008) [31] O. Theodoly, L. Cascao-Pereira, V. Bergeron, C.L. Radke, A combined streamingpotential optical reflectometer for studying adsorption at the water/solid surface, Langmuir 21 (2005) [32] S.M. Notely, S. Biggs, V.J.J. Craig, L. Wågberg, Adsorbed layer structure of a weak polyelectrolyte studied by colloidal probe microscopy and QCM-D as a function of ph and ionic strength, Phys. Chem. Chem. Phys. 6 (2004) [33] R.E. Hostetler, J.W. Swanson, Diffusion into and adsorption of polyethyleneimine on porous silica gel, J. Polymer Sci. 12 (1974) 29. [34] L. Wågberg, L. Ödberg, T. Lindström, R. Aksberg, Kinetics of adsorption and ion-exchange reactions during adsorption of cationic polyelectrolytes onto cellulosic fibers, J. Colloid Interface Sci. 123 (1988) 287. [35] H. Tanaka, L. Ödberg, in: Fundamentals of Papermaking, Trans Ninth Fundam. Res. Symp. at Cambridge, C.F. Baker, V.W. Punton (Eds.), Mech. Eng. Publ. Ltd., London, 1989, p [36] Y.G. Mishael, P.L. Dubin, R. De Vries, A.B. Kayitmazer, Effect of pores size on adsorption of a polyelectrolyte to porous glass, Langmuir 23 (2007) [37] L. Ödberg, H. Tanaka, A. Swerin, Nordic Pulp Paper Res. J. 8 (1993) 6. [38] L. Wågberg, Polyelectrolyte adsorption onto cellulose fibres a review, Nordic Pulp Paper Res. J. 15 (2000) 586. [39] S. Alkafeef, R.J. Gochin, A.L. Smith, The effect of double layer overlap on measured streaming currents for toluene flowing through sandstone cores, Colloids Surf. A 195 (2001) 77. [40] P.C. Hiemenz, R. Rajagopalan, Principles of Colloid and Surface Science, 3rd edition, Dekker, New York, 1997.