Surfactant-Modulation of the Cationic-Polymer-Induced Aggregation of Anionic Particulate Dispersions

Commodity formulations contain many chemically distinct components and their mutual interactions define the beneficial characteristics of the formulation. Mixing oppositely charged polymers and surfactants invariably induces macroscopic phase separation, to a degree dependent on the prevailing polymer and surface charge densities, and the interaction can be modulated by added ionic surfactants. Here, it is shown that a general universality exists between the charge present on a series of cationic-modified cellulose polymers—the charge being controlled either by the degree of cationic modification of the polymer itself or through the subsequent level of anionic surfactant binding—and its capacity to remove anionic colloidal material from solution, be that silica particles or polystyrene-butadiene lattices. Particulate material not removed from solution bears no adsorbed polymer, i.e., the particle surface is bare. Addition of nonionic surfactant does not negate this universality, implying that the nonionic surfactant is largely a spectator molecule or structure (micelle) in these systems, and that the dominant force is an electrostatic one.


Introduction
The adsorption of polymers to particulate surfaces continues to attract much attention due to its diverse industrial relevance, and depending on the specific application, the role of the polymer may be to promote stabilization of the dispersion, to induce flocculation or otherwise destabilize the dispersion, or a combination of both, dependent on the polymer concentration [1][2][3][4][5]. The adsorption of neutral homopolymers and copolymers onto surfaces is driven by a complex and subtle interplay between molecular weight and concentration-dependent enthalpic and entropic factors in defining the amount of adsorbed polymer, and inter alia its conformation at the particle surface [6,7]. The presence of charge on the polymer leads to additional contributions to the thermodynamics of adsorption when exposed to both oppositely and like-charged surfaces, and usually, the electrostatic force dominates [8][9][10]. The structure of the resultant polyelectrolyte-particle complex is defined by the polyelectrolyte and surface charge densities, polymer molecular weight and architecture, salt concentration/ionic strength of the medium, nonelectrostatic polyelectrolyte-surface interactions and the quality of the solvent [11][12][13].
Flocculation-inducing adsorption of polyelectrolytes to particle surfaces has been studied in some detail (e.g., [14][15][16][17][18]), and generally, there are three dominant mechanisms dependent on the molecular weight and/or charge density of the polyelectrolyte: (i) neutralization of the charge on the particle surface leading to attractive van der Waals-driven aggregation; (ii) interparticle bridging by The cationic cellulosic polymers are commercial in nature, Scheme 1 (Dow Chemical Company), and will be referred to by their nominal degrees of modification, expressed in terms of the nitrogen content; N = 0.5% (±0.1)%, N = 0.95% (±0.15)% and N = 1.8% (±0.2)%. The parent polymer has also been included for comparison, and is labelled N = 0. The manufacturer's literature suggests these polymers have molecular weights in the vicinity of 500,000 g·mol −1 . Scheme 1. Generic structure of quaternary ammonium salts of hydroxyethyl cellulose polymers. Adapted from ref [31].
Order of mixing is known to be a significant factor in these systems, and therefore all samples were prepared by adding polymer stock solutions (or their dilutions) to the particulate dispersions, and where surfactant (SDS or C12E6) is additionally present, this was added to the polymer stock solution prior to adding to the silica. The importance of order of addition highlights the nonequilibrium effects often observed in these systems.
The solvent-relaxation measurements were determined on a Zigo Acorn Drop bench-top NMR spectrometer operating at 13.0 MHz. Samples were equilibrated at 25 °C. The instrument's in-built configuration macros and data-analysis routines were used to optimize and obtain the experimental relaxation rates. All results are the average of at least triplicate measurements and frequent duplicate samples.

Results and Discussion
Mixing oppositely charged polymers and particles leads to complex phase separation, characterized by two discrete phases, one phase being opaque, and therefore very rich in large(r) structures, the second showing various degrees of opalescence due to the presence of dispersed nanoparticles. Figure 1a-d presents photographs of a large matrix of a series of samples in which the concentration of silica has been increased from 0 to 10 wt % in the presence of fixed amounts of polymer (typically 1000 ppm) at selected values of SDS concentration (0 ≤ [SDS] ≤ 1 mM). Our previous electrophoretic NMR characterization of the binary polymer/SDS systems indicates that the polymer/SDS solution phases separate (due to charge neutrality) around typically 4-8 mM for a polymer concentration Cpolymer = 1 wt %, suggesting that over the concentration window studied here (0-1 mM SDS), the polymer retains some cationic character, except perhaps at the higher SDS values studied [32]. Figure 2 presents a subset of these systems additionally in the presence of the nonionic surfactant, hexaethylene glycol monododecyl ether (C12E6). It should be noted that the uncharged polymer showed no measurable interaction with SDS.
Traversing the row of samples from left to right, i.e., the only variable increasing being the silica concentration, the samples become increasingly opaque, with the volume of the opaque phase largely increasing. A similar pattern is observed in the uncharged and charged cases. Where SDS has been added, the volume of the opaque phase also depends on [SDS]. In some cases, the opaque phase is the supernatant, and in some cases the subnatant. As one moves vertically through the series, the Scheme 1. Generic structure of quaternary ammonium salts of hydroxyethyl cellulose polymers. Adapted from ref [31].
Order of mixing is known to be a significant factor in these systems, and therefore all samples were prepared by adding polymer stock solutions (or their dilutions) to the particulate dispersions, and where surfactant (SDS or C 12 E 6 ) is additionally present, this was added to the polymer stock solution prior to adding to the silica. The importance of order of addition highlights the nonequilibrium effects often observed in these systems.
The solvent-relaxation measurements were determined on a Zigo Acorn Drop bench-top NMR spectrometer operating at 13.0 MHz. Samples were equilibrated at 25 • C. The instrument's in-built configuration macros and data-analysis routines were used to optimize and obtain the experimental relaxation rates. All results are the average of at least triplicate measurements and frequent duplicate samples.

Results and Discussion
Mixing oppositely charged polymers and particles leads to complex phase separation, characterized by two discrete phases, one phase being opaque, and therefore very rich in large(r) structures, the second showing various degrees of opalescence due to the presence of dispersed nanoparticles. Figure 1a-d presents photographs of a large matrix of a series of samples in which the concentration of silica has been increased from 0 to 10 wt % in the presence of fixed amounts of polymer (typically 1000 ppm) at selected values of SDS concentration (0 ≤ [SDS] ≤ 1 mM). Our previous electrophoretic NMR characterization of the binary polymer/SDS systems indicates that the polymer/SDS solution phases separate (due to charge neutrality) around typically 4-8 mM for a polymer concentration C polymer = 1 wt %, suggesting that over the concentration window studied here (0-1 mM SDS), the polymer retains some cationic character, except perhaps at the higher SDS values studied [32]. Figure 2 presents a subset of these systems additionally in the presence of the nonionic surfactant, hexaethylene glycol monododecyl ether (C 12 E 6 ). It should be noted that the uncharged polymer showed no measurable interaction with SDS.
Traversing the row of samples from left to right, i.e., the only variable increasing being the silica concentration, the samples become increasingly opaque, with the volume of the opaque phase largely increasing. A similar pattern is observed in the uncharged and charged cases. Where SDS has been Polymers 2020, 12, 287 4 of 11 added, the volume of the opaque phase also depends on [SDS]. In some cases, the opaque phase is the supernatant, and in some cases the subnatant. As one moves vertically through the series, the same gross pattern is observed, but it is clear that the switch from super-to subnatant is displaced, and the phase volumes vary.
Further interpretation of these visual phase diagrams is challenging, as the composition of each phase is unknown, and that of the opaque phase continues to change with time. Given the density of the silica, it is slightly surprising that the opaque phase creams at all. If one removes the dense phase, and allows it to stand, it will further phase separate, leading to additional volume of clear phase and more dense phases. Therefore, to accelerate this phase separation, the samples were gently centrifuged.
In order to understand the composition of the separated phases, the less dense (termed the equilibrium phase) has been characterized by both solvent relaxation and by dry weight analysis. Making the assumption that the bulk of the weight of the dried sample recorded in this manner arises from the particles, we therefore define this as an equilibrium particle concentration, and thus provides a convenient parameter (x-axis) on which to base solvent relaxation data, and thereby to enable in parallel the comparison with bare silica dispersions over the same initial concentration range. For the silica comparators, no differences between initial and equilibrium concentrations were observed in the dry weight analysis. The solvent relaxation in these no-polymer systems can therefore be compared with the with-polymer data.
Polymers 2020, 12, x FOR PEER REVIEW 4 of 11 same gross pattern is observed, but it is clear that the switch from super-to subnatant is displaced, and the phase volumes vary. Further interpretation of these visual phase diagrams is challenging, as the composition of each phase is unknown, and that of the opaque phase continues to change with time. Given the density of the silica, it is slightly surprising that the opaque phase creams at all. If one removes the dense phase, and allows it to stand, it will further phase separate, leading to additional volume of clear phase and more dense phases. Therefore, to accelerate this phase separation, the samples were gently centrifuged.  In order to understand the composition of the separated phases, the less dense (termed the equilibrium phase) has been characterized by both solvent relaxation and by dry weight analysis. Making the assumption that the bulk of the weight of the dried sample recorded in this manner arises from the particles, we therefore define this as an equilibrium particle concentration, and thus provides a convenient parameter (x-axis) on which to base solvent relaxation data, and thereby to enable in parallel the comparison with bare silica dispersions over the same initial concentration range. For the silica comparators, no differences between initial and equilibrium concentrations were observed in the dry weight analysis. The solvent relaxation in these no-polymer systems can therefore be compared with the with-polymer data.
Solvent relaxation provides a convenient method to characterize any near-surface structures via the dynamics of the solvent molecules, in this case, water [33][34][35][36][37][38][39]. The relaxation rate is usually enhanced if a polymer layer is present, compared with the same surface area, due to the presence of a volume of trapped water. Figure 3 presents the solvent-specific relaxation rate, i.e., the rate normalized to pure water, plotted in terms of the equilibrium silica concentration. All samples studied here have been reproduced on this single figure. As may be seen, there is a universality to this curve that relates the measured specific solvent relaxation rate to the equilibrium silica concentration, this universality holding for particles that have been exposed to polymer, to surfactant and to mixtures of polymer and surfactant, as well as those that have not been exposed. concentration is 1000 ppm throughout, the silica concentration is increased from 0 to 10 wt %. The first row has SDS = 0 mM, the second 0.50 mM. In the third and fourth rows, the total surfactant concentration is 0.50 mM, but the composition of the surfactant mixtures is 75%:25% (anionic: nonionic). The last row has 0.5 mM C 12 E 6 . Photographs were taken approximately 2 days after preparation.
Solvent relaxation provides a convenient method to characterize any near-surface structures via the dynamics of the solvent molecules, in this case, water [33][34][35][36][37][38][39]. The relaxation rate is usually enhanced if a polymer layer is present, compared with the same surface area, due to the presence of a volume of trapped water. Figure 3 presents the solvent-specific relaxation rate, i.e., the rate normalized to pure water, plotted in terms of the equilibrium silica concentration. All samples studied here have been reproduced on this single figure. As may be seen, there is a universality to this curve that relates the measured specific solvent relaxation rate to the equilibrium silica concentration, this universality holding for particles that have been exposed to polymer, to surfactant and to mixtures of polymer and surfactant, as well as those that have not been exposed.
Making the reasonable assumption that any added polymer or surfactant that adsorbs to the surface would perturb the solvent-specific relaxation rate [33][34][35][36][37][38][39], this universality strongly suggests that in all cases the particles detected in the solvent relaxation experiment are uncoated. In other words, the opaque phase consists of a concentration of polymer and particle with a composition to saturate that interaction, and any excess polymer or particle is displaced to the less-dense phase. When there is excess polymer present, all of the silica is removed, and one observes a solvent-specific relaxation rate that is characteristic of a simple polymer solution which is identical to the pure solvent, R 2sp tends to 0. When there is excess silica present, the polymer removes a fraction of the silica, and the solvent-specific relaxation rate is consistent with a bare silica dispersion commensurate with the remaining equilibrium concentration.   Making the reasonable assumption that any added polymer or surfactant that adsorbs to the surface would perturb the solvent-specific relaxation rate [33][34][35][36][37][38][39], this universality strongly suggests that in all cases the particles detected in the solvent relaxation experiment are uncoated. In other words, the opaque phase consists of a concentration of polymer and particle with a composition to saturate that interaction, and any excess polymer or particle is displaced to the less-dense phase. When there is excess polymer present, all of the silica is removed, and one observes a solvent-specific relaxation rate that is characteristic of a simple polymer solution which is identical to the pure solvent, R2sp tends to 0. When there is excess silica present, the polymer removes a fraction of the silica, and the solvent-specific relaxation rate is consistent with a bare silica dispersion commensurate with the remaining equilibrium concentration.
In those samples that also contain SDS, it is clear that the same universality between mass fraction and surface area exists. In this case, the anionic surfactant binds to the cationic polymer, thereby reducing the effective charge on the polymer, such that for a given polymer concentration, the polymer-surfactant complex removes less silica from the dispersion compared to the polymer alone. Nonetheless, all the detected silica surface is bare. In other words, in terms of the photographs in Figures 1 and 2, addition of SDS moves the phase diagram to the left. Similarly, one can also estimate the concentration of SDS required to nullify the charge on the cationic polymer, and therefore "turning off" any adsorption to the anionic surface.
Further confirmation of this simple binary phase separation may be sought by considering the zeta-potentials of the equilibrium (phase separated) silica phases (Table 1), these values being consistent with the original bare silica dispersion. Table 1. Zeta-potential analysis of the equilibrium phases obtained from the centrifugation of the twophase silica/polymer/surfactant systems studied here. All measurements were conducted in triplicate.

Zeta Potential/mV
Stock, as-supplied Ludox, 1-in-10 dilution  In those samples that also contain SDS, it is clear that the same universality between mass fraction and surface area exists. In this case, the anionic surfactant binds to the cationic polymer, thereby reducing the effective charge on the polymer, such that for a given polymer concentration, the polymer-surfactant complex removes less silica from the dispersion compared to the polymer alone. Nonetheless, all the detected silica surface is bare. In other words, in terms of the photographs in Figures 1 and 2, addition of SDS moves the phase diagram to the left. Similarly, one can also estimate the concentration of SDS required to nullify the charge on the cationic polymer, and therefore "turning off" any adsorption to the anionic surface.
Further confirmation of this simple binary phase separation may be sought by considering the zeta-potentials of the equilibrium (phase separated) silica phases (Table 1), these values being consistent with the original bare silica dispersion. Table 1. Zeta-potential analysis of the equilibrium phases obtained from the centrifugation of the two-phase silica/polymer/surfactant systems studied here. All measurements were conducted in triplicate. As an aside, a similar observation may be drawn from analogous studies using polystyrene-butadiene latex rather than silica (Figure 4).

Analysis of the Equilibrium
Consider the impact of addition of a nonionic surfactant to the polymer/particle/anionic system. The two surfactants would be expected to interact strongly and synergistically, and it is known that the presence of the nonionic often leads to an enhancement in the counterion dissociation of the anionic surfactant, effectively increasing the charge on the mixed surfactant complex. Dialysed Ludox plus polymer N = 1.8% and SDS −32.6 (±2.8) As an aside, a similar observation may be drawn from analogous studies using polystyrenebutadiene latex rather than silica (Figure 4). Consider the impact of addition of a nonionic surfactant to the polymer/particle/anionic system. The two surfactants would be expected to interact strongly and synergistically, and it is known that the presence of the nonionic often leads to an enhancement in the counterion dissociation of the anionic surfactant, effectively increasing the charge on the mixed surfactant complex. Figure 5 presents the solvent-specific relaxation rate for a select set of experiments in which the cationic polymer has been first exposed to mixed anionic-nonionic (SDS-C12E6) surfactant blends, before being added to the silica dispersion. Phase separation still occurs, with the visual pattern not too dissimilar to that observed in the absence of the nonionic surfactant, and reminiscent of that observed at lower anionic surfactant concentrations. As may be seen, the same universality exists, between the dry weight analysis and the nature of the silica surface, implying the nonionic surfactant is merely a spectator in the polymer/SDS/particle interaction.  Figure 5 presents the solvent-specific relaxation rate for a select set of experiments in which the cationic polymer has been first exposed to mixed anionic-nonionic (SDS-C 12 E 6 ) surfactant blends, before being added to the silica dispersion. Phase separation still occurs, with the visual pattern not too dissimilar to that observed in the absence of the nonionic surfactant, and reminiscent of that observed at lower anionic surfactant concentrations. As may be seen, the same universality exists, between the dry weight analysis and the nature of the silica surface, implying the nonionic surfactant is merely a spectator in the polymer/SDS/particle interaction.
As an aside, a similar observation may be drawn from analogous studies using polystyrenebutadiene latex rather than silica (Figure 4). Consider the impact of addition of a nonionic surfactant to the polymer/particle/anionic system. The two surfactants would be expected to interact strongly and synergistically, and it is known that the presence of the nonionic often leads to an enhancement in the counterion dissociation of the anionic surfactant, effectively increasing the charge on the mixed surfactant complex. Figure 5 presents the solvent-specific relaxation rate for a select set of experiments in which the cationic polymer has been first exposed to mixed anionic-nonionic (SDS-C12E6) surfactant blends, before being added to the silica dispersion. Phase separation still occurs, with the visual pattern not too dissimilar to that observed in the absence of the nonionic surfactant, and reminiscent of that observed at lower anionic surfactant concentrations. As may be seen, the same universality exists, between the dry weight analysis and the nature of the silica surface, implying the nonionic surfactant is merely a spectator in the polymer/SDS/particle interaction.  Figure 5. Solvent-specific relaxation rate as a function of equilibrium particle concentration for polymer N = 0.95% at 1000 ppm in the presence of anionic-nonionic surfactant blends; 0 (simple SDS) red diamonds, 0.75 blue triangles and 1.0 (simple nonionic) purple triangles. Additionally shown for comparison is the silica only dataset (black circles) and the SDS = 0 dataset. The total surfactant concentration is 0.5 mM.
The uncharged (parent) polymer induces flocculation of the silica dispersion via a bridging mechanism, most likely through the hydroxyl groups. The grossly similar behavior of the three