Poly(ethylene oxide) (PEO) and Poly(vinyl pyrolidone) (PVP) Induce

Aug 4, 2010 - University of Delaware, Newark, Delaware 19716 ... Corporate Research Department, The Procter & Gamble Company, Cincinnati, Ohio 45252...
0 downloads 0 Views 3MB Size
pubs.acs.org/Langmuir © 2010 American Chemical Society

Poly(ethylene oxide) (PEO) and Poly(vinyl pyrolidone) (PVP) Induce Different Changes in the Colloid Stability of Nanoparticles Naa Larteokor McFarlane and Norman J. Wagner* Center for Molecular and Engineering Thermodynamics, Department of Chemical Engineering, University of Delaware, Newark, Delaware 19716

Eric W. Kaler Departments of Materials Science and Chemistry, Stony Brook University, Stony Brook, New York 11794

Matthew L. Lynch Corporate Research Department, The Procter & Gamble Company, Cincinnati, Ohio 45252 Received May 12, 2010. Revised Manuscript Received June 29, 2010 The phase behavior of model polymer-colloid mixtures is measured for solutions approaching the “protein limit”, that is, when the radius of gyration of the polymer (Rg) is greater than or approximately equal to the radius of the colloid (R). Cationic nanoparticles are mixed with poly(ethylene oxide) (PEO) or poly(vinyl pyrolidone) (PVP) at size ratios of Rg/R = 0.7 and 1.8. The addition of PEO to stable nanoparticle dispersions leads to depletion flocculation in both deionized water and buffer solutions. The instability mechanism for the PVP-nanoparticle system depends on the suspension medium. In water, bridging occurs below the saturation adsorption of PVP, whereas depletion phase separation is evident at concentrations exceeding those necessary to saturate the particle surface. In acidic buffer, PVP addition results in depletion phase separation. The difference between bridging and depletion is distinguished by both visual appearances and rheological measurements. There is no trend (within error bars) in the polymer concentration required to induce instability with increasing Rg/R in contrast with theoretical predictions. This is most likely due to adsorption of polymer onto the particle surface.

1. Introduction Polymers are often added to colloidal-based products to improve their tactile and rheological behavior; however, the addition of polymers can either enhance colloidal dispersion or induce flocculation. This behavior depends on whether the polymer adsorbs to the particle, the relative size of the polymer to the colloid, and the polymer concentration relative to the polymer entanglement concentration (c*).1,2 For nonadsorbing polymercolloid mixtures at low particle concentrations, there are four distinct physical regimes that depend on the relative characteristic sizes of the polymer (determined by its radius of gyration, Rg) to the colloid (of radius R) and the ratio of the concentration of polymer (c) to c*.3 In dilute solutions of nonadsorbing polymers and colloids, that is, in the “colloid” limit (R . Rg,), it is possible to predict the conditions for phase separation using existing theories.2-10 The phase behavior of polymer-colloid systems in *Corresponding author. E-mail: [email protected]. (1) Russel, W. B.; Saville, D. A.; Schowalter, W. R. Colloidal Dispersions; Cambridge University Press: New York, 1989. (2) Ramakrishnan, S.; Fuchs, M.; Schweizer, K. S.; Zukoski, C. F. J. Chem. Phys. 2002, 116, 2201. (3) Fuchs, S.; Schweizer, K. S. J. Phys. (Paris) 2002, 14, R239. (4) Asakura, S.; Oosawa, F. J. Polym. Sci. 1958, 33, 183. (5) Gast, A. P.; Hall, C. K.; Russel, W. B. J. Colloid Interface Sci. 1983, 96, 251. (6) Tehver, R.; Maritan, A.; Koplick, J.; Banavar, J. R. Phys. Rev. E 1999, 59, R1339. (7) Lekkerkerker, H. N. W.; Poon, W. C. K.; Pusey, P. N.; Stroobants, A.; Warren., P. B. Europhys. Lett. 1992, 20, 559. (8) Lekkerkerker, H. N. W.; Stroobants, A. Lect. Notes Phys. 1993, 415, 1. (9) Illett, S. M.; Orrock, A.; Poon, W. C. K.; Pusey, P. N. Phys. Rev. E 1995, 51, 1344. (10) Vrij, A. Pure Appl. Chem. 1976, 48, 471.

Langmuir 2010, 26(17), 13823–13830

the so-called “protein” or “nanoparticle” limit, where Rg . R and c > c*, is the subject of active research.2,11-13 For strongly adsorbing polymers at low polymer concentrations, destabilization due to bridging flocculation is probable.14 As the polymer concentration increases and the surfaces of the particles are fully covered with polymer, steric stabilization dominates.15 Further addition of polymer may induce depletion phase separation.16 The goal of this project is to study the experimental stability boundaries of nonionic polymer-colloid aqueous dispersions in both the nanoparticle and colloidal limit. This work presents a systematic study of the phase and stability behavior and mechanism(s) of phase separation at two asymmetry (Rg/R) ratios for two polymers with differing adsorption energies. In addition, the stability boundary is tuned with added electrolyte, which contrasts with previous studies that modified the particle surface by surfactant adsorption.17 To study nanoparticle-polymer mixture stability, the adsorption of two nonionic polymers, PEO and PVP, onto positively charged nanoparticles is reported here. PEO is highly soluble in water because of its ability to form weak hydrogen bonds with the lone pair of electrons on the oxygen in its backbone. PEO is widely (11) Chen, Y. L.; Schweizer, K. S.; Fuchs, M. J. Chem. Phys. 2003, 118, 3880. (12) Zhang, Z.; van Duijneveldt, J. S. Langmuir 2006, 22, 63. (13) Hennequin, Y.; Evens, M.; Quezada Angulo, C. M.; van Duijneveldt, J. S. J. Chem. Phys. 2005, 123, 1. (14) Wong, K.; Lixon, P.; Lafuma, F.; Lindner, P.; Aguerre Charriol, O.; Cabane, B. J. Colloid Interface Sci. 1992, 153, 55. (15) Lafuma, F.; Wong, K.; Cabane, B. J. Colloid Interface Sci. 1991, 143, 9. (16) Snowden, M. J.; Clegg, C. M.; Williams, P. A.; Robb, I. D. J. Chem. Soc., Faraday Trans. 1991, 87, 2201. (17) Otsubo, Y. Nihon Reoroji Gakkaishi 1995, 23, 75.

Published on Web 08/04/2010

DOI: 10.1021/la101907s

13823

Article

McFarlane et al.

used for fundamental studies of the interactions of nonadsorbing polymers with colloidal particles and proteins and the resulting effects on phase behavior15,18,19 and protein crystallization.20 PEO is also used commercially as an additive in personal care products because of its high solubility and good wetting properties. PVP is a hygroscopic polymer with excellent water solubility and biocompatibility.21 PVP is also inert to humans and has excellent wetting properties, which allows for wide use in the pharmaceutical and healthcare industries. Other uses are in personal care products, paints, adhesives, and as a food additive. Nonpharmaceutical research on PVP currently focuses on PVP adsorption on particles, cross-linking, polymerization of novel PVP-based polymers, and properties of PVP blends.22,23 PEO-induced phase separation in colloidal suspensions has been studied.15,18,19 The study of PVP-induced phase separation in colloidal suspensions is relatively unexplored. In all cases,24 observations are of the effect of polymer on the colloidal stability of a nanoparticle dispersion, which is a two-phase (solid and liquid) system. However, as is frequently the case for the description of nanoscale colloidal dispersions, we may use the terms “phase transition” or “phase boundary” as a short-hand to denote changes in the nature of the colloidal dispersion as a function of composition variables.

2. Theoretical Background The well-studied Asakura-Oosawa (A-O) theory provides an expression for the potential of mean force between a pair of hard spheres due to depletion forces.3-6 The A-O potential adequately describes the phase behavior of polymer-colloid mixtures at dilute polymer concentrations (c < c*) when polymerpolymer interactions are negligible and when the particles are larger than the polymer coils in solution Rg < R. However, the A-O theory is inapplicable when Rg > R because the particle is able to penetrate into the volume occupied by the polymer.6 Improvements have been made to include explicitly the polymer concentration in calculating the phase boundaries.9 The polymer reference interaction site model (PRISM)3,25 for athermal suspensions incorporates a more realistic description of the excluded volume of the polymer and predicts an decrease in the miscibility gap (i.e., greater miscibility and thus stability) with increasing Rg/R in the nanoparticle limit, which is opposite to the predictions of the classical theories. The essential differences have been discussed11,25 and include the need to properly account for polymer-polymer and polymer-colloid correlations as well as the loss of conformational entropy at close polymer-particle separations. These issues become especially important at large Rg/R and c > c*. In these limits, the particles diffuse through a mesh of polymer, and the relevant length scale for the depletion interaction becomes the polymer mesh size.2 Solutions to the PRISM model can be obtained for a wide range of asymmetries and polymer concentrations, but quantitative comparisons with the experimental data are limited because of limitations of the Percus-Yevick closure approximations as used in the theory.2 (18) Liu, S. F.; Legrand, V.; Gourmand, M.; Lafuma, F.; Audebert, R. Colloids Surf., A 1996, 111, 139. (19) Olsson, M.; Joabsson, F.; Piculell, L. Langmuir 2005, 21, 1560. (20) Kulkarni, A.; Zukoski, C. J. Cryst. Growth 2001, 232, 156. (21) Wu, Z.; Gong, S.; Li, C.; Zhang, Z.; Huang, W.; Meng, L.; Lu, X.; He, Y. Eur. Polym. J. 2005, 41, 1985. (22) Parnas, R.; Chaimberg, M; Taepaisitphongse, V.; Cohen, Y. J. Colloid Interface Sci. 1989, 129, 441. (23) Cohen Stuart, M. A.; Fleer, G. J.; Bijsterbosch, B. H. J. Colloid Interface Sci. 1982, 90, 310. (24) Sakellariou, P. Colloid Polym. Sci. 1995, 273, 279. (25) Schweizer, K. S.; Curro, J. G. Adv. Polym. Sci. 1994, 116, 321.

13824 DOI: 10.1021/la101907s

Recent studies of nonadsorbing polymers2,12,13 have tested the applicability of the AO and PRISM theories. The trends in the phase boundary observed for dispersions containing nonadsorbing mixtures of polystyrene and hydrophobically modified silica nanoparticles2 and polystyrene and stearyl-alcohol-coated silica particles in the protein limit13 were compared with theoretical predictions. In both of these cases, there was a decrease in the size of the two-phase region (increasing miscibility) with increasing Rg/R (when viewed on a plot of c/c* versus φc) in agreement with the PRISM theory.2 In other reported work involving poly(isoprene) and sterically stabilized silica nanoparticles,12 the measured phase boundary lies between that theoretically predicted for interacting and noninteracting systems. Here a qualitative comparison will be made between the trends in phase boundaries predicted by the theories2,5,7,8,11 and our experimental trends.

3. Experimental Section 3.1. Materials and Characterization Methods. Solutions were formulated from Klebosol 30cal25 suspensions of nonporous colloidal silica stabilized at pH ∼3.67 (AZ Electronics, Lamotte, France). These particles have a proprietary surface coating and are positively charged as supplied with a point of zero charge measured to be at pH 8.6.26 These were mixed with solutions of poly(ethylene oxide) (35 000 Da, Mw/Mn = 1.24 and 400 000 Da PEO, Mw/Mn = 1.29, Aldrich) and poly(vinyl pyrrolidone) (55 000 Da, Mw/Mn = 1.73 and 360 000 Da PVP, Mw/Mn = 1.02, Aldrich) in either deionized water (Barnstead NANOpure deionization system, 18 to 18.3 MΩ cm resistivity) or in 100 mM pH 4.63 sodium acetate buffer (NaAc buffer, Fisher Scientific). All materials were used as received. Polymer solutions were prepared gravimetrically and stirred overnight to ensure complete dissolution following a published procedure.26 The vials were tumbled end-over-end and allowed to stand in a constant temperature bath at 25 ( 0.1 C for 7 days without any further agitation (except when noted). The phase boundaries were determined by visual observation of samples containing varying nanoparticle and polymer concentrations as well as by turbidity. (See the Supporting Information.) The polymer adsorption isotherm was determined from samples with a nanoparticle concentration of 0.36 wt % and varying polymer concentrations. The samples were centrifuged at 10 000 rpm for 3 h in a centrifuge (Sorvall RC-6 ultracentrifuge) to settle and separate the nanoparticles as a dense sediment, after which the polymer concentration in the supernatant was determined by either UV-vis spectroscopy or gel permeation chromatography (GPC). (See the Supporting Information.) The surface saturation concentration of polymer adsorbed the nanoparticles (Γsatn) and the equilibrium constant (K) was determined from a linearization of the Langmuir adsorption isotherm, discussed further in the Supporting Information. Dynamic light scattering (DLS) was used to determine the hydrodynamic radii (Rh) using standard methods. The intrinsic viscosity [η] was measured using capillary viscometry and used to define the overlap concentration c*. Following the method of Otsubo et al.17 the flow behavior of the dense phase obtained by decanting the supernatant of the phase-separated samples was measured at 25 C on an AR-G2 rheometer (40 mm parallel plate geometry) to assist in the identification of the phase separation mechanism. Creep measurements were applied for 60 s, followed by a relaxation time of 600 s. Further details of the experimental methods used can be found in the Supporting Information. (26) McFarlane, N. L.; Wagner, N. J.; Lynch, M. L.; Kaler, E. W. Langmuir 2010, 26, 6262.

Langmuir 2010, 26(17), 13823–13830

McFarlane et al.

Article Table 1. Summary of PEO and PVP Characterization in Water

polymer

3

[η] (cm /g)

Mv (Da)

Rg (nm)

Rh water (nm)

Rh buffer (nm)

Rg/Ra

c* (wt%)

35 kDa PEO 50.9 ( 0.1 34 000 ( 11b 16 ( 1c 4.3 ( 0.1 4.9 ( 0.3 2.0 ( 0.1 37 ( 1b 33.3 ( 2.0 23.3 ( 6.3 0.30 ( 0.01 400 kDa PEO 370 ( 3 540 000 ( 5500b b c 15 ( 1 7.8 ( 1.3 6.7 ( 0.8 3.8 ( 0.1 55 kDa PVP 21.2 ( 0.7 41 000 ( 1900 38 ( 1c 28.6 ( 5.1 24.2 ( 2.3 0.68 ( 0.01 360 kDa PVP 155 ( 1 340 000 ( 1100b a Radius of the nanoparticles (R) is 21 nm. bMv and Rg calculated from the intrinsic viscosity ([η]). cRg calculated from static light scattering.

0.76 1.8 0.72 1.8

Figure 1. Adsorption isotherm for (a) PEO and (b) PVP on nanoparticles from water and buffer. The solid line is the Langmuir fit to the experimental data. Table 2. Surface Saturation Concentration (Γsatn) of Polymers on Nanoparticles in Water and Acetate Buffer solvent polymer

water Kads (m3/mg)

Γsatn (mg/m2)

NaAc buffer ΔH (kBT/segment)a

Kads (m3/mg)

Γsatn (mg/m2)

ΔH (kBT/segment)a

35 kDa PEO 5.4 ( 0.6 0.50 ( 0.01b -0.23 ( 0.01 8.9 ( 1.1 0.69 ( 0.01b -0.10 ( 0.01 -0.27 ( 0.01 5.1 ( 1.1 1.20 ( 0.06b -0.15 ( 0.01 400 kDa PEO 8.3 ( 2.2 0.49 ( 0.04b c b -1.6 ( 0.1 10.4 ( 5.6 0.75 ( 0.08 -0.74 ( 0.01 55 kDa PVP 2100 ( 1800 0.40 ( 0.05 -1.0 ( 0.1 4.9 ( 0.7 1.1 ( 0.0b -0.60 ( 0.01 360 kDa PVP 920 ( 630 0.46 ( 0.04c a Heat of adsorption (ΔH) values obtained from isothermal titration calorimetry.29 b Surface saturation amount measured by gel permeation chromatography. c Surface saturation amount measured by UV-vis spectroscopy.

4. Results and Discussion 4.1. Material Characterization. The hydrodynamic radius of the nanoparticles (Rh, 21 ( 1 nm), ζ potential (þ37 ( 1 mV), and density (2.2 ( 0.1 g/cm3) of the nanoparticles were determined in water at 25.00 ( 0.01 C. There was no measurable aggregation of the nanoparticles in the 70 mM NaAc buffer or in DI water. (See the Supporting Information.) The surface area of the nanoparticles (67 ( 2 m2/g) was calculated from the value of Rh. Over the polymer concentration range studied, the pH of the aqueous PEO solutions remains constant as expected (35 kDa, 7.5 ( 0.1; 400 kDa, 7.8 ( 0.1), whereas the pH of aqueous 55 kDa PVP was observed to decrease from 7.2 to 3.6 (at 30 wt % concentration), and the pH of aqueous 360 kDa PVP decreases from 5.8 to 3.4 (at 6 wt % concentration) with increasing PVP concentration. The pH of the nanoparticle dispersions is 3.7 ( 0.2 in water, independent of concentration. NaAc buffer (10 mM) was sufficient to control the pH of both polymer solutions to (0.05 of the pH of the buffer (4.63). Buffer solutions were prepared with 100 mM NaAc buffer throughout this study. Table 1 summarizes the results of the polymer characterization. Guinier plots showed no dependence of Rg on polymer concentration below the entanglement concentration (c*), which is Langmuir 2010, 26(17), 13823–13830

expected for nonionic polymers. Table 1 also shows that the ratio Rg/R varies from 0.72 to 1.8, such that the experiments probe the interesting region where the polymer and colloid are of comparable dimensions. 4.2. Polymer Adsorption. The adsorption isotherms of PEO and PVP from water and buffer are shown in Figure 1 along with the fits to the Langmuir isotherm (eq S1 in the Supporting Information). The surface saturation, Γsatn, and the adsorption constant, Kads, are reported in Table 2. The measured plateau adsorption for 35 kDa PEO and 400 kDa PEO (Figure 1a, Table 2, 0.50 ( 0.01 and 0.49 ( 0.04 mg/m2, respectively) is statistically independent of polymer molecular weight, as expected,22 but larger than reported values for PEO adsorption onto alumina (0.04 mg/m2 for 5000 kDa PEO at pH 9.527 and 0.13 mg/m2 for 10 kDa PEG at pH 528). The adsorbed polymer increases the effective size of the nanoparticle,26 resulting in Rg/Reff of 0.63 and 0.87 for 35 and 400 kDa PEO, respectively. The measured polymer adsorption isotherms for PVP (Figure 1b) show a slight increase in the surface saturation (reported in Table 2) with increasing polymer molecular weight (27) Moudgil, B. M.; Vasudevan, T. V. J. Colloid Interface Sci. 1989, 127, 239. (28) Esumi, K.; Nakaie, Y.; Sakai, K.; Torigoe, K. Colloids Surf., A 2001, 194, 7.

DOI: 10.1021/la101907s

13825

Article

McFarlane et al.

Figure 2. Phase diagram for (a) 35 kDa PEO and nanoparticle and (b) 400 kDa PEO and nanoparticle at 25 C. The blue shaded region bound by the solid line represents the two-phase region in water, whereas the gray shaded region bound by the dashed line represents the twophase region in buffer. The plateau adsorption lines in water (solid blue line) and in buffer (dashed blue line) are also shown for reference.

for both water and NaAc buffer solutions, as expected.22,23,29 These values are also larger than those reported in the literature, but this may be due to differences in pH (0.1 mg/m2 for 10-300 kDa PVP adsorption onto alumina, no pH given30). The pH of both PEO-nanoparticle and PVP-nanoparticle dispersions explored here is around the pH of the nanoparticle dispersion itself (pH 3.7). The effective size of the colloid increases with adsorbed polymer,26 resulting in an effective Rg/Reff of 0.50 and 0.76 for 55 and 360 kDa PVP, respectively. PEO adsorbs onto oxide surface hydroxyls via a hydrogen bond with the basic ether oxygen in the backbone.30,31 PVP is capable of forming a resonance structure as a result of the carbonyl bond. Therefore, in aqueous solution, PVP acts as a Lewis base and adsorbs on an oxide surface via the carbonyl bond.30 This acid-base bond has been confirmed by observation of an IR shift from 1690 to 1645 cm-1 of the carbonyl stretching vibration of PVP on silica.32 Because ionic bonds are generally stronger than hydrogen bonds, PVP is expected to have a stronger affinity for the nanoparticle surface than does PEO. Although the saturated adsorbed amounts for PVP and PEO are similar, the adsorption equilibrium constant (Kads, Table 2) is two to three orders of magnitude larger for PVP than for PEO, confirming that the PVP-nanoparticle interactions are much stronger than the PEO-nanoparticle interactions for this system. There is an increase in Γsatn for both polymers when the solvent is changed from water to NaAc buffer (pH 4.63). This trend cannot be simply explained by the effects of buffer on the surface potential, which increases slightly and is more likely due to a decrease in the solvent quality of the buffer for the polymer. The polymer contracts in size in buffer (Rh, Table 1) and consequently is expected from thermodynamic considerations to adsorb more readily to the nanoparticle in buffer as compared with DI water. Γsatn also increases more significantly for the higher molecular weight polymers as solvent quality effects become more pronounced with increasing molecular weight.33 (29) Robinson, S.; Williams, P. A. Langmuir 2002, 18, 8743. (30) Pattanaik, M.; Bhaumik, S. K. Mater. Lett. 2000, 44, 352. (31) Mathur, S.; Moudgil, B. M. J. Colloid Interface Sci. 1997, 196, 92. (32) Cohen Stuart, M. A.; Fleer, G. J.; Bijsterbosch, B. H. J. Colloid Interface Sci. 1982, 90, 321. (33) Bailey, F. R.; Koleske, J. V. Poly(ethylene oxide); Academic Press: New York, 1976.

13826 DOI: 10.1021/la101907s

Independent isothermal titration calorimetry (ITC) measurements were performed to verify that the strength of adsorption varies with pH and polymer type. These measurements yield the molar enthalpy of adsorption, summarized in Table 2.26 The segmental adsorption enthalpy is higher for PVP than for PEO by a factor of ∼10 (Table 2). However, for both polymers, the adsorption energy is lower in buffer than in water by a factor of ∼2. Both trends in adsorption energy (ΔH) agree with trends seen in Kads. This is in contrast with the higher amounts adsorbed in buffer and is not what is reported for bare silica particles.18 This difference between the adsorbed amount and the adsorption energy suggests that there is in fact multilayer adsorption onto the particle in buffer. This associated layer apparently sediments with the particles and results in the appearance of a higher adsorption amount in buffer, as determined by the depletion method.26 4.3. Phase Behavior. PEO-Nanoparticle Phase Behavior. PEO-induced phase separation of colloidal silica has been reported to be poorly reproducible because of slow and irreversible PEO adsorption onto the silica particles.15 The reproducibility of floc formation is reported to depend on the method of sample preparation, namely, whether the polymer is added to the particle suspension or vice versa. Specifically, if concentrated PEO is added to nanoparticle dispersions, irreversible bridging can occur before sufficient polymer is dispersed in solution to saturate the surface of the particles and provide steric stabilization. For these cationic nanoparticle dispersions, however, the location of the phase boundary and the nature of the coexisting phases were found to be independent of the order of addition of the components. At all nanoparticle concentrations studied, phase separation for PEO and nanoparticles (Figure 2) occurs at polymer concentrations above the concentration of polymer in solution required for surface saturation of the nanoparticles, Γsatn (also shown in Figure 2), suggesting that phase separation is due to a depletion attraction between polymer-covered nanoparticles. Phase separation was not immediate at polymer concentrations around the phase boundary, unlike separation at high polymer concentrations. Samples near the phase boundary were turbid for a few hours before separating into a translucent dense dispersion and a cloudy supernatant. There is no significant change in the turbidity of 5 wt % nanoparticle solutions as a function of increasing 35 kDa PEO Langmuir 2010, 26(17), 13823–13830

McFarlane et al.

Article

Figure 3. Phase diagram for (a) 55 kDa PVP and nanoparticle and (b) 360 kDa PVP and nanoparticle at 25 C. The blue shaded region bound by the solid line represents the two-phase region in water, whereas the gray shaded region bound by the dashed line represents the two-phase region in buffer. The plateau adsorption lines in water (solid blue line) and in buffer (dashed blue line) are also shown for reference.

solution concentrations leading up to phase separation, as expected. This observation is not consistent with bridging, but rather supports depletion flocculation as the dominant phase separation mechanism for PEO-induced nanoparticle phase separation. The phase boundary at higher nanoparticle concentration is nearly independent of nanoparticle concentration and shifts to lower polymer concentration, in both water and buffer, with increasing polymer molecular weight. Similarly, at low nanoparticle concentrations, increasing PEO Mw leads to less stability. Therefore, increasing PEO Mw reduces the nanoparticle dispersion stability. The behavior upon switching solvents from water to buffer is more complicated but is qualitatively similar for both molecular weights of PEO. At low nanoparticle concentrations, the solutions are miscible until higher concentrations in buffer as compared with water. However, at higher nanoparticle concentrations, there is less stability, such that lower concentrations of polymer are required to drive instability. In buffer, the polymers are of smaller hydrodynamic size because they are in a poorer solvent, and as shown in Figure 1a, the adsorbed amount increases substantially in buffer because of multilayer adsorption.26 These results differ from reports of PEO and silica nanoparticle mixtures, where Liu et al.18 reported a loss of stability for 2000 kDa PEO-silica (Rh = 40 nm) at polymer concentrations around the entanglement concentration but below that needed for nanoparticle surface saturation. They attributed PEO-nanoparticle phase separation to a polymer bridging flocculation where aggregates are “necklaces of spheres adsorbed onto chains”. This phenomenon was reported to occur for very high molecular weight polymer (2  106 Da) at low ionic strength and high pH.14,15 Therefore, the phase separation observed on these cationic nanoparticles is not similar to the bridging flocculation reported for PEO on anionic silica nanoparticles, which may be expected from the lower interaction energies for the PEO-Klebesol system explored here.26 Rather, the loss of stability is qualitatively consistent with depletion-induced phase separation. PVP-Nanoparticle Phase Behavior. The PVP-nanoparticle phase map is shown in Figure 3. In marked contrast with the PEO case, phase separation for PVP-nanoparticles in water occurs at polymer concentrations below those required for surface saturation at all but the lowest nanoparticle concentration. This Langmuir 2010, 26(17), 13823–13830

Figure 4. Phase map for 360k Da PVP: 4 wt % nanoparticle in different concentrations of buffer solution. The phase boundary in water is also shown for reference. Single phase (open symbols), bridging flocculation (filled gray circles/shaded region), and depletion flocculation (filled dark circles).

implies bridging of the nanoparticles by adsorbed PVP molecules. Turbidity measurements of aqueous 5 wt % nanoparticle dispersions with increasing PVP concentrations show an increase in the turbidity of the one-phase samples, which further supports bridging. A significant difference in the phase behavior of the PVP-nanoparticle system is observed when changing the solution from water to 100 mM NaAc buffer. Phase separation in buffer is observed only at polymer concentrations above those required for nanoparticle surface saturation. At polymer concentrations above plateau adsorption, the surface of the nanoparticle is expected to be completely covered with PVP such that any phase separation observed at those concentrations is expected to be due to depletion flocculation. We studied phase maps of 360 kDa PVP in a dispersion of 4 wt % nanoparticles in various concentrations of buffer to investigate the effect of buffer concentration on polymer adsorption and the resulting phase behavior (Figure 4). The onset of bridging flocculation in 100 mM buffer occurs at polymer concentrations an order of magnitude higher than those observed with water. Recall that PVP adsorption on metal oxides occurs by an acid-base reaction between the PVP carbonyl group and the metal surface hydroxyl group.33 Acetate also has a carbonyl DOI: 10.1021/la101907s

13827

Article

McFarlane et al.

Figure 5. Creep recovery tests conducted on PVP and PEO phase samples to differentiate between the methods of phase separation. (a) 10 wt % 35 kDa PEO-6 wt % nanoparticles at 0.01 Pa (open green triangles); pure 10 wt % 360 kDa PVP at 0.01 Pa (red crosses). (b) 0.1 wt % 360 kDa PVP-6 wt % nanoparticles at 5 Pa (open blue squares); sheared 0.1 wt % 360 kDa PVP-6 wt % nanoparticles at 5 Pa (blue crosses); 4 wt % 360 kDa PVP-6 wt % nanoparticles at 5 Pa (open black diamonds); sheared 4 wt % 360 kDa PVP-6 wt % nanoparticles at 5 Pa (black crosses).

group and therefore can also adsorb onto the nanoparticle. The number of carbonyl groups from each source is in fact comparable, as a 1 wt % 360k Da PVP solution contains ∼90 mM of carbonyl groups whereas there are 100 mM of carbonyl groups in the NaAc buffer. Therefore, there is a strong competition for hydroxyl group adsorption sites between the buffer and the PVP. The observed shift of the bridging phase boundary to higher polymer concentration with increasing buffer concentration is likely the result of the increasing amount of PVP needed to overcome the competitive adsorption of the buffer, which increases with increasing buffer concentration. Upon stirring of the bridged samples shown in Figure 4 in a vortex mixer, the samples became homogeneous and did not show evidence of bridging flocculation. DLS measurements of the size particles in the stable dispersions showed that the nanoparticles are covered with adsorbed polymer. This suggests that the partially covered, bridged samples are in a nonequilibrium state, which can be relaxed to equilibrium by vigorous stirring. Stirring enables the initially bridge adsorbed polymer to rearrange into a polymer brush that imparts steric stability. If sufficient free polymer is also present, then depletion flocculation is observed. (Figure 4, filled dark circles). This metastable behavior is only evident in buffer solution. To investigate further the mechanism of colloidal instability for nanoparticles in PVP, we studied the adsorption of PVP onto the nanoparticles as a function of increasing polymer concentration. Figure S1a of the Supporting Information shows the appearance of 360 kDa PVP in solution with 6 wt % nanoparticles in water after 7 days of equilibration. It is clear that there is a change in separation mechanism when the polymer concentration increases beyond the plateau adsorption amount (Figure S1a (C-D) of the Supporting Information). At polymer concentrations below that required for surface saturation, where bridging is the proposed separation mechanism (Figure S1a (A-C) of the Supporting Information), the suspensions separate into dense white flocs, which sediment with gravity, and a less dense supernatant with a distinct phase boundary. For samples above plateau adsorption (Figure S1a D-F of the Supporting Information), separated samples are visually similar to PEO-nanoparticle samples, which separated because of depletion flocculation. A sharp boundary is observed between the nanoparticle-rich phase at the bottom and 13828 DOI: 10.1021/la101907s

the top solution. The nanoparticle-rich bottom phase appears translucent rather than the cloudy supernatant observed for the lower concentration (bridge flocculated) samples. These results also demonstrate that the adsorption isotherms are important for interpreting the mechanism of polymer-induced phase separation. Figure S1b of the Supporting Information shows the appearance of the 360 kDa PVP samples shown in Figure S1a 1 week after they were sheared in a vortex mixer. The bridge-flocculated samples (Figure S1b (A-C) of the Supporting Information) remained unstable while the samples initially bridge flocculated, but with polymer concentrations above that required for saturation adsorption, they were peptized by shearing (Figure S1b (D-F) of the Supporting Information). Therefore, shearing the samples at polymer concentrations above that required to fully coat the nanoparticles can reverse the bridging flocculation and lead to stabilization. Hence, a path-dependent phase separation mechanism is observed. 4.4. Flow Properties of Bridged and Depleted Flocs. We further explored the mechanism of loss of phase stability by bridging flocculation or depletion attraction by measuring the elasticity of the dense phase of phase-separated samples. Following Otsubo,17 samples that separate by bridging are expected to exhibit an elastic recovery that will not be present in samples that phase separate by depletion interactions. This is because the polymer bridging two colloids can store accumulated strain in the sample and release that energy as elastic recovery upon relaxation, whereas samples phase-separated by depletion do not have a physical connection between the colloids to store elastic energy. Creep and recovery studies (Figure 5a) show that the dense phase of the 35 kDa PEO-nanoparticle mixtures exhibit viscous flow with little recoverable strain, similar to that for a simple polymer solution (360 kDa PVP) shown for reference. This suggests the sample phases separate by depletion attractions. The sediment from the 360 kDa PVP-nanoparticle mixtures at 0.1 and 4 wt % polymer does not exhibit much flow even under much higher stresses (500 times higher), and the small amount of deformation is largely recovered (Figure 5b). This suggests that these samples are dominated by bridging flocculation. Depleted samples that initially bridged before depletion (4 wt % 360 kDa PVP-nanoparticle mixture) exhibit an elastic characteristic similar to bridged samples prior to shearing . Langmuir 2010, 26(17), 13823–13830

McFarlane et al.

Article

Figure 6. (a) Fluid-solid phase boundary for 35 kDa PEO and nanoparticles from water (]) and buffer ([) and for 400 kDa PEO and nanoparticles from water (0) and buffer (9). (b) Fluid-solid phase boundary for 55 kDa PVP and nanoparticles from water (4) and buffer (2) and for 360 kDa PVP and nanoparticles from water (O) and from buffer (b).

After mixing (Silverson L4RT, 3,000 rpm for 1 min then 5000 rpm for 1 min) and re-equilibration, the dense phase of the 0.1 wt % 360 kDa PVP-nanoparticle mixture shows an even higher modulus (lower creep) and nearly complete elastic recovery, suggesting that mixing enhanced polymer bridging. The 4 wt % 360 kDa PVP-nanoparticle mixture resuspends upon mixing, and the solution rheology is now comparable to that observed for the sediment in the PEO mixtures. This supports the visual observation noted above (Figure 4 and Figure S1 of the Supporting Information) wherein shearing PVP-nanoparticle dispersions that contain more polymer than needed to saturate the particles surface leads to rearrangement of the adsorbed polymer favoring brush formation. This reduces bridging flocculation and can facilitate depletion flocculation if the free polymer concentration is sufficient. 4.5. Comparison of Experimental Phase Boundary Trends to Theory. The Flory-Huggins χ parameters for PEO and PVP in water and buffer (0.367, 0.486, 0.498, and 0.498) were calculated using published correlations.12 Although these values indicate slightly interacting systems, they are still within the range known to exhibit athermal good solvent phase behavior, as predicted by the PRISM method.12 The phase maps for polymer-colloid systems are plotted as the ratio of the polymer concentration to the polymer overlap concentration (c/c**) against the nanoparticle concentration in Figure 6. For systems that separated by depletion flocculation (PEO in both water and buffer and PVP in buffer), there is no trend in the phase boundary with increasing polymer molecular weight within the error bars, in contrast with the classical2,7,9 or PRISM2,11 models. This is most likely due to the adsorption of polymer onto the particle surface. Using the hydrodynamic radii of the polymer-coated nanoparticles (Reff), the Rg/Reff calculated are 0.63 (35 kDa PEO), 0.95 (400 kDa PEO), 0.50 (55 kDa PVP), and 0.76 (360 kDa PVP). These values are toward the “colloidal limit” (Rg < R) where the phase boundary is expected to shift to lower c/c** with increasing Rg/R. Figure 7 shows the phase boundaries for the 35 and 400 kDa PEO and the theoretical phase boundaries predicted by the PRISM theory.11 The amount of free polymer (cfree = ctotal in solution cadsorbed on particle surface) is plotted against the effective nanoparticle volume fraction, calculated using Reff. The experimental phase boundary shifts to lower c/c** with increasing nanoparticle concentration, which is consistent with theoretical trends (Figure 7), Langmuir 2010, 26(17), 13823–13830

Figure 7. Comparison of the experimental phase boundary of PEO in water to the boundary predicted by the PRISM theory in the limits where Rg/R f 0 (blue short-dashed line) and Rg/R f ¥ (blue solid line) and the free volume theory of Fleer and Tuinier (2008) (green dashed line). Also shown are the phase boundaries for 35 kDa PEO when Rg/Reff = 0.63 (]) and 400 kDa PEO when Rg/Reff = 0.95 (0).

but the experimental phase boundary does not fall between the two extremes predicted by the theory (Rg/R f 0 or Rg/R f ¥). An alternative prediction by Fleer and Tuinier34 based on a free volume theory is also show for Rg/R = 1. This has qualitatively similar trends to those of the experimental data but predicts that less polymer is required for phase separation. This may be a consequence of the penetrability of the adsorbed polymer layer on the particles by the free polymer, which is not considered in the model. This comparison shows that more theoretical work is needed on depletion flocculation for adsorbing polymers interacting with nanoparticle dispersions because such systems do not simply map onto predictions for nonadsorbing systems.

5. Conclusions Both PEO and PVP are observed to adsorb on the cationically modified silica nanoparticles and at sufficient concentrations destabilize the dispersions. The mechanism for loss of stability is different for the two systems and can be understood from the relative strength of polymer adsorption, with PVP having a (34) Fleer, G. J.; Tuinier, R. Adv. Colloid Interface Sci. 2008, 143, 1.

DOI: 10.1021/la101907s

13829

Article

stronger enthalpy of adsorption than PEO. PEO-nanoparticle solutions do not exhibit significant bridging flocculation; therefore, phase separation proceeds by depletion flocculation at polymer concentrations above surface saturation, independent of choice of solvent. In contrast, PVP in water initially looses stability by bridging at polymer concentrations well below those required for surface saturation. At higher concentrations above those required for surface saturation, however, strong mixing leads to depletion phase separation. This demonstrates that the mechanism of destabilization critically depends on the location of the phase boundary relative to the surface adsorption saturation line. It is also cautionary in that the result (bridging flocculation versus depletion phase separation) can be path-dependent, as

13830 DOI: 10.1021/la101907s

McFarlane et al.

demonstrated rheologically. Finally, a qualitative comparison of the measured depletion phase boundary to available theory does not show agreement, most likely because of polymer adsorption on the nanoparticle that is not accounted for in current theories applicable to this parameter range. Acknowledgment. This work was funded by the National Science Foundation GOALI program (CBET-0625047) and The Procter and Gamble Co. Supporting Information Available: Further details of experimental methods. This material is available free of charge via the Internet at http://pubs.acs.org.

Langmuir 2010, 26(17), 13823–13830