Effectiveness of Poly(vinylpyridine) Block Copolymers as Stabilizers of

Langmuir 1999, 15, 7145-7156

7145

Effectiveness of Poly(vinylpyridine) Block Copolymers as Stabilizers of Aqueous Titanium Dioxide Dispersions of a High Solid Content Serge Creutz† and Robert Je´roˆme* Center for Education and Research on Macromolecules (CERM), University of Lie` ge, Sart-Tilman, B6, B-4000 Lie` ge, Belgium Received July 29, 1998. In Final Form: May 27, 1999 Diblock copolymers of 4-vinylpyridine (4VP) and sodium methacrylate (MANa) have been synthesized and tested as dispersants for aqueous dispersions (80 wt % solid) of alumina-coated titanium dioxide. P4VP is strongly anchored onto the pigment surface. The molecular composition of the block copolymers has an effect on the amount of copolymer required for imparting good stability to the slurry. Where the two parent triblock copolymers are concerned, triblocks consisting of PMANa outer blocks have a stabilizing efficiency comparable to the diblock copolymers, whereas the reverse structure is much less efficient. In the case of a nonblock distribution of 4VP and MANa in the copolymer chains, the stabilization effect is lost. Finally, the comparison of different stabilizing blocks attached to the same anchoring block confirms that the electrosteric stabilization mechanism is superior to the steric stabilization mechanism. A decrease in the anchoring strength of the diblocks has a deleterious effect on the dispersion stability.

1. Introduction The production of a large range of commercially available products, such as coal,1 ore,2 paints,3 printing inks,4 ceramics,5,6 pesticides, and pharmaceutics,7,8 requires the use of well-defined and very stable solid dispersions. For obvious economical reasons, high solid dispersions are by far preferred, although more difficult to be prepared and stabilized. The dispersion stability, which ultimately controls the quality of the final product, is commonly imparted by appropriate dispersants, which actually build up either a steric or an electrostatic barrier against flocculation.9,10 In the case of solvent-based dispersions, i.e., nonaqueous dispersions, steric stabilization4 has proved to be most efficient. Nowadays, very strong environmental incentives give preference to aqueous dispersions at the expense of formerly used nonaqueous ones. This evolution requires more efficient dispersants, since the stabilization mechanisms, which are effective in water, are sensitive to external parameters, such as water hardness, pH, ionic strength, and temperature. It is worth noting that depending on the dispersant used, two stabilization mechanisms have to be distinguished. For instance, low molecular weight ionic surfactants, such as sodium dodecyl sulfate, contribute to electrostatic stabilization, compared to nonionic oligo- or polymeric † Present address: Dow Corning S.A., Parc Industriel C, B-7180 Seneffe, Belgium. * To whom correspondence should be addressed.

(1) Tadros, Th. F.; Taylor, P.; Bognolo, G. Langmuir 1995, 11, 4678. (2) Gebhardt, J. E.; Fuerstenau, D. W. Colloids Surf. 1983, 7, 221. (3) Patton, T. C. Paint Flow and Pigment Dispersion, 2nd ed.; Wiley: New York, 1979. (4) Leemans, L.; Fayt, R.; Teyssie´, Ph.; Uytterhoeven, H. Polymer 1990, 31, 106. (5) Lee, H. J. Colloid Interface Sci. 1993, 159, 210. (6) de Laat, A. W. M. Ph.D. Thesis, Wageningen Agricultural University, The Netherlands, 1995. (7) Hunter, R. J. Foundations of Colloid Science; Oxford University Press: New York, 1989. (8) Alexandridis, P. Colloid Interface Science 1996, 1, 490. (9) Napper, D. H. Polymeric Stabilization of Colloidal Dispersion; Academic Press: New York, 1983. (10) Everett, D. H. Basic Principles of Colloid Science; Royal Society of Chemistry: London, 1988.

surfactants, such as nonylphenylpoly(ethylene oxide) or poly(ethylene oxide-block-propylene oxide-block-ethylene oxide),11 which are commonly used as steric stabilizers of aqueous dispersions.12,13 In the 1980s, it was, however, shown that these two stabilization mechanisms could be advantageously combined into a unique one, known as electrosteric stabilization9 and typically promoted by block copolymers containing a polyelectrolyte component. The stabilizing polyelectrolyte block is definitely less sensitive to ionic strength14 than ionic low molecular weight surfactants and to temperature than nonionic surfactants. Nowadays, macromolecular engineering of a steadily increased number of polymers is a very instrumental tool for tailoring block copolymers, in terms of molecular architecture, chemical composition, weight composition, and molecular weight. Despite this remarkable opportunity, no systematic study of the electrosteric stabilization has been reported in a close relationship to the structural characteristic features of ionic block copolymers. It is the purpose of this paper to fill in this gap and to make useful guidelines available. A series of block copolymers consisting of an anchoring poly(4-vinylpyridine) block and a stabilizing sodium poly(methacrylate) one have thus been synthesized, while changing the molecular architecture (AB diblocks, ABA and BAB triblocks) and the molecular weight and composition in the case of diblocks (Table 1). A tapered block has also been synthesized by anionic polymerization, and a random copolymer of these two comonomers has been prepared by radical polymerization. The former copolymer consists of two main PMANa and P4VP blocks connected to each other through a transient block of regularly changing composition, in contrast to the latter, which is composed of MANa and 4VP units randomly distributed along the chain. Actually, tert-butyl (11) Luckham, P. F.; Bailey, A. I.; Miano, F.; Tadros, Th. F. ACS Symp. Ser. 1995, 615, 166. (12) Tadros, Th. F. Solid/Liquid Dispersions; Academic Press: London, 1987. (13) Tadros, Th. F. Surfactants; Academic Press: London, 1984. (14) Leemans, L.; Fayt, R.; Teyssie´, Ph.; de Jaeger, N. C. Macromolecules 1991, 24, 5922.

10.1021/la980951e CCC: $18.00 © 1999 American Chemical Society Published on Web 08/20/1999

7146

Langmuir, Vol. 15, No. 21, 1999

Creutz and Je´ roˆ me

Table 1. Characteristics of the Different Copolymers Used in This Contribution code PMANab D19 D30 Tin Tout Tp Rd DEO D2VP DSt a

4VP-b-tBMA 4VP-b-tBMA tBMA-b-4VP-b-tBMA 4VP-b-tBMA-b-4VP 4VP/tBMA (tapered) 4VP/tBMA (radical) 4VP-b-BO 2VP-b-tBMA St-b-tBMA

monomer unit

Mn (first block)

64 27-85 30-53 60-33-60 10-53-10

2800 3200 3500 7500

34-136 22-58 24-65

3600 2300 2500

NMR composition (wt %)

Mw/Mn

cmc (mg/L)

6900 14800 10700 20500 9500 7000c 7400c 9600 10500 11800

19-81 30-70 17-83 21-79 16-84 18-82 38-62 22-78 21-79

1.07 1.2 1.2 1.2 1.25 1.2 1.6 1.2 1.1 1.1

11 5 35 87 10 50 5 25 5

Mn(total) ) Mn(SEC) of first sequence divided by the NMR weight percentage of the first sequence. b Polysciences. c Mn(total) ) Mn(SEC).

methacrylate was used as a precursor for the sodium methacrylate units, because of a well-controlled anionic polymerization under mild conditions (0 °C) and an easy hydrolysis into acids, which are then neutralized with sodium hydroxide. For the sake of comparison, hydrophilic poly(ethylene oxide) has been substituted for the sodium poly(methacrylate) block, whereas two different blocks have been associated with the sodium poly(methacrylate) stabilizing block, i.e., poly(2-vinylpyridine) and polystyrene. The molecular characteristics of all these block copolymers are listed in Table 1. 2. Experimental Section 2.1. Anionic Polymerization. 4-Vinylpyridine (4VP), 2-vinylpyridine (2VP), R-methylstyrene (RMeSt), styrene (St), and tert-butyl methacrylate (tBMA) were first vacuum distilled from calcium hydride and then stored under a nitrogen atmosphere at -20 °C. Before polymerization, 4VP, 2VP, and tBMA were diluted with toluene, added with 1 M triethylaluminum in toluene until a yellowish green color appeared, and finally distilled under reduced pressure just prior to polymerization. Styrene previously diluted with toluene was dropwise added with fluorenyllithium until a persistent orange color was observed and then distilled under reduced pressure. 1,1-Diphenylethylene was dried and redistilled over sec-butyllithium just prior to the polymerization. Ethylene oxide (EO) was twice distilled over n-BuLi just before use. AIBN (Janssen) was used as received. (Diphenylmethyl)lithium and potassium were prepared at room temperature by reacting diphenylmethane with lithium and potassium naphthalenide, respectively, that were previously prepared by naphthalene metalation in THF at room temperature. Lithium chloride was flame dried under vacuum just prior to polymerization and stored under nitrogen. THF was purified by refluxing over a freshly prepared sodium benzophenone complex. Pyridine was purified by distillation over sec-butyllithium. Polymerization was carried out under dry nitrogen in flasks equipped with three-way stopcocks capped with a rubber septum. All glassware was flamed under vacuum before use. Solutions were transferred through stainless steel capillaries or with glass syringes. 2.1.1. Synthesis of Poly(4VP-b-tBMA) (D19 and D30 in Table 1).15 A 9/1 v/v mixture of pyridine and THF was cooled to 0 °C and dropwise added with (diphenylmethyl)lithium until a persistent yellow/orange color was observed. Then the required amount of this initiator was transferred to the reaction medium, followed by 4VP, which was allowed to polymerize for 20 min. An aliquot was withdrawn for characterization, and tBMA was then injected, the final copolymer concentration being 50 g/L. After an additional hour, copolymerization was stopped by addition of degassed methanol. Complete conversion was reached. The copolymers were recovered by precipitation into water. 2.1.2. Synthesis of 4VP/tBMA-Tapered Copolymer (Tp in Table 1).15 A 9/1 v/v mixture of pyridine and THF was cooled to 0 °C and dropwise added with (diphenylmethyl)lithium until (15) Creutz, S.; Teyssie´, Ph.; Je´roˆme, R. Macromolecules 1997, 30, 1.

Mn (total)a

a persistent yellow/orange color was observed. Then the required amount of this initiator was transferred to the reaction medium. 4VP and tBMA were mixed together and then added, the final copolymer concentration being 50 g/L. After 1 h, copolymerization was stopped by addition of degassed methanol. Complete conversion was reached. The copolymer was recovered by precipitation into water. 2.1.3. Synthesis of Poly(tBMA-b-4VP-b-tBMA) (Tin in Table 1). A THF solution of RMeSt was dropwise added with lithium naphthalenide at 0 °C until a persistent red color was observed, followed by a molar amount half that of RMeSt. A 9/1 v/v pyridine/THF mixture, cooled at 0 °C, was separately added with lithium naphthalenide until a persistent red color was also observed, followed by the required amount of the RMeSt initiator. 4VP was then added and allowed to polymerize for 15 min. An aliquot was withdrawn for characterization, before tBMA was injected and allowed to polymerize for half an hour, the final copolymer concentration being 50 g/L. Copolymerization was finally quenched with degassed methanol. Complete conversion was reached and the copolymer recovered by precipitation into water. 2.1.4. Synthesis of Poly(4VP-b-tBMA-b-4VP) (Tout in Table 1).16 A THF solution of RMeSt was dropwise added with lithium naphthalenide at 0 °C until a persistent red color was observed, followed by a molar amount half that of RMeS. tBMA was then added and allowed to polymerize for 15 min. An aliquot was withdrawn for characterization. 4VP, activated with triethylaluminum (1/30th-fold the molar concentration of 4VP), was then injected, the final copolymer concentration being 50 g/L. After 5 min, the copolymerization was quenched with degassed methanol. Complete conversion was reached and the copolymer recovered by precipitation into water. 2.1.5. Synthesis of Poly(4VP-b-EO) (DEO in Table 1). A 9/1 v/v mixture of pyridine and THF was cooled to 0 °C and dropwise added with (diphenylmethyl)potassium until a persistent yellow/ orange color was observed, followed by the required amount of this initiator. 4VP was then dropwise added and allowed to polymerize for 5 min. An aliquot was withdrawn for characterization, and EO was then injected, the final copolymer concentration being 50 g/L. The polymerization was allowed to proceed overnight, and the temperature was allowed to increase slowly to room temperature. Complete conversion was reached and the copolymer recovered by precipitation into heptane. 2.1.6. Synthesis of Poly(2VP-b-tBMA) (D2VP in Table 1). THF was added with a 10-fold molar excess of LiCl compared to the initiator, cooled to -78 °C, and dropwise added with (diphenylmethyl)lithium until a persistent yellow/orange color was observed. The required amount of this initiator was then transferred to the medium, followed by 2VP, which was polymerized for 1 h. An aliquot was withdrawn for characterization. tBMA was then added and allowed to polymerize for 2 h more, the final copolymer concentration being 50 g/L. Copolymerization was stopped by addition of degassed methanol. Complete conversion was reached and the copolymer recovered by precipitation into water. 2.1.7. Synthesis of Poly(St-b-tBMA) (DSt in Table 1). THF was added with a few drops of RMeSt and a 10-fold molar excess (16) Creutz, S.; Teyssie´, Ph.; Je´roˆme, R. Macromolecules 1997, 30, 5596.

Poly(vinylpyridine) Block Copolymers as Stabilizers of LiCl with respect to the initiator. It was then cooled to -78 °C and dropwise added with sec-butyllithium until a persistent orange/red color was observed. The required amount of this initiator was transferred to the reaction medium, followed by styrene, which was polymerized for half an hour. An aliquot was withdrawn for characterization. Polystyrene chains were endcapped with 1,1-diphenylethylene prior to tBMA addition, the final copolymer concentration being 50 g/L. After 2 h more, the copolymerization was quenched with degassed methanol. Complete conversion was reached and the copolymer precipitated into water. 2.2. Radical Copolymerization of 4VP and tBMA (Rd in Table 1). The random copolymerization was carried out in toluene at 65 °C under a nitrogen atmosphere. AIBN (0.9 g, 5.5 mmol) was solubilized into 40 mL of toluene and then added with 3.6 mL of 4VP (33 mmol) and 24.6 mL of tBMA (150 mmol). This solution was transferred into 40 mL of toluene preheated at 65 °C, at a rate of 15 mL/h. When the injection was complete, the polymerization was continued for an additional half an hour. The final yield was 52%. The copolymer was precipitated into water. 2.3. (Co)polymer Characterization. Size-exclusion chromatography (SEC) of 4VP copolymers was carried out in N,Ndimethylformamide added with 10% triethylamine and 10% pyridine at 45 °C,15 by using a Hewlett-Packard 1050 liquid chromatograph equipped with a mixed C PLGel column and a Hewlett-Packard 1047A refractive index detector. THF added with 1% triethylamine was used as the eluent at 35 °C for the analysis of 2VP copolymers. Styrene copolymers were characterized by SEC in THF at 35 °C. Poly(2-vinylpyridine) and polystyrene standards were used for calibration. 1H NMR spectra were recorded at 400 MHz with Bru¨ker AN 400 superconducting magnet equipment. 2.4. Hydrolysis. The poly(tert-butyl methacrylate) block was hydrolyzed by refluxing the copolymer overnight in a 5/1 v/v dioxane/37% HCl solution. The hydrolyzed copolymers were recovered by solvent distillation under vacuum and redissolution in water upon neutralization with NaOH. Finally, copolymers were purified by dialysis against demineralized water. 2.5, Critical Micelle Concentration Determination. The critical micelle concentration (cmc) of each block copolymer in water was measured by light scattering with a BI-200 photogoniometer (Brookhaven Instruments) equipped with a BI-2030 128 channel correlator (Brookhaven) and an Ar ion laser (Lexel Model 95), emitting vertically polarized light at 488 nm. All the measurements were performed at 25 °C. A solution of each copolymer was prepared at pH 9 with a 0.01 M Borax buffer, which was previously relieved of dust by centrifugation at 20 000 rpm for 15 min. This solution was left standing for at least 24 h and then diluted by incremental amounts of a previously centrifuged Borax solution, so as to make solutions of decreasing concentrations. All these solutions were aged for at least 24 h before measurement. The scattered light intensity was measured at 90° as a function of the copolymer concentration. The concentration at which the light scattering dropped off as a result of the micelles disappearance was reported as the cmc. 2.6. Dispersions Preparation. Titanium dioxide, RCL 535 from SCM Chemicals, was of the rutile type. It was precoated with alumina and characterized by a surface area of 12 m2/g and a density of 4.2 g/cm3. To 50 mL of an aqueous solution of copolymer (in demineralized water at pH 9), 200 g of titanium dioxide was added and ground with a dissolver disk at a speed rate of 3700 rpm for 15 min (volume fraction of TiO2 ) 0.48). The copolymer content was changed and referred to as the weight ratio with respect to titanium dioxide (e.g., 1 g of dispersant for 200 g of TiO2 is reported as 0.5%). 2.7. Rheological Measurements. Dispersion rheology was measured with a Rheotest 2 equipped with a Couette geometry at 25 °C. The samples were presheared in the Couette cell at 0.54 s-1 for 20 min. An incremental shear rate sweep (from 0.54 to 1312 s-1 and then back to 0.54 s-1) was carried out in order to monitor the τ vs γ˘ curves. Data reported at each shear rate in the upward scanning were steady-state values. The dispersion was maintained under shear at 1312 s-1 for 1 min. Data were

Langmuir, Vol. 15, No. 21, 1999 7147 Scheme 1

recorded 5 s after stabilization at each shear rate during the downward sweep. The Bingham model was used:

τ ) τβ + ηplγ˘

(1)

where τ is the shear stress measured at the shear rate γ˘ . The apparent yield, τβ, and the plastic viscosity, ηpl, were determined from the linear part of the τ vs γ˘ curve, as the value extrapolated to the origin and the slope, respectively. The hysteresis area was determined by subtraction of the integral of the downward curve from that of the upward curve.

3. Results and Discussion As exemplified in Table 1, the cmc of the copolymers is strongly influenced by the molecular architecture. Indeed, the triblock copolymers exhibit a much higher cmc than the diblock copolymers, highlighting their more difficult aggregation. This is especially true for Tout, where the PMANa block has to bend for the two outer hydrophobic blocks to be localized in the micellar core. It is worth noting that the tapered copolymer (Tp) exhibits characteristics closer to the diblock equivalent than the random one. As far as the diblock copolymers are concerned, their cmc’s are similar, with the exception of the poly(2VP-b-MANa) copolymer. The higher cmc of the latter might be related to the lower molecular weight of the hydrophobic block. Furthermore, the alignment of the 2VP dipoles might be less regular compared to the 4VP dipoles and lead to a looser packing within the micellar core (Scheme 1). In this study, the copolymer concentration range (ca. 5-40 g/L) is far above the cmc (at least 55 times). Therefore, unimers are expected to play a minor role in the adsorption process, in contrast to the exchange between micelles, as will be discussed further for the poly(St-b-MANa) copolymer. In a preliminary study,17 the measurement of adsorption isotherms was unsuccessful. Indeed, the complete recovery of the titanium dioxide particles required centrifugation at such a high speed (>15 000 rpm) that micelles of the nonadsorbed copolymer were also sedimented, thus preventing any reliable characterization. In line with many research groups,1,5,11-13,18-26 rheology has finally been chosen as the most appropriate characterization for high (17) Leclef, Y.; Creutz, S.; Je´roˆme, R. Unpublished results. (18) Luckham, P. F.; Ansarifar, M. A. Adv. Fine Particles Process. 1990, 145. (19) Jones, D. A. R.; Leary, B.; Boger, D. V. J. Colloid Interface Sci. 1992, 150, 84. (20) De Silva, G. P. H. L.; Luckham, P. F.; Tadros, Th. F. Colloids Surf. 1990, 50, 263. (21) Choi, G. N.; Krieger, I. M. J. Colloid Interface Sci. 1986, 113, 101. (22) Sung, A.; Piirma, I. Langmuir 1994, 10, 1393. (23) Sato, T.; Kohnosu, S. J. Colloid Interface Sci. 1992, 152, 543. (24) Schroeder, J. Prog. Org. Coat. 1988, 15, 337. (25) Tadros, Th. F. Adv. Fine Particles Process. 1990, 71. (26) Liang, W.; Bognolo, G.; Tadros, Th. F. Langmuir 1995, 11, 2899.

7148


Figure 1. Plot of the shear stress vs shear rate at D19 content of 0.32%. Upward scanning (a) and downward scanning (b) of shear rates.

solid dispersions. Rheology is indeed a very efficient tool for studying the stability of dispersions and the dispersant concentration required for the full surface coverage of the solid particles1,5,20 or “pigment demand”.3 The shear rate dependence of the steady-state shear stress of TiO2 dispersions has been measured in order to learn about the effect of the molecular structures of polymeric dispersants and the criteria for their design. Five main rheological characteristics of the dispersions have been considered : (i) the viscosity at low shear rate (0.54 s-1), (ii) the apparent yield value (τβ) (see eq 1), (iii) the hysteresis area, (iv) the shear-thinning factor, arbitrarily defined as the ratio of shear viscosities measured at 0.54 and 364.5 s-1, respectively, and (v) the plastic viscosity (ηpl) (eq 1). Typical rheological curves are shown in Figure 1, while scanning shear rates upward and downward, respectively. The reproducibility of the data is better than 15%.17 The shear viscosity of dispersions of solid particles is basically influenced by the particle volume fraction.13 When this volume fraction approaches the maximum packing, a dramatic increase in viscosity occurs. Depending on the type of lattice, the maximum packing for monodisperse spherical particles lies between ca. 0.52 and 0.74.27 In this study, the TiO2 volume fraction, i.e., 0.48, has been chosen in this critical region so that a small change in the dispersion state may have a dramatic effect on the shear viscosity. In this respect, the effective volume fraction must be considered rather than the theoretical one,26-28 since polymer adsorption onto the particle surface results in an increase of the effective particle volume and thus in an increase of viscosity compared to the same dispersion of bare particles. Moreover, in the absence of dispersant, flocculation may occur, which results in a sharp increase in the effective volume fraction and in the final viscosity. Compared to a stable dispersion, flocculation is at the origin of a high yield value, a higher viscosity at low shear rate, a large hysteresis, and a more pronounced shear-thinning effect.1,5,11-13,18-26 Indeed, a higher initial stress (yield value) is required for disrupting the flocs in the case of severe flocculation. Under shear, the flocs dissociate, which leads to a sharp drop in viscosity (shear thinning). As long as some shear is maintained, flocculation remains limited or may be prevented from occurring, which accounts for an important hysteresis. (27) Toussaint, A. Prog. Org. Coat. 1992, 21, 255.


Figure 2. Plot of the viscosity at 0.54 s-1 vs the dispersant/ pigment wt/wt %: (b) PMANa, (9) D19, ([) D30, (2) Tin, (4) Tout.

All the molecular characteristics of the synthesized copolymers are reported in Table 1. Poly(4-vinylpyridine) has been chosen as a block that might have strong propensity to adsorb onto the pigment. Indeed, TiO2 used in this study is coated with alumina, which is a common practice for preventing any photolytic degradation that could be catalyzed by titanium dioxide. Amines, such as vinylpyridine, are expected to adsorb onto alumina by hydrogen bonding with the surface hydroxyl groups and by Lewis acid-base interactions with the aluminum cations, as well. These acid-base interactions should be the major contribution to amine anchoring onto aluminacoated TiO2.29 4VP has been initially preferred to 2VP, because a better accessibility of the nitrogen atom to interaction sites was expected, giving rise to more efficient dispersants. A second reason for the choice of 4VP rather than 2VP has to be found in the inability of the poly(tBMA) anion to initiate the 2VP anionic polymerization30 in contrast to 4VP,15,16 which precludes the synthesis of both triblock copolymers consisting of 2VP outer blocks and tBMA-2VP-tapered copolymers. The poly(4VP-bMANa) copolymer, D19, will be used as a reference for the following discussion. 3.1. Homopolymer. Since the poly((meth)acrylic acid) salt block of the polymeric dispersants is known for its dispersing properties,2,3,5,6 this polyelectrolyte has been first tested and compared to the reference diblock (D19) (Figures 2-6). Upon increasing the concentrations of sodium poly(methacrylate), the shear viscosity of the TiO2 dispersion dramatically decreases and goes through a minimum. Actually, the adsorption of increasing amounts of polyelectrolyte prevents steadily more efficiently flocs from being formed, until the pigment surface is saturated. This minimum in viscosity is commonly designated as the “pigment demand”. The addition of an exceeding amount of dispersant (0.7%) results in flocculation. As discussed by Napper,9 this polymer excess may contribute to the depletion of the polymer layer in the boundary region between two particles and be responsible for the so-called (28) Liang, W.; Tadros, Th. F.; Luckham, P. F. J. Colloid Interface Sci. 1992, 153, 131. (29) van der Beek, G. P.; Cohen Stuart, M. A.; Fleer, G. J.; Hofman, J. E. Macromolecules 1991, 24, 6600. (30) Ishizone, T.; Hirao, A.; Nakahama, S. Macromolecules 1993, 26, 6964. (31) Liang, W.; Tadros, Th. F.; Luckham, P. F. J. Colloid Interface Sci. 1993, 158, 152. (32) Liang, W.; Tadros, Th. F.; Luckham, P. F. J. Colloid Interface Sci. 1993, 155, 156.

Poly(vinylpyridine) Block Copolymers as Stabilizers

Figure 3. Plot of the yield value vs the dispersant/pigment wt/wt %: (b) PMANa, (9) D19, ([) D30, (2) Tin, (4) Tout.

Figure 4. Plot of the hysteresis area vs the dispersant/pigment wt/wt %: (b) PMANa, (9) D19, ([) D30, (2) Tin, (4) Tout.

Figure 5. Plot of the shear-thinning factor vs the dispersant/ pigment wt/wt %: (b) PMANa, (9) D19, ([) D30, (2) Tin, (4) Tout.

depletion flocculation.31,32 This phenomenon does not, however, occur when the polymer anchoring onto the pigment surface is strong enough, as will be confirmed in the case of the diblock copolymers under consideration. It must be mentioned that the dispersion containing 0.7% sodium poly(methacrylate), or PMANa, partly flocculates and breaks down in the Couette cell at intermediate

Langmuir, Vol. 15, No. 21, 1999 7149

Figure 6. Plot of the plastic viscosity vs the dispersant/pigment wt/wt %: (b) PMANa, (9) D19, ([) D30, (2) Tin, (4) Tout.

shear rates (>2.3 s-1). This explains why only the low shear rate viscosity has been measured and not the complete rheological behavior. Weak polyacids, such as poly(methacrylic acid), are not fully neutralized. Their degree of neutralization, R, actually depends on the pH and concentration as well. At pH 9, R is of the order of ∼0.9.33,34 PMANa may thus be considered as a copolymer of methacrylic acid and sodium methacrylate. The former groups may anchor onto solid particles by hydrogen bonding and the latter ones by electrostatic interactions with an oppositely charged surface. At pH 9, the alumina surface is close to neutrality5,12 so that anchoring should primarily occur by hydrogen bonding.2 The polyelectrolyte is more likely adsorbed with a loop-train conformation, in such a way that loops of PMANa protrude into the bulk solution and provide the solid particles with an electrosteric stabilization. In the extreme situation where only one chain end of the polyelectrolyte is anchored to the solid surface, the thickness of the protective barrier (length of the stretched chains) would be ca. 16 nm (0.25 nm per monomer35). This thickness should be decreased by a factor of at least 2 or 3, taking the ca. 10% anchoring units and the loop-train conformation into consideration. This situation may explain the limited dispersion efficiency of polyelectrolytes, which cannot prevent some flocculation from occurring even when the particle surface is completely covered. There is an analogy between polyelectrolytes and poly(vinyl alcohol). Both homopolymers behave as copolymers. Indeed, poly(vinyl alcohol) results from the partial hydrolysis of poly(vinyl acetate): the residual vinyl acetate units being the anchoring groups and the poly(vinyl alcohol) segments the stabilizing blocks.9,36 In case of the diblock copolymer, D19, a rapid drop in viscosity is followed by a slow increase upon increasing concentration (Figure 2). Compared to sodium poly(methacrylate), the minimum in viscosity is much lower, which is the signature of a better stabilization,1,3,5,20 and it is observed at a smaller content (0.21% or a maximum surface coverage of 0.18 mg of stabilizer/m2), which indicates a higher efficiency. This improved stabilization (33) Liu, G.; Guillet, J. E.; Al-Takrity, E. T. B.; Jenkins, A. D.; Walton, D. R. M. Macromolecules 1991, 24, 68. (34) Bohm, J. Th. C. Ph.D. Thesis, Wageningen Agricultural University, The Netherlands, 1974. (35) Patrickios, C. S.; Hertler, W. R.; Abbott, N. L.; Hatton, T. A. Macromolecules 1994, 27, 930. (36) Scholtens, B. J. R. Ph.D. Thesis, Wageningen Agricultural University, The Netherlands, 1977.

7150

Langmuir, Vol. 15, No. 21, 1999 Scheme 2

is confirmed by a strong reduction in hysteresis and shear thinning (Figures 4 and 5). It must be pointed out that D19 is a polyampholyte. Depending on the pH, poly4VP may be protonated or not and thus water-soluble or not. Moreover, at a constant composition, the pKa and thus the protonation degree37-39 of the 4VP and MANa co-units depend on their distribution within the copolymer, as a result of nearest-neighbor and long-range electrostatic interactions. To our best knowledge, there is no report on the solution properties of poly(4VP-b-MANa) copolymers in relation to pH, in contrast to the poly(2VP-b-MAA)37 and poly(2VP-b-AA)38 copolymers (MAA and AA stand for methacrylic and acrylic acid, respectively). According to Kamachi et al.,37 the isoelectric point of poly(2VP-b-MAA) copolymers of a composition comparable to D19 lies between ca. pH 2 and 3. Since the pyridinium ions are more acidic than the carboxylic groups, the predominant species at the isoelectric point would be the uncharged molecular form (A) rather than the zwitterionic form (B)39 (Scheme 2). By analogy, poly4VP would be deprotonated at pH 9 and thus no more water-soluble, while polyMAA would be essentially neutralized and thus negatively charged. Micelles, which were reported by Briggs et al.38 for poly(2VP-b-AA), are also observed for the D19 copolymer. When the poly4VP and polyMANa blocks are segregated within micelles, the effects of the nearest-neighbor and long-range electrostatic interactions must be strongly reduced. In addition, the high local concentration of polyMANa in the micelles has to decrease slightly the degree of neutralization compared to that of the homopolymer at the same total concentration.40 For this reason, a gradient of increasing neutralization degree should manifest itself in the polyelectrolyte brush, while going from the micellar core to the bulk solution. Since at pH 9 the polyMAA block is mostly neutralized and therefore stretched out,40 sea urchin-like micelles, composed of a poly4VP core and polyMANa spines, are expected to be formed. It is, however, beyond the scope of this paper to substantiate further this hypothesis. (37) Kamachi, M.; Kurihara, M.; Stille, J. K. Macromolecules 1972, 5, 161. (38) Briggs, N. P.; Budd, P. M.; Price, C. Eur. Polym. J. 1992, 28, 739. (39) Bekturov, E. A.; Kudaibergenov, S. E.; Rafikov, S. R. J. Macromol. Sci., Rev. Macromol. Chem. Phys. 1990, C30, 233. (40) Kiserow, D.; Prochazka, K.; Ramireddy, C.; Tuzar, Z.; Munk, P.; Webber, S. E. Macromolecules 1992, 25, 461.


The basic question is to know how the dispersant is adsorbed onto the pigment surface. Indeed, it may be anchored by either the hydrophobic poly4VP block or the hydrophilic polyMANa block or both of them. The selective adsorption of the polyMANa block is not a reasonable hypothesis, since then the poly4VP would have to interact directly with water, which is quite unfavorable.1 Although formation of a double layer could dramatically decrease the poly4VP-water contactssas often occurs in the case of adsorption of cationic surfactants onto anionically charged surfaces12sthe minimum in viscosity should then be shifted toward higher concentrations compared to sodium poly(methacrylate), which was used as dispersant. As previously mentioned, the adsorption of poly4VP onto the pigment surface is highly favorable because of possible hydrogen bonding and Lewis acid-base interactions. This situation might account for a higher gain in energy compared to sodium poly(methacrylate), which only contains ca. 10% anchoring units, and thus for the lower concentration of diblock copolymer needed to reach the minimum in viscosity. Furthermore, polyMANa will lose less entropy by protruding into the bulk solution than when being adsorbed with a loop-train conformation. So, although one may not rule out that polyMANa is not adsorbed at all, this block should contribute to building up an electrosteric stabilizing layer. Assuming a length of 0.25 nm35 for the monomer unit and the complete stretching of the polyMANa chains,40 a layer thickness of ca. 21 nm may be calculated, which is much larger than the ca. 5-nm thickness, which is usually recommended in order to overcome the van der Waals attraction.9,20 The yield stress, which is a measure of the strength and number of flocs,32 is reduced but not completely suppressed by the diblock D19 (Figure 3). Although some residual flocculation has to be considered, most of the yield stress more likely results from the long-range electrosteric repulsion of the stabilizing layers.19,22,26,28,41 Indeed, interaction and overlap of the electrical double layers of particles cause electroviscous effects, which tend to increase the effective excluded volume and thus the rheological parameters of stable dispersions. An increase in viscosity, less marked than with polyMANa, is observed at higher diblock concentrations. The question is to know whether a depletion flocculation occurs or not. The excess of dispersant compared to the minimum (ca. 0.2-0.3%) is ca. 0.6% (24 g/L), i.e., 2000-fold the cmc. At this concentration, micelles formed by the dispersant excess might have a meaningful effect on the water viscosity and may account for the smooth increase in viscosity with concentration of the diblock. The small hysteresis and shear-thinning factor, which remain almost constant, are also in favor of an increase in the water viscosity rather than flocculation. The plastic viscosity tends to corroborate this hypothesis. At high shear rates, the flocs are broken down into smaller units, which are likely very similar for all the dispersions. Then, the viscosity is controlled by the hydrodynamic interactions that result from the applied shear. So the plastic viscosity only depends on the viscosity of the liquid phase and on the effective volume fraction of the solid particles,1,32,42 which would be constant in this case. Therefore, the increase in the dispersion viscosity at higher diblock concentrations would basically reflect a more viscous dispersion phase rather than a depletion flocculation. (41) Hirtzel, C. S.; Rajagopalan, R. Colloidal Phenomena; Noyes: Park Ridge, NJ, 1985. (42) van der Hoeven, Ph. C. Ph.D. Thesis, Wageningen Agricultural University, The Netherlands, 1991.

Poly(vinylpyridine) Block Copolymers as Stabilizers Scheme 3

The charged micelles are expected to distribute throughout the pigment dispersion in such a way that the organization that results from the mutual repulsion of the stabilized TiO2 particles is not lost but rather reinforced into a more dense particles network responsible for the more pronounced increase in yield stress (Scheme 3). The radius of the micelles as measured by light scattering (ca. 60 nm) completely fits within the interstices left by the pigment particles (ca. 300 nm). Hence, the final viscosity has to result from the mutual repulsion of micelles and pigment particles organized according to an ordered lattice. This means that the contribution of the micelles and the stabilized TiO2 particles to the dispersion viscosity would not be additive but that some synergism might occur. The conclusion that no depletion flocculation occurs is consistent with strong anchoring of the diblock to the pigment surface through the poly4VP block rather than through the polyMANa one. This situation, which is the most favorable to the pigment stabilization as previously explained, might allow us to predict still an acceptable stability in the case for an overshoot in concentration, which is a regular practice in the paint industry. 3.2. Influence of the Copolymer Composition. That the molecular composition of a diblock copolymer has to be optimized for imparting the best stability to a dispersion has been shown both theoretically43-45 and experimentally.46-48 This general conclusion will be supported in a forthcoming paper dealing with the dispersing capability of poly(di(methylamino)ethyl methacrylate) containing diblock copolymers.49 In this paper, only one composition richer in P4VP compared to the D19 diblock has been considered. According to previous studies,43-48 a too low P4VP composition should result in the weaker anchoring of the dispersant and, thus, to a poor stability under shear and possibly to depletion flocculation. The 4VP content of the diblock D30 (30 wt %) has been increased with respect to the copolymer D19 (19 wt %) essentially by decreasing the molecular weight of the polyelectrolyte block. A priori, this modification has to result in a thinner electrosteric barrier. As a result, ca. 2 times more polymeric dispersant is required to reach the minimum in the (43) Evers, O. A. Ph.D. Thesis, Wageningen Agricultural University, The Netherlands, 1990. (44) Evers, O. A.; Scheutjens, J. M. H. M.; Fleer, G. J. Macromolecules 1990, 23, 5221. (45) Evers, O. A.; Scheutjens, J. M. H. M.; Fleer, G. J. J. Chem. Soc., Faraday Trans. 1990, 86, 1333. (46) Wu, D. T.; Yokoyama, A.; Setterquist, R. L. Polym. J. 1991, 23, 709. (47) Hoogeveen, N. G.; Cohen Stuart, M. A.; Fleer, G. J. Macromol. Chem. Phys. 1996, 197, 2553. (48) Orbay, M.; Laible, R.; Dulog, L. Makromol. Chem. 1982, 183, 47. (49) Creutz, S.; Je´roˆme, R. In preparation.

Langmuir, Vol. 15, No. 21, 1999 7151

viscosity and the other properties of the dispersion (Figures 2-6). As far as the minima in viscosity for the two diblocks are compared, the dispersion stability seems to be quite comparable. Viscosities and shear-thinning factors are also similar, whereas the hysteresis area and the yield value are somewhat smaller for D30 compared to D19. In the diblock concentration range beyond the minimum in viscosity, this property, the yield value, the hysteresis area, and the plastic viscosity increase less rapidly in the case of D30, the shear-thinning factors being comparable. This observation could result from the smaller molecular weight of the polyMANa block in this diblock copolymer. Since the stabilizing block of D30 protrudes less deeply into the aqueous solution, the effective volume fraction of the pigment must accordingly be smaller. The interparticle distance should accordingly increase and their mutual interactions decrease consistently with the reported rheological data. In agreement with the higher content of the poly4VP block and, thus, the stronger anchoring of the dispersant, there is no indication in favor of a depletion flocculation. In the limits of comparison of the diblocks D19 and D30, it appears that the copolymer composition has mainly an effect on the minimum amount of dispersant required for imparting stability to the dispersion, and not on the stability. Adsorption of block copolymers, which combine a strongly charged block and a hydrophobic block, has been theoretically examined by Dan et al.50 on the basis of a scaling model. Above the cmc, the surface density of the adsorbed layer is predicted to scale as γ18/25Φs4/5NA12/25/ NP6/5, where γ, Φs, NA, and NP stand for the surface tension of the micellar core-corona interface, the salt concentration, the number of hydrophobic monomer units, and the number of charged monomer units, respectively. This model assumes that micelles act just as a reservoir and remain unadsorbed. Whether the micelles adsorb onto the pigment particles51-54 or merely act as a reservoir of free chains55-58 is still a matter of controversy. The conclusion by Dan et al. is believed to hold for the system under study, as will be further developed for the poly(St-b-MANa) copolymer. On the basis of the molecular characteristics of D19 (NA ) 27 and NP ) 85) and D30 (NA ) 30 and NP ) 53), a surface density ca. 2 times higher (ca. 1.85) in the case of D30 ((3012/25 × 856/5)/(536/5 × 2712/25)) is predicted, which is consistent with the displacement of the minimum in viscosity to a 2 times higher concentration. 3.3, Influence of the Molecular Architecture. The molecular architecture of polymeric dispersants is a structural parameter that is often disregarded in the study of dispersion stabilization. The well-controlled anionic polymerization of 4VP and tBMA, however, allows us to design the block copolymers, i.e., diblocks of the AB type and triblocks of the two ABA and BAB structures. Therefore, the poly(MANa-b-4VP-b-MANa) (Tin) and poly(4VP-b-MANa-b-4VP) (Tout) copolymers have been synthesized and tested as dispersants. It is worth noting that the former triblock has the same hydrophilic-hydrophobic-hydrophilic block distribution as the commonly used (50) Dan, N.; Tirrell, M. Macromolecules 1993, 26, 4310. (51) Cao, T.; Yin, W.; Armstrong, J. L.; Webber, S. E. Langmuir 1994, 10, 1841. (52) Huguenard, C.; Pefferkorn, E. Macromolecules 1994, 27, 5271. (53) Huguenard, C.; Elaissari, A.; Pefferkorn, E. Macromolecules 1994, 27, 5277. (54) Munch, M. R.; Gast, A. P. Macromolecules 1990, 23, 2313. (55) Ouali, L.; Pefferkorn, E. Macromolecules 1996, 29, 686. (56) Marques, C.; Joanny, J. F.; Leibler, L. Macromolecules 1988, 21, 1051. (57) Johner, A.; Joanny, J. F. Macromolecules 1990, 23, 5299. (58) Zhan, Y.; Mattice, W. L. Macromolecules 1994, 27, 683.

7152

Langmuir, Vol. 15, No. 21, 1999 Scheme 4

poly(ethylene oxide-block-propylene oxide-block-ethylene oxide) nonionic surfactant. The main characteristics of the TiO2 dispersion prepared in the presence of the poly(MANa-b-4VP-b-MANa) copolymer are comparable to those reported for the D19 dispersant (Figures 2-6). No depletion flocculation is observed in contrast to what happened for PMANa, which highlights a stronger dispersant anchoring. The rheological behavior at concentrations higher than the minimum in viscosity would again reflect the thickening of water by the excess of dispersant, as supported by the plastic viscosity (Figure 6). Since the previously mentioned scaling law is only valid for diblock copolymers, the structure of the diblock (NA ) 33 and NP ) 120) equivalent to Tin (NA ) 33 and NP ) 2 × 60) has been considered to be a rough approximation. The minimum in viscosity for the theoretical diblock equivalent should be observed at a concentration smaller by ca. 27% compared to D19. In contrast, a higher concentration (by ca. 10%) is reported. Since the actual polyelectrolyte thickness, which governs mainly the adsorption,50 is ca. 2 times less extended than for the theoretically equivalent diblock copolymer, this discrepancy is not surprising. If the model for the triblock copolymer is a diblock consisting of one PMANa block of the same length and one P4VP block 2 times shorter compared to the triblock (NA ) 16.5 and NP ) 60), a 20% increase in surface density is predicted. This slight overestimation might be related to the fact that the two modeled chains, which originate from one triblock copolymer, do not adsorb independently from each other. Therefore, more bare spots on the particle will be left, since they may accommodate the adsorption of only one modeled chain and not two (Scheme 4). As first approximation, the triblock consisting of polyMANa outer blocks is as efficient a dispersant as the diblock of ca. the same composition (17 vs 19% poly4VP). This conclusion is far from being general; e.g., when poly4VP is substituted by poly((dimethylamino)ethyl methacrylate)49 and poly(aminoalkyl methacrylate),59 respectively, triblock copolymers with external polyelectrolyte blocks are commonly less efficient than the diblock equivalents in stabilizing dispersions. In the case of the reverse triblock copolymer, Tout, the minimum in viscosity is observed at a copolymer concentration comparable to D19. Nevertheless, the main rheological characteristics are about 2 times higher compared to those of dispersions stabilized by the diblock, although still better than in the case of the PMANa dispersant (Figures 2-6). Anchoring onto the pigment surface can occur either through one of the two outer P4VP blocks or through these two blocks simultaneously. In the latter case, the PMANa block will have to bend, resulting in an additional entropic penalty that has to be compensated (59) Creutz, S.; Je´roˆme, R. In preparation.

Creutz and Je´ roˆ me Scheme 5

by the adsorption enthalpy. If only one P4VP block is adsorbed, the second one will have to protrude into the bulk solution. In the case of micelles formed by this type of triblock copolymers, it has been predicted that the chains tend to bend in order to avoid this very unfavorable contact with the aqueous solution.60 Therefore, loops are expected to be formed upon adsorption, which has been confirmed experimentally.61-63 Nevertheless, the adsorption process is concentration dependent. At low concentrations, where there are plenty of bare spots on the particle surface, the adsorption of one of the outer blocks will be followed by the adsorption of the second one and the simultaneous chain bending. At high concentrations, the second outer block will no longer have the opportunity to adsorb, because of the lack of adjacent bare spots on the surface.61-63 This block, which then protrudes into the bulk solution, could either adsorb onto a second particle or form a mixed network with the Tout micelles (Scheme 5). In the former case, much higher rheological characteristics, typical of a flocculated dispersion, would be expected.64 In the latter case, a gellike structure, characterized by a very high viscosity at low shear rate, would appear due to the tridimensional network formed by the linked micelles and particles.63 Both cases are minor if not absent, since no such dramatic increase in the rheological characteristics has been observed. Therefore, the adsorption seems to occur mainly through loop formation even at high concentrations. This conclusion would be consistent with the adsorption of unimers55-58 and the low concentrations of “adsorbable” chains in relation to a low cmc value. Compared to D19, the copolymer Tout appears to be a less efficient dispersant. Although most of the chains are thought to be adsorbed through both outer blocks, the adsorption of a few chains through a single outer block might occur and be responsible for some particle bridging and thus a poorer dispersion. Moreover, the loops formed by the PMANa block result in a thinner protective barrier (less than ca. 6 nm) than in case of the nonfolding of that block on the surface. The plastic viscosity, which is smaller than for D19, tends to corroborate the hypothesis of some flocculation and/or particle bridging. Whatever the basic reason for a lower efficiency, this type of triblock structure is not recommended as the dispersant. As a result of the bending of the PMANa block, desorption of one of the two adsorbed P4VP blocks might occur under high shear and lead to shear-induced flocculation. In the present case, the copolymer anchoring has proved to be strong enough to preserve the stability under shear (from 0.54 to 1312 s-1). Within the limits of validity of the model of Dan et al., Tout adsorption has been scaled by using an equivalent diblock copolymer, made of one P4VP outer block and half (60) Balsara, N. P.; Tirrell, M.; Lodge, T. P. Macromolecules 1991, 24, 1975. (61) Dorgan, J. R.; Stamm, M.; Toprakcioglu, C.; Je´roˆme, R.; Fetters, L. J. Macromolecules 1993, 26, 5321. (62) Dorgan, J. R.; Stamm, M.; Toprakcioglu, C. Polymer 1993, 34, 1554. (63) Jenkins, R. D.; Durali, M.; Silebi, C. A.; El-Aasser, M. S. J. Colloid Interface Sc. 1992, 154, 502. (64) Otsubo, Y. Langmuir 1994, 10, 1018.


Langmuir, Vol. 15, No. 21, 1999 7153

Scheme 6

the polyMANa central block of the triblock (NA ) 10 and NP ) 26.5). A minimum in viscosity at a ca. 2.5 times higher concentration than D19 is predicted. This overshoot qualitatively agrees with the formation of loops, which prevents some of the anchoring sites from being available (Scheme 6).62 3.4. Influence of the Copolymer Blockiness. All the block copolymers studied up to now have been synthesized by anionic polymerization, which is not the most straightforward technique, in contrast to radical polymerization. The main drawback of radical polyadditions is their inability to lead to blocky comonomer distributions. How far is the blockiness of the copolymer required for imparting to it dispersant properties? To answer this question, a copolymer (Rd) has been synthesized by radical polymerization of the comonomers; the mixture was slowly added into the reaction medium held at 65 °C. This technique allows the mixture to reach high comonomer conversion, while keeping the comonomer feed ratio constant.48 Although the comonomer reactivity ratios are unknown, we may assume a more or less random distribution of the 4VP and tBMA units along the chain, by analogy with the MMA/4VP pair (rMMA ) 0.54 and r4VP ) 1.05)48 and the MMA/tBMA pair (rMMA ) 0.96 and rtBMA ) 1.35),65 respectively. As mentioned above, the pKa values of 4VP and MANa and thus their protonation degrees37-39 depend on their distribution in the copolymer, as a result of the nearestneighbor and long-range electrostatic interactions. In the case of poly(2VP-r-AA),38 the random distribution of 2VP along the chain makes it more basic than in the homopolymer. The same situation should also prevail in the case of the poly(4VP-r-MANa) copolymer. Nevertheless, since the isoelectric point of the poly(2VP-b-MAA)37 copolymer of a comparable composition lies in the range of pH ) 2-3 and since the pyridinium ions are more acidic than the carboxylic acid groups,39 the P4VP should still be deprotonated at pH ) 9 and, thus, no more watersoluble. Since the random distribution also increases the acidity of methacrylic acid, a higher degree of neutralization should be expected. Therefore, the PMANa units should be adsorbed to a lesser extent in favor of the randomly distributed 4VP monomer units with formation of a loop-train conformation, responsible for a thin stabilizing layer. As expected, the dispersion stability is poor. The poly(4VP-r-MANa) copolymer (Rd) provides the dispersion with quite comparable properties to those of PMANa, except for a shift of the minimum in viscosity toward a lower concentration (Figures 6-11). Depletion flocculation is also observed at higher concentrations, as shown by the pronounced increase in viscosity, hysteresis, yield stress, and shear-thinning factor. Indeed, since the plastic viscosity is low at high concentrations (Figure 11), the increase in the main rheological characteristics may not be accounted for by the water thickening by an excess of dispersant. (65) Yuki, H.; Okamoto, Y.; Shimada, Y.; Ohta, K.; Hatada, K. J. Polym. Sci., Polym. Chem. Ed. 1979, 17, 1215.

Figure 7. Plot of the viscosity at 0.54 s-1 vs the dispersant/ pigment wt/wt %: (b) PMANa, (9) D19, ([) Rd, (2) Tp, (1) DEO, (4) D2VP.

Figure 8. Plot of the yield value vs the dispersant/pigment wt/wt %: (b) PMANa, (9) D19, ([) Rd, (2) Tp, (1) DEO, (4) D2VP.

Figure 9. Plot of the hysteresis area vs the dispersant/pigment wt/wt %: (b) PMANa, (9) D19, ([) Rd, (2) Tp, (1) DEO, (4) D2VP.

Both the homoPMANa and the poly(4VP-r-MANa) copolymer are expected to adsorb in a comparable looptail conformation, in qualitative agreement with the comparable stability imparted to the dispersions. The observation of the minimum in viscosity at a lower concentration for Rd indicates a stronger affinity for the pigment surface, in relation to the higher adsorption

7154


Figure 10. Plot of the shear thinning factor vs the dispersant/ pigment wt/wt %: (b) PMANa, (9) D19, ([) Rd, (2) Tp, (1) DEO, (4) D2VP.

Figure 11. Plot of the plastic viscosity vs the dispersant/ pigment wt/wt %: (b) PMANa, (9) D19, ([) Rd, (2) Tp, (1) DEO, (4) D2VP.

enthalpy of P4VP. In the case of PMANa, the lower affinity for the surface has to be compensated by a higher concentration so that the particle coverage is equivalent. In conclusion, the random distribution of the comonomers is not able to build up an appropriate electrosteric barrier and, thus, to avoid flocculation. Now the question is raised to know whether a pure diblock copolymer is a stringent requirement or if some blockiness in the comonomer distribution is acceptable. In this respect, 4VP and tBMA have been mixed together and then anionically copolymerized. According to the reactivity ratios, rtBMA ) 13.3 ( 6.6 and r4VP ) 0.47 ( 0.3,15 the hydrolyzed copolymer may be viewed as a tapered copolymer consisting of two main PMANa and P4VP blocks connected to each other through a transient block of regularly changing composition. This tapered copolymer (Tp) thus has a molecular structure intermediate between a random and a pure diblock copolymer. It has the advantage of a more straightforward synthesis than a pure diblock copolymer. Compared to D19, the minimum in the rheological characteristics is shifted toward higher concentrations, ca. 1 1/2 times higher (Figures 6-11). Excepted at high dispersant contents, these two copolymers provide the dispersions with a comparable stability. A plateau is observed in the case of Tp as soon as the minimum is


reached. This behavior results from a lower thickening of the aqueous solution, as shown by the plastic viscosity (Figure 11). The absence of any depletion flocculation must be pointed out. The high stability imparted by Tp makes it comparable to diblock copolymers, which confirms the blockiness of the comonomer distribution. The most surprising observation is the higher adsorption density of Tp on the pigment surface compared to that of D19. According to the model of Dan et al.,50 which is, however, only valid for diblock copolymers, a molecular weight ca. 2 times smaller for Tp would fit the experimental surface density data. Although, the molecular weights measured by SEC for Tp (7000) and D19 (7300) are only apparent, the blockiness of the comonomer distribution in the two copolymers and the close agreement in the apparent molecular weights make it unreasonable to consider that a difference in molecular weight might account for the complete difference observed in the adsorption density of the dispersants. On the other hand, if the central block of the tapered copolymer blockswhich is of a continuously changing compositionsis adsorbed, the predicted surface density could then fit with the experimental data.66 Most likely, the higher surface density of the tapered copolymer compared to the pure diblock would result from some differences in the molecular weight and from the partial or complete adsorption of the tapered block. 3.5. Influence of the Stabilizing Block. The outstanding properties of diblock copolymers containing a polyelectrolyte block have been compared to the stabilizing properties of a nonionic equivalent, e.g., poly(ethylene oxide) containing an amphiphilic diblock. This comparison is worth being made, since most oligo- and polymeric commercial dispersants, such as nonylphenylpoly(ethylene oxide) or poly(ethylene oxide-block-propylene oxideblock-ethylene oxide) copolymers, are typically nonionic. To our best knowledge, such a comparison of polymeric dispersants, which only differ in their stabilizing block, has not been reported yet. The anionically synthesized poly(4-vinylpyridine-blockethylene oxide) copolymer, DEO, has been compared to D19 (Figures 7-11). A poorly stable dispersion is observed, although the surface density is ca. 3 times higher. It is noteworthy that the plastic viscosity is low. This confirms that at high shear rates, flocs are deaggregated and that the much higher viscosity is not due to the water thickening by dispersant excess. Because of the instability of the dispersion added with 0.5% dispersant, only the viscosity at low shear has been measured. Once again, both P4VP and poly(ethylene oxide) can a priori anchor onto the surface. At the very early adsorption stage, the two blocks are expected to adsorb onto the bare surface until this surface is completely covered. In a second step, additional diblock would adsorb as a result of the displacement of the PEO block of previously adsorbed diblock by their P4VP block. This behavior has been observed when a poly(styrene-block-ethylene oxide) copolymer is adsorbed onto silica.67 Furthermore, pyridine has proved to be an efficient displacer of PEO (Mw ) 246 000) originally adsorbed onto alumina.29 The PEO (66) If we assume that the ratio between the molecular weights measured by SEC (7000/7300 Tp/D19) is similar for the real molecular weight determined by combining SEC and NMR (14 800 for D19), the real molecular weight of the tapered copolymer would be 14 200 (N4VP ) 22 and NMANa ) 84). The observed surface density would be predicted, if both the poly4VP and the tapered blocks are adsorbed and if the latter contains roughly 14 MANa units (NA ) 36 and Np ) 70). (67) Tripp, C. P.; Hair, M. L. Langmuir 1996, 12, 3952.


block of the poly(4VP-b-EO) copolymer would then protrude into the bulk solution with the formation of a steric barrier. The rheological characteristics of the dispersions stabilized by the poly(4VP-b-EO) copolymer would also agree with a stable dispersion close to the maximum packing of the dispersed particles.11,26 Indeed, the interpenetration (interdigitation) of the stabilizing layers could account for a high viscosity at low shear rates and a high yield value as well. The compression of these layers under shear can then decrease the effective volume fraction and, thus, the viscosity. In parallel, a high shear-thinning factor and a large hysteresis should be observed. Nevertheless, the hypothesis of the PEO block interpenetration is very unlikely. Luckham et al.11 have measured a layer thickness of ca. 10 nm for PEO-based dispersants, consisting of ca. 145 EO units, adsorbed onto carbon black. Therefore, in the most optimistic hypothesis, a ca. 10-nm layer thickness is expected for the poly(4VP-b-EO) copolymer (DEO), consisting of ca. 136 EO units. In contrast, despite a 2 times thicker stabilizing layer, the D19 copolymer exhibits a stable dispersion with lower rheological characteristics. Interpenetration of the PEO chains is thus most unlikely, and the high rheological characteristics have to reflect a poor stability. It can be concluded that the TiO2 dispersion under consideration in this study is poorly stabilized by a nonionic diblock copolymer. As a comparison, a layer thickness as small as 6 nm has been shown to be sufficient in the case of ionic diblock copolymers.49 This comparison highlights the long-range electrostatic repulsion promoted by these copolymers, despite a much lower surface density. As expected, the electrostatic repulsion also influences strongly the surface density. According to the Dan et al. model,50 the ionic copolymer equivalent to DEO should exhibit a surface density ca. 40% smaller than D19, thus almost 5 times smaller than the actual value. If we refer to the results published by de Silva et al.20,68 and to the results of this study, the adsorption of a large amount of dispersant is not necessarily the signature of a highly stable dispersion, as is usually accepted. Indeed, the stabilizing efficiency of a dispersant may not be merely anticipated from adsorption measurements. This restriction should be as severe as the solid content is high, since stabilization thus requires a strong repulsion. In case of high solid contents, the dispersion rheology is the most appropriate basis for discussing the dispersion stabilization. 3.6. Influence of the Anchoring Block. As mentioned earlier, 4VP has been preferred to 2VP not only for more favorable polymerization conditions but also because of the better accessibility of the nitrogen atom, which was thought to give rise to more efficient dispersants. For the sake of confirmation, a poly(2VP-b-MANa) copolymer, D2VP, has been anionically synthesized. The rheological characteristics of the TiO2 dispersion at the minimum in viscosity remain comparable when D2VP is substituted for D19. This minimum is, however, observed at a concentration ca. 1 1/2 times higher in case of D2VP (Figures 7-11). At higher concentrations, the increase in the rheological properties is smoother as a result of the less important thickening of the bulk solution, as exemplified by the plastic viscosity (Figure 11). This effect should reflect the lower molecular weight of D2VP. Therefore, P2VP is anchored onto TiO2 strongly enough to stabilize the dispersion and prevent any depletion flocculation from occurring. (68) De Silva, G. P. H. L.; Luckham, P. F.; Tadros, Th. F. Colloids Surf. 1990, 50, 251.

Langmuir, Vol. 15, No. 21, 1999 7155

If we assume that the surface tension of the micellar core-corona interface is similar for 4VP and 2VP, the minimum in viscosity observed at ca. 0.3% is predicted by the scaling model of Dan et al.50 Very surprisingly, the position of the nitrogen atom does not influence significantly the adsorption and the efficiency of poly(vinylpyridine) copolymers. To confirm that the polyaminated block has an important role, polystyrene (PSt) has been substituted for poly(vinylpyridine). So the poly(St-b-MANa) copolymer, DSt, has been anionically synthesized and tested as the dispersant up to a 1% concentration. At higher concentrations, i.e., ca. 1.25%, solutions of this diblock form a gel, which is currently studied by SAXS and SANS.69,70 Whatever the concentration of DSt, no stable dispersion is formed, although the scaling model of Dan et al.50 predicts a minimum in viscosity at ca. 0.27% in the case of the same adsorption enthalpy and surface tension as for poly4VP. The conclusion that the phenyl rings of PSt have no marked tendency to adsorb onto the pigment surface is consistent with an adsorption enthalpy of PSt onto alumina, which is at least 4 times smaller compared to polar polymers, such as poly(methacrylates), and even 8 times smaller than PEO.29 Although these measurements were conducted in carbon tetrachloride, this general conclusion should be valid in water. Similarly, PSt is easily displaced from alumina by polar solvents, such as dioxane, alkyl acetate,29 etc. Therefore, there is no reason for water not to displace PSt from the pigment surface. This poor affinity of PSt for the TiO2 surface can explain why a minimum in the dispersion viscosity is not observed at copolymer contents smaller than 1.25%. As shown by SEC,71 sedimentation velocity,72 and fluorescence,73 poly(St-b-MANa) forms very stable micelles in water. Indeed, at room temperature, no unimer-micelle exchange is observed.73 This observation is consistent with that of Cao et al.,51 who have shown that in water, poly(St-b-MAC) micelles are stable and do not adsorb onto a PSt surface because of the “glassy” character of the micellar core. According to these authors, adsorption may only occur in the presence of an organic solvent, which softens the micellar core to allow some contacts with the surface. Moreover, as a result of low cmc (∼5 mg/L), only a very limited amount of unimers is available. So TiO2 does not appear to compete efficiently enough with the micelles for being stabilized by a layer of adsorbed unimers. Would micelles adsorb onto TiO2, this mechanism would not contribute to the dispersion stability. This conclusion is not surprising, since the adsorption would be promoted by the polyMANa corona, and as a result of the sea urchinlike shape of the micelles, only the outer monomer units can be in contact with the surface, among which only a few units are in the acid form and prone to anchoring. The only way for the micelles to be strongly adsorbed would suppose an important flattening of the polyMANa brush so as to increase the TiO2/polyMANa contacts. This hypothesis is, however, very unlikely as theoretically predicted by Johner et al.57 The flattening of the micelle corona may be viewed as the extreme formation of a copolymer double layer. Then a copolymer concentration (69) Gaspard, J.P.; Creutz, S.; Bouchat, Ph.; Je´roˆme, R.; Cohen Stuart, M. A. Phys. B 1997, 234, 268. (70) Bouchat, Ph. Master Thesis, University of Lie`ge, Belgium, 1996. (71) Creutz, S.; Je´roˆme, R. In preparation. (72) (a) Munk, P.; Ramireddy, C.; Tian, M.; Webber, S. E.; Prochazka, K; Tuzar, Z. Makromol. Chem., Macromol. Symp. 1992, 58, 195. (b) Tian, M.; Qin, A.; Ramireddy, C.; Webber, S. E.; Munk, P.; Tuzar, Z.; Prochazka, K. Langmuir 1993, 9, 1741. (73) van Stam, J.; Creutz, S.; De Schryver, F. C.; Je´roˆme, R. Submitted for publication.

7156


ca. 2 times higher than in the case of homopolyMANa, i.e., 0.6%, should be predicted for reaching a maximum in stability. This situation is not observed even at a concentration 3 times higher than for polyMANa. According to the results of this study, only the adsorption of unimers contributes to the dispersion stability, the micelles acting as reservoirs. 4. Conclusion Block copolymers containing a polyelectrolyte block have been observed to be efficient dispersants for high solid dispersions. As a result of their amphiphilic structure, they become anchored onto the pigment surface as strongly as the adsorption enthalpy of the hydrophobic block is high. In this respect, poly4VP and poly2VP blocks have proved to be strongly anchoring blocks toward TiO2. They accordingly stabilize a polyelectrolyte brush, which is actually an electrosteric barrier against coalescence. In contrast to homopolymer dispersants, no depletion flocculation then occurs. As far as the copolymer architecture, i.e., diblock vs triblock copolymers, is concerned, the stabilization efficiency of poly(4VP-b-MANa) and poly(MANa-b-4VP-bMANa) copolymers is quite comparable. Nevertheless, this conclusion should not be taken for granted, since contradictory results have been reported in the case of copolymers of a different chemical composition.49,59 As a rule, diblock copolymers are at least as efficient assif not superior tosthe triblock equivalents, in which the outer blocks are stabilizing blocks. This might indicate that substitution of poly(EO-b-PO), where PO stands for propylene oxide, for Pluronics, i.e., the commercially available triblock dispersants poly(EO-b-PO-b-EO), could be advantageous. Although triblock copolymers with anchoring outer blocks can stabilize dispersions, they should be disregarded, since the bridging of two particles cannot be ruled out. Moreover, the simultaneous adsorption of the two outer blocks on the same particle has to result in the bending of the inner stabilizing block so that the thickness of the stabilizing layer is decreased and the (74) Creutz, S.; Je´roˆme, R. In preparation.


dispersion stability as well. Finally, an overshoot in concentration, which is a common practice in industry, is favorable to the adsorption of only one anchoring block per particle and, thus, to the dispersion flocculation. This study has also emphasized that the blockiness of the comonomer distribution in a binary copolymer has a decisive effect on the dispersant efficiency. Since most of the commercially available copolymers have no pronounced blockiness, the “controlled” radical polymerization is now raising optimistic prospects for the synthesis of new polymeric dispersants of an attractive costperformance balance. The higher efficiency of the electrosteric stabilization compared to the steric stabilization has also been highlighted in this study. Therefore, the commonly used PEObased dispersants, which are of a limited efficiency in stabilizing high solid dispersions, might be advantageously replaced by ionic copolymers. Finally, the stability of the micelles formed by polymeric dispersants is a key parameter in the competition of the solid surface toward micelles for being protected and thus stabilized by unimers that appear to be the actual stabilizing agent. In addition to block copolymers of vinylpyridine, poly((dimethylamino)ethyl methacrylate),49 poly(aminoalkyl methacrylate),59 and poly(methyl methacrylate)74 containing copolymers have also been studied, and the results will be reported in forthcoming papers. Acknowledgment. We are grateful to Akzo Nobel Coatings Technology Center, Pigments and Pigment Dispersing Agents, for financial support. Dr. J. Akkerman (Akzo Nobel Coatings Technology Center), Dr. H. van Haak (Akzo Nobel Central Research), and Dr. A. Toussaint from the Belgian Coatings Research Institute (CORI) are thanked for fruitful discussions. We also thank the “Services Fe´de´raux des Affaires Scientifiques, Techniques et Culturelles” for general support in the frame of the “Poˆles d’Attraction Interuniversitaires 4/11: Supramolecular Chemistry and Catalysis”. LA980951E

Effectiveness of Poly(vinylpyridine) Block Copolymers as Stabilizers of

Recommend Documents