Langmuir 2008, 24, 9989-9996
9989
Dispersant Adsorption and Viscoelasticity of Alumina Suspensions Measured by Quartz Crystal Microbalance with Dissipation Monitoring and in Situ Dynamic Rheology Lisa Palmqvist*,† and Krister Holmberg‡ Swerea IVF AB, Box 104, SE-431 22 Mo¨lndal, Sweden, Applied Surface Chemistry, Department of Chemical and Biological Engineering, Chalmers UniVersity of Technology, SE-412 96 Go¨teborg, Sweden ReceiVed March 7, 2008. ReVised Manuscript ReceiVed June 11, 2008 Adsorption behavior and water content of adsorbed layers of four dispersants for aqueous ceramic processing were studied by quartz crystal microbalance with dissipation monitoring (QCM-D) on alumina surfaces. The dispersants were a poly(acrylic acid), a lignosulfonate, and two hydrophilic comb copolymers with nonionic polyoxyethylene chains of different molecular weights. A Voigt model was applied to analyze the viscoelastic behavior of the adsorbed dispersant layers. The results from QCM-D were compared with viscoelastic properties determined by in situ dynamic rheology measurements of highly concentrated alumina suspensions during slip casting. The QCM-D results showed that both the poly(acrylic acid) and the lignosulfonate adsorbed in low amounts and in a flat conformation, which generated thin, highly rigid layers less than 1 nm thick. The water content of these layers was found to be around 30% for the lignosulfonate and 35% for the poly(acrylic acid). High casting rate and strength in terms of storage modulus were observed in the final consolidate of the suspensions with the two polyelectrolytes. In contrast, the high molecular weight comb copolymer adsorbed in a less elastic layer with a thickness of about 6 nm, which is enough to provide steric stabilization. The viscous behavior of this layer was attributed to high water content, which was calculated to be around 90%. Such a water-rich layer gives a lubrication effect, which allows for reorientation of particles during the consolidation process, resulting in a high final strength of the ceramic material. During consolidation, the suspension showed a slow casting rate, most likely due to rearrangement facilitated by the lubricating layer. The short-chain comb copolymer adsorbed in a 1.5 nm thick, rigid layer and gave low final strength to the consolidated suspension. It is likely that the poor consolidation behavior is caused by flocculation due to insufficient stabilization of the dispersion.
Introduction In aqueous processing systems for ceramic materials, dispersants are essential for obtaining stable suspensions and for successful forming of components with high strengths and homogeneity. The role of the dispersant is to create a welldispersed ceramic suspension through proper repulsion between the particles. The dispersants are often polyelectrolytes, which give rise to electrostatic stabilization, or comb-type copolymers with an ionizable anchoring backbone and grafted noncharged chains, which provide steric stabilization. The stabilization efficiency of poly(acrylic acid)s as dispersants for aqueous ceramic systems is well-established.1-4 Understanding the interactions in the ceramic particle-dispersant system is important because * Corresponding author. Tel: +46 317066000. Fax: +46 31276130. E-mail:
[email protected]. † Swerea IVF AB. ‡ Chalmers University of Technology. (1) Briscoe, B. J.; Khan, A. U.; Luckham, P. F. J. Eur. Ceram. Soc. 1998, 18, 2141–2147. (2) Tari, G.; Ferreira, J. M. F.; Lyckfeldt, O. J. Eur. Ceram. Soc. 1998, 18, 479–486. (3) Baklouti, S.; Pagnoux, C.; Chartier, T.; Baumard, J. F. J. Eur. Ceram. Soc. 1997, 17, 1387–1392. (4) Cesarano, J.; Aksay, I. A. J. Am. Ceram. Soc. 1988, 71, 1062–1067. (5) Guldberg-Pedersen, H.; Bergstro¨m, L. J. Am. Ceram. Soc. 1999, 82, 1137– 1145. (6) Laarz, E.; Meurk, A.; Yanez, J. A.; Bergstro¨m, L. J. Am. Ceram. Soc. 2001, 84, 1675–1682. (7) Palmqvist, L.; Lyckfeldt, O.; Carlstro¨m, E.; Davoust, P.; Kauppi, A.; Holmberg, K. Colloids Surf. A 2006, 274, 100–109. (8) Greenwood, R.; Bergstro¨m, L. J. Eur. Ceram. Soc. 1997, 17, 537–548. (9) Pagnoux, C.; Serantoni, M.; Laucornet, R.; Chartier, T.; Baumard, J.-F. J. Eur. Ceram. Soc. 1999, 19, 1935–1948. (10) Palmqvist, L. M.; Lange, F. F.; Sigmund, W.; Sindel, J. J. Am. Ceram. Soc. 2000, 83, 1585–1591.
these govern the adsorption behavior and the conformation of the polymer on the surface, as well as the charge distribution on the particle. Adsorption of dispersants in aqueous ceramic systems has been studied by various techniques, such as AFM5-7 and ζ-potential measurements.2-4,7-10 In addition to the above-mentioned characteristics, the viscoelastic properties of a ceramic particle-dispersant system may influence the tribological properties and, thus, the lubricity of the coated ceramic particles. This, in turn, may be of importance for the rearrangement and stress release properties during the consolidation sequence of a ceramic body. The lubricating effect of a polymer layer is largely dependent on the water content of the layer.11 It is therefore of interest to gain more understanding of the nature of the adsorbed layer in terms of water content, lubricity, and viscoelasticity. A technique for studying the adsorption of macromolecules on solid surfaces is the quartz crystal microbalance with dissipation monitoring (QCM-D). This technique can be used to monitor the formation of a dispersant layer at a surface that resembles the surface of a ceramic particle. Moreover, in contrast to other methods for determining adsorption, such as ellipsometry and surface plasmon resonance, the adsorbed mass detected by QCM-D includes that of bound solvent. This opens the possibility to relate viscoelastic film properties to water content. By applying an analytical model to fit the dissipation data, it is possible to calculate the viscoelastic properties of the dispersant layer adsorbed at the ceramic surface. QCM was first used in vacuum (11) Mu¨ller, M. T.; Yan, X.; Lee, S.; Perry, S. S.; Spencer, N. D. Macromolecules 2005, 38, 3861–3866.
10.1021/la800719u CCC: $40.75 2008 American Chemical Society Published on Web 08/16/2008
9990 Langmuir, Vol. 24, No. 18, 2008
PalmqVist and Holmberg
Figure 1. Schematic representation of the QCM-D crystal and the dynamic rheology setup.
or gaseous media as an ultrasensitive weighing device,12 and the technique was later extended for use in liquid media13,14 and proven very useful as a biological sensor.15 The major part of the reports deals with biopolymers.16-23 QCM-D studies have also been done on polyelectrolytes24-28 and copolymers,29-32 as well as on starch.33,34 The majority of the studies deal with hydrophobic gold surfaces but also hydrophilic surfaces such as silica and titania have been used.11,26,28,29,34-36 To our knowledge, only one QCM-D study with alumina as substrate has been reported, and that concerned adsorption of dextran.37 For determining the viscoelastic properties of adsorbed layers, models have been successfully developed for QCM-D38,39 and applied to various systems.22,30,40,41 The use of the QCM-D technique to study dispersants at ceramic surfaces is new. In the present study, the adsorption behavior, (12) Czanderna, A. W.; Lu, C. In Applications of Piezoelectric Quartz Crystal Microbalances; Czanderna, A. W., Lu, C., Eds.; Elsevier: Amsterdam, 1984; pp 1-18. (13) Rodahl, M.; Ho¨o¨k, F.; Kasemo, B. Anal. Chem. 1996, 68, 2219–2227. (14) Rodahl, M; Ho¨o¨k, F.; Krozer, A.; Brzezinski, P.; Kasemo, B. ReV. Sci. Instrum. 1995, 66, 3924–3930. (15) Ebersole, R. C.; Ward, M. D. J. Am. Chem. Soc. 1988, 110, 8623–8628. (16) Ho¨o¨k, F.; Kasemo, B.; Nylander, T.; Fant, C.; Sott, K.; Elwing, H. Anal. Chem. 2001, 73, 5796–5804. (17) Ho¨o¨k, F.; Rodahl, M.; Brzezinski, P.; Kasemo, B. Langmuir 1998, 14, 729–734. (18) Ho¨o¨k, F.; Rodahl, M.; Kasemo, B.; Brzezinski, P. Proc. Natl. Acad. Sci. U.S.A. 1998, 95, 12271–12276. (19) Molino, P. J.; Hodson, O. M.; Quinn, J. F.; Wetherbee, R. Biomacromolecules 2006, 7, 3276–3282. (20) Patel, A. R.; Frank, C. W. Langmuir 2006, 22, 7587–7599. (21) Reimhult, E.; Larsson, C.; Kasemo, B.; Ho¨o¨k, F. Anal. Chem. 2004, 76, 7211–7720. (22) Stengel, G.; Ho¨o¨k, F.; Knoll, W. Anal. Chem. 2005, 77, 3709–3714. (23) Wang, X.; Ho, C.-T.; Huang, Q. J. Agric. Food Chem. 2007, 55, 4987– 4992. (24) Azzaroni, O.; Moya, S.; Farhan, T.; Brown, A. A.; Huck, W. T. S. Macromolecules 2005, 38, 10192–10199. (25) Moya, S. E.; Brown, A. A.; Azzaroni, O; Huck, W. T. S. Macromol. Rapid Commun. 2005, 26, 1117–1121. (26) Notley, S. M.; Eriksson, M.; Wågberg, L. J. Colloid Interface Sci. 2005, 292, 29–37. (27) Plunkett, M. A.; Claesson, P. M.; Ernstsson, M.; Rutland, M. W. Langmuir 2003, 19, 4673–4681. (28) Wågberg, L.; Pettersson, G.; Notley, S. J. Colloid Interface Sci. 2004, 274, 480–488. (29) Domack, A.; Prucker, O.; Ru¨he, J.; Johannsmann, D. Phys. ReV. E 1997, 56, 680–689. (30) Dutta, A. K.; Belfort, G. Langmuir 2007, 23, 3088–3094. (31) Nunalee, F. N.; Shuli, K. R.; Lee, B. P.; Messersmith, P. B. Anal. Chem. 2006, 78, 1158–1166. (32) Yan, Y.; Zhou, X.; Ji, J.; Yan, L.; Zhang, G. J. Phys. Chem. B 2006, 110, 21055–21059. (33) Merta, J.; Tammelin, T.; Stenius, P. Colloids Surf. A 2004, 250, 103–114. (34) Tammelin, T.; Merta, J.; Johansson, L.-S.; Stenius, P. Langmuir 2004, 20, 10900–10909. (35) Irwin, E. F.; Ho, J. E.; Kane, S. R.; Healy, K. E. Langmuir 2005, 21, 5529–5536. (36) Morand, R.; Noworyta, K.; Augustynski, J. Chem. Phys. Lett. 2002, 364, 244–250. (37) Kwon, K. D.; Green, H.; Bjo¨o¨rn, P.; Kubicki, J. D. EnViron. Sci. Technol. 2006, 40, 7739–7744. (38) Voinova, M. V.; Rodahl, M.; Jonson, M.; Kasemo, B. Phys. Scr. 1999, 59, 391–396. (39) Johannsmann, D.; Mathauer, K.; Wegner, G.; Knoll, W. Phys. ReV. B 1992, 46, 7808–7815. (40) Larsson, C.; Rodahl, M.; Ho¨o¨k, F. Anal. Chem. 2003, 75, 5080–5087. (41) Munro, J. C.; Frank, C. W. Macromolecules 2004, 37, 925–938.
water content, and viscoelastic properties of four different dispersants for aqueous ceramic suspensions were measured by the QCM-D technique on alumina substrates. Dispersants providing different types of stabilization, such as electrostatic, steric, or electrosteric, were chosen. Alumina is a very important ceramic material for electronics, structural applications, and biomedical implants. The results from the QCM-D measurements were compared to rheological measurements done in an oscillatory setup, as illustrated in Figure 1. The gradual build-up of elasticity and rigidness was monitored in situ during slip casting of concentrated alumina suspensions. Together with the dynamic rheology measurements of concentrated ceramic systems during consolidation, the QCM-D results provide information on how the adsorption, lubrication, and viscoelasticity of the dispersants may influence the casting properties in real ceramic systems.
Theory Quartz Crystal Microbalance with Dissipation (QCM-D). The QCM-D response is recorded as changes in f and D of an AT-cut quartz crystal at the resonance frequency, fn (n ) 1), and different overtones (n > 1). The QCM-D response is sensitive to any mass change of the quartz crystal, including the mass of bound water that is associated with the adsorbed solute. The density and thickness of the adsorbed layer will have an effect on the response, as well as on the viscoelastic properties of adsorbed species. For flat, uniform, and rigid films, the change in resonance frequency, ∆f ) f - f0, is directly proportional to the adsorbed mass, ∆m, as demonstrated by the Sauerbrey relationship17
∆m ) -
C∆f n
(1)
where C is the mass sensitivity constant of the crystals (here 17.7 ng/cm2 Hz at 5 MHz) and n is the overtone number (n ) 1, 3, 5, 7 in this work). However, in the case of macromolecule adsorption, there may be some deviation from the Sauerbrey relationship due to the viscoelastic nature of the adsorbed layer and to the presence of coupled water that is sensed as a mass uptake.26 Deposited polymer films, biomolecular films and biosensors are examples of species that often do not behave like elastic masses on the QCM surface and, thus, do not obey the Sauerbrey relationship.27,42 From the change in dissipation, information about the viscoelastic behavior of the layer may be gathered. The applied voltage is cut off and the decay of the amplitude of the crystal is measured as a function of time. From this, the dissipation factor, D, as defined by
D )
Edissipated 2πEstored
(2)
can be determined. For an adsorbed layer with high rigidity, no change in dissipation will be observed as a function of adsorption. (42) Marx, K. A. Biomacromolecules 2003, 4, 1099–1120.
Adsorption and Viscoelasticity of Alumina Suspensions
Langmuir, Vol. 24, No. 18, 2008 9991
However, for an adsorbed viscoelastic layer, the energy dissipated through the layer will increase. Therefore, by observing the change in dissipation, ∆D, a semiquantitative measure of the relative stiffness or conformation of an adsorbed layer may be determined. The adsorbed layer can be represented by a homogeneous viscoelastic film, which is characterized by a shear viscosity, ηf, and a shear modulus, µf, as well as a thickness and a density. By assuming an analytical Voigt model, the viscous and elastic components of the adsorbed layer can be determined from the complex shear modulus, G*, defined as
G* ) G ′ + iG ′′ ) µf + i2πfηf
(3)
where G′ is the storage modulus, G′′ the loss modulus, and f the frequency. Measurements at multiple harmonics are required to decompose the measured dissipation into viscous and elastic components. In this study, the third, fifth, and seventh overtones were used to calculate the data in a one-layer model according to the method of Voinova et al.38 Dynamic Rheology of Suspensions. In general, a material can be characterized by two types of rheological behavior.43 It can be described as a solid showing elastic behavior if the deformation is fully recovered after removal of the applied stress. The material can be described as a liquid having a viscous response if it flows under a very small stress. Ceramic suspensions with high solids loadings are examples of materials that show a combined viscoelastic response.44 Oscillatory measurements can provide important information regarding the rheological properties of suspensions that will be subjected to small strains during a forming operation. The viscoelastic properties are given by the complex modulus, G*, both in the QCM-D experiments and in dynamic rheology measurements. In the oscillatory experiments, the material, i.e. the ceramic slip in this case, is subjected to a continuously oscillating strain as the peak value of the stress, σ0, and the phase shift, δ, between the stress and the strain are recorded. The measurements are often performed in the linear viscoelastic (LVE) region, where the viscoelastic response is independent of strain, and it can be assumed that no irreversible changes of the structure of the suspensions take place. In the LVE region, the stress to strain ratio is given by the complex modulus, G*, which is independent of the magnitude of stress or strain since
G/ )
σ0 ) G′ + iG′′ γ0
(4)
where σ0 is the peak stress and γ0 is the strain. The storage modulus, G′, represents the in-phase stress-to-strain ratio and is a measure of the elastic properties. The loss modulus, G′′, represents the out-of-phase stress-to-strain ratio and is a measure of the viscous properties. The phase shift, δ, defined as
tan δ )
G′′ G′
(5)
will vary between 0° for a completely elastic body and 90° for a purely viscous material. It should be noted that, whereas QCM-D measurements are done at high frequencies, ∼107 Hz, with polymers adsorbing freely on a surface, slip casting measurements are normally carried out at 1 Hz in a high solids system, which involves particle interactions. Taken together the two methods can give useful (43) Bergstro¨m, L. In Surface and Colloid Chemistry in AdVanced Ceramics Processing; Pugh, R. J., Bergstro¨m, L., Eds.; Surfactant Science Series 51; Marcel Dekker: New York, 1994; Chapter 5. (44) Lewis, J. A. J. Am. Ceram. Soc. 2000, 83, 2341–2359.
knowledgeabouthowdispersantadsorptionanddispersant-surface interactions at the molecular level play a role in ceramic slip casting.
Materials and Methods Chemicals. The dispersants were commercial products provided by the manufacturers. They were used as received. The poly(acrylic acid) (Dolapix PC21, Zschimmer & Schwartz GmbH & Co.) is linear with NH4+ as counterion and has a molecular weight of about 4500. The lignosulfonate (Vanisperse, Borregard Lignotech) is based on highly refined sodium oxylignin and has a molecular weight of about 6500. It contains sulfonate and carboxylate groups with Na+ as counterion. The long-chain comb copolymer (Conpac 30, Perstorp AB) consists of a polycarboxylate backbone modified with sulfonate groups and grafted polyoxyethylene (PEO) chains. Each PEO chain has 60-80 oxyethylene units. The total molecular weight of Conpac 30 is about 100 000. The short-chain comb copolymer has, according to the supplier (Hypermer KD7, Uniqema Chemie BV), much shorter PEO chains and a total molecular weight of only about 3000. The relative amount of oxyethylene in this molecule was determined by 1H NMR, using a 500 Mz Varian Unity Inova spectrometer (see Results and Discussion). Alumina powder (AKP30, Sumitomo Chemicals Ltd.) was used as received for all rheology experiments. The powder has a mean particle diameter of 270 nm and a specific surface area of 6.8 m2/g. Analytical grade NH3 and HCl were used for all pH adjustments. QCM-D Measurements and Modeling. The instrument used was a commercial quartz crystal microbalance with dissipation monitoring (D300, Q-Sense AB, Go¨teborg, Sweden), as described in detail elsewhere.14 Circular AT-cut quartz crystals (from Q-Sense AB) with a fundamental frequency of 5 MHz were coated with gold and an alumina top layer. The alumina coating was confirmed by ESCA. Prior to measuring, the crystal was cleaned by submerging it into an aqueous solution of a nonionic surfactant (Sabopal LM7). It was then rinsed in deionized water followed by ethanol and put in an ultrasonic bath for 10 min. It was rinsed again in deionized water, dried by flowing nitrogen gas, and then exposed to UV/ozone for 10 min. The whole procedure was then repeated, after which the crystals were put in sterile containers prior to use. All experiments were carried out at pH 8.5 in deionized water with 0.01 M KCl added to avoid fluctuations in ionic strength. The temperature of the measurement chamber was kept at 25 ( 0.1 °C, and all solutions were adjusted to 25 °C in a heating bath prior to injection. In a typical experiment, a baseline was first established for a 0.01 M KCl solution of pH 8.5. At t ) 3 min, 1 mL of a solution of the polymeric dispersant with a concentration of 26 mg/mL was injected into the crystal chamber. Adsorption was allowed to take place during 7 min, after which another 1 mL injection was done at t ) 10 min. The changes in frequency and dissipation upon adsorption were monitored throughout the experiment. At t ) 20 min, when no further adsorption could be seen, rinsing of the chamber was done by injecting KCl solution. This was repeated at t ) 23 min and the desorption behavior was studied. Most experiments were discontinued at t ) 30 min. A viscoelastic Voigt model described in detail elsewhere16,38 was used to analyze the frequency and dissipation data. A commercial software program (Q-Tools, Q-Sense AB, Go¨teborg, Sweden) developed for QCM-D was used for the calculations. In the Voigt modeling, bulk density and bulk viscosity of 1.0 g/cm3 and 0.001 Pa s respectively were used for all fits. The layer density was approximated to 1.2 g/cm3. Viscosity Measurements and Dynamic in Situ Measurements. Samples for rheological evaluations were prepared by planetary milling for 1 h in containers with Si3N4 lining and Si3N4 balls and measured after overnight equilibration. The processing experiments were done at the natural pH of 9.5 ( 0.5 of the suspensions. Equilibrium apparent viscosity was obtained through controlled stress measurements using a Stress Tech CS rheometer (Rheologica Instruments) with cup and cylinder at different shear rates. The samples were presheared at 400 s-1, at rest for 60 s, and then
9992 Langmuir, Vol. 24, No. 18, 2008
PalmqVist and Holmberg
Figure 2. Normalized frequency shift and change in dissipation vs adsorption time for the (a) long-chain comb copolymer, (b) short-chain comb copolymer, (c) poly(acrylic acid), and (d) lignosulfonate. Black lines ) experimental, gray lines ) model fit.
equilibrated at each shear rate for 60 s followed by a measurement period of 60 s. Dynamic rheology measurements were done in situ during slip casting in a modified cone-and-plate setup, schematically shown in Figure 1. A plastic ring with diameter 3.4 cm and a height of 4 cm was filled up to 2 cm height with plaster of Paris, which was allowed to harden overnight. The slip to be investigated was poured into the ring, after which a 2° cone with diameter 3 cm was immediately lowered into it, leaving a 3 mm gap between the cone center and the hardened plaster of Paris. A stress of 0.07 Pa was applied at 1 Hz in an oscillating mode during the measurement. The viscoelastic behavior of the slip was recorded during the gradual build-up of the cast up to a final consolidation height of 3 mm.
Results and Discussion Adsorption and Dissipation Measured by QCM-D. The natural pH of an aqueous alumina suspension is usually 9-10, i.e. close to the isoelectric point (iep), which is around pH 9.4.10 Around the iep, the charge profile of the alumina surface is not ideal for adsorption of an anionic polymer. There are relatively few positive sites available for adsorption. Nevertheless, the adsorption of poly(acrylic acid)s and the consequential increase of negative charges at the surface are well-established1-5 and have also been reported for lignosulfonates.7,45,46 We have shown in a previous work that the ζ-potential at pH 9 increases from almost zero for a bare alumina surface to approximately -40 and -30 mV through adsorption of the poly(acrylic acid) and the lignosulfonate, respectively, that are used as dispersants in the present work.7 These dispersants were consequently referred (45) Byman-Fagerholm, H.; Mikkola, P.; Rosenholm, J. B; Lide´n, E.; Carlsson, R. J. Eur. Ceram. Soc. 1999, 19, 41–48. (46) Galassi, C.; Rastelli, E.; Roncari, E.; Ardizzone, S; Cattania, M. G. J. Mater. Res. 1995, 10, 339–344.
to as dispersants providing electrostatic stabilization. In the same study, the long-chain comb copolymer, also used in this work, gave very low ζ-potential over the whole pH range measured, and the stabilization observed was therefore attributed to pure steric repulsion, as was also confirmed by force measurements made by AFM. The short-chain comb copolymer, the fourth dispersant of the present work, was found to give a low, negative net charge of approximately -10 mV when adsorbed on alumina in water at pH 9.47 This dispersant was then assumed to act by both electrostatic and steric, or electrosteric, stabilization. In Figure 2a-d typical curves of changes in frequency and dissipation for three different overtones (n ) 3, 5, and 7) are given for the four dispersants. Data from the first overtone (5 MHz) is not included due to high scatter, as observed also by others.30 The figures include fits from the viscoelastic Voigt modeling. As seen, all of the fits follow the experimental data closely. For a nonrigid, viscoelastic layer, the Sauerbrey mass will be underestimated compared to a calculated mass based on the Voigt model. As a rule of thumb, the linear Sauerbrey relation is valid if the difference between the normalized curves for the frequency shifts is less than 5 Hz.35 Figure 2a shows data from the long-chain comb copolymer. The difference between the curves for the normalized frequency shifts is fairly large, and hence the Sauerbrey relationship is not valid. However, from the Voigt model a mass can be calculated, and the values are given in Table 1. Data for the short-chain comb copolymer are shown in Figure 2b. Clearly, this polymer adsorbs at the alumina surface, but in lower amount per surface area compared to the long-chain comb (47) Lyckfeldt, O.; Palmqvist, L.; Carlstro¨m, E. J. Eur. Ceram. Soc. Submitted.
Adsorption and Viscoelasticity of Alumina Suspensions
Langmuir, Vol. 24, No. 18, 2008 9993
Table 1. Mass Calculated from the Sauerbrey Relationship, ∆mf,Sauerbrey, and from the Voigt Model, ∆mVoigt; Layer Thickness Calculated from the Voigt Model, df,Voigt; Shear Viscosity, ηf; and Shear Modulus, µf, of the Adsorbed Layera ∆mf,Sauerbrey (ng/cm2) dispersant
n)1
n)2
n)3
∆mVoigt(ng/cm2)
df,Voigt (nm)
ηf (10-3 Pa s)
µf (105 Pa)
long copolymer short copolymer lignosulfonate poly(acrylic acid)
590 ( 28 164 ( 52 60 ( 15 34 ( 2
514 ( 28 152 ( 65 51 ( 27 29 ( 1
462 ( 28 146 ( 70 62 ( 19 30 ( 1
765 ( 35 186 ( 26 71 ( 12 57 ( 17
6.4 1.5 0.6 0.5
1.7
2.0
a
The values are averages from three experimental runs and include max and min variations between the measurements.
copolymer, which is expected considering the difference in molecular weight of the grafted chains. The total dissipation is less than 1 × 10-6, which indicates a high degree of elasticity.34 Such a layer can be regarded as rigid. Hence, the Sauerbrey equation should hold reasonably well for this polymer. Calculations of Sauerbrey mass and Voigt mass are given in Table 1. As seen, the Voigt mass is slightly higher than the Sauerbrey mass for this polymer. For layers that are almost fully elastic, some deviation between the Voigt mass and the Sauerbrey mass can be expected due to the simplicity of the Voigt model. The shifts in frequency and dissipation for the poly(acrylic acid) and the lignosulfonate are given in parts c and d of Figure 2, respectively. These two charged dispersants show similar adsorption behavior, with small changes in both frequency and dissipation. Hence, both the amount of adsorbed mass and the viscous contributions of the layers are small compared to, in particular, the long-chain comb copolymer. This can be explained by differences in adsorption characteristics and was also observed in previous work.27,34 At pH 8.5, both the poly(acrylic acid) and the lignosulfonate can be regarded to be completely dissociated.48 Highly charged linear polyelectrolytes are known to adsorb in a flat conformation at surfaces of opposite charge, as is the case here.49 A small change in dissipation, indicating high rigidity, has been observed earlier for PAA adsorbed at a silica surface.26 The lignosulfonate is highly branched and its characteristics with respect to rigidity of the adsorbed layer seem not to have been investigated before. Apparently, the adsorption behavior and viscoelasticity of the lignosulfonate layer is similar to the poly(acrylic acid) as seen in Table 1. Layer Structure. Figure 3 shows the change in dissipation vs frequency for the third overtone for the different dispersants. The steepness of the slope of the curve is an inverse measure of the rigidity of the layer.30 Also, a change in slope indicates coverage-induced structural changes in the adsorbed layer. A decrease in the slope (as seen beyond -17 Hz for the long-chain comb copolymer) is an indication that the layer becomes more rigid due to increased packing density. While the polyelectrolytes and the short comb copolymer all show low dissipation and moderate slopes during the whole adsorption process, indicating rigid layers that give small viscoelastic response, the long-chain comb copolymer shows a steeper slope, indicating a more viscous behavior, particularly at low coverage. The overall dissipation for the latter polymer is high, which is also indicative of a thick, viscous layer. This is not unexpected, since densely packed, long PEO chains are known to protrude out into an aqueous bulk solution, bringing steric stabilization to particulate systems50 and protein-repellency to macroscopic surfaces.51,52 (48) Walczak, W. J.; Hoagland, D. A.; Hsu, S. L. Macromolecules 1992, 25, 7317. (49) Dahlgren, M. A. G.; Waltermo, Å.; Blomberg, E.; Claesson, P. M; Sjo¨stro¨m, L.; Åkesson, T.; Jo¨nsson, B. J. Phys. Chem. 1993, 97, 11769–11775. (50) Napper, D. H. J. Colloid Interface Sci. 1977, 58, 390–407. (51) Holmberg, K.; Tiberg, F.; Malmsten, M.; Brink, C. Colloids Surf. A 1997, 123-124, 297–306.
Figure 3. Change in dissipation vs frequency for the 3rd overtone for the dispersants.
The high rigidity of the short-chain comb copolymer is probably a result of formation of a thin layer. One should keep in mind that this graft copolymer is a relatively small molecule with a molecular weight of only 3000. 1H NMR shows that the oxyethylene content is around 60%, which means a total PEO molecular weight of 1800. Assuming five PEO chains on the molecule, each chain would have a molecular weight of 360, which corresponds to around eight oxyethylene units per chain. One may anticipate that with such short PEO chains the electrostatic charge of the ionic anchoring chain will not be fully screened, and this was in fact shown earlier by ζ-potential measurements.47 This is likely to lead to less water entrapment and, thus, to a more rigid layer. Hence, the short-chain comb copolymer will behave like an intermediate between the highly charged polyelectrolytes and the long-chain comb copolymer, where the charges on the backbone are fully screened, as discussed above. Viscoelastic Response by Voigt Modeling of Dissipation Data. As shown in Figure 3, the long-chain comb copolymer gave a strong increase in dissipation upon adsorption, indicating a considerable viscoelastic contribution from this layer. The viscoelastic components were calculated by applying the Voigt model; see Table 1. At adsorption equilibrium the shear viscosity, ηf, was 1.7 × 10-3 Pa s and the shear modulus, µf, was 2.0 × 105 Pa. These values are in good agreement with previous values for adsorption of PEO brushes30 and for starch adsorption.34 In Figure 4, the shear modulus and the shear viscosity are plotted against the calculated Voigt mass for the long-chain comb copolymer. The high values observed at very low surface coverage are probably due to flow variations when the sample solution is injected. The changes in the slopes at a mass coverage of approximately 450 ng/cm2 coincide with the structural change (52) Malmsten, M.; van Alstine, J. M. J. Colloid Interface Sci. 1996, 177, 502–512.
9994 Langmuir, Vol. 24, No. 18, 2008
Figure 4. Shear viscosity and shear modulus vs adsorbed mass for the long-chain comb copolymer.
in the layer observed in Figure 3, which divides the adsorption into two stages. A modest increase in µf at both stages of adsorption has previously been observed for polymers containing PEO chains.30 Layer Thickness. From the Voigt model, assuming a layer density of 1.2 g/cm3, the effective layer thickness was estimated; see Table 1. The density value was chosen on the basis of measurements on similar systems.26,30 The thickness of the longchain comb copolymer, around 6 nm, is close to the Flory radius, which was estimated to around 5 nm in a previously study of the same copolymer.7 The value for the short-chain comb copolymer, 1.5 nm, is reasonable, considering the short side chains of this graft copolymer. As discussed above, it is reasonable that such a thin PEO layer will not be enough to shield the charges of the underlying surface. The layer extension of only 1.5 nm is not enough to give a viscous contribution. From the dissipation measurements, it was concluded that the poly(acrylic acid) and the lignosulfonate gave highly rigid, almost fully elastic layers. This is compliant with the thin layers, 0.5 and 0.6 nm, respectively, obtained for these dispersants. Such small values indicate adsorption in a very flat configuration at the surface, which is the expected behavior for a polyelectrolyte in fully dissociated state. Thin layers due to flat adsorption have previously been observed for polyelectrolytes in QCM-D measurements under similar conditions.34 Water Content of Adsorbed Layers. Surface-bound, brushlike copolymers such as poly(acrylic acid)-PEO comb copolymers are known to modify friction forces.11,53 Water bound in the PEO layer is believed to play an important role for this lubrication effect.11 It is therefore of interest to know the water content of adsorbed layer of the dispersant, and the QCM-D technique is one way of estimating it. The values of total adsorbed amount obtained from the QCM-D measurements were related to the values previously obtained for optimizing the amount of dispersant, i.e., finding the amount that gives the highest stability in ceramic slips. The latter optimization gives values of adsorbed mass without coupled water. The optimization was done by viscosity measurements in aqueous alumina suspensions with 50 vol % solids using the poly(acrylic acid), the lignosulfonate, and the long-chain comb copolymer as dispersants.7 For the short-chain comb copolymer, which is not an efficient dispersant in aqueous solution, data from studies of alumina in water/alcohol were used.47 The data are shown as the weight percent of total solids in Table 2 and recalculated to ng/cm2. The (53) Klein, J.; Kumacheva, E.; Mahalu, D.; Perahla, D.; Fetters, L. J. Nature 1994, 370, 634–636.
PalmqVist and Holmberg
values of Voigt mass from the QCM-D experiments (already presented in Table 1) are also given in the table and the differences between the values should correspond to the water content of the adsorbed layers. In Table 2 the water content of the layers is given as a percentage of the total, i.e. polymer with bound water. The areal solvation,11 i.e., the mass of bound solvent per unit area, was also estimated. The estimated water content was 90% for the long-chain comb copolymer, which is in good agreement with a reported value of 84% for PEG brushes.11 The areal solvation for this polymer is 66 ng/cm2. The high water content is not unexpected, since PEO chains are known to entrap substantial amounts of water due to hydrogen bonding.11 The water content in the short-chain comb copolymer layer was estimated to about 60%, giving an areal solvation of 44 ng/cm2. The lower water content in combination with the much thinner layer explains why this polymer showed higher rigidity. The lowest water content was found for the charged polyelectrolytes. The poly(acrylic acid) had a water content of 35% and an areal solvation of 13 ng/cm2. The lignosulfonate had almost the same areal solvation, 14 ng/cm2, and a calculated water content of 28%. For surface layers of poly(acrylic acid), values for water content between 16 and 31% have been given.26 Hence, our findings are in good agreement. It has also been shown that as the charge density of a polyelectrolyte decreased from 100% to 1%, the bound water content increased from almost zero to 80% of the total detected mass.27 Data for water binding capacity of surface layers of lignosulfonates seem not to be available in the literature, but the value obtained here, which is on the same order as that for poly(acrylic acid), seems reasonable considering that both polymers are highly charged and that they both form very thin adsorbed layers. Rheology of Slips during Casting. During slip casting water is removed from the slip through capillary forces in the mold, which is typically made of porous plaster of Paris. Hence, the solids content will increase gradually upon consolidation. When enough water has been removed for all particles to be settled, the slip has transformed into a consolidated state. The structural changes, casting rates, and final strengths of the different slips during casting were studied. The measurements were done with a cone-and-plate setup, with the lower plate replaced by hardened plaster of Paris. The rheological response depends on how much of the cone area that is in actual contact with the sample to be measured. In the present case, the contact area of the cone will vary during the casting sequence until the point where the whole sample is consolidated. At that point a plateau level is reached, at which the value of the final strength in terms of storage modulus can be assessed. Moreover, it is possible to determine the casting rate, as well as the phase shift, δ, during the whole casting sequence since tan δ ) G′′/G′ is independent of the cone contact area. The results from in situ casting of 50 vol % alumina slips with optimized amounts of dispersants are shown in Figures 5 and 6. The inset in Figure 5 shows viscosity curves from slip optimization in previous work.7 The viscosity levels at the initial stage of casting, i.e. in the slip, show how well-dispersed the system is. A stable, dispersed system will settle into a dense, incompressible consolidated layer.54 Figure 5 shows that both the poly(acrylic acid) and the lignosulfonate initially have low values of the storage modulus, G′, as well as low values of the viscosity of the suspensions. This indicates that these systems are well-dispersed. Moreover, the casting rate is high for both (54) Persson, M. In Surface and Colloid Chemistry in AdVanced Ceramics Processing; Pugh, R. J, Bergstro¨m, L., Eds.; Surfactant Science Series 51; Marcel Dekker: New York, 1994; Chapter 7.
Adsorption and Viscoelasticity of Alumina Suspensions
Langmuir, Vol. 24, No. 18, 2008 9995
Table 2. Optimum Amounts of Adsorbed Mass from Slip Viscosity and from the Voigt Model, ∆mVoigt; Estimated Water Content; and Calculated Areal Solvation optimum amount in slip7,47 dispersant
wt %
ng/cm2 a
∆mVoigt (ng/cm2)
water content (wt%)
areal solvationb (ng/cm2)
long copolymer short copolymer lignosulfonate poly(acrylic acid)
0.5 0.5 0.35 0.25
74 74 51 37
765 186 71 57
90 60 28 36
66 44 14 13
a
Calculated from wt % solids in alumina slips with a specific surface area of 6.8 m2/g.
Figure 5. Storage modulus vs time from oscillatory rheological measurements during in situ slip casting of the long-chain comb copolymer, the short-chain comb copolymer, the poly(acrylic acid), and the lignosulfonate. The inset shows initial slip viscosities (poly(acrylic acid) and lignosulfonate curves overlap).
Figure 6. Phase shift from oscillatory rheological measurements during in situ slip casting of the long-chain comb copolymer, the short-chain comb copolymer, the poly(acrylic acid), and the lignosulfonate.
polyelectrolytes, with a steep slope at around 700 s, indicating the onset of consolidation. The resulting wet strengths of the consolidated bodies are high, as seen from plateau values of the storage modulus of about 1 MPa for both the poly(acrylic acid) and the lignosulfonate. Almost the same high wet strength is obtained with the long-chain comb copolymer. However, the casting sequence is considerably slower for this dispersant. A different behavior is observed for the short-chain comb copolymer. This polymer works well as dispersant in mixed solvent systems,47 but its dispersing efficiency in aqueous slips with high solids content is insufficient. Thus, it was not possible to optimize the amount of dispersant in the slip. However, to allow for viscosity and casting measurements, an arbitrary amount of 1.5 wt % of dispersant was chosen, which resulted in reasonably low viscosity, although still much higher than for the other dispersants. This indicates that the suspension is flocculated and
b
Calculated according to ref 11.
unstable. Figure 5 shows that this system has a slow casting rate and that the final strength of the consolidated body is much lower than obtained with the other three dispersants. This is an indication of a loosely packed structure of aggregates caused by poor stabilization of the initial slip. Apparently, neither the charges of the anchoring chain nor the protruding PEO chains are enough to provide stabilization. As discussed above, a probable reason for the poor performance of this dispersant is that the short-chain comb copolymer does not provide a thick enough layer to stabilize the particles. The 1.5 nm thick layer is not enough to overcome the van der Waals attraction forces, which are known to operate over the range of 5-10 nm for colloidal particles.55 The low degree of water entrapment may be another reason for the lack of lubrication, as was also discussed before. Influence on Casting Rate. The time from when the casting begins until completion of the consolidation is important in ceramic processing because much time can be saved in the production step with rapidly consolidating systems. Long PEO chains have slow dynamics, and dispersants based on such chains require a long time to adopt the final conformation upon consolidation. This is probably one reason for the long casting time observed for the long-chain comb copolymer, as seen in Figure 5. Another reason may be the lubrication effect induced by the high water content of the adsorbed layer, as demonstrated by the QCM-D experiments. It is generally believed that lubrication by polymers plays an important role in achieving good particle packing in ceramic systems.56The polymer lubricity allows flow to occur during consolidation due to reduced friction.57 More recently, the lubricity of polymer layers has been attributed to high amounts of entrapped solvent in the adsorbed layer.11 The lubricity caused by the water bound in the adsorbed polymer layer may allow for rearrangement over an extended time period, leading to stress relaxation. This could explain the long casting time and the high final wet strength observed for the long-chain comb copolymer system. The poly(acrylic acid) and the lignosulfonate show high casting rates. This is typical for systems with low viscosities, where little relaxation is necessary for consolidation. Phase Shifts during Consolidation. When the PEO grafts of the long-chain comb copolymer were freely protruding into the bulk solution in the QCM-D measurements, there were few restraints and the viscous contribution of the polymer layer was obvious. However, in a particle system such as the alumina slip, the elastic nature will dominate at high solids loadings. This is due to compression of the layer, which will lead to interpenetration of the PEO chains in the sterically stabilized system.43 The phenomenon is reflected in the phase shift of the slips during casting, as seen in Figure 6. The sterically stabilized slips with the two comb copolymers show elastic behavior, seen as a phase (55) Tadros, T. F. In Handbook of Applied Surface and Colloid Chemistry; Holmberg, K., Ed.; Wiley: New York, 2001; Vol. 1, Chapter 16. (56) Pugh, R. J. In Surface and Colloid Chemistry in AdVanced Ceramics Processing; Pugh, R. J., Bergstro¨m, L., Eds.; Surfactant Science Series 51; Marcel Dekker: New York, 1994; Chapter 4. (57) Yin, T. K.; Aksay, I. A.; Eichinger, B. E. In Ceramic Powder Science II (Ceramic Transactions, B); Messing, E. R., Fuller, G. L., Hausner, H., Eds.; American Ceramic Society: Westerville, 1988; Vol. 1, pp 654-662..
9996 Langmuir, Vol. 24, No. 18, 2008
angle well below 45° during the whole experiment. In contrast, the electrostatically stabilized slips show viscous behavior with phase angles around 70° at the initial stage of casting. Upon consolidation, values below 10° are obtained, as expected from solid, elastic materials.
Conclusions This work has shown that the QCM-D technique is very useful for studying adsorption and viscoelasticity of dispersants on ceramic surfaces. Combining values of adsorbed mass and, in particular, dissipation from the QCM-D measurements with dynamic rheology measurements, as illustrated in Figure 1, is a powerful approach. We have demonstrated that the combination of QCM-D and viscosity optimization of colloidal slips is useful as a way to determine the amount of water entrapped in the adsorbed layers of the different dispersants. The amount of bound water is
PalmqVist and Holmberg
important because it is decisive of the viscoelastic properties of the layer, which, in turn, influence the lubricity of the particles. Good lubricity is needed in order to attain high packing density of the dispersion. The water contents found, ranging from 28 and 36% for the lignosulfonate and the poly(acrylic acid), respectively, to 60 and 90% for the comb copolymers with short and long PEO chains, respectively, are reasonable and in accordance with literature values of similar systems. To our knowledge this is the first time that these two methods are combined in a colloidal system. Acknowledgment. Financial support for this research work was gratefully received from The Swedish Knowledge Foundation. The help of Ola Lyckfeldt with designing the experimental setup for in situ dynamic rheology measurements is much appreciated. LA800719U
