Biomacromolecules 2004, 5, 276-283
276
Effect of Shear Flow on the Phase Behavior of an Aqueous Gelatin-Dextran Emulsion Y. A. Antonov, P. Van Puyvelde,* P. Moldenaers, and K. U. Leuven Department of Chemical Engineering, W. de Croylaan 46, B-3001 Leuven, Belgium Received April 23, 2003
A rheo-optical methodology, based on small angle light scattering and transmitted light intensity measurements, has been used to study in situ and on a time resolved basis the shear induced morphology in ternary two-phase water-gelatin-dextran mixtures. Emulsions close to the binodal line as well as far from it have been investigated. It is shown that above a critical shear rate, shear-induced mixing occurs at the length scales probed by the laser light. It is hypothesized that the shear-induced homogenization is due to the shear forces that exceed the intermolecular forces of the self-association process of the gelatin. The isothermal phase diagram at a fixed shear rate has been determined. In addition, the structure evolution after cessation of flow has been studied. When flow is stopped after homogenization, phase separation occurs almost instantaneously. When subsequently applying a low shear rate, the structure coarsens due to coalescence of the dispersed droplets. The kinetics of this coalescence process is strain controlled. I. Introduction A change in phase state, structure, and properties of a system close to its critical physicochemical parameters (temperature, pressure, volume, and composition) is normally denoted as a critical phenomenon. Such phenomena are characterized by a change in specific volume and internal energy of the system and are the subject of many investigations dealing both with low and high molecular weight systems.1-4 Because the relaxation times in polymer solutions are large, critical phenomena in these systems are more pronounced as compared to low molecular weight mixtures. Flow fields have a strong effect on critical phenomena in polymer mixtures, and studying them has both an academic as well as an industrial interest. On one hand, such studies may lead to a better understanding of nonequilibrium statistical physics. On the other hand, because processing of materials involves flow, the influence of flow on critical phenomena has a large impact on technological processes in industry. Observations concerning the effect of shear on the phase behavior of synthetic polymer-containing systems have a long history. There are reports on a pronounced extension of the homogeneous region under flow (shear induced mixing)5-7 as well as on phase separation under shearing (shear induced demixing).8-10 The influence of shear flow on the phase equilibrium of synthetic polymer systems has been studied both theoretically11,12 as well as experimentally, especially by applying a wide variety of scattering techniques. For example, binary polymer blends under shear have been studied by small-angle neutron scattering by Nakatani et al.13 In the latter investigation, the experiments were focused on the phase transition near the critical temperature for blends consisting of high molecular weight components. For low molecular weight polymer blends * To whom correspondence should be addressed.
(deuterated polystyrene and polybutadiene), Hobbie et al.14 established that shear inhibits phase separation at the critical molecular composition: flow disrupts the long-range critical fluctuations when the inverse of the shear rate becomes comparable to the relaxation rate of these fluctuations. Changes in the phase state of a synthetic polymer system, consisting of polystyrene and polybutadiene in a common solvent DOP, have also been observed by Hashimoto and co-workers15-17 using small-angle light scattering. These studies show a complex behavior with several regimes depending on the magnitude of the shear rate. At low shear rates, the mixture could be regarded as a conventional twophasic system. At high shear rates, flow is believed to suppress the surface ondulations that develop on extended droplets. This allows extended cylindrical domains to form, known as string phases.18-19 The diameter of these fibrils decreases when shear rate is gradually increased until eventually the fibril diameter equals the interfacial thickness. In this case, shear-induced homogenization occurs. Biopolymer systems may differ from synthetic polymer systems in various ways. Unlike synthetic polymers with flexible chains, many proteins are known to be relatively symmetric compact molecules and are usually able to form solutions that can still be considered dilute for concentrations 10-fold higher than for synthetic polymers of the same molecular weight.20 In addition, two-phasic aqueous biopolymer mixtures are characterized by a very low interfacial tension.21,22 These peculiarities may have a strong effect on the phase equilibrium of biopolymer mixtures.23 Although most of the investigations have been performed using mixtures of synthetic polymers,24,25 the study of the phase behavior of biopolymer mixtures under flow is of significant technological interest because food emulsions are typically subjected to flow when being processed.26 However, quantitative studies on the effect of shear on the phase behavior
10.1021/bm0300352 CCC: $27.50 © 2004 American Chemical Society Published on Web 12/24/2003
Phase Behavior of a Gelatin-Dextran Emulsion
of biopolymer systems are limited.27 In this study, we focus on the phase equilibrium under shear flow of a biopolymer emulsion at isothermal conditions with compositions close to and far from the binodal line. A rheo-optical methodology, based on small angle light scattering (SALS) and turbidity measurements, is employed to investigate the structure. A water-gelatin-dextran system is chosen as a model system because of its low compatibility at the selected conditions and the liquid state of the coexisting phases over a wide range of compositions. It is assumed that by using gelatin and dextran, which are characterized by a strong difference in self-association of the macromolecules in water,28-30 one can obtain information concerning the role of the protein-solvent interaction on the thermodynamical behavior of the emulsion during flow. II. Materials and Methods The gelatin sample used is an ossein gelatin type A 200 Bloom produced by SBW Biosystems, France. The Bloom number, weight average molecular mass and the isoelectric point of the sample, reported by the manufacturer are respectively 207, 99.3 kD, and 8-9. The high molecular weight dextran T-2000 sample was purchased from Amershan Pharmacia Biotech AB. Its intrinsic viscosity in water at 293 K and weight average molecular mass, reported by the manufacturer, are 0.9 dL/g and 2 × 106 respectively. To prepare molecularly dispersed gelatin solutions, an acetic acid-sodium hydroxide buffer (pH ) 5.0, ionic strength µ ) 0.002) was gradually added to the gelatin and stirred first at 333K for 20 min and then at 318 K for 1 h. The required pH values of the solutions (5.0) were adjusted by addition of 0.1-0.5 M NaOH or HCl. These were required in small amounts, avoiding changes in ionic strength. The solutions were centrifuged at 50 000 g for 1 h at 313 K to remove insoluble particles. Subsequently, the concentration of the biopolymer was determined by measuring the dry weight residue. Dextran solutions were prepared by dispersing the gum in the acetic acid-sodium hydroxide buffer (pH ) 5.0, µ ) 0.002) under stirring for 1 h at 318 K. Subsequent manipulations were the same as those described for the preparation of the gelatin solutions. Both the gelatin and dextran solutions show almost Newtonian behavior (at temperatures of 313 K and above) up to concentrations of 30% where the hydrodynamic volumes occupied by the chains overlap.31-33 The ternary water-gelatin-dextran systems were prepared by mixing solutions of each biopolymer at 318 K. After mixing for 1 h, the systems were centrifuged at 50 000 g for 1 h at 318 K using a temperature-controlled rotor in order to separate the phases. The phase diagram of the ternary system was determined at quiescent, isothermal conditions (318 K) at pH 5.0, µ ) 0.002 and is shown in Figure 1. The procedure to obtain the phase diagram is adapted from Koningsveld and Staverman34,35 and Polyakov and colleagues.36,37 The compatibility of gelatin and dextran is characterized by the coordinates of the binodal, threshold, and critical points. Small angle light scattering experiments (SALS) have been performed on a Rheometrics Optical Analyzer (ROA) that
Biomacromolecules, Vol. 5, No. 2, 2004 277
Figure 1. Isothermal phase diagram of the dextran-gelatin-water system under quiescent conditions (T ) 318 K, ionic strength ) 0.002, pH ) 5.0). O ) critical point (gelatin: 9.5%, dextran: 2.5%). ∆ ) treshhold point (gelatin: 6.5%, dextran: 5.35%). Emulsion A has been studied extensively in this paper.
has been modified to perform SALS measurements. The flow cell consists of a parallel plate device in which the shear flow results from the rotation of the upper glass plate of the cell. The rotational speed is adapted to have the desired shear rate at the observation area. The gap between the plates has been set at 1 mm, and the temperature was kept constant at 318 K by means of a thermostatised water bath. A He-Ne laser (wavelength 633 nm) has been used as the light source, and the scattered light is intercepted on a screen of semitransparent paper with a beam stop. The image is recorded using a CCD camera (Ikegami ICI-810P), which is mounted under the screen. The CCD camera is connected either to a frame grabber (Data Translation DT 3851) or to a video recorder to collect the scattering patterns. In addition, the ROA has been used to measure the transmitted light intensity. It has to be noted that the samples are rather transparent, allowing relatively high volume fractions to be investigated. III. Results and Discussion A. Shear-Induced Homogenization. The experimental results shown in this section have been obtained on a water (83.72%)-gelatin (2.44%)-dextran (13.84%) emulsion. It contains 99 wt % of the dextran enriched phase and 1 wt % of the gelatin enriched phase which have been mixed by hand, typically resulting in a very fine morphology. This emulsion is located in the two-phase region far from the critical point and close to the binodal line (emulsion A in Figure 1). The coexisting phases have about equal Newtonian viscosities at 318 K, i.e., 0.1 Pa s and 0.08 Pa s for the dextran enriched and the gelatin enriched phase, respectively. The interfacial tension of this emulsion, measured by a rheooptical methodology, amounts to 2 × 10-5 N/m at this temperature.21 To study the effect of flow on the phase behavior, a flow history consisting of three shear zones is used. First, a preshear of 0.5 s-1 is applied for 1000 s (500 strain units). It has been verified that this procedure leads to a reproducible initial morphology which is due to the fast coalescence process in this particular blend. The resulting droplet radius is of the order of 35 micron.38 Subsequently, this preshear is stopped and the slightly deformed droplets are allowed to retract to a spherical shape. Finally, the shear rate is suddenly
278
Biomacromolecules, Vol. 5, No. 2, 2004
Antonov et al.
Figure 2. Evolution of the SALS patterns as a function of time at different shear rates (emulsion A). Seconds refer to time after application of flow.
increased and the evolution of the SALS patterns and transmitted intensity are monitored. The evolution of the SALS patterns upon applying different shear rates is shown in Figure 2, starting from an isotropic scattering pattern at rest. In each experiment, a freshly loaded sample has been used. When a shear rate of 15 s-1 is applied, the SALS patterns become very anisotropic: a streaklike pattern appears perpendicular to the flow direction which reflects the deformation of the droplets. As was demonstrated by Van Puyvelde et al., the anisotropy reaches a maximum during flow, then decreases with time, and finally a steady-state elliptic pattern is observed.38 It has been demonstrated that this evolution at relatively low shear rates could be explained on the basis of deformation, breakup, and coalescence of the dispersed droplets. Hence, under these conditions, the biopolymer emulsion behaves according to the theoretical relations for conventional emulsions and similar to synthetic blends.38 At somewhat higher shear rates (e.g., 20 and 40 s-1 on Figure 2), a streak-like pattern develops but gradually loses its intensity with time. When steady state conditions are reached, a particular scattering pattern emerges: its shape is of the “butterfly” type in the sense that the pattern has a strong intensity along the flow direction and a dark streak normal to it (e.g., the pattern at 20 s-1 after 80 s). It is clear from these observations that the scattering mechanism at these shear rates has changed: the scattering pattern cannot be explained anymore on the basis of deformed droplets. The data seem to suggest that the correlation length of the concentration fluctuations along the flow direction, which corresponds to scattering perpendicular to the direction of flow, becomes extremely large, which could be indicative of the onset of mixing. The higher the shear rate, the weaker the pattern becomes which also implies that the correlation length perpendicular to the flow direction increases.
Figure 3. Evolution of the transmitted intensity (Idc) as a function of time at different shear rates (emulsion A). O: Idc of the continuous phase.
In this sample, the most interesting transition occurs at a shear rate of 60 s-1 (see Figure 2) where the evolution of the SALS patterns is different. It is seen that during shearing the anisotropic streak-like pattern gradually loses its intensity until finally no scattered light is observed anymore, except for some low intensity at small angles (critical scattering). This indicates that the sample is homogenized on the length scales probed by light scattering. The transmitted intensities (Idc) under the same flow conditions are shown in Figure 3. At a shear rate of 15 s-1, the evolution of Idc is similar to the evolution observed for blends of synthetic polymers:39 an increase in Idc is followed by a decrease and finally Idc evolves toward a steady-state value. This evolution has been explained on the basis of deformation, breakup, and coalescence of droplets under the influence of the applied shear rate. The time at which the Idc value reaches its maximum value is in agreement with the time at which the SALS pattern reaches its maximum
Biomacromolecules, Vol. 5, No. 2, 2004 279
Phase Behavior of a Gelatin-Dextran Emulsion
anisotropy. At 20 s-1, the lowest shear rate at which a butterfly type of SALS pattern develops in Figure 2, a slight maximum in the Idc versus time curve is still observed. However, at still higher shear rates, it is seen that the transmitted intensity gradually evolves toward a steady-state value. At a shear rate of 60 s-1, the transmitted intensity reaches the same value as the one measured for the continuous (dextran rich) phase. These observations support the conclusion drawn from the scattering experiments: there are no interactions of the structure with the laser light, and hence, at this shear rate, the sample can be considered to be homogenized on the length scales probed by the laser light. In the previous experiments, shear-induced homogenization occurred when the shear rate was suddenly increased from 0.5 to 60 s-1. Hence, the morphological history consisted of fibril formation and subsequent breakup and coalescence. To investigate whether the observed shearinduced homogenization is influenced by the shear history and consequently the morphological history of the sample, experiments have been performed in which the shear rate was gradually increased from 0.5 to 60 s-1. Each time, an increase of a factor 2 was used. In such experiments, it is anticipated that binary breakup is the dominant breakup mechanism without the formation of fibrils. In addition, it can be expected that the diameter evolution can be explained on the basis of the critical capillary number. Also in this case, it is observed that no scattering pattern was found anymore at 60 s-1 and that the Idc gradually increased to finally reach the Idc of the dextran enriched phase at 60 s-1. These observations support the conclusions put forth above: at 60 s-1, the system is in a single phase state, at least at the length scales probed by the laser light. In addition, it can be noted that, if the system would still be in a regular two-phasic state, the expected droplet radius at 60 s-1 based on the critical capillary number would be of the order of 0.5 micron when these stepwise experiments are performed. This is well within the sensitivity range of the laser light used (wavelength 633 nm). It can be argued that the scattering power of small structures decreases significantly and therefore the sensitivity of the camera becomes insufficient to pick up the scattering patterns. However, even with structures of 0.1 micron, the transmitted intensity should be affected by this two-phasic structure. Hence, because no interaction is seen since the transmitted intensity reaches the value of the continuous phase, the sample can be considered to be in a homogeneous state at 60 s-1 at the length scales probed by the laser light. The phenomenon of shear-induced homogenization is known for two-phasic systmes of synthetic polymers but to our knowledge has not yet been observed in liquid protein-polysaccharide systems. Compared to synthetic polymeric systems, a phase transition in the watergelatin-dextran system occurs at a higher total concentration of biopolymers and at smaller shear rate values.15,18,19 B. Shear-Induced Homogenization: Thermodynamical Interpretation. The effect of shear on the phase transition in aqueous biopolymer mixtures can be qualitatively analyzed within the general theoretical concepts of phase equilibria in a ternary polymer-polymer-solvent system.40-42 It is known that the thermodynamical state of a ternary system
is determined by the sign of the determinant43 ∂2Gm ∆)
∂m22
∂2Gm ∂m2m3
2 ∂2Gm ∂ Gm ∂m2m3 ∂m23
(1)
where Gm is the free energy of mixing and m2 and m3 are the molar concentration of the polymeric components. The system is stable provided ∆ < 0. When ∆ > 0, a phase transition from the two-phase state to the single-phase state can occur. The free energy Gm can be expressed in terms of second virial coefficients, as suggested by Edmond and Ogston, to determine the coordinates of the critical point of the ternary system41 1 + A22 ) K2/3 m2c 1 + A33 ) K-2/3A23 m3c
(2)
where m2c and m3c are the molar concentrations of components 2 and 3 at the critical point. Aij are the second virial coefficients on the molar scale, related to pair interactions of similar and dissimilar macromolecules, respectively, and K is a parameter that characterizes the phase diagram asymmetry. K is defined as K)
m3c m2c
(3)
According to Edmond and Ogston, the condition for a system to undergo a phase separation is given by41
(
)(
)
1 1 + A22 + A33 - A232 < 0 m2c m3c
(4)
A simple algebraic manipulation leads to 1 + A22m2c + A33m3c < (A232 - A22A33)m2m3
(5)
In a good solvent (as is the case for water in these dextrangelatin systems), the left part of eq 5 is always positive because the virial coefficients A22 and A33 are positive. Consequently, the right-hand side of eq 5 is also positive leading to a necessary and sufficient condition for the phase separation of a system A23 > xA22A33
(6)
Hence, the condition corresponding to a homogenization of a ternary system becomes A23 < xA22A33
(7)
an inequality that holds whenever the virial coefficients A22 and A33 are sufficiently large and A23 small. The reasoning put forth above will now be applied to the biopolymeric system under study in this paper. To focus the discussion, the following notation will be used: component
280
Biomacromolecules, Vol. 5, No. 2, 2004
1 is the solvent water (denoted as S), component 2 is the protein (pr), and component 3 the polysaccharide (ps). The cross virial coefficient Apr-ps can change from positive values, typical of the excluded volume effect of mutually repulsive macromolecules, to negative values corresponding to an enhancement of the formation of protein-polysaccharide complexes. This tendency in Apr-ps sign changes is equivalent to a decrease in the Flory-Huggins interaction parameter χpr-ps that characterizes protein-polysaccharide interactions. Values of the virial coefficients Apr-pr and Aps-ps are determined by interactions of the biopolymer molecules with each other and with the solvent. A positive value of the virial coefficient characterizes the excluded volume effect of macromolecules meaning that the biopolymer-solvent interaction is more favorable than the interaction between the biopolymer molecules. The virial coefficients Apr-pr and Aps-ps are controlled by the intensity of self-association of the protein and polysaccharide. These coefficients may change for biopolymer molecules from positive to negative values. This tendency corresponds to a decrease in solvent quality or an increase in the Flory-Huggins interaction parameters χpr-s and χps-s describing the interactions of protein and polysaccharide with the solvent (S). Obviously, the polymer concentrations being constant, homogenization of the system (∆ > 0) can be attained whenever Apr-pr and Aps-ps increase or (and) when Apr-ps decreases which means that the process of self-association of similar macromolecules is suppressed or (and) interaction between dissimilar macromolecules is intensified. Therefore, when analyzing the phenomenon of homogenization, one should first of all pay attention to the possible changes of these interaction parameters during shear. Specific interactions of proteins and polysaccharide, which are responsible for the single-phase state of the gelatin-PS solutions, occur often.31,44,45 These interactions are strongly dependent on pH, on the ionic strength of the system, as well as on the structural features of the biopolymers. This concerns also the behavior of the gelatin-dextran solution.31,45 In the acid pH range, from ∼2 to 3.8, gelatin is compatible with dextran even at high total concentrations due to their interaction which results in formation of water soluble gelatin-dextran complexes. The precise mechanism of this complex formation is not yet fully clarified. However, at pH 5.0 (i.e., the conditions studied in this work), the interaction between gelatin and dextran molecules is fully suppressed.31,46 The Apr-ps value determined for the gelatindextran system at pH 5.0 is positive and equal to 2.8 × 104 cm3 mol/g2.46 This means the absence of a specific affinity between these macromolecules at pH 5.0. It seems plausible to assume that when the specific affinity between biopolymers is absent in the quiescent state shear will not affect this property. On the other hand, dextran molecules are characterized by a negligibly small capacity to self-associate in water due to its unregular branched structure and the absence of functional groups which are capable to form hydrogen bonds.33 Gelatin molecules on the other hand possess a pronounced capacity to form associates and aggregates31,32,47 and mechanical energy can have a strong influence on the
Antonov et al.
Figure 4. Evolution of the transmitted intensity of a 30% gelatin solution under a shear rate of 60 s-1 and after cessation of flow.
associative behavior of gelatin molecules. Dissociation of the gelatin chain under shear was studied many years ago by Bourgoin and Joly48,49 using flow birefringence experiments. Gelatin concentrations were low (
