Elsevier

Polymer

Volume 42, Issue 23, November 2001, Pages 9601-9610
Polymer

The influence of gelation on the mechanism of phase separation of a biopolymer mixture

https://doi.org/10.1016/S0032-3861(01)00479-7Get rights and content

Abstract

The influence of the gelation of one component (gelatin) on the phase separation and morphology of an aqueous mixture of gelatin and dextran has been investigated. Small angle light scattering and confocal microscopy experiments show that the mechanism of phase separation is similar to spinodal decomposition, even in the presence of a rapid gelation. At temperatures well below the gelation temperature the phase separation kinetics are halted by the gelation, resulting in an immobile, interconnected morphology. This effect is seen as a pinning of the peak in the light scattering data. We also show that the gelation affects the position of the scattering peak, effectively deepening the quench as the gelation proceeds, through an apparent increase in molecular weight.

Introduction

Recently much experimental and theoretical work has concentrated on phase separation of polymers in the presence of other ordering phenomena such as crystallisation or glass transitions. Physical gelation of one of the separating components is another similar effect influencing the phase ordering process. A system in which phase separation occurs alongside gelation can exhibit a wide range of morphologies. The rate and onset of the different ordering mechanisms and the effect of the competing processes (a glass transition would tend to oppose phase separation, for example), dictate the morphology of the material produced. For instance, by halting the phase separation at an early stage a percolating network can be formed. The use of phase separation to produce such structures is of great importance [1], say, to alter the mechanical properties of materials. Novel electronic and optical properties can also be produced. Photodiodes are an illustration of this: these can be made from two different polymers that are in close proximity — hence, an interpenetrating network of two phases, each containing different polymers is the ideal form since there is a large area where the two polymers are in close proximity [2]. Physical properties of structures produced depend sensitively on the interactions of the different mechanisms involved; understanding the relationships between the various phenomena will allow control over the properties of the material produced. Thus the interplay and effects of the various competing mechanisms affecting the conventional phase separation of polymer mixtures have attracted attention of late.

There are several well-established theories describing phase separation and the subsequent coarsening of the structure (see, for example the reviews by Binder [3] and Gunton [4]). In many cases, including those we examine here, the initial phase separation proceeds via spinodal decomposition, where a characteristic length scale is seen throughout the sample. This leads to, if each phase is present in sufficient quantities, an interconnected structure spanning the sample. At this point the compositions of the two phases are (theoretically) almost at equilibrium. There are various mechanisms of subsequent coarsening whereby this structure can evolve, forming larger regions of each phase. Such mechanisms are distinct by different rates of change of the dominant length scale in the sample, r. For example, if the coarsening mechanism is that of evaporation–condensation, whereby material from small droplets of one phase diffuses through the sample to larger droplets, then a growth rate of rt1/3 would be expected, similarly for the case where droplets can move around and coalesce. If, on the other hand the coarsening is via a hydrodynamic mechanism then one expects a growth rate of rt.

In addition, there has recently been interest in the effect of viscoelasticity on phase separation [5], [6], [7], [8], [9], [10]. Work done by Keller et al. [11] on systems close to the glass transition of one of the pure components indicated how the phase separation process could be halted by a change in mobility of one component. Several authors have investigated the effect of a glass transition on spinodal decomposition. Sappelt and Jackle [7] modelled the problem by using a mobility that decreased rapidly with increasing concentration of the glass-forming component. Barton et al. [12] also included a mobility dependent on the state of the phase, and found that spinodal decomposition proceeds until the glass composition is reached, after which the growth proceeds much more slowly such that it appears arrested on experimental timescales. The images obtained by Isayeva et al. [13] show this in a handsome manner — they examined a system in which simultaneous crystallisation and phase separation occurred, both were arrested by an interjecting glass transition, resulting in very different morphologies depending on the initial composition and quench temperatures. The effect of the glass transition on phase separation was also examined by de Graaf et al. [14], who studied mixtures of polystyrene and methacrylate, which separated upon cooling, and reported a physical arrest of the phase separation as the polystyrene rich phase underwent a glass transition.

More recently, work has concentrated on systems in which an asymmetry in the bulk and shear moduli of the phases leads to changes in the separation dynamics. Onuki and Taniguchi [9], [15] have done a number of computer simulations of spinodal decomposition in polymer solutions. They include an elastic energy term in the free energy to model the effect of a mechanical imbalance. Following an incubation time, they find that ‘holes’ of the solvent appear, then the polymer-rich regions become thin, forming a ‘sponge-like’ network. This structure coarsens with time, eventually breaking up into bulk polymer rich domains. These simulation findings correspond well with experimental results of Tanaka et al. [16] who studied the system of polystyrene and poly(vinyl methyl ether). This system separates on increasing the temperature, the poly(vinyl methyl ether) rich phase being less viscoelastic than the polystyrene rich phase. This asymmetry was found to result in the phase separation pattern of such a mixture, when close to its critical composition, being characterised by a sponge-like continuous structure of the more viscoelastic phase. Tanaka et al. suggest that the system is initially dominated by elastic energy, behaving as an elastic gel. During the later stages of separation the rate of deformation slows down, allowing the stresses to relax, thus the domain shapes can transform to the shape of lowest interfacial energy. Tanaka has suggested that an asymmetry in the moduli of the two components would generally be expected to produce such viscoelastic phase separation [8], [16].

The majority of the examples so far have examined synthetic polymer systems. Bansil et al. [17], [18] used a system of gelatin, water and methanol to look at the effect of gelation on phase separation. When the quench depth was such that the sample was below the gel temperature for pure gelatin, Tgel, the gelation of the gelatin was found to halt the phase separation process. For quenches to higher temperatures the phase separation was found to be similar to classical, non-gelling mixtures.

The biopolymer system of gelatin and dextran in a solution of sodium chloride has been studied previously [19], [20], [21]. Both gelatin and dextran are soluble in water and the addition of sodium chloride screens the weakly charged gelatin and adjusts the upper critical phase separation temperature, Tc, and gelation temperature to within a range suitable for experimentation [22]. The dependence of the phase boundary on the salt concentration was not explicitly studied, however previous work [19] found that 0.5 M sodium chloride brought the system into a regime where phase separation and gelation occurred at rates and temperatures which were convenient to study. The simplest model of the gelation of gelatin is that in solution at high temperatures the conformation of the gelatin chain is a random coil and, as the solution cools, the gelatin undergoes a coil to helix transition — this is sometimes referred to as the ‘frustrated renaturation’ of the collagen triple helix. The helices associate and a gel is formed with a structure of connected triple helices. When the temperature is lowered below the temperature of the coil–helix transition a gel will be formed provided the concentration is above a certain level. Below this concentration the coil–helix transition will still occur, but a percolating network (and hence elastic gel) will not be formed. In order to study the gelatin/dextran/water system as a pseudo-binary system the total polymer concentration in solution is fixed and only the ratio of gelatin and dextran is altered. Tromp et al. [19] examined such a mixture above and below the gelation temperature for a concentration of 4.2% gelatin, 4.2% dextran using small angle light scattering and optical microscopy. They found that the early stages were similar for both gelling and non-gelling situations (cooling to just above and just below Tgel). Classically, a quench of a mixture of such a ratio of gelatin to dextran is expected to phase separate via the mechanism of spinodal decomposition, as predicted by the linear Cahn–Hilliard theory. Microscopy experiments revealed a composition variation of a characteristic wavelength throughout the sample, as would be expected for spinodal decomposition, and a peak in the light scattering data was observed. However, in contrast with the predictions of the Cahn–Hilliard theory, the peak position in reciprocal space was not stationary and moved to lower values of the wavevector, q, at the earliest experimentally accessible times. The Cahn–Hilliard plot, R(q) vs. q2, was non-linear in all cases. The morphology as observed by optical microscopy matches the light scattering data quite well: the change in the power-law exponent for the growth of the dominant length scale occurs at the same time as the interconnected sample appears to break up into isolated domains of gelatin-rich droplets, which go on to coalesce. This transition to droplets does not occur for the quench to low temperatures and the gelation ‘freezes in’ the interconnected structure. The rheology of the system has also been studied [21]. The dynamic shear modulus of a phase separating solution was followed as a function of time. It was found that samples quenched to a low temperature, where optical microscopy showed that a spinodal structure was frozen in by the gelation, had a high shear modulus, reflecting the connectivity of the gelatin rich phase.

These studies demonstrate the variety of structures that can be obtained simply by adjusting the rates and onsets of various mechanisms. How these mechanisms interact with and affect each other is, however, less clearly understood. It is likely that, as with the glass transition case mentioned earlier, the changing mobility of one component will affect the separation kinetics. Here we investigate the phase separation of gelatin and dextran further, essentially considering the phase behaviour of a two-component mixture of polymers, with the fixed-concentration water playing the role of a suspending matrix only. The effects are likely to depend on the onset and the rate of the gelation process compared to that of phase separation. It is possible that a slightly faster gelling rate would produce a more enhanced effect on the separation mechanism; indeed it is not obvious that the classical mechanisms of phase separation (spinodal decomposition, for example) will remain under such circumstances. If the mixture is quenched below Tc but above Tgel one might expect classical phase separation behaviour. Below Tgel, the formed elastic network would have a profound effect on the kinetics and morphology.

The compositions under study are in the range of 3.8–5.0% gelatin, with a fixed total polymer concentration of 8.4%. For such a high degree of dilution it is assumed the system is pseudo-binary, that is, the water partitions equally between gelatin rich and dextran rich phases. The values of Tc and Tgel for the range of mixtures used are shown in Fig. 1. The ratios examined here are indicated. The cloud point temperature, Tc, is determined by measuring the temperature at which a mixture, cooled slowly from the homogeneous state, becomes an opaque, scattering liquid. The cloud point curve essentially determines the equilibrium compositions to which the mixture will separate, the volumes of the phases are governed by the initial composition. The gel curve shows the temperatures, Tgel, below which a pure gelatin solution, at given concentrations, would form a gel. These two temperature curves define the region relevant for our present findings.

The temperature at which the coil–helix transition occurs is, for gelatin, in the region of 30°C. The cloud point temperatures are higher than this, thus it is likely that the phase separation (seen at the cloud point) is due to the unfavourable interaction between the gelatin coil (the high temperature state) and the dextran coil. The gel line measured is lower than the coil–helix transition because the gel takes time to form, thus even though the coil–helix transition may occur quite quickly, the slower gel formation means that within the experimental time scale a gel does not appear to form at higher temperatures, even though the gelatin may be in the helix form. Thus the gel line should be taken as an indication of gelation occurring within the experimental timescale. The quench depth, that is, how far below Tc the mixture is cooled, affects the driving force for phase separation. Similarly the concentration of gelatin affects if and how quickly the mixture gels. Note that the mixtures are all quenched to temperatures below the coil–helix transition, thus we might expect a change in the phase separation character due to this effect, that is, there was an initial unfavourable interaction between the gelatin coil and the dextran coil, but below the coil–helix transition the interaction between gelatin helix and dextran coil must be considered. Similarly, when the gelatin helices associate to form a gel the interaction between the associated helices and the dextran coil has to be taken into account.

Various mechanisms by which phase separation can take place have been detailed in some classical reviews [3], [4]. These theories offer a framework with which the light scattering data can be analysed. The classical Cahn–Hilliard theory predicts that the composition difference, δφ, evolves as:δφ(q,t)=δφ0expR(q)2twhereR(q)=2Mq2[|f″|+gq2]where q is the wavevector, δφ0 is the composition difference at time t=0s. M is the mobility constant, φ is the composition and g is an elastic constant. The equilibrium free energy density, f, is assumed to depend on composition, φ, with the gradient term, gq2, penalising the variation of φ on interfaces; f″ denotes a second derivative with respect to φ. Eq. (1) predicts that a characteristic lengthscale appears in the sample, the composition difference between the two phases growing with time. That the composition evolution is written in terms of wavevector is convenient since data from small angle light scattering can be analysed easily in terms of this theory. This characteristic length scale (or the equivalent wavevector, q) is determined by the driving force to phase separate and the energy penalty caused by having a composition gradient in the sample. This characteristic wavevector, qmax, is given by:qmax=|f″|T,φ02g

A critical wavevector above which no composition variations grow, can also be defined asqcrit=|f″|T,φ0g

This corresponds to the critical wavelength below which the energy penalty due to the composition gradient is so great that the composition variation is unstable and is suppressed. Thus, Eq. (1) predicts a peak in the scattered light at a value q=qmax, the intensity, I(q), at each wavevector increasing exponentially with the exponent R(q), and a critical wavevector, qcrit, above which no increase in intensity is expected. As the composition difference between the phases increases, the linear theory breaks down and the sample starts to coarsen, the characteristic wavelength increases and the two phases become more different.

Although complications such as the presence of a gelling component indicate that the classic, linear Cahn–Hilliard theory is unlikely to be completely successful, it is nonetheless worthwhile to see how well the initial sample evolution can be described by this theory. The first and most obvious complication is the formation of the elastic network. This itself has its complexities. The gelatin molecules first undergo a coil–helix transition and then link up with other helices to form a network. This initial configuration change of the molecules may alter the interaction strength or range and will introduce a dynamic asymmetry in the elasticity of the separating components, thus altering the dynamics of subsequent separation. This transformation takes place quickly and is followed by a slower change, thought to be a refinement of the structure [23]. Also to be considered are the approximations inherent in the classical Cahn–Hilliard theory. We use the simplest linear Cahn–Hilliard theory, ignoring, for example, the compositional dependence of the parameter g describing the gradient term in Eq. (1), and indeed other effects frequently found to lead to, for example, lower critical temperature behaviour. This approach allows us to examine how far the linear Cahn–Hilliard theory can be stretched to predict phase separation behaviour of low-concentration polymer mixtures having an additional elastic degree of freedom.

We show that although gelation is present from the initial stages and strongly influences the final morphology of the sample the phase separation process retains many of the characteristics of spinodal decomposition, being affected only by the apparent increase in molecular weight of the gelatin chains and the final gelling of the structures.

Section snippets

Experimental

The gelatin and dextran samples used in this work were obtained from Aldrich. The gelatin was Type A, extracted from porcine skin. It had a bloom number of 175 and a molecular weight of 200 000. The dextran was produced by Leuconostoc mesenteroides with an average molecular weight of 167 000. Both polymers are expected to have a high polydispersity. In all cases, samples were made by mixing the appropriate masses of polymers and salt solution in a vial. The polymers were allowed to swell

Results

The samples examined contain different ratios of gelatin to dextran and are labelled A (gelatin poor sample) through to D (a gelatin rich sample). The total polymer concentration in each sample is 8.4%. For convenience we summarise the compositions of the samples below; the corresponding ratios of gelatin to dextran are given in brackets:

  • A:3.8% gelatin:5.0% dextran (0.45:0.59);

  • B:4.2% gelatin:4.2% dextran (0.50:0.50);

  • C:4.6% gelatin:3.8% dextran (0.55:0.45);

  • D:5.0% gelatin:3.4% dextran (0.60:0.40).

Conclusions

From the data we can distinguish several regimes:

  • 1.

    First an early stage in which microscopy of simultaneously phase separating and gelling samples indicate that the initial mechanism of phase separation is that of spinodal decomposition. Small angle light scattering data also shows many universal characteristics of classical spinodal decomposition [19]. For all samples at early times the light scattering results are similar to those found for samples quenched well above the gelation temperature.

Acknowledgements

We would like to thank Eugene Terentjev for valuable discussions and critical reading of this manuscript. Financial support from the BBSRC and Unilever is gratefully acknowledged.

References (29)

  • H. Berghmans et al.

    Polymer

    (1998)
  • A. Keller et al.

    Polymer

    (1998)
  • I. Isayeva et al.

    Polymer

    (1998)
  • J. Lal et al.

    Physica A

    (1992)
  • J.J. van Aartsen

    European Polymer Journal

    (1970)
  • L.H. Sperling et al.

    Polymers for Advanced Technologies

    (1996)
  • J.J. Halls et al.

    Nature

    (1995)
  • K. Binder

    Spinodal decomposition

  • J.D. Gunton et al.

    The dynamics of first-order phase transitions

  • R. Ahluwalia

    Physical Review E

    (1999)
  • S. Puri et al.

    Physical Review E

    (1997)
  • D. Sappelt et al.

    Europhysics Letters

    (1997)
  • H. Tanaka et al.

    Physical Review Letters

    (1997)
  • T. Taniguchi et al.

    Physical Review Letters

    (1996)
  • Cited by (0)

    View full text