Kolloid-Z. u. Z. Polymere 243, 14-20 (1971 ) From the AKZO Research Laboratories Arnhem (The ~Vetherlands) Liquid-liquid phase separation in concentrated solutions of non-crystallizable polymers by spinodal decomposition By C. A. Smolders, J. J. van Aartsen, and A. Steenbergen With 11 figures (Received June 15, 1970) Introduction We will describe here under what condi- In the past, investigations of phase separa- tions spinodal decomposition can be expect- tion in concentrated polymer solutions cen- ed for polymer solutions. A number of ob- tered first of all on the crystallization kinet- servations on phase separation in concen- ics (1) and morphology (2) of polymer crys- trated solutions of poly(2,6-dimethyl-l,4- tals growing from solutions and, secondly, phenylene ether), (PPO| Tg = 210 ~ and on the equilibrium thermodynamic proper- of an ethylene-vinyl acetate copolymer ties of liquid-liquid phase separation phe- (Elvax-40), Tg ~ 0 ~ will be reported, which nomena (3). will make it clear that spinodal decompo- Papkov and Yefimova (4) have recently sition can be realized in polymer solutions. given a classification of different types of phase separation for polymer solutions, essen- Thermodynamic considerations tially the two types mentioned above, in which they also pay some attention to the Fig. 1 a and b, gives a schematic represen- difference in final state of separation (i. e. tation of the thermodynamics of liquid- "gel" or two liquid layers) and to the mechanism of separation itself. For liquid- A Fm [~_ liquid phase separation of solutions of non- crystallizable polymers the difference in I LE -r final state is ascribed to a combined effect of the thermodynamics of the system (the width of the miscibility gap in the temperature-com- position diagram) and the flow properties of the concentrated polymer phase. In their own words: "For amorphous or barely crystal- lizing polymers, gel formation is only to be expected if the flow temperature of the poly- mer is high enough and the low molecular weight component does not reduce this point too much". The authors' view on the mecha- nism of separation can be summarized as T -- -- -~ -- TI follows: the gel is formed by separation and growth of microsections of the dilute poly- mer phase, leaving the concentrated poly- mer phase as a gradually stiffening matrix, until its final equilibrium concentration is reached. This is the well-known mechanism of nucleation and growth of one of the equi- c E I F\ ,~ librium phases. ', Recently, Cahn and others (5, 6) have de- scribed a second type of mechanism of phase separation: the spinodal decomposition of a homogeneous solution. This mechanism may lead to typical network structures in isotropic Fig. 1. Diagrams of the free energy of mixing (AFro) ~nd media, as has been shown for inorganic temperature (T) versus composition (q3) of a two com- glasses (7). ponent system polymer-solvent with a miscibility-gap Smolders et al., Liquid-liquid phase separation in concentrated solutions 15 liquid phase separation in polymer solutions polymer, would be formed. This is in accord- (two components). Lowering the temperature ance with the picture usually given (3, 4) for of a homogeneous solution of composition phase separation in solutions of amorphous r at T 1 to below the temperature T r (on the polymers. binodial curve) makes phase separation ener- Is should be realized that, especially at getically favourable. When equilibrium has polymer concentrations higher than a few been reached the separated phases will have percent, undercooling below the binodial curve attained the compositions CA and qSB at T2, can easily occur before effective nucleation is q~c and q)D at T 3, these compositions being possible. This may be a reason why so-called given by the common tangents of the AF,~-q) cloud point curves lie below the binodial curves at these temperatures. The binodial branches for high polymer concentrationsl). curve in the temperature-composition dia- gram thus collects all the equilibrium com- 2. Spinodal decomposition positions. In the composition region within the spi- Another set of characteristic points on the nodal curve in the T-~b diagram the solution AFrn-~b curves are the inflection or spinodal will not be stable to fluctuations in concen- points, e.g. points E and F at T a. These tration, of however small an amplitude. The points mark the composition region within theory of spinodal decomposition as given by which the system, after rapid cooling to T3, is Cahn (5) predicts that small fluctuations in unstable as a homogeneous system, whereas concentration of wavelengths above a certain in the composition regions CE and FD the critical value, Ac, will spontaneously increase system is metastable at T~. The inflection in amplitude. This means that the difference points in the zlF~-q) diagram at different in concentration in adjacent regions in the temperatures collect to give the spinodal solution, at fixed distances apart, increases curve of the temperature-composition dia- in a continuous way. Cahn also showed that, gram (dotted line in fig. 1 b). although all concentration fluctuations of A > Ae do increase in amplitude with time, Two mechanisms of liquid-liquid phase separa- the rate of increase depends on the fluctuation tion wavelength. Fluctuations of a wavelength N~ow apparently two different mechanisms Dm ~ As 1/2 will grow fastest. of phase separation are possible: The theory describes the earliest stages of the process only. If the spinodal decomposi- 1. Nucleation and growth tion pattern is conserved in later stages, In the regions between the binodial and which can be expected for strongly increasing spinodal curves in the T-r diagram the solu- viscosity of the regions increasing in polymer tion is stable to ordinary small amplitude concentration, a fairly uniform structure will fluctuations in concentration, but unstable to result, with a length parameter Dn occurring concentration fluctuations (which may be most frequently as the distance between called nuclei) having compositions close to polymer-rich regions. The morphological the other extreme of the miscibility gap. structure to be expected from the spinodal Such a nucleus will spontaneously grow in mechanism is a more or less rigid mass con- size as soon as its radius has surpassed a sisting of interconnected regions of the con- critical value, following the familiar pattern centrated polymer phase, embedded in the known from liquid separation from the gas dilute polymer phase (i. e. practically pure phase and from crystallization phenomena. solvent.) Cahn suggests that the relative Eventually the solution will become a disper- volume fraction of the minor phase present sion of droplets of one of the equilibrium ~) We would like to point out here that as a model for phases in the other one. concentrated polymer solutions such a system should For concentrated polymer solutions with preferably be considered as an assembly of polymer (J~ > (Jbcr (while q~cr ~< 10% by weight), one segments and solvent molecules with a practically expects nuclei of the practically pure solvent homogeneous distribution of segments over the solution. Concentration fluctuations can then be described as phase to be formed. Depending on the flow fluctuations in segment density or as fluctuations in the properties of the polymer phase matrix, concentration of solvent molecules. Hence the influence which increases in concentration, a more or of the properties of the individual polymer molecules (e. g. their m. w. distribution) on nucleation phenomena less rigid mass with solvent-filled holes or, and on the cloud point curve - as is experiencedat low at the other extreme, two liquid layers, one polymer concentrations- becomes unimportant at high practically pure solvent, the other rich in polymer concentrations. 16 Kolloid-Zeitschri]t und Zeitschri/t /i~r Polymere, Band 243 Heft 1 should at least be 15% in order to maintain Survey of observations interconnectivity of this phase. For polymer solutions with a typical asymmetrical binod- We have made a number of observations ial curve this means that this interconnected (among which light scattering measurements, structure can already be expected to be dilatometry, normMand electron microscopy) realizable above a few percent of polymer, by on concentrated polymer solutions, upon weight or volume. As an example: taking the quenching these solutions to sufficiently low binodial compositions at about 0~o and 30% temperatures for phase separation to occur. volume of polymer respectively, the separated The results of these observations can be mixture contains more than 15% by volume explained by a spinodal decomposition mecha- of the polymer phase if the total polymer nism at early stages of the phase separation content of the mixture is above 4.5% by process, followed by other, slow phenomena volume. At a lower polymer content spinodal at high enough quenching temperatures. decomposition would give separated polymer- Since not enough thermodynamic data rich regions, a distance/)m apart. were available to compute the spinodM tem- If the polymer-rich phase has sufficient peratures at different compositions, we had mobility at all concentrations, the primary to rely on other methods for estimating these regions may form larger aggregates and even- properties (see below). tually the splitting up into two bulk liquid We want to stress again that concentrated phases becomes possible. Also the case of a solutions may easily be undercooled below contracting gel phase (8) covered by a layer the binodial curve without phase separation of the "solvent" phase of changing composi- of the nucleation and growth type taking tion can easily be explained on the basis of place. On the other hand, phase separation the gel structure formed in an early stage and always occurs below the spinodal tempera- at relatively low total polymer concentration ture, since then the homogeneous system is by the spinodal decomposition mechanism.