Miscibility and Crystallization in Polycarbonate/Poly(ǫ-caprolactone) Blends: Application of the Self-Concentration Model Dinorah Herrera, † Jean-Carlos Zamora, ‡ Alfredo Bello, † Mario Grimau, ‡ Estrella Laredo,* ,† Alejandro J. Mu 1 ller, ‡ and Timothy P. Lodge § Physics Department and Materials Science Department, Universidad Simo ´ n Bolı ´var, Apartado 89000, Caracas 1080-A, Venezuela, and Department of Chemistry and Department of Chemical Engineering & Materials Science, University of Minnesota, Minneapolis, Minnesota 55455-0431 Received March 7, 2005; Revised Manuscript Received April 22, 2005 ABSTRACT: Polycarbonate/poly(ǫ-caprolactone) (PC/PCL) blends are found to be miscible when extruded samples are studied without any further thermal treatment. PCL crystallizes in blends containing 60% or less polycarbonate, a component that remains amorphous for all blend compositions under these conditions. Single, broad calorimetric glass transitions together with distinct component dynamics determined by thermally stimulated depolarization current experiments indicate the miscibility of the blends and the existence of different average local compositions. The Lodge-McLeish model is applied to the compositional variation of the two effective glass transition temperatures. Quantitative agreement is obtained for both components by adjusting the self-concentration values to best fit the experimental points. The relevant length for PCL is very close to its Kuhn length, whereas for PC the best fit leads to a slightly shorter characteristic length. It is shown that upon annealing at sufficiently high-temperature PC undergoes crystallization and thereby induces phase segregation in the otherwise amorphous regions of the blends. 1. Introduction Mixtures of bisphenol A polycarbonate (PC) and poly- (ǫ-caprolactone) (PCL) have been extensively studied, and many apparently conflicting results have been obtained on this system, to the point that it has been labeled the “Gordian blend system” by Varnell et al. 1 The blends have been regarded by many authors as fully miscible on account of the finding of one calorimetric glass transition temperature, T g , that varies with composition within a range encompassed by the T g ’s of the pure components. 2-8 However, the temperature range for the T g calorimetric step in the blends is broader than in the neat components, and this broaden- ing is nonsymmetric with composition. In a preceding work on this system, 9 we demonstrated that blends that had been stored at room temperature for 18 months and that were originally regarded as miscible 8 (since they had exhibited a single T g as determined by differential scanning calorimetry, DSC) were now either partially mixed or phase separated, depending on their thermal treatment. Phase separation was demonstrated by DSC, transmission electron microscopy, spherulitic growth rate measurements, and thermally stimulated depolar- ization currents (TSDC). The driving force behind phase segregation was attributed to the slow PCL crystalliza- tion during storage. The only way to restore miscibility in these phase-separated blends was to subject them to extrusion or dissolution and reprecipitation. This particular system is interesting in that both components may crystallize. Consequently, at any tem- perature the blend might form an amorphous, one-phase mixture or undergo liquid-liquid phase separation. Below the crystallization temperature of PC (ca. 230 °C) or PCL (ca. 60 °C) crystallization of either (or both) components could occur, leading to two or three coexist- ing phases, with interfacial zones of varying composi- tion. Furthermore, the glass transitions of the compo- nents are extremely widely separated (140 °C for PC, -66 °C for PCL). Both the structure and the dynamics in such a system are expected to be quite rich and thermal history dependent. In a miscible blend, in addition to the intermediate and broad calorimetric glass transition, distinct com- ponent dynamics have often been found. This “dynamic heterogeneity” has been widely studied, and several explanations have been proposed. One approach is that the local concentration variations result from the effect of thermal concentration fluctuations. 10-12 Another ap- proach by Lodge and McLeish 13 emphasizes chain con- nectivity effects or “self-concentration”, which result in effective glass transitions for each component, T g eff (φ), based on the average local composition perceived by each component. These may be quite different from that of the bulk composition. Various experiments, such as dielectric spectroscopy if the two blend components are polar, are able to distinguish different dynamics, i.e., different T g eff (φ) for each component, in agreement with the dynamic heterogeneity explained by the self- concentration model. The model is able to rationalize the compositional variations of the T g eff (φ)’s. The origi- nal calculations were based on the Kuhn segment length, l K , as the relevant length scale for the self- concentration application. Once the self-concentrations φ SA and φ SB are calculated from the volume fraction occupied by a Kuhn length’s worth of monomer within a volume V ∼ l K 3 , the effective local concentration is estimated for each component, φ effA and φ effB . The Fox equation is then used with the effective concentrations to calculate the effective glass transition temperatures for components A and B, T gA eff (φ) and T gB eff (φ). One im- portant consequence of this approach is that a fully miscible, molecularly mixed blend can still exhibit two † Physics Department, Universidad Simo ´n Bolı ´var. ‡ Materials Science Department, Universidad Simo ´n Bolı ´var. § University of Minnesota. * Corresponding author. E-mail elaredo@usb.ve. 5109 Macromolecules 2005, 38, 5109-5117 10.1021/ma050481c CCC: $30.25 © 2005 American Chemical Society Published on Web 05/18/2005