Effect of PBT Molecular Weight and Reactive Compatibilization on the Dispersed-Phase Coalescence of PBT/SAN Blends

Transcription

1 Effect of PBT Molecular Weight and Reactive Compatibilization on the Dispersed-Phase Coalescence of PBT/SAN Blends NELSON M. LAROCCA, 1 EDSON N. ITO, 2 CARLOS TRIVEÑO RIOS, 3 LUIZ A. PESSAN, 1 ROSARIO E. S. BRETAS, 1 ELIAS HAGE Jr. 1 1 Department of Materials Engineering, Federal University of São Carlos, Rod Washington Luiz, Km 235, , São Carlos, SP, Brazil 2 Department of Materials Engineering, Federal University of Rio Grande do Norte Campus universitário s/n, , Natal, RN, Brazil 3 Department of Mechanical Engineering, Federal University of Mato Grosso Rodovia Guiratinga, km 06, Rondonópolis, MT, Brazil Received 1 March 2010; revised 22 June 2010; accepted 4 July 2010 DOI: /polb Published online 2 September 2010 in Wiley Online Library (wileyonlinelibrary.com). ABSTRACT: Poly(butylene terephthalate) (PBT)/styrene-acrylonitrile copolymer (SAN) blends were investigated with respect to their phase morphology. The SAN component was kept as dispersed phase and PBT as matrix phase and the PBT/SAN viscosity ratio was changed by using different PBT molecular weights. PBT/SAN blends were also compatibilized by adding methyl methacrylate-co-glycidyl methacrylate-co-ethyl acrylate terpolymer, MGE, which is an in situ reactive compatibilizer for melt blending. In noncompatibilized blends, the dispersed phase particle size increased with SAN concentration due to coalescence effects. Static coalescence experiments showed evidence of greater coalescence in blends with higher viscosity ratios. For noncompatibilized PBT/SAN/MGE blends with high molecular weight PBT as matrix phase, the average particle size of SAN phase does not depend on the SAN concentration in the blends. However noncompatibilized blends with low molecular weight PBT showed a significant increase in SAN particle size with the SAN concentration. The effect of MGE epoxy content and MGE molecular weight on the morphology of the PBT/SAN blend was also investigated. As the MGE epoxy content increased, the average particle size of SAN initially decreased with both high and low molecular weight PBT phase, thereafter leveling off with a critical content of epoxy groups in the blend. This critical content was higher in the blends containing low molecular weight PBT than in those with high molecular weight PBT. At a fixed MGE epoxy content, a decrease in MGE molecular weight yielded PBT/SAN blends with dispersed nanoparticles with an average size of about 40 nm. VC 2010 Wiley Periodicals, Inc. J Polym Sci Part B: Polym Phys 48: , 2010 KEYWORDS: blends; coalescence; compatibilization; morphology; PBT; SAN INTRODUCTION The physical properties of immiscible polymer blends are highly dependent on the blend s morphology, that is, the size, shape and concentration of the dispersed phase. 1 It is well-known, for instance, that to produce a very tough blend of a rubber material and a brittle polymer, the rubber material must be well-dispersed in the brittle matrix and the dispersed particles should have an optimal size and interparticle distance. 2 The final morphology of immiscible blends depends on the competing processes of breakup and coalescence of droplets during blending of the melt. 3 These processes, in turn, depend on the blend s composition, viscosity ratio, interfacial tension and processing conditions. Thus, knowing the degree to which each of these parameters influence the breakup and coalescence processes enables one to achieve the desired morphology by finely adjusting these parameters. Theoretical Insights About the Effects of Droplet Breakup and Coalescence The deformation of droplets in a matrix is governed by the balance between the external shear stress of the matrix, s m, which tends to deform the droplets, and the interfacial stress, given by the interfacial tension, C, which tends to keep the cohesion of the particle. The capillary number, Ca, quantifies the ratio between these two stresses: Ca ¼ g m _cr (1) C where g m is the matrix viscosity, _c is the shear rate and R is the radius of the dispersed particle. If the capillary number exceeds a critical capillary number, Ca c, the external shear stress on the particle is too large to be compensate by the interfacial tension, and the particle becomes unstable and Correspondence to: E. Hage Jr. ( elias@ufscar.br) Journal of Polymer Science: Part B: Polymer Physics, Vol. 48, (2010) VC 2010 Wiley Periodicals, Inc WILEYONLINELIBRARY.COM/JOURNAL/JPOLB

2 ARTICLE can breakup. For Newtonians fluids in a simple shear flow, Taylor 4,5 has shown that Ca c becomes: Ca c ¼ 0:5 16g r þ 16 19g r þ 16 where g r is the viscosity ratio, that is, the ratio between the droplet viscosity and the matrix viscosity. This equality, however, holds true only for g r < 1. 6 Grace s experimental data 7 has shown that in shear flows the Ca c decreases from 0.1 < g r < 1, reaches a minimum at g r ¼ 1 and becomes infinity when g r reaches the value of 4. This behavior can be described by the empirical equation of De Bruijn, 8 which fits the Grace s data: log Ca c shear ¼ 0:506 0:0995 log g r þ 0:124ðlog g r Þ 2 0:115 log g r log 4:08 The infinite value of Ca c at viscosity ratio of about 4 means that the breakup in shear flows is not possible for g r > 4 due to the rotation of the viscous droplets in the shear flow. However, in an elongational flow the breakup occurs for any g r value. 6,7 The critical capillary number in purely elongational flow can be forecasted trough the following empirical equation, which fits well the experimental Ca c data available in literature for this kind of flow: 9 log Ca c elong ¼ 0: :02442 log g r þ 0:02221ðlog g r Þ 2 0:00056 log g r log 1:015 An estimation of the droplet size in a shear or elongational flow can be achieved by using eqs 3 or 4, respectively, combined with the following equation: (2) ð3þ ð4þ R ¼ Ca c g m c=c 2 1=3 (5) However, this estimation becomes more inaccurate as the dispersed phase concentration is increased due to the increasingly influence of another process: the coalescence. The coalescence process in blends can be summarized as a two-step process. The first step involves collision of the droplets during the melt flow. Droplet collision frequency can be described by Smoluchowsky s theory, 10,11 which predicts that this frequency is proportional to the flow s shear rate, the volume of droplets, and the squared number of droplets per unit volume. Thus, the higher the concentration of the dispersed phase the larger the number of droplets per unit volume, and hence, the higher the collision frequency. In the second step, the droplets are deformed during collision and a matrix phase film is formed between the droplets. This film is continually squeezed and drained out of the gap until it reaches a critical thickness, h c. In this condition, rupture of the interface film may occur, leading to fusion of the droplets. From these processes it can be inferred that not all collisions of droplets result in coalescence, since the interaction time between the droplets in the second step must also be longer than the time required to drain the liquid matrix film. Because of the drainage step, coalescence is dependent on the viscosity of the matrix, which affects the drainage time. Droplet size is also important in the drainage process, since the larger the droplets the larger the area covered by the film between the droplets, and therefore, the longer the time needed to drain the liquid matrix film. 3,10 13 Besides the adjustment of the parameters that control the breakup and coalescence steps, such as g m, g r, _c, among others, another efficient way to control blend morphology is by incorporating interfacial copolymers or compatibilizers into polymer blends. These copolymers can be prepared previously as block copolymers and added to the blend, or they can be generated in situ at the interface by specific chemical reactions during mixing of the blend melt. Compatibilization can reduce the size of the dispersed phase considerably by reducing the interfacial tension between the phases and/or by increasing the coalescence suppression. Coalescence can be suppressed by steric hindrance of the dispersed phase due to the compatibilizer molecules at the interface However, it has been proposed more recently that interfacial copolymers can also suppress coalescence by delaying the rupture process of the matrix film between the droplets. 19 The aim of this study was to examine the influence of some variables such as matrix molecular weight, dispersed phase content and compatibilizer functionality on the coalescence characteristics of PBT/SAN blends. The copolymer used as reactive blend compatibilizers was prepared from methyl methacrylate (MMA), glycidyl methacrylate (GMA) and ethyl acrylate (EA) monomers, and is hereafter referred to as MGE. The compatibilization mechanism is triggered by the miscibility of MGE molecules containing up to 20 wt % of GMA in SAN copolymers containing 25 wt % of acrylonitrile. 20 Furthermore, the epoxy groups of the GMA comonomer are known to react with the carboxyl end groups of polyester molecules, such as PBT Thus, during mixing of the blend melt, the MGE molecules, which are miscible with the SAN phase, can react with the PBT matrix molecules, forming in situ graft PBT-g-MGE molecules, which remain preferentially at the blend interface. As a result, these graft molecules may decrease the interfacial tension and suppress the effects of coalescence. The PBT/SAN blend has been studied because it is a good model to observe the interfacial properties of PBT/ABS blends, and of the more recent PBT/AES blends. 32,33 ABS and AES are two-phase polymer systems in which a rubber phase polybutadiene in the case of ABS and ethylene-propylene-diene in that of AES is dispersed in a SAN matrix. In PBT/ABS and PBT/AES blends, an interface PBT/ SAN is generated between the SAN phase from ABS or AES and the PBT matrix phase. In PBT/SAN blend this interface and the coalescence characteristics can be studied without the influence of the ABS and AES rubber phase. COALESCENCE OF PBT/SAN BLENDS, LAROCCA ET AL. 2275

3 JOURNAL OF POLYMER SCIENCE: PART B: POLYMER PHYSICS DOI /POLB TABLE 1 Molecular Characteristic of the PBT Grades Used in This Work Polymers M n (Kg/mol) a [COOH] (l eq/g) PBT 20 (Valox 195) a 29 PBT 40 (Valox 315) a 55 a By diluted solution viscosimetry. EXPERIMENTAL Materials PBT resins with two different molecular weights were used in this work. These resins, whose commercial names are Valox 195 and 315, are produced by SABIC Innovative Plastics. These PBT grades were used to explore the effects of molecular weights and carboxyl end group concentrations. Their main characteristics are shown in Table 1. The SAN grade CN77E, with a number-average molecular weight of 47 Kg/mol and a acrylonitrile content of 25 wt %, was supplied by Bayer Polymer S.A. This SAN grade was chosen due to its molecular similarities with the SAN phase of a higher rubber content ABS, which is normally used to toughen PBT/ABS blends. 25 The MGE compatibilizers were synthesized by bulk polymerization using MMA, GMA, and EA as comonomers. The latter comonomer was incorporated at 2 wt % to prevent unzipping, which usually occurs in acrylic polymers during high temperature melt processing. 28 To prepare the MGE with several GMA contents having the same molecular weight, the compatibilizers were synthesized by low conversion rate copolymerization (15%), using azobisisobutyronitrile (AIBN) as initiator. Details of the synthesis of MGE are described elsewhere. 20,33 Table 2 shows the results of the GPC and 1 H NMR analysis of MGE. The MGE 0 represents the MGE copolymer without GMA, which is expected not to react with PBT end groups. The measured amounts of GMA in the MGE samples were kept close to the nominal values. It can be also seen that the amount of GMA does not influences significantly the number-average molecular weights of the copolymers. The MGE molecular polydispersity index, Mw M n, obtained was very close to 1.5 for each sample. In addition to synthesizing these MGE samples, a low molecular weight sample containing 10 wt %ofgma(mge 10 -LMW) was also synthesized by fractionating the MGE 10 samples using methanol and chloroform as nonsolvent and solvent, respectively. As Table 2 indicates, the MGE 10 and MGE 10 -LMW samples have quite distinct number-average molecular weights. These MGE samples, both containing 10 wt % of GMA, were used to study the effect of the compatibilizer s molecular weight on the PBT/SAN morphologies. It should be pointed out that, based on the results reported by Gan and Paul, 20 all the synthesized MGE samples described in Table 2 areexpectedtobefullymiscibleinthesanphaseofpbt/san blends. To ensure that the PBT and MGE molecules would react in situ at the PBT/SAN interface, MGE was previously blended with SAN by solution blending before melt blending with PBT in the batch mixer. This procedure was expected to increase the molecular miscibility between MGE and SAN. The solution was blended by dissolving the SAN and MGE, in a weight proportion of SAN/MGE 75/25, in chloroform, followed by precipitation of the mixture into powder form using an excess of methanol. The resulting powder was then melt-blended with the PBT in an internal mixer. The SAN/MGE solution blend is referred to here as (SANþMGE). Jeon and Kim 34 used a similar procedure to prepare PBT/polystyrene (PS) blends with PS-GMA copolymer, first premixing the reactive compatibilizer PS-GMA in the PS phase by solution blending. They concluded that the PBT-graft-PS-co-GMA graft molecules were more efficiently located at the interface PBT/PS blend than in blends prepared without the solution premixing procedure. Compatibilized PBT/PS blends prepared by adding PS-GMA through melt blending showed micelles of PBT-graft-PS-co-GMA graft molecules dispersed in the PBT phase. Melt Processing All the materials were vacuum dried overnight at 60 C before being melt-blended. Melt blending was carried out in an internal mixer from Haake, designed as a System 90 torque rheometer with a series 600 batch mixer and roller rotors. The temperature of the intensive mixer plates was set at 240 C. The blends were prepared using a rotor speed of 50 rpm. PBT was always put into the mixing chamber first, after which the proper amount of SAN or (SANþMGE) was added. Most of the blends were mixed for 10 minutes, after which samples were quickly removed from the internal chamber and immediately quenched in liquid nitrogen. In the case of a few blends, to better analyze the coalescence process, the rotor speed was reduced to 5 rpm after 10 min of mixing and this speed was maintained for another 20 min. The samples were then collected as described earlier. This sample collecting procedure was also used by Sundararaj and Macosko 3 and Wildes et al. 35 The blends were dubbed as follows: PBTx/(SANþMGE y ) (A/B), where x indicates the PBT molecular weight (20 or 40 Kg/mol), y indicates the nominal GMA weight % content in the MGE, A represents the weight % of PBT, and B represents the weight % of the mixture (SANþMGE). TABLE 2 Molecular Characteristics of Synthesized MGE Terpolymers Copolymer Designation a GMA (wt %) b M n (kg/mol) c MGE MGE MGE MGE MGE MGE 10 -LMW MGE a The underline MGE numbers represent the GMA previously established weight % content. b Measured by 1 H NMR. c Measured by GPC. Mw Mn 2276 WILEYONLINELIBRARY.COM/JOURNAL/JPOLB

4 ARTICLE Analysis of the Samples A scanning electron microscope (SEM, Philips XL-30FEG) was used to examine the blend morphologies of most of the samples. Brittle fractures of the samples were obtained by cryofracturing them in liquid nitrogen. The dispersed phases, SAN or SANþMGE were extracted from the samples fractured surfaces using THF solvent. The morphology of blends containing low molecular weight MGE was examined by transmission electron microscopy (TEM, Philips CM120). These samples were cryo-ultramicrotomed at 50 C into 40 nm slices and stained with ruthenium tetroxide (RuO 4 ) vapor for 2 hours. The particle size was analyzed using an Image-Pro Plus 4.5 image analyzer. The number-average particle size was determined from about 300 particles. RESULTS AND DISCUSSION FIGURE 1 SEM micrographs of fracture surfaces of PBT/SAN blends prepared at 240 C and 50 rpm for 10 minutes in the Haake intensive mixer. PBT20/SAN blends with the following SAN wt % contents: (a) 30, (c) 20, (e) 10, (g) 5, (i) 2.5. PBT40/ SAN blends with the following SAN wt % contents: (b) 30, (d) 20, (f) 10, (h) 5, (j) 2.5. PBT/SAN Binary Blends SEM micrographs of the fracture surfaces of the noncompatibilized binary PBT/SAN blends are shown in Figure 1. Most of the dispersed SAN particles show a regular spherical shape. The increase in PBT molecular weight led to a marked decrease in SAN particle size, mainly in blends with lower SAN contents. Note that as the PBT molecular weight increased the size of these particles decrease correspondingly. As Table 3 indicates, the higher the molecular weight of PBT the higher its melt viscosity, g*, and elasticity, which is proportional to the melt s shear modulus, G 0. Thus, the higher the molecular weight of PBT the higher the melt stress on SAN particles, since PBT is the matrix in PBT/SAN blends. An increasing stress level, in turn, led to a more intense breakup of the SAN phase, decreasing its particle size. Moreover, as the SAN content increased, the SAN particle size increased steadily as a result of poorer dispersion, mainly due to coalescence effects. This tendency can be explained by Smoluchowski s coalescence theory, 10,11 which predicts that the greater the number of dispersed particles the greater the probability of collision between them. Figure 2 quantifies the aforedescribed effects, showing curves of the particle sizes as a function of SAN content, which were taken from the micrographs in Figure 1. The number-average SAN particle size was plotted as a function of the SAN content for PBT/SAN blends with PBT20 and PBT40, respectively. The curves in this figure indicate a broadening of the particle size distribution, represented by the taller standard deviation bars as the SAN concentration increased. This may be the result of significant coalescence and breakup of droplets occurring simultaneously at higher SAN concentrations, since small particles may be formed by breaking droplets in the mixer s high shear regions, while large particles are formed by coalescing droplets in its low shear regions. 3 TABLE 3 Rheological Parameters for PBT/SAN Blends Matrix of (99/1) PBT/SAN Blends C (mn/m) a g* m G 0 (Pa s) b (Pa) c g r d D shear (lm) e D elong (lm) f Average Diameters of SAN Particles at 1 wt % SAN in the Blend (lm) PBT PBT ,071 17, a PBT/SAN Interfacial tension, measured by the droplet retraction method. 36 b Matrix viscosity, measured by oscillatory rheometry at an angular frequency of 68 rad/s. c Matrix Shear Modulus, measured by oscillatory rheometry at an angular frequency of 68 rad/s. d Viscosity ratio at an angular frequency of 68 rad/s. e Calculated by combining eqs 3 and 5. f Calculated by combining eqs 4 and 5. COALESCENCE OF PBT/SAN BLENDS, LAROCCA ET AL. 2277

5 JOURNAL OF POLYMER SCIENCE: PART B: POLYMER PHYSICS DOI /POLB FIGURE 2 Comparison of the number-average diameter of SAN particles versus wt % dispersed phase of the PBT20/SAN and PBT40/SAN blends. Since the curves in Figure 2 appear to level off when the SAN content is decreased, it can be assumed that coalescence is milder at a SAN concentration close to 1 wt %. Therefore, in blends with this very low SAN content, SAN particle size depends mostly on the breakup process and it would be interesting to compare the measured average size of SAN particles at this concentration against the estimated diameters calculated by using the eqs 3 and 4, which consider only the breakup process by a shear flow and by an elongational flow, respectively. We assume that besides shear flow the internal mixer also allows for elongational flow in the blends because the breakup process took place even in the blend PBT20/SAN, which have a g r of 6.5. As described earlier, in a shear flow the breakup is not predicted to take place for g r higher than 4, but in an elongational flow the breakup can occurs at any g r value. 6,7 Other authors have described similar observations in other systems at high viscosity ratios. 3,37 39 We calculated the theoretical average particle diameter in a shear flow, D shear, and in a elongational flow, D elong, by using eqs 3 and 4, respectively, combined with the eq 5. Taking into account the geometry of the internal mixer used here, the maximum calculated shear rate applied to the melt was 68 s 1 at 50 rpm. g m and g r were determined at that shear rate from the viscosity versus shear rate curves, and the C values were obtained from the results of previous work. 36 Table 3 summarized all these data and the calculated values of D shear and D elong. It should be observed that D shear for the blend PBT20/SAN could not be determined since, as described earlier, no breakup is predicted in shear flow for a blend with such high viscosity ratio. However, the elongational flow of the internal mixer is predicted to lead to an average particle diameter D elong of 0.05 lm. For the PBT40/SAN blend, because of its low viscosity ratio, the breakup can takes place whether by shear or elongational flow and for both flows the same average particle diameter value of 0.01 lm is predicted. Therefore, the predicted average particle diameters for the blends with PBT20 and PBT40 are 0.05 and 0.01 lm, respectively which are significantly lower than the measured ones, as shown in Table 3. These differences might be due to the fact that eqs 3 and 4 are fittings for Newtonians systems. In the PBT/SAN system the components are non-newtonian viscoelastic fluids in the melt state. Viscoelastic effects tend to hinder the breakup process, since the particle s elasticity resists the shear forces responsible for breaking the droplet. 13,40,41 To gain further insights about the intensity of coalescence in the PBT/SAN blends, the PBT20/SAN and PBT40/SAN blends were subjected to quasi-static coalescence experiments. Applying the procedure described previously in the Experimental section, the evolution of coalescence was inferred by comparing the morphologies of the blends mixed for 20 minutes at low shear rates with the morphologies of the PBT/SAN blends depicted in Figure 1, which represent the morphologies of the blends at the beginning of the coalescence process. Figure 3(a,b) show the number-average particle diameter before and after quasi-static coalescence in the two PBT/SAN blends. Overall, there was a substantial increase in SAN average particle size after the quasi-static coalescence treatment. The coalescence level was estimated from the difference between D n at 20 minutes and at 0 minutes of coalescence. As the SAN concentration increased, the coalescence level also increased in both PBTs. These results were expected, since the probability of collisions between the SAN droplets would increase with increasing SAN concentrations. Another remarkable change was the wider particle size distribution of the SAN particles, which is reflected in the error bars due to the standard deviations plotted in Figure 3(a,b). Coalescence usually causes an increase in particle size dispersion. 10,42 However, if some small particles do not find a partner of equal size with which to coalesce at the beginning of the process, these particles remain stable for long times and can coexist with large particles at the end of the coalescence process. With regard to the effect of the molecular weight of the PBT matrix, the lower viscosity PBT matrix (PBT20) was found to promote greater coarsening of the SAN particles at a given SAN concentration, as shown in more detail in Figure 4 for the SAN content of 20 wt %. This suggests that blends with PBT20 have a larger coalescence rate than blends with PBT40, which is probably due to the lower viscosity of PBT20. As the PBT matrix viscosity decrease, the collision forces between SAN particles is reduced and therefore the particles deformation upon collision is decreased as well. As a result, the area of the matrix film between two SAN particles is expected to be smaller and the drainage time is thus shorter. Furthermore, the drainage time is expected to decrease further due to the lower viscosity of the matrix between SAN particles Compatibilized PBT/SAN/MGE blends Effect of SAN Concentration and PBT Type on Compatibilized Blends Figure 5(a,b) show the dependence of the number-average particle size on the amount of SAN phase containing 25 wt % 2278 WILEYONLINELIBRARY.COM/JOURNAL/JPOLB

6 ARTICLE FIGURE 3 Number-average diameter of the SAN phase before and after the experiments of static coalescence versus SAN concentration in the PBT/SAN blend; (a) PBT20/SAN blends; (b) PBT40/SAN blends. of MGE 10. Curves were also plotted for the noncompatibilized blends to facilitate comparisons. Among the PBT/SAN blends, the compatibilized ones showed less dependence of the dispersed phase concentration on the particle size diameter, and the PBT40/SAN compatibilized blends showed almost no dependence at all. To ascertain the role of the reactive compatibilizer on the breakup of SAN particles, the particles sizes were examined at the lowest dispersed phase concentration, that is, 1 wt % of SAN. Based on the assumption that coalescence is absent at the lowest concentration of dispersed phase, a reduction in particle size at this SAN concentration could be attributed only to an decrease of the interfacial tension by the compatibilizer, as one would deduced from eq 1. 3 An analysis of the behavior of PBT40/SAN compatibilized blends shown in Figure 5(b) indicates that, at the lowest SAN content, the dispersed phase particle size in the noncompatibilized blend is nearly the same as in the compatibilized one. This also holds true for the PBT20/SAN blends at 1 wt % of dispersed phase, considering overlapping of the error bars for the average particle size [Fig. 5(a)]. Thus, the compatibilizer has a negligible effect on the breakup of these blends, a behavior which can be ascribed to two possible effects as follows. The first possibility is that the density of graft copolymers formed at interface, R, is not high enough to lead to a meaningful decrease of interfacial tension and consequently to a decrease of particle size, since it has been shown that the decrease of interfacial tension depends on R The second possibility is that R is indeed high enough and the decrease of interfacial tension is meaningful but breakup is not proportional to the interfacial tension of the compatibilized SAN particles. In fact, some works have shown that even though interfacial compatibilizers do reduce the interfacial tension, compatibilized droplets can resist to deformation and breakup possibly because of development of gradients of interfacial tension (Marangoni stresses) induced by gradients of compatibilizer concentration at interface due to flow. 16,46 Since the role of the reactive compatibilizer MGE 10 on the breakup process of SAN particles in the blends seems to be negligible (at least for the concentration of the MGE 10 utilized in the blends), any change in the morphology of the binary blends due to the presence of MGE 10 can be ascribed to the suppression of coalescence. In the case of PBT40/SAN blends [Fig. 5(b)], it can be concluded that compatibilization effectively suppresses particle coalescence, since within their compositional range, the increase in D n values is negligible compared to the value at 1 wt % content. On the other hand, the D n values for the PBT20/SAN compatibilized blends [Fig. 5(a)] began to increase in compositions above 2.5 wt %, indicating that above this SAN content the compatibilizer did not suppress coalescence completely in these blends. The origin of these different effects of compatibilization on the coalescence of the blends with PBT20 and PBT40 will be discussed ahead. FIGURE 4 Comparison of the level of static coalescence for the PBT20/SAN and PBT40/SAN blends containing 20 wt % of SAN. COALESCENCE OF PBT/SAN BLENDS, LAROCCA ET AL. 2279

7 JOURNAL OF POLYMER SCIENCE: PART B: POLYMER PHYSICS DOI /POLB Effect of Epoxy Concentration on the MGE/ PBT In Situ Reactions To gain a better understanding of the chemical reactivity between the compatibilizer, MGE, and PBT in the compatibilized blends, further experiments were performed by varying the GMA content in the MGE copolymer and maintaining a constant SAN concentration in the PBT/SAN blend. Thus, the amount of GMA epoxy groups that react with PBT molecule end groups was varied. Figure 6(a,b) show the number-average particle diameters plotted as a function of GMA in the MGE copolymer in the PBT/SAN compatibilized blends. In addition, Figure 7 shows the effect of the epoxy content on the relative SAN particle diameter, established as the D n /D no ratio, where D no is the SAN particle diameter for MGE without GMA. As Figure 7 indicates, the D n /D no ratio in PBT40/ SAN blends rapidly decreases down to 5 wt % of GMA, after which it decreases very slowly and apparently levels off at 10 wt % of GMA, which corresponds to 2.5 wt % of GMA concentration in the dispersed phase. On the other hand, the D n /D no ratio in the PBT20/SAN blends decreases slowly and continually from 0 to 15 wt %. The D n values for the absence of coalescence are also indicated by a dashed line in Figure 6(a,b). These are the D n values of the compatibilized blends containing 1 wt % of dispersed phase, which were determined from Figure 5(a,b). As can be seen in Figure 6(b), coalescence in the blends with PBT40 containing 20 wt % of dispersed phase is suppressed with 10 wt % of GMA concentration in the MGE terpolymer, corresponding to 2.5 wt % of GMA concentration in the dispersed phase. However, Figure 6(a) shows that coalescence in the blends with FIGURE 5 Comparison of the number-average diameter of the SAN particles versus wt % dispersed phase for the noncompatibilized and compatibilized blends: (a) PBT20 blends; (b) PBT40 blends. The dispersed phase in the compatibilized blends contains 25 wt % of MGE 10. FIGURE 6 Effect of MGE epoxide content on the size of SAN particles in the compatibilized blends containing 20 wt % SAN: (a) PBT20 blends; (b) PBT40 blends WILEYONLINELIBRARY.COM/JOURNAL/JPOLB

8 ARTICLE FIGURE 7 Comparison of the effect of MGE epoxide content on the relative decrease of the size of SAN particles in blends with PBT20 and PBT40. PBT20 was not suppressed even with 15 wt % of GMA concentration, corresponding to 3.75 wt % of GMA concentration in the dispersed phase. The effect of the MGE epoxy content on the D n of dispersed phase in the PBT/(SANþMGE) blends can be explained by the dependence of the reaction rate on the concentrations of the reactive groups, and by the relative magnitude of the reaction rate to the coalescence rate. The kinetics of the interfacial reaction between the carboxyl end groups of PBT and the epoxy functional groups of MGE can be described by a second-order rate equation: 47,48 d½graft copolymerš=dt ¼ k½coohš½epoxyš (6) where k is the kinetics constant, and [COOH] and [epoxy] are, respectively, the concentrations of carboxylic and epoxy groups, at the interface PBT/SAN. According to eq 6, the increase in GMA content in the MGE mixed into the SAN phase leads to an increase of epoxy concentration at the PBT/SAN interface, and to a corresponding increase in the graft copolymer PBT-graft-MGE formation rate at the interface. Thus, the higher the epoxy concentration the faster the increase of density of graft copolymer at the interface, R. It has been shown that coalescence may be suppressed when the interface is covered by a minimum density of graft copolymer, R min. If the reaction rate at the interface is faster than the coalescence rate, then the droplets can be covered with R min before coalescence takes place, causing these droplets to stabilize against coalescence. The effect of the reaction rate on the morphology of PBT/SAN blends containing reactive compatibilizers has already been demonstrated by Oyama et al. 49 Instead of varying the concentration of the compatibilizers reactive groups, Oyama et al. modified the reaction rate by varying the kinetics constant through the incorporation of catalysts and modification of the reactive group in the compatibilizer. The increase in the reaction rate was found to lead to a more refined morphology of the SAN dispersed in the PBT matrix. Based on the earlier discussion, an analysis of Figure 6(b) indicates the likelihood that in compatibilized blends with PBT40 containing 10 wt % of GMA concentration in MGE (MGE 10 ), the concentration of epoxy groups at the PBT40/ SAN interface allows for a sufficiently fast reaction rate to produce a density of graft copolymer at the interface higher than R min before the beginning of the coalesce, which causes suppression of coalescence at this GMA content in MGE. Actually, Figure 5(b) indicates that the interfacial coverage promoted by this copolymer is high enough to cover fast even the larger interfacial area of a blend with 30 wt % of SAN, since at this SAN concentration the D n value is still the same of the blend with 1 wt % of SAN. In another hand, for the compatibilized blends containing PBT20, Figure 6(a) suggests that more than 15 wt % of GMA in MGE would be necessary for the plotted curve to reach the dashed line. This indicates that this blend requires a higher concentration of GMA in MGE to increase the reaction rate and produce a sufficiently high density of graft copolymer at the interface to stabilize the morphology against coalescence for the blend with 20 wt % of SAN. Likewise, Figure 5(a) suggests that the reaction rate in a blend with MGE copolymer containing 10 wt % of GMA (MGE 10 ) is not high enough to generate in time the R min at interface for SAN concentrations higher than 2.5 wt %. The earlier discussion indicates that the blends with PBT20 require a higher minimum amount of GMA in MGE to suppress coalescence than those with PBT40. This can be at least partially explained by the dependence of the reaction rate on the concentrations of reactive groups in the blend. As can be inferred from eq 6, the higher the PBT carboxyl chain end groups the higher the reaction rate. Martin et al. 50 found evidence of this dependence between the carboxyl concentration and reaction rate in PBT/Polyethylene-graft- GMA blends. They reported that in these reactive blends containing PBT with high carboxyl concentration, the time required to reach a balanced morphology during melt mixing was lower than in blends containing PBT with low carboxyl content. In the present work, a lower reaction rate in blends with PBT20 than in blends with PBT40 is likewise expected, since the concentration of carboxyl chain end groups is lower in the former PBT (Table 1). Because of this lower rate, it is likely that in the compatibilized blends with PBT20 containing MGE 10 a low amount of graft copolymer is produced at the PBT20/SAN interface from the moment this area is generated by breakup up to the onset of coalescence. It is instructive to estimate how lower is the reaction rate in PBT20 blends as compared with PBT40 blends based on the stochiometry of the reactive epoxy and carboxylic groups in the blends. By using only the eq 6 we can compare the initial reaction rates in the blends (at the very beginning of the interfacial reaction) by using the initial concentration of carboxylic and epoxy groups at interface. Furthermore, if we COALESCENCE OF PBT/SAN BLENDS, LAROCCA ET AL. 2281

9 JOURNAL OF POLYMER SCIENCE: PART B: POLYMER PHYSICS DOI /POLB assume that the kinetic constant k is nearly the same for both PBT s, eq 6 indicates that blends with PBT20 and PBT40 have the same initial reaction rate when the initial concentrations of epoxy groups in the blends have the following correlation: ½epoxy PBT20 Š¼ ½COOH PBT40 Š ½COOH PBT20 Š ½epoxy PBT40 Š (7) where [epoxy PBT ] and [COOH PBT ] are the initial concentrations of epoxy and carboxyl groups in blends with the respective PBT grades. The values of [COOH PBT20 ] and [COOH PBT40 ] can be considered to be 29 and 55 lmol/g, as shown in Table 1. By comparing blends with the same concentration of dispersed phase and with the same MGE type, the ratio [COOH PBT40 ]/[COOH PBT20 ] in the blends is about 2. Therefore, it is estimated that it is necessary to double the concentration of epoxy groups in PBT20 blends to reach the same initial reaction rate of the PBT40 blends. As already stated, Figure 6(b) shows that coalescence in the blends with PBT40 containing 20 wt % of dispersed phase is suppressed when the GMA concentration in the dispersed phase is at least 2.5 wt %. Then, considering the previous discussion about the competition between reaction and coalescence rate, it can be said that the reaction rate of this blend is the minimum reaction rate necessary to suppress the coalescence in a blend with 20 wt % of dispersed phase. According to eq 7, in order to PBT20 blends reach this same initial reaction rate the estimated concentration of epoxy groups in the SAN phase should be about 5 wt %. In other words, the estimated concentration of GMA in a MGE copolymer should be at least about 20 wt % when the concentration of MGE in the SAN phase is 25 wt %. However, following the trend of the curve depicted in Figure 6(a), it seems that even at this GMA content the coalescence in these PBT20 blends would not be completely suppressed. This suggests that the initial reaction rate of PBT20 blends should be even higher than the PBT40 blends to suppress coalescence. This indicates that besides the lower concentration of carboxylic groups, others characteristics of the PBT20 also contributes to decrease the efficiency of the reactive compatibilization in suppressing the coalescence. These possible characteristics are discussed later. In addition to the low carboxyl concentration in PBT20, blends with this matrix also have a higher coalescence rate than the blends with PBT40, as shown previously. If the coalescence rate increases, the reaction rate should also increase to generate a coverage R min of graft copolymer at the interface before the onset of coalescence, and then stabilize the morphology against coalescence. Furthermore, according to the Lyu s theory of coalescence suppression by interfacial copolymers, 19 R min is inversely proportional to polymerization degree of the copolymer segment in the matrix (in this case, the PBT segment) and therefore it is estimated that the R min for PBT20 blends is about twofold larger than for PBT40 blends, what means that the reaction rate should be even further increased to compensate for this effect. This increase of reaction rate is achieved by further increasing the concentration of epoxy, as indicated by eq 6. Therefore, the GMA concentration in MGE should be sufficiently higher in blends with PBT20 to increase the reaction rate to a level which can compensate for the lower carboxyl content and the larger coalescence rate and R min of these blends. Estimation of Maximum Interfacial Coverage Some concerns can arise regarding the possibility that the coalescence took place in compatibilized blends with PBT20 due to an insufficient concentration of MGE chains available to produce R min. To address this issue, we have done an estimation of the maximum possible interfacial coverage, R max, which can be generated from the MGE concentration in the blend. By assuming that all MGE chains in the blend react with PBT chains and originate a graft copolymer and remain at interface, we can estimate R max from the following equation: MGE chains=vol R max ¼ Interface area=vol ¼ N Aq b C MGE D (8) 6M MGE u d where N A is Avogadro s number, q b is the density of the blend, 51 C MGE is the MGE concentration in the total blend, D is the diameter of the dispersed domain, M MGE is the molecular weight of MGE and / d is the volume fraction of the dispersed SAN phase. For all the blends in this work, the estimated R max values from eq 8 are about 0.2 and 0.05 chains/nm 2 for the blends with PBT20 and PBT40, respectively. To evaluate if the R max are enough to suppress coalescence, we should also estimate R min. By considering that copolymers suppress coalescence by preventing the rupture process of the matrix film between the droplets, Lyu 19 proposed that R min can be estimated as R min ¼ h c /Na 3, where h c is the critical film thickness of the matrix below which the rupture takes place and N and a are respectively the polymerization degree and the statistical segment length of the copolymer segment in the matrix (in this case, the PBT segment). By considering a value of h c about 4 nm 19 and a value of a for PBT of 1.33 nm, 34 it turns out from this expression that the values of R min are about 0.01 and chains/nm 2 for blends with PBT20 and PBT40, respectively. However, it has been shown that R min calculated by Lyu s theory is about 10 times lower the real values. 17,19 Thus, we can expect that the real values of R min are about 0.1 and 0.05 chains/nm 2 for blends with PBT20 and PBT40, respectively. By comparing the estimated value of R max in blends with PBT20 (0.2 chains/nm 2 ) with R min values (0.1 chains/nm 2, respectively), we would expect that in these blends the concentration of MGE available is more than enough to lead to coalescence suppression. This is also holds true for blends with PBT40, since for these blends is estimated that R max R min. Even though for both blends it would be expected coalescence suppression based solely on the available amount of MGE, in the PBT20 blends this suppression did not take place, which indicates that the coalescence is not 2282 WILEYONLINELIBRARY.COM/JOURNAL/JPOLB

10 ARTICLE due to stochiometric issues. Instead, this supports the argument that besides the availability of enough MGE chains in the blends to generate enough graft copolymers chains at interfaces, these graft copolymer should also be generated fast enough at interface before the coalescence takes place, that is, the suppression of coalescence depends on the ratio between the reaction rate and coalescence rate. It turns out that for the PBT20 blends this ratio is likely to be lower than for the PBT40 blends. Additional Considerations About the Interfacial Reactions It should be stressed that since PBT20 has a lower molecular weight than PBT40 the concentration of end chains is higher in the former than in the latter PBT and thus it would be expected that the concentration of carboxyl groups were higher in the former PBT, assuming that the carboxyl groups are chains end groups. However, PBT chains are also expected to have hydroxyl groups at the chain ends besides carboxyl groups. The relative concentration of hydroxyl/carboxyl groups for thermoplastics polyesters like PBT depends mainly on the type and concentration of catalysts used in synthesis during the steps of transesterification and polycondensation. 54,55 Thus, it is likely that in PBT20 the concentration of hydroxyl groups is larger than in PBT40. Despite the fact that hydroxyl groups can also react with the epoxy groups of the MGE copolymers, this reaction is far slower (10 20 times slower) than the reaction between the carboxyl-epoxy groups 56,57 and therefore it is expected that these hydroxyl end groups have a negligible role on the interfacial reactions between PBT and MGE. Some works have shown evidence that the interfacial reaction rate depends on the molecular weight of the reactive polymer, whereas it was observed that the lower the molecular weight of reactive polymers the higher the interfacial reaction rate. 58,59 Based solely on this finding, we would expect the reaction rate in blends containing low molecular weight PBT20 to be higher than in blends with high molecular weight PBT40, and therefore the suppression of coalescence in the former blends to be higher than in the latter. As shown previously, suppression of coalescence is actually higher in compatibilized blends containing PBT40 than PBT20. This suggests that if the reaction rate increases as the molecular weight of PBT decreases, this effect is eliminated by the slower reaction rate in response to the decrease of carboxyl content and by the increase of the coalescence rate. This hypothesis is corroborated by the work of Martin et al., 50 which demonstrates that the effect of the molecular weight of PBT chains on the reaction kinetics of reactive PBT/Polyethylene-graft-GMA blends is small. By comparing blends containing PBTs of different molecular weights and similar carboxyl content, the authors demonstrated that there is no significant difference in the time required to reach an equilibrium morphology during melt mixing. Up to now we have discussed only about the early stage of the morphology development of PBT/SAN blends, when the reaction rate should be fast enough to cover the fresh interface (which was created by breakup) with a concentration of graft copolymers before the advent of the coalescence process. If the reaction is slow, the coalescence takes place until the interfacial area reach a value high enough to be stabilized by the concentration of graft copolymers already formed at the interface. It is expected that after this equilibrium is reached the size of the covered particle is not changed any longer, that is, neither breakup or coalescence could take place, although the reaction at interface could continue until the interface becomes crowed of graft copolymers. It is not expect breakup even if the covered particles are too large that the capillary number becomes higher than Ca c, because, as mentioned previously, it has been showed that the compatibilized droplets can resist to deformation and breakup possibly because of development of gradients of interfacial tension (Marangoni stresses). 16,46 In addition, because of the evidences of crosslinking reactions as described below, it is also likely that the compatibilized interface is crosslinked by some extent. As showed elsewhere, 60 interfacial crosslinking in reactive systems lead to highly stable droplets against breakup and coalescence. Some concerns can also arise about the leaving of the graft copolymers from interface, due to a possible dragging effect promoted by the shear stress in the PBT matrix, that is, a hydrodynamic effect of pull-out of the copolymers. 61,62 However, it is unlikely that these copolymers can leave the interface toward the PBT phase due to the shear stress in the PBT matrix, because the high molecular weight of the MGE copolymers surely imposes a high entanglement degree with the SAN molecules, promoting a good anchoring of the graft copolymers at interface. This anchoring is expected to be even stronger when the interface is crosslinked, which seems to be likely for the compatibilized PBT/SAN blends, as discussed below. Variation of Melt Viscosity During the Formation of PBT/SAN/MGE Blends To gain further insights about the differences of reactivity between the compatibilized blends with PBT20 and PBT40, Figure 8 depicts the torque versus time curves of neat PBTs and of mixtures of PBTs with 5 wt % of MGE 10 compatibilizer. Assuming that the measured torque in this mixer is a nearly linear function of viscosity 55 and that viscosity increases during the reactions between PBT and the compatibilizer due to the increased molecular weight of reaction products (mainly PBT-g-MGE molecules), 63,64 a higher torque is indirect evidence of the reaction in progress. Thus, Figure 8 indicates that while the incorporation of MGE 10 does not promote any increase of torque in PBT20, it causes a substantial large increase in PBT40, suggesting that PBT40 is more reactive to the copolymer than PBT20. The apparent invariance of torque in the PBT20/MGE mixture is likely due to the low molecular weight and low concentration of carboxyl groups of PBT20 when compared with PBT40. We believe that the considerable increase of torque in the PBT40/MGE mixture is due to the formation of higher molecular weight graft copolymers and to crosslinked COALESCENCE OF PBT/SAN BLENDS, LAROCCA ET AL. 2283

11 JOURNAL OF POLYMER SCIENCE: PART B: POLYMER PHYSICS DOI /POLB The decrease in SAN particle size and the increase in melt viscosity are strong evidence of the formation of graft copolymers in compatibilized blends containing PBT20 and PBT40. We have attempted to remove only the homopolymer phase PBT to quantify the amount of copolymer formed in the blends by nuclear magnetic resonance (NMR) and/or infrared analysis (FTIR) in a procedure similar to that reported by Jeon and Kim 34 However, it was not possible to remove only the PBT with a solvent without also removing the copolymer. We have performed tests with various solvents that dissolve PBT, such as trifluoroacetic acid (TFA), hexafluoroisopropanol (HFIP), phenol and tetrachloroethane, but they also dissolve SAN and the copolymer. The selective removal of SAN would be easier (for instance, with chloroform), but in the present system it would be impossible to remove all the SAN, since this is the dispersed phase. FIGURE 8 Curves of torque in the internal mixer versus time for neat PBTs and the PBT/MGE 10 95/5 mixtures (240 C, 50 rpm). material formed by reactions between PBT and MGE chains in response to the high concentration of carboxyl groups of the PBT40 molecules. As a result, very large crosslinked molecules are probably formed, giving rise to high melt viscosity. Further evidence of such crosslinking reactions has been reported by Martin et al. 65,66 in the system with PBT and rubber containing epoxy groups. These crosslinking reactions are not expected to be substantial in the PBT20/MGE mixture due to the low concentration of carboxyl groups of PBT20 molecules. Although the PBT20/SAN/MGE system did not show the formation of crosslinked material, it is very likely that graft copolymers were formed, since the SAN particle size decreased with the incorporation of MGE in this system, as indicated in Figure 5(a). Effect of the Molecular Weight of MGE on PBT/SAN Compatibilized Blends To investigate the effect of the molecular weight of the reactive copolymer MGE, PBT/SAN/MGE (80/15/5) blends were prepared using MGE 10 and MGE 10 LMW, as high and low molecular weight MGE types, respectively. The morphologies for these ternary blends are shown in Figure 9(a,b). Using MGE 10 LMW as compatibilizer, it was possible to obtain a morphology of tiny particles of 40 nm dispersed among larger particles in the blend. Such particles were not present in the blend with MGE 10. We speculate that the microemulsion shown in Figure 9(b) is formed by interfacial roughening during the melt mixing process, a mechanism described as follows. When the interface is crowded with interfacial copolymers, the interfacial tension approaches zero, which can induce the formation of a very rough pinchoff interface It is likely that the shear stress of the matrix during processing is able to detach these pinchoffs from the interface and disperse them in the matrix, forming the FIGURE 9 TEM photomicrographs of the blends: (a) PBT/SAN/MGE 10 (80/15/5), (b) PBT/SAN/MGE 10 LMW (80/15/5) WILEYONLINELIBRARY.COM/JOURNAL/JPOLB