Instability of styrene/polystyrene/polybutadiene/polystyrene-b-polybutadiene emulsions that emulate styrene polymerization in the presence of polybutadiene.
The bulk process for the production of high-impact polystyrene (HIPS) polymerizes styrene (St) in the presence of a polybutadiene (PB). The reaction mixture is a heterogeneous oil-in-oil emulsion, with polymerization taking place in two or more phases. The reaction generates both free polystyrene (PS) homopolymer and grafted PS branches (by attack onto the initial PB chains). Before the phase inversion (that occurs at around 10% conversion), the continuous phase is rich in PB (the PB-rich phase), and the dispersed phase is rich in PS (the PS-rich phase). After the phase inversion, the opposite occurs. The final thermoplastic typically exhibits a continuous PS phase with dispersed rubber particles that in turn contain vitreous occlusions (the so-called "salami" morphology). In the final material, the graft copolymer (GC) molecules tend to phase separate into a PS-rich phase and a PB-rich phase. Rubber grafting plays a crucial role in the development of the salami morphology during the phase inversion period.
The continuous industrial bulk (or quasi-bulk) HIPS process involves a stirred prepolymerization stage (at 90[degrees]C, and up to about 30% conversion), followed by an unstirred finishing stage (at 130[degrees]C, and up to about 75% conversion). The evolution of the oil-in-oil morphology along the prepolymerization is difficult to investigate; and so far the real monomer-swollen morphologies have never been observed. Instead, only unswollen morphologies have been reported (1), (2). The polymerization kinetics and the partition of reagents and products between the phases are strongly dependent on the temperature and system viscosity. In turn, these variables depend on the stirring conditions and molecular characteristics of the evolving polymers (i.e., free PS, unreacted PB, and generated GC). To simplify this highly multivariate system, the stability and morphology of St/PB/PS/St-butadiene copolymer blends with no chemical reaction and at room temperature have been investigated. The room temperature is required to avoid the thermal polymerization of St that occurs at 70[degrees]C or higher. Ternary phase diagrams of St/PS/PB blends at room temperature have been developed to estimate the approximate phase volumes and compositions along a bulk HIPS process without rubber grafting (3). So far, thermodynamic equilibrium diagrams that include the GC have not been reported.
The physical properties of HIPS depend on the system morphology and molecular characteristics of the final polymers. The development of particle morphology along a bulk HIPS process is still a matter of controversy. Fischer and Hellmann (4) suggested a thermodynamically driven process. Before phase inversion, the continuous PB-rich phase contains dispersed PS particles and GC micelles. After the phase inversion, the GC micelles coalesce and are transformed into the vitreous occlusions of the dispersed rubber particles. In addition, while the more highly branched GC molecules tend to concentrate at the outer interface of the rubber particles, the less branched GC molecules concentrate at the internal interface of the vitreous occlusions (4). Contrary to such interpretation, and from the simple observation of unswollen morphologies, Leal and Asua (2) concluded that thermodynamic equilibrium is never met along the polymerization, and that particle morphology is essentially a fluid-dynamics driven process. In addition, they observed that before the phase inversion, GC micelles are not observed in the continuous PB-rich phase, and that the degree of grafting importantly affects the final particle morphology. The grafting efficiency is defined as the mass of grafted St with respect to the total mass of polymerized St. A high grafting efficiency along the prepolymerization increases the stability of the PS--PB interfaces and favors the development of the salami morphology (2).
Many attempts have been made to generate HIPS and ABS-like materials with new morphologies by simple blending of the polymer components (5-7); and such studies have spurred basic research on polymer miscibility and morphology development. Jiang and Xie (8) reviewed the thermodynamics of homopolymer/diblock copolymer (BC) mixtures. According to composition, temperature, concentration, solvent (when present), and molecular characteristics, the possible morphologies are: (i) a single homogeneous phase; (ii) a phase-separated copolymer with the homopolymer dissolved in one of the phases; (iii) a macrophase-separated homopolymer and copolymer, with the copolymer not microphase segregated; or (iv) a macrophase separated homopolymer and copolymer, with the copolymer microphase segregated. In this last (and most common) situation, some of the polymers could be partially dissolved in the other coexisting phases. For blends involving a generic PA homopolymer and a generic PA-b-PB di-BC with PA and PB blocks of equal length, Hashimoto et al. (9), (10) discussed the stability limits of their micro- and macrophase transitions. Two cases were theoretically investigated. When the molecular weight of the PA homopolymer is much lower than the molecular weight of the PA block in the copolymer, then microphase transition dominates macrophase transition; and unless the BC concentration are very low, microdo-main structures are developed without macrophase separation. In contrast, when the molecular weight of the PA homopolymer doubles more than the molecular weight of the PA block in the copolymer, then macrophase separation dominates microphase separation. In addition, for generic homopolymer/di-BC blends, Shull and Winey (11) and Matsen (12) developed theoretical models for predicting phase behavior. The model by Shull and Winey (11) calculates the distribution of the PS homopolymer across the PS-rich microdomains; and it was validated with measurements of PS/PS-b-polyisoprene blends exhibiting lamellar structures of known equilibrium repeat periods in strong segregation regimes. Matsen (12) investigated weakly segregated PA/PA-b-PB blends of similar degrees of polymerization. The relative stability of the numerous possible phases was examined, and phase diagrams were constructed with the help of a self-consistent field theory. Focusing on the lamellar microstructure of a symmetrical di-BC, it was found that the addition of a high-molecular-weight homopolymer leads to macrophase separation (between a diblock-rich lamellar phase and a homopolymer-rich disordered phase), due to attractive interaction between the diblock bilayers. In contrast, a low-molecular-weight homopolymer induces a repulsive interaction between the bilayers. with indefinite addition of the homopolymer into the lamellar phase without macrophase separation. Furthermore, the homopolymer was seen to stabilize other morphologies, such as double-diamond, hexagonally-perforated lamellar, and close-packed spherical phases (12).
Several publications have analyzed the effect of molecular weights in blends of PS-b-PB with either PS or PB. Selb et al. (13) investigated the morphology of micelles generated when mixing 0.5-10% in weight of a PS-b-PB with PB of three different molecular weights (1600, 3300, and 4500 g/mol). The resulting spherical micelles exhibited a core of PS blocks and a shell of PB blocks with dissolved PB chains. Kinning et al. (14), (15) investigated the formation of micelles in PS-b-PB/PS blends. The critical micelle concentration increased when decreasing any of the following: (a) the free PS molecular weight; (b) the PS composition in the copolymer; or (c) the molecular weight of the BC (14), (15). Mathematical models on the morphology of homopolymer/di-BC blends have mainly considered the most common situation of spherical micelles. However, transitions into cylindrical, vesicular, and lamellar micelles have also been reported (16-20). For PS-b-PB/PS mixtures, Kinning and Thomas (21) suggested that transitions from spherical into nonspherical micelles are favored when increasing: (a) the molecular weight of the aggregating block, (b) the molecular weight of the homopolymer matrix, and (c) the BC concentration.
Bates et al. (22), (23) investigated PS/PS-b-PB blends with volume ratios of PS homopolymer with respect to PS blocks as high as three. They reported that randomly distributed spherical PB domains in PS matrixes are observed when the free PS molecular weight is lower than the molecular weight of the PS block. Furthermore, mesophase structures without macrophase separation were still observed with average molecular weights for the free PS and PS blocks of 116 and 77 kDa, respectively. For PS/PB-b-PS blends, Selb et al. (13) reported similar results for randomly distributed PS spheres in a continuous PB matrix.
Zhao and Feng (24) investigated the behavior of a SBS tri-BC and a four-arm star block copolymer (SB4) in blends with either PS or PB. The homopolymer molecular weights were similar to the molecular weights of their corresponding blocks. The SB4 copolymer exhibited a center of crosslinked AB blocks and branch ends of PS blocks. The solubility of free PS in the PS blocks of the mentioned copolymers were similar to the solubility of equivalent PS/PS-b-PB blends, and greater than the solubility of free PB in the central PB blocks of SB4 (24).
Several articles have investigated the effect of adding minor fractions of a St--butadiene copolymer into PS/PB blends, with PS as the major component. Jiang and Huang (25) used simple PB-g-PS copolymers with an average of a single PS branch per molecule. For varying molecular weights of the free and grafted PS chains, the following was observed by transmission electron microscopy (TEM): (a) the two phases were totally immiscible when the molecular weight of free PS was much larger than the molecular weight of the grafted PS branches; (b) the degree of mixing improved when the PS molecular weight became closer to the molecular weights of the grafted PS branches, and (c) for free PS molecular weights lower than the PS branches, then such chains dissolve into each other, with PB microdomains dispersed in the PS matrix. These observations are consistent with the observations by Inoue et al. (26) and Jiang and Xie (8) for homopolymer/di-BC blends.
Cavanaugh et al. (27) compared the IZOD impact strength of PS/PB blends containing several commercial St--butadiene copolymers. Although short di-BC proved ineffective as interface agents, the best performance was that of an asymmetric di-BC of sufficiently high molecular weights so as to generate entanglements in the main interfaces. Vranjes et al. (28) investigated the compatibilization capacity of di-, tei-, and pentablock linear St--butadiene copolymers in PS/PB blends. Tri-BCs proved to be the best from the point of view of system morphologies and stress transfer characteristics.
Consider the case of St/PB/PS blends. Keskkula (29) investigated blends containing St, PB, six different PS samples, and a PS-b-PB of 60% in weight of St units. The blends emulated St conversions of 28% (i.e., after the phase inversion). The stability of the resulting oil-in-oil emulsions decreased when the molecular weight of the free PS was reduced, and therefore when the system viscosity was decreased. Thus, while emulsions containing a low molar mass PS (of viscosity between 7 and 48 cP at 25[degrees]C and 19.6 [s.sup.-l]) readily demixed, no demixing was observed in the emulsions with viscosities over 136 cP. Also, an increased stability and smaller droplets were observed when increasing the BC concentration with a constant PS concentration. For a bulk polymerization of St in the presence of a rubber and varying amounts of a chain transfer agent, Keskkula (28) found that a reduction in the PS molecular weights destabilized the reacting emulsions and increased the rubber particles size. With increasing grafting efficiencies, the size of the spherical rubber particles were initially reduced, but an excessive grafting produced a complex mixture of spherical, rod, and laminar rubber particles (29), (30).
Rigler et al. (31) investigated the demixing of St/F'S blends containing three different rubbers: a PB, a polypropylene oxide, and an ethylene--propylene copolymer; and their observations were interpreted with the Flory--Hug-gins theory (32) and Scott's approximation (33). In addition, Rigler et al. (31) measured the partition of St between the rubber-rich and PS-rich phases. For volume fractions of PS in the PS-rich phase between 0.001 and 0.5, the volume ratio of St in the PB-rich and PS-rich phases varied between 1 and 1.10, respectively. Ludwico and Rosen (34) and Luciani et al. (3) also determined the partition of St in St/PB/PS blends. As before, the monomer exhibited a slight preference toward the PB-rich phase; and the St partition coefficient was seen to slightly increase with increasing PS concentrations (34), (3).
This work investigates the room temperature instability and phase morphologies of St/PB/PS/star BC blends. The blends contained 6% in weight of butadiene units; and emulate 12 hypothetical bulk polymerizations of St in the presence of PB with constant grafting efficiencies (varying between 0 and 25.5%), and at St conversions of 10%, 15%, and 20% (in principle, close to or immediately after the phase inversion).
The experiments involved mixtures of St, a PB, a low molar mass PS (PSI), a high molar mass PS (PSH), and a star BC. The monomer and the PSH (in pellets) were provided by Estizulia C.A. (Venezuela), and were used as received. The monomer contained tert-butyl catechol as inhibitor. The medium-cis PB rubber was from Eni (Italy), grade Intene 50A; and contained an antioxidant. Before its use, the PB was grated into thin flakes. The low molar mass PS (PSL) was synthesized in our laboratories as follows. First, 1.81 g of benzoyl peroxide (initiator) and 4.55 g of' tert-dodecyl mercaptan (transfer agent) were dissolved in 1359 g of St, and the solution was loaded into a 2-L stainless steel reactor (Buchi-glassuster 280 BEP). Then, a stirred polymerization at 250 rpm was carried out at 90[degrees]C for 12 h under a nitrogen blanket. Finally, the polymer was precipitated in cold methanol and vacuum-dried at 50[degrees]C. The resulting solid was in the form of an irregular powder and flakes.
The star BC was from Phillips Petroleum, grade KK38. This star copolymer nominally contains four-branches per molecule, with each branch exhibiting a central PB block and an external PS block. Its global composition was determined by proton-nuclear magnetic resonance ([.sup.1]H-NMR: at room temperature and with [CDCI.sub.3] as solvent), in a Bruker AV1 300 MHz, yielding 64 mol% ([+ or -]4%) of St units, 36 mol% ([+ or -]4%) of butadiene units, with 7 mol% ([+ or -]1%) of 1,2-PB and 29 mol% ([+ or -]2%) of 1.4-PB. This represents 77% in weight of St units in BC.
All the polymers were analyzed by size exclusion chromatography (SEC). The chromatograph was a Waters Breeze (Milford, MA), fitted with a full set of six it-Styragel columns, a differential refractometer (DR) Waters 2414, a ultraviolet (UV) detector at 256 nm, and a Wyatt Dawn multiangle light scattering (LS) photometer. The PS molecular weights were determined from a direct calibration with PS standards. The PB molecular weights were determined applying the universal calibration concept. The star BC molecular weights were directly measured by LS/DR. The resulting molecular weight distributions and averages are shown in Fig. 1 and Table 1. The star BC was analyzed by triple detection (Fig. 2). The UV signal senses the phenyl groups of the PS chains, lout exhibits a very weak sensitivity toward the butadiene repeating units. The DR and UV chromatograms of the BC are multimodal, and almost proportional to each other. This yielded an essentially constant composition with the molecular weight; and a global mass fraction of St units of 67%. The difference between this result and the St composition by NMR could be due to dispersion in the different analyzed batches of the commercial copolymer. The multimodal DR chromatogram of the BC suggests a mixture of molecular species containing 4, 3, 2, or 1 branches/molecule of molecular weights 1,90,000, 1,40,000, 85,000, and 45,000 g/mol, respectively. This is consistent with an anionic synthesis procedure; with initial generation of linear di-BC anions, and their later crosslinking with a tetrafunctional terminator agent. The chain lengths of the external PS blocks in the star BC can be estimated from the global BC composition and the molecular weight of a single branch; yielding around 35,000 g/mol. This molecular weight is close to that of [PS.sub.L], but lower to that of [PS.sub.H]. Thus, micromixing is to be expected between [PS.sub.L] and the BC, but not necessarily between [PS.sub.H] and the BC.
TABLE 1. Average molecular weights of the polymer components. [bax.M.sub.n] [bax.M.sub.n](g/mol) [bax.M.sub.w]/ (g/mol) [bax.M.sub.n] [PS.sub.L] 36,100 75,900 2.10 [PS.sub.H] 119,100 222,100 1.86 PB 81,600 136,700 1.68 Star BC 114,000 175,000 1.54
Twelve room temperature prepolymerizations of St in the presence of PB were emulated. The following assumptions were adopted for such hypothetical reactions: (a) the initial solutions only contained St monomer and dissolved PB; (b) the St polymerization produces fixed proportions of free and grafted PS chains, thus yielding constant grafting efficiencies; (c) the generated polymers exhibit constant molecular characteristics; (d) the increasing conversions are emulated by reducing the monomer mass and simultaneously increasing the polymerized St mass (contained in the free PS and in the copolymer); and (e) the star BC behaves as an hypothetical GC from the points of view of their contributions toward emulsion instability and final demixed morphologies.
The recipes are presented in Table 2. To investigate the effect of the free PS molecular weight, six of the emulated prepolymerizations (Exps [1.sub.L] - [6.sub.L]) contained [PS.sub.L], whereas the other six (Exps [1.sub.H] - [6.aub.H]) contained [PS.sub.H]. The masses of St and butadiene units in the BC were calculated employing the global mass fraction of St determined by NMR. The following characteristics were common to all the blends: (1) the nominal total mass was 100 g; (2) the nominal total mass of Bd units (contained in the PB and in the BC) was 6 g; (3) the major component was always the monomer, followed by the PS homopolymer; and (4) the mass of free PB was always higher than that of the star BC, except for Exps [6.sub.L] and [6.sub.H] with the highest grafting efficiency. The St grafting efficiency ([E.sub.g]) was fixed at 0, 5.1%, 10.2%, 15.3%, 20.4%, and 25.5%; by imposing a constant ratio between the mass of St in the BC and the total mass of bound St (in the free PS and in the BC). Clearly, the blends with [E.sub.g] = 0 did not contain the star BC. In each experiment, three hypothetical conversions were emulated: x = 10%, 15%, and 20%; as given by the ratio between the mass of bound St (in the PS and in the BC) and the total mass of free and bound St. The mass fractions of total polymer in the blends at 10%, 15%, and 20% conversion were 15.4%, 20.1%, and 24.8%, respectively.
Experimental Procedure and Results
All the experiments were carried out at room temperature, and involved two stages: (i) a mixing stage during 1 day, for the preparation of the oil-in-oil emulsions and (ii) a demixing stage by simple decantation during 30 days, with determination of the interface levels and analysis of the final demixed phases.
For the mixing stage, all the solids were simultaneously loaded into the monomer, and the mixture was stirred for 24 h at 500 rpm and in the dark (to prevent St photoinitiation and PB photooxidation); by means of a Heidolph R6L70 stirrer. Immediately after, the oil-in-oil emulsion viscosities were measured at 25[degrees]C and at a shear rate of 14 [s.sup.-1] (see Fig. 3). To this effect, 20 mL of the emulsions were loaded into the cylindrical cup a Brookfield RVTDV-I viscometer (internal diameter 2.3 cm, and height 7 cm). In addition, around 11 mL of the oil-in-oil emulsions were loaded into cylindrical plastic tubes (internal diameter 1.3 cm), for the decantation stage.
The decantation stage first involved the determination of an initial demixing time until appearance of a clear interface level (Table 2). Thereafter, the interface levels were almost daily recorded (Figs. 4 and 5). The initial periods until appearance of a clear interface level varied from a few minutes in all the experiments with [PS.sub.L], to several days in Exps [5.sub.H] and [6.sub.H] The decantation process is driven by a difference in the phase densities; and this yielded an upper PB-rich phase and a lower PS-rich phase. In some of the higher viscosity experiments, a third intermediate BC-rich phase was also observed. The final demixed phases are illustrated in Fig. 6.
TABLE 2. Emulated polymerization recipes and approximate demixing times until appearance of clear interface levels. Star BC Exp no. [E.sub.g](%) x(%) St [PS.sub.L] PS PB (g) or blocks blocks [PS.sub.H] (g) (g) (g) [1.sub.L], 0 10 84.6 9.4000 -- -- [1.sub.H] 15 79.9 14.100 -- -- 20 75.2 18.800 -- -- [2.sub.L], 5.1 10 84.6 8.9300 0.4826 0.1441 [2.sub.H] 15 79.9 13.395 0.7238 0.2162 20 75.2 17.860 0.9650 0.2883 [3.sub.L], 10.2 10 84.6 8.4600 0.9650 0.2883 [3.sub.H] 15 79.9 12.690 1.4476 0.4324 20 75.2 16.920 1.9302 0.5765 [4.sub.L], 15.3 10 84.6 7.9900 1.4476 0.4324 [4.sub.H] 15 79.9 11.985 2.1714 0.6486 20 75.2 15.980 2.8952 0.8648 [5.sub.L], 20.4 10 84.6 7.520 1.9302 0.5765 [5.sub.H] 15 79.9 11.280 2.8952 0.8648 20 75.2 15.040 3.8602 1.1531 [6.sub.L], 25.5 10 84.6 7.050 2.4126 0.7207 [6.sub.H] 15 79.9 10.575 3.6140 1.0810 20 75.2 14.100 4.8254 1.4413 Demixing time Exp no. PB (g) Blends Blends w/[PS.sub.L] w/[PS.sub.H] [1.sub.L], 6.0000 Few mins. Few nuns. [1.sub.H] 6.0000 Few mins. 2 days 5.0000 Few mins. 6 days [2.sub.L], 5.8433 Few mins. Few mins. [2.sub.H] 5.7650 Few mins. 1 day 5.6867 Few mins. 3 days [3.sub.L], 5.6867 Few mins. Few mins. [3.sub.H] 5.5300 Few mins. 1 day 5.3733 Few mins. 1 day [4.sub.L], 5.5300 Few mins. Few mins. [4.sub.H] 5.2950 Few mins. 1 day 5.0680 Few mins. 1 day [5.sub.L], 5.3733 Few mins. Few mins. [5.sub.H] 5.0660 Few mins. 1 day 4.7467 Few mins. 1 day [6.sub.L], 5.2167 Few mins. Few mins. [6.sub.H] 4.8250 Few mins. 1 day 4.4333 Few mins. 4 days Exps [1.sub.L] - [6.sub.L] involved the low-molar mass PS ([PS.sub.L]), and Exps [1.sub.H] - [6.sub.H] the high-molar mass PS ([PS.sub.H]). The mass of star BC is divided into the expected masses of its blocks. [E.sub.g] and x, respectively, represent the grafting efficiency and St conversion.
At the end of the decantation period, and for a reduced number of blends, the phases were carefully isolated by means of a syringe, and the following was determined: (a) the St partition coefficient, for samples that did not exhibit the intermediate BC-rich phase and (b) the unswollen morphologies of some of the isolated phases, by TEM.
The St partition coefficient was defined by [K.sub.st] [xi] [w.sub.st, PB]/[w.sub.st, PS]; where [w.sub.st, PB] and [w.sub.st, PS] are the mass fractions of St in the PB-rich and PS-rich phases, respectively. To determine such concentrations, all polymers in each phase were precipitated in cold methanol and vacuum-dried until constant mass. The monomer mass was then calculated from the difference between the initial phase mass and the final dry mass. The results are shown in Table 3 and Fig. 3.
TABLE 3. Gravimetric determination of the St partition coefficient ([K.sub.St]) between the upper (PB-rich) phase and the lower (PS-rich) phase. PB-rich PB-rich phase phase Exp no. [E.aub.g] (%) x (%) Total St Total St mass mass muss mass (g) (gl (g) (g) [1.sub.L] 0 10 5.1853 4.47 4.8032 3.92 15 4.7766 3.96 6.2675 4.79 20 4.3821 3.35 5.8467 4.18 [2.sub.L] 5.1 10 4.9906 4.31 4.6038 3.75 15 3.5425 2.92 6.5991 5.07 20 3.8194 3.01 6.0042 4.27 [6.sub.L] 25.5 10 4.1938 3.64 5.0994 4.21 15 2.8611 2.35 6.4719 5.04 20 1.9271 1.48 7.4609 5.47 [1.sub.H] 0 10 5.4386 4,70 4.4529 3.58 15 3.7135 3.05 6.9763 5.28 20 2.9121 2.56 7.0291 4.95 [2.sub.H] 5.1 10 4.6829 4.06 5.4242 4.37 15 4.2037 3.38 5.6607 4.30 20 3.2107 2.51 7.3115 5.18 Exp no. [K.sub.St] (a) Error (b) (%) [1.sub.L] 1.05 -0.7 1.08 -0.7 1.07 -2.0 [2.sub.L] 1.06 -0.7 1.07 -1.4 1.11 -1.5 [6.sub.L] 1.05 -0.1 1.05 -1.0 1.05 -1.5 [1.sub.H] 1.07 -1.0 1.09 -2.4 1.25 0.6 [2.sub.H] 1.08 -1.4 1.06 -2.4 1.10 -2.8 The [K.sub.St], values are also represented in Fig. 3. (a) Defined by [K.sub.St], = [w.sub.St].PB-rich/[w.sub.St].PS-rich (b) Fractional error in the total St mass with respect to the theoretical formulations of Table 2.
The TEM measurements were carried out in a JEOL JEM 1220 at 100 kV. For the samples preparation, thin slices of the dry phases (width = 80 nm) were cut at -190[degrees]C in a Leica Ultracut cryomicrotome, fit with a diamond knife. Then, the slices were stained with osmium tetraoxide vapor for 15 s. Figures 7 and 8, respectively, present the unswollen morphologies of the upper (PB-rich) phases of Exps [2.sub.L] and [2.sub.H] at x = 10%. Figure 9 presents the morphologies of the two final phases of Exp [6.aub.L] at x = 20%. Figure 10 presents the morphologies of the three final phases of Exp [6.sub.H] at x = 20%.
Finally, the initial demixing periods until appearance of a clear interface level were determined for four additional blends containing St, the star BC, and each of the base homopolymers ([PS.sub.L], [PS.sub.H], and PB) (Table 4). All these recipes contained 75.2 g of St. Blends [7.sub.L] and [7.sub.H] did not contain free PB and emulate reactions where all the initial PB was transformed into BC, at x = 23.9% and [E.sub.g] = 20.4%. Blends 8 and 9 did not contain free PS, and emulate a (highly improbable) reaction with [E.sub.g] = 100% at x = 1.3 and 6.0%, respectively. The results are shown in the last columns of Table 4. Experiment [7.sub.L] with [PS.sub.L] and BC (in the absence of PB) showed no clear demixing along the 30 days period. The equivalent Exp [7.sub.H] with [PS.sub.H] and BC showed a clear interface (with an upper BC-rich phase and a lower PS-rich phase) after 13 days of decantation time. Experiment 8 with PB and a high BC concentration showed an almost instantaneous demixing into an upper PB-rich phase and a lower BC-rich phase. In contrast, Exp 9 with PB and a low concentration of BC showed no demixing along the 30 days period.
The demixing process and final morphologies of blends containing St, PB, PS, and star BC were analyzed. The blends emulate bulk prepolymerizations of St in the presence of PB at St conversions of 10% and higher.
Let us first consider the St partition coefficients for final blends without the intermediate BC-rich phase. As seen in Table 3 and Fig. 3, the resulting [K.sub.St] values are all slightly higher than unity (between 1.05 and 1.25). This indicates a small preference of the monomer toward the upper PB-rich phase irrespective of conversion and also whether or not it included the BC. The partition coefficients of Exps 1L and 1H without BC are very close to previous determinations by Rigler et al. (31), Ludwico and Rosen (34), and Luciani et al. (3) for St/PS/PB blends. Clearly, the [K.sub.St] values are little affected by the presence of the BC: and this seems reasonable, remembering that the St partition between the homopolymers is close to unity.
Figure 3 shows the emulsion viscosities immediately after the mixing stage. As expected, the viscosity generally increases with solid content, and therefore with conversion. Also, the viscosity is expected to be strongly determined by the molecular weight of the main (continuous) phase. The major polymer component is in all cases PS, and the St partition is close to unity. Thus, the PS-rich phase is expected to exhibit the largest phase volume, and therefore to constitute the main continuous phase (especially, at conversions of 15% and 20%). Accordingly, a high increase in viscosity is observed in equivalent experiments when only the molecular weight of the free PS is increased.
For a clear determination of the main interface level, the demixing periods presented in Tables 2 and 4 should be considered. As expected, the demixing periods were longer for the mixtures that exhibited the highest viscosities; that is in the blends containing [PS.sub.H], and at the higher conversions. Therefore, while rapid demixing periods of only few minutes were observed for all the experiments of Table 2 performed with [PS.sub.L], the blends with [PS.sub.H] at 15% and 20% conversion required between 1 and 6 days to demix (Table 2). However, the opposite effect was observed when comparing blends [7.sub.L] and [7.sub.H] without PB (Table 4): while the blend with [PS.sub.H] demixed after 13 days, the blend with [PS.sub.L] showed no demixing in a 30 days period. This behavior can be explained by the higher affinity of the BC toward [PS.sub.L] when compared with [PS.sub.H]. Finally, Blends 8 and 9 without PS exhibited two limiting behaviors (Table 4): while Blend 9 with similar amounts of PB and BC underwent a rapid demixing, no demixing was observed in Blend 8 with a small amount of BC.
TABLE 4. Final blends containing St, the BC, and each of the homopolymers: recipes and approximate demixing times until observation of a clear interface. Star BC Blend no, [E.sub.g] x St [PS.sub.L] PS PB PB (g) (%) (g) or blocks block-s [PS.sub.h] (g) (g) (g) [7.sub.L] 20.4 23.9 75.2 18,800 4.8254 1.4413 -- (a), [7.sub.H] (a) 8 (b) 100 1.3 75.2 -- 0.9650 0.2883 6.0000 9 (b) 100 6.0 75.2 -- 4.8254 1.4413 6.0000 Demixing time Blend no, Blend Blends w/[PS.sub.l] w/[PS.sub.H] [7.sub.L] No demixing 13 days (a), [7.sub.H] (a) 8 (b) No demixing 9 (b) Few mins. (a) Experiments without PB. (b) Experiments without PS.
Figures 4 and 5 present the evolution of the interface levels after appearance of a clear interface. The levels remained essentially unchanged along the decantation period, except when an intermediate BC-rich phase was separated (in the blends with [PS.sub.H] with the higher grafting efficiencies). The generation of the intermediate BC-rich phase coincides with a simultaneous reduction of the PS-rich phase. This suggests migration of the BC from the lower [PS.sub.H]-rich phase into a new BC-rich phase. Additionally, it suggests an initial preference of the BC toward the lower PS-rich phase, even though it contained [PS.sub.H].
Figure 6 illustrates the final demixing characteristics of the formed isolated phases. In the blends with [PS.sub.L], only two phases were observed, and (as expected) the lower PS-rich phase increased with conversion level. The volumes of the final upper PB-rich phases were little affected by the molecular weight of the free PS; thus suggesting strong incompatibility between the homopolymers, even with [PS.sub.L]. In some of the blends with [PS.sub.H], a third intermediate BC-rich phase was observed; and as expected, the BC-rich phase volume generally increased with conversion and grafting efficiency. An increased grafting efficiency increased the mass of BC with respect to that of PB; and this can explain the reduction observed in the upper PB-rich volumes. Note however, that the middle BC-rich phase was generated by migration from the lower PS-rich phase, and not from the upper PB-rich phase.
Figures 7 and 8 present the morphologies of the upper PB-rich phases of Exps [2.sub.L] and [2.sub.H] at 10% conversion, whereas Figs. 9 and 10 present the morphologies of all the isolated phases in Exps [6.sub.L] and [6.sub.H] at 20% conversion and with the highest grafting efficiency (i.e., in blends containing more BC than PB).
The micrographs of all the upper PB-rich phases show dispersed PS particles, and most of such particles contain lamellar occlusions of BC; thus reaffirming the idea that the BC prefers the PS-rich phase rather than the PB-rich phase. This seems reasonable, bearing in mind the mass fraction of St units in the star BC (=77% in weight), and the molecular topology (with PS branch ends facilitating its dispersion in a PS matrix).
The upper (PB-rich) phase of Exp [6.sub.L] with [PS.sub.L] exhibits large macrodomains of PB homopolymer (Fig. 9a) and of lamellar BC (Fig. 9b). All the phases of Exps [6.sub.L] and [6.sub.H] show either continuous or dispersed lamellar BC regions (Figs. 9 and 10). The observed lamellar morphology for BC with 77 wt% of St units was attributed to the characteristic star molecular architecture of BC, as reported in the literature (35). In Exp [6.sub.L], the high affinity of the BC toward [PS.sub.L] determines that most of the BC remains in the lower PS-rich phase as large lamellar regions (Fig. 9c and d). In Exp 6H, the lower affinity of the BC toward [PS.sub.H] determines its separation into the intermediate BC-rich phase. Note however, that the intermediate BC-rich phase also contains large domains of free PS (Fig. 10c and d), whereas the lower PS-rich phase contains dispersed particles of lamellar BC (Fig. 10e and f).
All the upper PB-rich phases contained dispersed PS particles with internal lamellar BC occlusions. This suggests that given a sufficiently long decantation time, all the emulsions with significant amounts of the three polymeric components will eventually separate into three phases (a PS-rich, a PB-rich, and a BC-rich). Clearly, this separation does not occur in the real HIPS process due to: (a) mechanical agitation during the prepolymerization stage; (b) high reaction rates and viscosities during the finishing and devolatilization stages; and (c) low diffusivity of macromolecules. Furthermore, molecular diffusion between phases is expected to be considerably hindered if the graft or BC molecules placed themselves at the main interfaces. This effect also prevents particle coalescence, through combination of steric effects with the development of a Marangoni stress in the gap between approaching droplets (36).
All the emulated reactions with [PS.sub.L] exhibited a rapid demixing. However, no demixing was observed in Blend 8 (containing [PS.sub.L] and BC in the absence of PB). The reason for this is a high affinity of BC toward [PS.sub.L]; due to the similar molecular weights of [PS.sub.L] and the PS block. The stability/instability of an oil-in-oil emulsion depends on the system viscosity and on the affinity between the different polymer chains. A low molecular weight PS reduces viscosity and favors demixing. But additionally, a low molecular weight PS increases its compatibility toward the other polymer components, thus increasing demixing times (or even totally preventing demixing, as in Blend [7.sub.L].
After isolation of the phases by decantation, all the phase morphologies resulted heterogeneous, with macrophase separation from the other phases. An important result of this investigation is the separation of an independent BC-rich phase in the blends with [PS.sub.H]. This somehow modifies the (perhaps naive) idea of the graft or BC molecules located at the interface of large PS-rich and PB-rich phases. In the HIPS process, most of the original PB becomes GC, and the small fraction of unreacted PB is expected to exhibit a low molecular weight. Thus, PB-rich phases only containing PB homopolymer are nonrealistic. Instead, most PB chains belong to GC that generate independent lamellar structures; with a small fraction of free PB possibly dissolved into PB microphase of such lamellar structures.
Polymer blends can be prepared by casting from a mutual solvent, by melt mixing, or by in situ polymerization. In the bulk HIPS process, both the initial PB and the generated PS chains are dissolved by the St monomer in approximately equal proportions, and the reaction temperature is above the glass transition temperature of PS. Thus, all the three mentioned methods of preparation of polymer blends coexist in the investigated HIPS process. This work is a contribution toward a better understanding of the stability/instability of the oil-in-oil emulsions generated in a bulk HIPS process immediately after the phase inversion. In the real (high temperature) process, the system viscosities are considerably lower than at room temperature, thus favoring system instability and demixing. But at the same time, the affinity between chains is expected to increase with increasing temperatures, thus favoring emulsion stability and hampering demixing. If these two opposite effects were approximately compensated when increasing the system temperature, then the main observations presented in this work could be applied to the real polymerization process.
The authors greatly acknowledge Prof. V. Abetz (Helmholtz-Zentrum, Geesthacht, Germany) for his help with the NMR analysis; J.L. Castaneda and M. Brandolini (CONICET, Argentina) for the SEC analysis. Finally, our thanks to Estizulia C.A (Venezuela) for providing some of the base materials.
Correspondence to: Haydee Oliva; e-mail: email@example.com
Contract grant sponsors: Universidad del Zulia, Condes-LUZ, Venezuela and Universidad Nacional del Litoral, Conicet, Argentina.
Published online in Wiley Online Library (wileyonlinelibrary.com).
[c] 2013 Society of Plastics Engineers
(1.) G. Soto, E. Nava, M. Rosas, M. Fuenmayor, I.M. Gonzalez, G.R. Meira, and H.M. Oliva, J. Appl. Polym. Sci., 92, 1397 (2004).
(2.) G. Leal and J. Asua, Polymer, 50, 68 (2009).
(3.) C. Luciani, D. Estenoz, G. Meira, N. Garcia, and H. Oliva, Macromol. Theory Simul., 16, 703 (2007).
(4.) M. Fischer and G.P. Hellmann, Macromolecules, 29, 2498 (1996).
(5.) G. Riess, in Thermoplastic Elastomers, A Comprehensive Review, N.R. Legge, G. Holden, and H.E. Schroeder, Eds., Hanser Publishers, Munich (1987).
(6.) A. Etch, Angew. Makromolek. Chem., 58, 175 (1977).
(7.) A. Etch, Angew. Makromolek. Chem., 90, 95 (1980).
(8.) M. Jiang and H. Xie, Prog. Polym. Sci., 16, 977 (1991).
(9.) T. Hashimoto, T. Tanaka, and H. Hasegawa, Molecular Conformation and Dynamics of Molecules in Condensed Systems (M. Nagasawa, Ed.), 2, 257 Elsevier Pub. (1987).
(10.) R. Mori, H. Tanaka, and T. Hashimoto, Macromolecules, 20, 381 (1987).
(11.) K. Shull and K. Winey, Macromolecules, 25, 2637 (1992).
(12.) M. Matsen, Macromolecules, 28, 5765 (1995).
(13.) J. Selb, P. Marie, A. Rameau, R. Duplessix, and Y. Gallot, Polym, Bull., 10, 444 (1983).
(14.) D. Kinning and E. Thomas, J. Chem. Phys., 90, 5806 (1989).
(15.) D. Kinning, K. Winev, and E. Thomas, Macromolecules, 21, 3502 (1988).
(16.) K. Sardelis, H.J Michels, and G. Allen, Polymer, 28, 244 (1987).
(17.) S. Aggarwal, Polymer, 17, 938 (1976).
(18.) G. Eastmond and D. Philips, Polymer, 20, 1501 (1979).
(19.) E. Bradford, Colloidal and Morphological Behavior of Block and Graft Copolymers, Plenum, NY, 21 (1971).
(20.) F. Bates, C. Berney, and R. Cohen, Macromolecules, 16, 1101 (1983).
(21.) D. Kinning and E. Thomas, Macromolecules, 17, 1712 (1984).
(22.) F. Bates, C. Berney, and R. Cohen, Macromolecules, 16, 1101 (1983).
(23.) F. Bates, R. Cohen, and A. Argon, Macromolecules, 16, 1108 (1983).
(24.) W. Zhao and Z. Feng, Proc. 1st China-Italy Polym. Meet., Genova (1989).
(25.) M. Jiang, X. Huang, and T. Yu, Polymer, 24, 1259 (1983).
(26.) T. Inoue, T. Soen, T. Hashimoto, and H. Kawai, Macromolecules, 3, 87 (1970).
(27.) T. Cavanaugh, K. Buttle, J. Turner, and B. Nauman, Polymer, 39, 4191 (1998).
(28.) N. Vranjes, F. Lednicky, J. Kotek, J. Baldrian, V. Rek, 1. Fortelny, and Z. Horak, J. Appl. Polym. Sci., 108, 466 (2008).
(29.) H. Keskkula, Plast. Rubbers Mater. Appl., 16(5), 66 (1979).
(30.) H. Keskkula, Rev. Rep., 1(5) (1999).
(31.) J. Rigler, L. Muller, and B. Wolf, Angew. Makromolek, Chem., 74, 113 (1978).
(32.) P. Flory, Principles of Polymer Chemistry, Cornell Univ. Press, Ithaca (1953).
(33.) R. Scott, J. Chem. Phys., 17, 279 (1949).
(34.) W. A. Ludwico and S.L. Rosen, J. Appl. Polym. Sci., 19, 757 (1975).
(35.) R. Adhikari and G. Michler. Prog. Polym. Sci., 29, 949 (2004).
(36.) E. Van Hemelrijck, P. Van Puyvelde, Ch. W. Macosko, and P Moldenaers, J. Rheol., 49, 783 (2005).
Eliezer Velasquez, (1) Haydee Oliva, (1) Alejandro J. Muller, (2) Juan V. Lopez, (2) Jorge Vega, (3) Gregorio R. Meira, (3) Mona Wambach (4)
(1) Laboratorio de Polimeros y Reacciones, Escuela de Ingenieria Ouimica, Universidad del Zulia, Maracaibo, Venezuela
(2) Grupo de Polimeros, Departamento de Ciencia de los Materiales, Universidad Simon Bolivar, Apartado 89000, Caracas 1080-A, Venezuela
(3) INTEC (CONICET and Universidad Nacional del Litoral), Guemes 3450, Santa Fe (S3000GLN), Argentina
(4) Helmholtz-Zentrum Geesthacht, Zentrum fur Material- und Kustenforschung GmbH, Institut fur Polymerforschung, Max-Planck-Strasse 1,21502 Geesthacht, Germany
|Printer friendly Cite/link Email Feedback|
|Author:||Velasquez, Eliezer; Oliva, Haydee; Muller, Alejandro J.; Lopez, Juan V.; Vega, Jorge; Meira, Gregori|
|Publication:||Polymer Engineering and Science|
|Date:||Sep 1, 2013|
|Previous Article:||Effect of coating method on gas separation by PDMS/PES membrane.|
|Next Article:||On thermal conductivity of micro- and nanocellular polymer foams.|