Inline Rheological Behavior of Dispersed Water in a Polyester Matrix With a Twin Screw Extruder.
Solid-liquid dispersions consisting of sub-micron polymer particles have many industrial applications such as adhesives, paints, or textile, and paper coatings. Conventional emulsification techniques are losing prevalence due to the high costs surrounding solvent recovery and increasingly stringent environmental regulations, especially pertaining to the health and safety of workers and the consumer. Exclusive use of water as an environmentally benign medium is preferred in green manufacturing processes. This article looks at aspects of a new green manufacturing process utilizing a conventional twin-screw extruder that is intent on producing an aqueous dispersion of high viscosity polymers internally by a top-down approach.
The top-down approach for preparing submicron polymer particles is generally considered when polymerization in water is implausible. However, mechanical top-down methods, like rotor-stator homogenization or milling, experience considerable difficulties in controlling the product particle size especially below the micron scale and hence, solvent emulsification is generally preferred as it is a phase-inversion technique [1-3]. In solvent emulsification, the solvent is responsible for lowering the system viscosity by dissolving the polymer, as well as acting as a surface-active agent. Particles are ultimately formed on dilution with water. Solvent-free emulsification is a less robust phase-inversion technique as it experiences greater difficulties in countering the high viscosity of many industrially interesting polymers (namely 1-1,000 Pa s) due to the absence of organic solvents, but is emerging to show its effectiveness with several viscous systems [4-9]. For example, Ashrafizadeh et al.  demonstrated the emulsification of crude oil by homogenization, which had a viscosity of 2.7 Pa-s, with water and no hazardous solvents at 25[degrees]C. They produced oil domains with a Sauter mean diameter varying between 8 and 25 [micro]m among their samples. These are quite large domain sizes relative to standard emulsions but should be expected as the viscosity of the dispersed phase increases in such mechanical dispersion processes . Song et al.  created a waterborne, pressure-sensitive adhesive with a multi-impeller batch mixer from rosin ester and partially hydrogenated wood rosin acid having a blend viscosity of 5 Pa x s from 43.3[degrees]C to 54.4[degrees]C. In their case, the mean particle size of the rosin emulsion varied between 0.6 and 1.5 [micro]m for different processing conditions, smaller in their case by the better compatibility of rosin with water than crude oil. However, a closer example to this study would be the work of Akay et al. [6-8], who studied aqueous phase inversion in a batch mixer for different polymer melts with relatively high viscosities, including low density polyethylene which had a viscosity of approximately 50 Pa s. Moreover, the mean particle size changed between 1.03 and 4.19 [micro]m for different content of hydrophobically modified water-soluble polymers. As the viscosity of the polymer intended to be emulsified increases, demands on the processing method to bring water uniformly into an oleophilic phase grow more difficult. For viscosities above 10 Pa s, a twin-screw extruder seems well suited as process equipment to the task of dispersion but its continuous nature introduces challenging constraints on emulsification since the available residence time for adequately mixing that may result in phase inversion is limited to seconds.
The use of a twin-screw extruder to emulsify polymers in the manner described above is referred to as solvent-free extrusion emulsification (SFEE) and was first patented by Abe et al.  in 1982. The SFEE process is exceedingly complicated and currently runs with a very narrow window of stable operation [12, 13], but understanding the mechanisms related to the process will permit improved processability. One of the most critical steps in the process is increasing the surface area between the polymer melt and water phases prior to phase inversion, through intensive mixing and appropriate chemistry. For Song et al. , base neutralization of the rosin acid promoted incorporation of water into the oleophilic matrix, leading to suitable conditions for phase inversion. For higher viscosity polymers like the polyester under study in this work, this neutralization reaction can produce chain degradation, which was not a point addressed in their study. In SFEE, this neutralization must occur simultaneously with incorporation of water to reduce the interfacial energy as the dispersion of water domains increases the interfacial area. In this study, the changes to molecular weight and generation of surface-active endgroups will be separated to understand which factor has a more significant influence on improved water dispersion within an oleophilic matrix.
The purpose of this article is to better understand this mechanism of water dispersion into a high viscosity polymer melt, which is prior to phase inversion in SFEE; emulsions will not be produced in this work as a result of its focus but presented in a future paper on SFEE. As mentioned above, improving the understanding of SFEE will be done by decoupling two factors which have been noted to occur during the initial water dispersion step, which are mixing and the polar group contributions to both surface energy and chain length reduction by utilizing a novel inline rheometer. The work builds on a previous study  that was a rheological study of the dispersion of water into a high viscosity molten polyester using a batch reactor.
MATERIALS AND METHODS
A polyester synthesized from fumaric acid (FA) and dipropoxylated bisphenol A (pBPA) was provided by the Xerox Corporation (Webster, NY). The resin was a high flow grade ([M.sub.w] = 17081 g/mol; [M.sub.W]/[M.sub.N] = 4.2) with an acid number of 17.7 [+ or -] 1.7 mg/g KOH. Sodium hydroxide (NaOH) was purchased from Caledon Laboratories Ltd. (Georgetown, ON). Deionized water (>0.1 [micro]S/[cm.sup.2]) were used in the trials. Samples of the monomers, FA, and pBPA, were provided by Xerox Corporation to prepare a processing modifier resembling the polyester.
Processing Modifier Preparation
The modifier to adjust viscosity and interfacial properties of the polyester with water, was prepared by reacting the two monomers at a 1:1 molar stoichiometry for one hour at 180[degrees]C with 0.1 wt% catalyst (Fascat 4100; PMC Group, Mount Laurel, NJ). The resulting mixture of monomer and oligomer species ground but otherwise used without post-treatment was referred to as non-converted (NC) modifier in the study. The grade referred to as converted (C) modifier had any existing acid endgroup neutralized with 10 wt% NaO[H.sub.(aq)] for three hours at 60[degrees]C under agitation by an impeller at 300 rpm. The neutralized modifier was subsequently washed with deionized water until the pH of the wash water had dropped below 8. An acid number of zero was found on completion of the neutralization reaction. Both NC and C modifiers were used to lower the matrix viscosity in this study.
Extrusion trials were done with a 40 L/D, 27 mm Leistritz ZSE-HP co-rotating twin screw extruder (TSE) supplied by the American Leistritz Extruder Corporation (Somerville, NJ). The barrel temperature profile was kept constant at 95[degrees]C in every zone. The extruder was operated at a fixed screw speed of 300 rpm. The polyester flake and processing modifiers were manually blended and then fed at 8 kg/h by a DDSR20 gravimetric feeder (Brabender Technologie; Mississauga, Ontario). Water was fed to a zone 26 L/D upstream of the die with a 260D high pressure syringe pump (Teledyne Isco; Lincoln, NE); this offers a much longer mixing time than is normal for SFEE but ensures the greatest chances that the water will be fully incorporated on reaching the inline rheometer. An intensive mixing screw design was used as similar by Neubauer and Dunchus , though the actual design cannot be provided due to proprietary restrictions.
The extruder die functioned as an inline rheometer in the study, based on an orifice plate flowmeter design. The main bore diameter of the die was 15 mm and included a 2 mm orifice plate midway along its length to increase the pressure difference, as seen in Fig. 1. The aim of the design was to partially restrict the exit of the polymer-water mixture so that it did not drool and to be able to accurately detect the pressure drop across the die for trials where high water content was used. For the present trials, the die was configured with two 21 MPa pressure transducers (model PT467E; Dynisco) at a distance of 17 mm either side of the orifice plate; both transducers were positioned to ensure that fully developed flow was being monitored A metal-sheathed, grounded K-type thermo couple was placed into the melt within 87 mm away from the orifice plate toward the die exit to take into account effects such as water cooling and viscous dissipation. The rheometer had an estimated measurement sensitivity to viscosities for a resin-to-water (R/W) ratio as low as 1.5 based on the 0.5% nameplate accuracy of the selected pressure transducers; this article will present the amount of water being added into the process as a resin-to-water ratio (R/W) rather than as a weight basis to be consistent with the industry to which the technology applies.
This inline rheometer was calibrated using the open-source numerical software package, OpenFOAM 2.0.1 (OpenCFD, Bracknell, UK), for the 3-D simulated environment shown in Fig. 1c. The flow domain of the die was meshed with 93,782 hexahedral elements using GAMBIT (Fluent Inc., Canonsburg, PA). Mass and momentum conservation equations were solved by the SIMPLE algorithm. A single-phase fluid of differing viscosities from 0.001-200 Pa s was approximated by solving the simulations with a power-law constitutive model with an index value of 0.96; the index value corresponded to the neat polyester as obtained from off-line parallel-plate and capillary rheometers . A non-slip boundary condition was assumed valid at all die walls. The inlet of the die was set to 2.22 x [10.sup.-6] [m.sup.3]/s and the outlet boundary was fixed at atmospheric pressure. Laminar steady shear flow under the isothermal conditions was used for the simulations.
Trial Procedure. Trials began by adding water at R/W = 135 (i.e., 0.74 wt%) to prevent the molten polyester from blocking the water injector. The polyester feedstock contained varying content (0, 0.5, 1.5, and 5 wt%) of either converted or non-converted modifier in a trial. The system was run for five minutes to reach steady conditions, at which point the fluctuations in monitored viscosity dropped well below [+ or -] 4.5%. After the first five minutes, the rate of water addition was increased to R/W = 5.
The melt temperature was measured in the die and as a result, the inline apparent viscosity data were corrected to a comparable 100[degrees]C using an Arrhenius relationship with flow activation energy of 32.7 kJ/mol (determined by off-line rheometry). This was necessary as the melt temperature varied [+ or -] 4[degrees]C for the different modifier conditions on account of viscous heat dissipation, making it necessary to standardize to a reference temperature for their comparison. Extrudate were collected for analysis after an additional eight minutes once steady-state conditions in the extruder had been confirmed by pressure and temperature measurements at the die. Most of the extrudate were dried under vacuum conditions at 60[degrees]C over night for characterizations. A small sample of the extrudate was not dried but rather immediately dispersed in water, as described below.
Offline Rheological Properties. Shear viscosities of the extruded polyester without water were measured in a parallel plate rheometer (ARES; TA Instruments, New Castle, DE) at 100[degrees]C, 110[degrees]C, 120[degrees]C, 130[degrees]C, and 140[degrees]C, respectively. The flow activation energy was determined based on an Arrhenius correlation to correct for viscous dissipation in the inline rheometer. In addition, complex viscosity curves of the dried extrudates containing different modifier types and contents were determined by an ARES parallel plate rheometer (TA Instruments, New Castle, DE) at 100[degrees]C. The diameter of the parallel plate was 25 mm and gap distance between plates was 1.5 mm. It was operated in oscillating mode over a frequency range of 0.1-100 rad/s at a strain of 18% (selected based on a strain sweep test).
Weighted modifier was dissolved in a pyridine-deuterated chloroform solution containing a known amount of a perfluorinated version of BPA (Bisphenol AF) as an internal reference. To the mixture, 2-chloro-4,4,5,5-tetramethyl-1,3,2-dioxaphospholane (CTDP) was added as the phosphorus derivitizing agent. CTDP reacts quickly and quantitatively with both acid and hydroxyl groups, and through the use of Bisphenol AF as an internal standard, the amount of each specific acid and hydroxyl group can be determined . Phosphorus-31([sup.31]P) NMR measurements were performed using a Bruker AV-400 NMR spectrometer.
The samples were dissolved in methanol and filtered through 0.2 [micro]m Teflon (PTFE) prior to analysis. The samples were analyzed by liquid chromatography (LC)/UV and LC/MS. The amount of pBPA plus FA/pBPA adduct were determined using an authentic sample of pBPA and assuming a similar response factor for the 1:1 adduct. The amount of pBPA was determined using LC/MS relative to the authentic standard. The amount of the 1:1 adduct was estimated as equal to the difference between the two methods.
LC/UV. 2 [micro]L of each sample was injected onto the LC system where the components were separated on a 3 mm x 50 mm Hypersil ODS (3 [micro]m) C-18 column using a 0.05% phosphoric acid/methanol gradient mobile phase with UV detection at 228 nm.
LC/MS. The samples were separated on a 2 x 30 mm Kinetex XB-C18 column at 300 [micro]L/min using a gradient from 100% water to 100% methanol over 8 min on the Accela High Speed LC system interfaced to the Q-Exactive mass spectrometer. Data were collected using positive and negative mode electrospray ionization.
Gel Permeation Chromatography
Samples were prepared by dissolving approximately 5 mg modifiers or extrudate samples/mL in spectral grade CH[Cl.sub.3] spiked with 0.1% (v/v) triethylamine (99% purity) followed by filtration through a 0.2 [micro]m syringe filter (Whatman PTFE) into 2 mL sample vials. The samples were injected using an auto-injector onto a Polymer Labs GPC-50 System equipped with 2 x 5 [micro]m PL Mixed Bed C GPC columns (300 x 7.5 mm) with a guard column and a Wyatt triple detection system (light scattering, viscometry and refractive index). Standard molecular weights were calculated using a 10-point polystyrene linear calibration and the absolute molecular weights were calculated using the triple detection data.
Scanning Electron Microscopy (SEM)
Morphology of the samples were evaluated using a Hitachi SU8000 scanning electron microscope (SEM). The accelerating voltage utilized was 2.0 kV. Samples were prepared by drying the extrudate samples out in a fume hood before mounting on a stage and sputter coated with platinum/palladium.
Thermal properties of dried polyester, extrudate samples, and neat modifiers were measured using a DSC Q200 differential scanning calorimeter (TA Instruments, New Castle, DE) with a heatcool-heat cycle over a temperature range of 0[degrees]C-200[degrees]C for a heating/ cooling rate of 10[degrees]C/min in an inert atmosphere (50 mL/min [N.sub.2]). Glass transition temperature ([T.sub.g]) and peak melting temperature ([T.sub.m]) were determined from the thermograms using Universal Analysis V1.7V software (TA Instruments, New Castle, DE).
To interpret dynamic viscosity changes seen in the trials as indicators of water incorporation into the polymer and hence the state of the polymer system in terms of readiness to emulsify, it was necessary to test for this possibility outside of the extruder; neither the extmder, screw design, nor die configuration were suitable for emulsification in this work as that was not the focus of the study. Approximately 3 grams of the extrudate was sampled at the die and immediately transferred to a 60[degrees]C heated water bath system fitted with a Polytron[R] PT 45/80 homogenizer (12 mm tip diameter). The sample was dispersed into 0.5 L of MilliQ water at 16,000 rpm over a three minutes period. Finally, 0.3 L of additional water was introduced to cool the dispersed extrudate samples before it was collected for particle size analysis. The premise to this test was that polymer samples containing thinner striated layers of water would disperse into smaller particles by homogenization. This assumes that the morphology of the polymer-water mixture demonstrates sufficiently slow stress relaxation that its state of mixedness changes little in the time taken to transfer samples from the die to the water bath.
Particle size distributions of the homogenized extrudate samples were measured with a Malvern Mastersizer 2000[TM] (Malvern, United Kingdom). It was capable of detecting particles in a range of 0.2-2000 pm. The mean particle size distributions were determined based on three replicate measurements (n = 3). The mean particle size ([D.sub.50]) was used to represent the relatively tendency of the water-polyester extrudate to be dispersed into a greater amount of water.
Determination of Surface Energy
Ground and dried extrudate samples were hot pressed in a 5 x 5 x 0.5 cm mold under 2 metric tons for 2 min at 100[degrees]C. Surface energy values of the samples were estimated using the sessile drop contact angle method at room temperature with 1-bromonaphthalene (dispersive component [[delta].sup.d] = 42.517 mJ/[m.sup.2], polar component [[delta].sup.p] = 0.283 mJ/[m.sup.2]) and glycerin ([[delta].sup.d] = 35.541 mJ/[m.sup.2], [[delta].sup.p] = 25.359 mJ/[m.sup.2]). Dispersive and polar contributions to the total solid surface energy were calculated according to the Wu  method.
RESULTS AND DISCUSSION
Validation of Viscosity Calculations by the Die Rheometer
Neat molten polyester was extruded, without added water or modifier, at barrel/die temperatures of 100-140[degrees]C to measure its melt temperature and corresponding pressure drop across the inline rheometer. It is assumed that the Cox-Merz rule was applicable between the measurements in the offline rheometers and the die. Open source numerical software was used to simulate the custom build die for all viscosities of the neat polyester at different temperatures (200 Pa s and less) but also for neat water (0.001 Pa s) to calculate pressure drops. Calculated pressure drops and measured pressure drops across the orifice plate for the neat polyester at the different temperatures differed by less than 5%. Figure 2 compares the results obtained from the numerical simulations with experimental data from a parallel plate rheometer for the neat polyester. A semiempirical pressure drop-viscosity calibration curve, discussed below, is shown in the plot to closely fit to the experimental data.
Examining the system at fixed flow rate and different temperatures produced a range of viscosities for the calibration of the die. According to Bond  who examined the entrance pressure drop (Pe) for viscous fluid flow through an orifice plate flowmeter, a linear relationship exists between pressure drop and viscosity. Bond proposed the following equation:
[P.sub.e] = ([mu]q/[R.sup.3]) * [v.sup.3] (1)
where [mu] is the generalized Newtonian viscosity, Q is the volumetric flow rate, R is the radius of the orifice plate, and [v.sub.3] is the dimensionless constant for a viscous fluid. A best-fit ([R.sup.2] = 0.98) semiempirical expression for apparent viscosity was determined with the inline rheometer based on measured pressure drop which assumes the validity of the linear correlation observed by Bond. That expression was found to be:
[eta] = 8.65 * [10.sup.-5] [DELTA]P (2)
where [DELTA]P (Pa) is the pressure difference between the two pressure transducers in the die and [eta] is the apparent viscosity (Pa s). This semiempirical model operates with less than 5% error for viscosities greater than 1 Pa-s. The error from the semiempirical model (Eq. 2) increases up to 40% when neat water was considered. The model was considered appropriate in the context of this study for converting a pressure difference into estimates of apparent viscosity since the lowest recorded viscosity was approximately 70 Pa x s.
Use of the inline rheometer and Eq. 2 in this study is recognized to be based on several significant assumptions. First and foremost, it was assumed that the water-polyester mixtures remained morphologically similar across the orifice plate such that both pressure transducers were measuring flow conditions for the same fluid system. The similarity in individual pressure profiles at the two transducers during the tests suggests that the mixtures were adequately dispersed such that the extensional flow event across the orifice did not noticably affect the morphological state of the mixture; differences in the mixed state of a fluid are often denoted by attenuation of fluctuations in pressure for flow systems. Secondly, it is assumed that the water was adequately dispersed into the polyester such that the fluid flow resembled a single-phase system at all conditions studied; visual observations at the die exit showed no evidence of the water separating from the mixture. Goger et al.  previously measured the viscosity of this multiphase system in a batch reactor and found that a uniform distribution of water phase in polyester matrix can be assumed to exist for the polymer/water (i.e., multiple phase) system under the shear and pressurized environment.
Plasticizing Versus Surface Active Influence of the Modifier
The NC and C modifier produced in the study was analyzed by Gel Permeation Chromatography (GPC), NMR, and HPLC to determine their composition. The estimated ratio of monomer, dimer and larger species for the two clean modifiers are summarized in Table 1. The two modifiers consisted mostly of dipropoxylated bisphenol monomer (with the FA either sublimating or being washed away during cleaning) and dimer. Ideally, a modifier with a consistent degree of polymerization suitable for plasticization of the polyester was sought in the study to simplify the comparison between surface tension and viscosity effects on the dispersion process. Since the necessary quantity of modifier used in the study made it too difficult to terminate the reaction to produce exclusively a dimer or trimer, it is necessary to quantify the differences in plasticization between the two modifiers before examining their effects on the extrusion process.
Table 2 shows glass transition temperature ([T.sub.g]) values for the dried extrudate samples blended with different modifier types and contents. The residual water content among the extruded samples in this analysis was consistent at 1.84 [+ or -] 0.07%, determined by a Mettler-Toledo HG63 moisture analyzer. In all cases, samples showed only a single [T.sub.g] in their DSC thermograms, pointing to miscibility of the modifier in the polyester matrix. With increasing modifier content from 0 to 5 wt%, the [T.sub.g] was reduced from 61[degrees]C to 54[degrees]C, respectively, with the NC modifier, and from 61[degrees]C to 58[degrees]C, respectively, for the C modifier. The decrease in [T.sub.g] was considered to be a result of increased free volume within the polymer matrix by modifier inclusion. The change in free volume was calculated based on these [T.sub.g] values for the different polyester blends, using the Flory-Fox method and are summarized in Fig. 3 to highlight the plasticizing effect of these modifiers [19, 20]. The free volume of the polymer was higher with the NC modifier compared to the C modifier for all three concentrations tested. GPC results of the polyester/modifier blends provided in Table 3 confirm that the NC modifier tended to slightly lower the number average molecular weight to a greater extent than the C modifier for 1.5 and 5 wt%. Overall, the NC modifier was a better plasticizer than the C modifier in polyester though ideally they should be equivalent, at least in this characterization where no water was present. The plasticization behavior of the differing modifiers blended into polyester was similarly reflected in the Newtonian melt viscosity curves measured by parallel plate rheometer, with the data included in Fig. 3.
Figure 4 depicts the change in apparent viscosity with respect to the different modifier type and its content blended into the polyester, as measured by the inline rheometer, for R/W = 135 (i.e., the minimum water condition used to prevent the injector from being blocked by melt). The plot includes for reference, the apparent viscosities of extruded neat polyester (when the injector was completely removed from the extruder) and the extruded neat polyester at R/W = 135, which were both higher than the modifier containing samples. The uncertainty in apparent viscosity (denoted by the included error bars representing one standard deviation) corresponding to the neat polyester at R/W = 135 (21 Pa s) was higher than the uncertainty when either NC or C modifier were present with water. This larger uncertainty was suggestive of poorer mixing between the neat polyester and the small amount of water being added, though no water was visible at the die exit or on the extrudate surface to suggest that the water was not at least being partially incorporated. The plot shows that the apparent viscosity decreased with increasing content of either NC or C modifier. For the NC modified blends, the reported apparent viscosity is similar to the parallel plate data reported in Fig. 3, confirming the accuracy of the inline device. The magnitude in apparent viscosity seen in Fig. 4 was always higher with the non-converted modifier compared to the converted modifier at equivalent concentration in the polyester, which is opposite to expectations based on the free volume and molecular weight of the blends but unlike those measurements the viscosity was determined in the presence of water.
While plasticization resulted from either NC or C modifier being used, as evident by the declining viscosity with increased modifier content, the affinity of the polyester blend for water was always significantly greater with the C modifier. Table 4 indicates that the polar surface energy contribution for the polymer/modifier blend significantly increased from 2.16 to 11.87 mN/m while the dispersive contribution was reduced from 41.2 to 30.56 mN/m, as the C modifier content was increased from 0 to 5 wt%. In comparison, the NC modifier content has a negligibly small effect on the polar versus dispersion contributions to the total surface energy. However, the total surface energy did not vary based on the type of modifier used. Based on this polar contribution, the acid groups of the NC modifier should have much lower affinity for water compared to the carboxylate of the C modifier. Therefore, even though the C modifier itself had a lesser plasticizing effect than the NC modifier, the apparent viscosity of the polyester blend with C modifier was lower than NC modifier at R/W = 135 due to improved dispersion of water into the former polymer matrix.
Dispersion of Water into the Polyester/Modifier System
The addition of a larger amount of water into the molten polyester/modifier blend at R/W = 5 produced a significant change in apparent viscosity, as monitored by the inline rheometer, that was determined based on the type and content of modifier included. Figure 5 illustrates a difference in apparent viscosity on a time basis after R/W was decreased from 135 to 5, with a gradual viscosity drop (5 wt% NC modifier) or rapid decline (5 wt% C modifier) shown based on the modifier type used. The plot demonstrates extremes in the transient behavior of the process that were seen once water was added to the process. Similar transient viscosity behavior has been reported by Akay et al.  while attempting to disperse water into a molten low density polyethylene containing a surfactant using a batch melt mixer. Our own published results with a batch system  reported on a rapid torque drop being observed as water was dispersed into the same molten polyester used in this study, though in that case, surfactants were included. Potential reasons to be considered for the decline are, as follows: (1) water hydrolysis, (2) mechanical degradation, or (3) morphological development of an immiscible polymer-water dispersion; hydrolytic and mechanical degradation are known issues with solvent free extrusion emulsification. Considering the possibilities of water hydrolysis or degradation, the chain length of the extruded polyester without modifier ([M.sub.n] = 4,156 [+ or -] 139 g/mol) changed little compared to the original (unprocessed) polyester ([M.sub.n] = 4,158 [+ or -] 91 g/mol), exhibiting a viscosity drop of only 7 [+ or -] 4 Pa s at R/W = 5. Comparatively, the viscosity dropped by 71.3 [+ or -] 5.1 Pa s for polyester with 0.5 wt% C modifier ([M.sub.n] = 4,168 [+ or -] 76 g/mol) and 40.7 [+ or -] 4.6 Pa s for polyester with 0.5 wt% NC modifier ([M.sub.n] = 4,118 [+ or -] 76 g/mol). The GPC results suggest that hydrolytic and mechanical degradation were both improbable causes of the aforementioned viscosity drop. It had been reported that hydrolytic degradation was not responsible for the rapid torque drops seen in our earlier studies in a pressurized batch reactor either . In both that earlier batch work and in this study, morphological development of polymer-water blend was the considered explanation for the viscosity phenomenon being discussed.
The magnitude and slope of the apparent viscosity corresponding to this region of decline were used to quantify the phenomenon. It should be noted that slope of the horizontal line is assumed to be zero for the slope calculation. With NC modifier used in the polyester blend, the magnitude in viscosity drop (i.e., the difference in viscosity between the two plateau states) increased to 61 Pa s as modifier content increased to its highest loading (5 wt%) in the trials, as seen in Fig. 6. Conversely, the magnitude in viscosity drop was much larger with the C modifier blends, reaching 104 Pa s at the highest modifier loading (5 wt%). Moreover, Fig. 7 indicates that the viscosity drop became very abrupt, based on having the largest slope, with the highest C modifier content (i.e., 131 Pa s/min.) at R/W = 5. Comparatively, the slope of viscosity drop with NC modifier was less (reaching 46 Pa s/min at 5 wt%) representing a slower decline compared to the blends with C modifier. However, the viscosity drop was still sharper with NC modifier than found with the neat polyester (i.e., 5 Pa s/min) for the R/W = 5 condition.
The plots in Figs. 6 and 7 included surface energy contributions to reflect the significance of surface active agent on the transient viscosity behavior being observed by the inline rheometer. The samples with no modifier and NC modifier had similar surface energy contributions though showed increasing magnitude and slope of the viscosity drop with increasing content, showing that plasticization had an influence on the transition. However, the plots better show that the magnitude and slope of the viscosity drop were significantly larger as the polar surface energy contribution increased, suggesting that this was a stronger effect on the transition even when we account for the difference in free volume afforded between the C and NC modifiers. The carboxylate ions of the C modifier would seem to be effective at the polymer-water interface in decreasing the interfacial tension [9, 21] such that water dispersed more rapidly in the process than solely relying on mechanical energy.
Homogenization of Extrudate Samples
To further affirm that the transient viscosity behavior noted by the inline rheometer in Figure 5 was an indicator of water being intimately dispersed within the polyester, as a prelude to emulsification, the extrudate samples were homogenized offline. A sample with a higher developed interfacial area of water and polymer should break down into smaller particles in an excess of water by homogenization. These particles will not resemble the proper emulsion ultimately sought by SFEE, but due to the absence of sufficient surfactant or water in the process, this approach seemed reasonable in seeking additional that the transient viscosity behavior was in fact indicating the morphological state of the polymer-water dispersion.
Figures 8 and 9 show the mean particle size diameter ([D.sub.50]) plotted against the magnitude and the slope of the previously discussed viscosity drop, respectively, for the different types and contents of modifier. Homogenization and particle size measurement was conducted for both immediately extruded samples as well as extrudates stored for three days prior to the test. [D.sub.50] of the stored extrudate samples was consistent at 940 [+ or -] 120 [micro]m and did not vary based on the modifier type and content. The large particles, relative to those immediately homogenized, likely reflected water separation on storage. Figure 8 shows that [D.sub.50] decreased for samples corresponding to an increasing magnitude for the viscosity drop. [D.sub.50] for the blend with 5 wt% C modifier had decreased to 70 [micro]m, while being quite large at 720 pm without modifier. Moreover, the content of C modifier in the polyester blend had a significant influence on [D.sub.50], greater than NC modifier, as seen in Figs. 8 and 9. The smaller particles obtained by homogenization for C modifier compare to NC modifier fits with the hypothesis that a higher interfacial area between polymer and water was being produced and detected within the extruder when the C modifier was used. It was therefore with reasonable confidence to infer that the viscosity drop detected by the inline viscometer could be considered as related to the state of water dispersion in polyester matrix, for the dispersion zone of SFEE. Having this process-related indicator will make it easier to test formulation and operational factors related to SFEE modeling in future studies.
Morphology of the Water Dispersed into Polyester
Finally, the morphology of the extruded polyester-water dispersion at R/W = 5 was examined. Stratified polyester fibrils are indicated with white arrows in the SEM image shown in Fig. 10. In the polymer blend field, the morphology of a dispersed phase follows a well-known, three step mechanism starting from lamellas, to fibrils, and finally transformed into droplets . This eventual droplet transformation requires a prolonged period of time for high viscosity ratios . It seems that a stratified lamella structure occurred in the polyester-water system produced in the extruder, but in the dispersion stage, the water failed to complete the first transformative step of forming fibrils due to the excessively high viscosity ratio between the polyester and water.
The new inline rheological measurement technique was developed with the validation of 3D numerical software Open-FOAM, as well as conventional parallel plate and capillary rheometers. It has been shown that this dynamic technique can be used to detect viscosity changes, which are related to the state of dispersion of water into a polyester matrix. The water forms stratified lamella-like structure in the polymer as proper physiochemical conditions arise. Recognizing the relationship between the transient viscosity behavior and the developed state of dispersion, this inline rheometer will be an invaluable tool in modeling the SFEE process. Changes in the SFEE process related to matrix molecular weight and the content of a surface active modifier were separated in the study to understand which factor had a greater effect on water dispersion in the polyester matrix; these two factors change simultaneously in the SFEE process normally and create a challenging system to control. Without surprise, higher polar surface energy contributions were most important but the relevance of lower viscosity implies that some chain degradation in the actual SFEE process is important.
Supporting Information related to the strain sweep for the parallel plate measurements is available for this article and is accessible for authorized users.
The authors wish to thank the Xerox Corporation for their generous funding of this work as well as for the provision of the resin.
[1.] M. Trotta, F. Debemardi, and O. Caputo, Int. J. Pharm., 247, 153 (2003).
[2.] H. Hamishehkar, J. Enami, A.R. Najafabadi, K. Glani. M. Minaiyan, H. Mahdavi, and A. Nokhodchi, Colloids Surf. B: Biointerfaces, 74, 340 (2009).
[3.] S.N. Sani, N.G. Das, and S.K. Das, J. Microencapsulation, 26, 556 (2009).
[4.] M.A. Norato, L.L. Tavlarides, and C. Tsouris, Can J. Chem. Eng., 76, 486 (1998).
[5.] D. Song, W. Zhang, R.K. Gupta, and E.G. Melby, AIChE J., 57, 96 (2011).
[6.] G. Akay, Chem. Eng. Sci., 53, 203 (1998).
[7.] G. Akay, and L. Tong, J. Colloid Interface Sci., 239, 342 (2001).
[8.] G. Akay, L. Tong, H. Bakr, R.A. Choudhery, K. Murray, and J. Watkins, J. Mater. Sci., 37, 4811 (2002).
[9.] S.N. Ashrafizadeh, E. Motaee, and V. Hoshyargar, J. Petrol. Sci. Eng., 86, 137 (2012).
[10.] W.D. Pandolfe, J. Dispersion Sci. Technol., 2, 459 (1981).
[11.] K. Abe, M. Tsuruoka, S. Kiriyama, and K. Nakamori, U.S. Patent, 4320041 A (1982).
[12.] J. Chung, C.M. Cheng, S.M. Malachowski, E.D. Godshall and J.L. Pearson, U.S. Patent, 9052625 B2 (2014).
[13.] K. Zhou, G. Sacripante, and E. Zwartz, U.S. Patent, 20070141494 A1 (2007).
[14.] A. Goger, M.R. Thompson, J.L. Pawlak, and D.J.W. Lawton, Ind. Eng. Chem. Res., 54, 5820 (2015).
[15.] A.C.P. Neubauer, and N.W. Dunchus, U.S. Patent, 20120238681 A1 (2016).
[16.] A. Spyros, J. Appl. Polym. Sci., 83, 1635 (2002).
[17.] S. Wu, J. Polym. Sci. Part C: Polym. Symp., 34, 19 (1971).
[18.] W.N. Bond, Proc. Phys. Soc. Lond., 34, 139 (1921).
[19.] M. Bakar, and F. Djaider, J. Thermoplast. Compos. Mater., 20, 53 (2007).
[20.] T.G. Fox, and P.J. Floory, J. Appl. Phys., 21, 581 (1950).
[21.] S. Acevedo, X. Gutierrez, and H. Rivas, J. Colloid Interface Sci., 242, 230 (2001).
[22.] S.C. Jana, and M. Sau, Polymer, 45, 1665 (2004).
A. Goger (iD),(1) M.R. Thompson, (1) J.L. Pawlak, (2) M.A. Arnould, (2) A. Klymachyov, (2) R. Sheppard, (2) D.J.W. Lawton (3)
(1) Department of Chemical Engineering, Me Master University, Hamilton, Ontario, Canada
(2) Xerox Corporation, Rochester, New York
(3) Xerox Research Center of Canada, Mississauga, Ontario, Canada
Additional Supporting Information may be found in the online version of this article.
Correspondence to: M.R. Thompson; e-mail: firstname.lastname@example.org
Contract grant sponsor: Xerox Corporation.
Caption: FIG. 1. (a) 3D view of customized die. (b) Dimensions of the customized die. (c) Pressure distribution in customized die at viscosity of 200 Pa s. * All dimensions are in mm.
Caption: FIG. 2. Comparison of numerical simulation and experimental result.
Caption: FIG. 3. Free volume fraction and Newtonian viscosity for different modifier types and contents. ((a) Non-converted (NC) modifier for free volume fraction, (b) Converted (C) modifier for free volume fraction, (c) Non-converted (NC) modifier for Newtonian viscosity and (d) Converted (C) modifier for Newtonian viscosity).
Caption: FIG. 4. Apparent viscosity for neat polyester without water, different modifier type and content at resin-to-water ratio (R/W) 135. ((a) Non-converted (NC) modifier, (b) Converted (C) modifier, (c) Neat polyester without water, (d) No modifier with water).
Caption: FIG. 5. Example graphs of viscosity versus time from inline rheometer for 0.5 wt%NC modifier and 5 wt% C modifier at R/W = 5. (-5 wt% C modifier, -5 wt% NC modifier).
Caption: FIG. 6. Dispersive and polar contributions to total solid surface energy versus magnitude of viscosity drop. ((a) Dispersive contribution, (b) Polar contribution).
Caption: FIG. 7. Dispersive and polar contributions to total solid surface energy versus slope of viscosity drop. ((a) Dispersive contribution, (b) polar contribution).
Caption: FIG. 8. Magnitude of viscosity drop versus D50 for various modifier types and contents.
Caption: FIG. 9. Slope of viscosity drop versus D50 for various modifier types and contents.
Caption: FIG. 10. Stratification of polyester domain by water at resin-to-water (R/W) ratio 5:1 in TSE.
TABLE 1. Content determination method and content of both NC and C modifier. Non-converted Converted modifier modifier Component (relative wt%) (relative wt%) Method FA 7-9% 2-4% NMR pBPA (includes all 25-30% 32-36% HPLC oligomers) FA-pBPA 15-20% 30-40% HPLC pBPA-FA-pBPA 5-10% 1-3% GPC pBPA-FA-pBPA-FA-pBPA 6-8% 1-3% GPC TABLE 2. Glass transition temperature ([T.sub.g]) and melting temperature ([T.sub.m]) of dried extrudate samples and modifier, FA, and bisphenol A. [T.sub.g] [T.sub.m] Sample name ([degrees]C) ([degrees]C) Neat polyester 61.3 [+ or -] 0.6 -- 0.5 wt% non-converted (NC) modifier 60.8 [+ or -] 0.3 -- 1.5 wt% non-converted (NC) modifier 59.1 [+ or -] 0.3 -- 5 wt% non-converted (NC) modifier 54.0 [+ or -] 0.4 -- 0.5 wt% converted (C) modifier 61.1 [+ or -] 0.4 -- 1.5 wt% converted (C) modifier 60.0 [+ or -] 0.4 -- 5 wt% converted (C) modifier 57.9 [+ or -] 0.3 -- Converted (C) modifier -- 130.8 [+ or -] 2.3 Non-converted (NC) modifier -- 136.1 [+ or -] 3.1 Bisphenol A -- 67.4 [+ or -] 0.4 Fumaric Acid (a) -- 287.0 NC, Non-converted; C, Converted. (a) Vendor supplied value. TABLE 3. Number average molecular weight for different modifier type and content. Sample Name [M.sub.n] [g/mol] Neat polyester 4158 [+ or -] 91 0.5 wt% non-converted (NC) modifier 4118 [+ or -] 76 1.5 wt% non-converted (NC) modifier 3960 [+ or -] 76 5 wt % non-converted (NC) modifier 3529 [+ or -] 76 0.5 wt% converted (C) modifier 4168 [+ or -] 76 1.5 wt% converted (C) modifier 4012 [+ or -] 76 5 wt% converted (C) modifier 3670 [+ or -] 76 TABLE 4. Dispersive and polar contribution to total surface energy for different modifier type and content. Dispersive Polar contribution contribution Sample name (mN/m) (mN/m) Neat polyester 41.2 [+ or -] 1.55 2.16 [+ or -] 0.46 0.5 wt% non-converted 41.05 [+ or -] 1.81 2.12 [+ or -] 0.38 (NC) modifier 1.5 wt% non-converted 40.67 [+ or -] 1.62 1.83 [+ or -] 0.42 (NC) modifier 5 wt% non-converted 40.57 [+ or -] 0.95 1.46 [+ or -] 0.53 (NC) modifier 0.5 wt% converted 37.62 [+ or -] 1.15 5.54 [+ or -] 0.32 (C) modifier 1.5 wt% converted 33.41 [+ or -] 1.27 7.65 [+ or -] 0.41 (C) modifier 5 wt% converted 30.56 [+ or -] 1.35 11.87 [+ or -] 0.54 (C) modifier Total surface Sample name energy (mN/m) Neat polyester 43.36 [+ or -] 1.76 0.5 wt% non-converted 43.17 [+ or -] 1.54 (NC) modifier 1.5 wt% non-converted 42.50 [+ or -] 1.74 (NC) modifier 5 wt% non-converted 41.93 [+ or -] 1.61 (NC) modifier 0.5 wt% converted 43.16 [+ or -] 1.73 (C) modifier 1.5 wt% converted 41.06 [+ or -] 1.62 (C) modifier 5 wt% converted 42.43 [+ or -] 1.59 (C) modifier
|Printer friendly Cite/link Email Feedback|
|Author:||Goger, A.; Thompson, M.R.; Pawlak, J.L.; Arnould, M.A.; Klymachyov, A.; Sheppard, R.; Lawton, D.J.W.|
|Publication:||Polymer Engineering and Science|
|Date:||May 1, 2018|
|Previous Article:||Quaternized Polysulfone-Based Nanocomposite Membranes and Improved Properties by Intercalated Layered Double Hydroxide.|
|Next Article:||Cure Behavior and Kinetic Study of Diglycidyl Ether of Bisphenol A With a Tertiary Amine Salt by Differential Scanning Calorimetry.|