DYNAMIC RHEOLOGICAL MEASUREMENTS OF AQUEOUS POLYESTER DISPERSION IN BATCH REACTOR AND TWIN SCREW EXTRUDER Ali Goger *, Michael R. Thompson *, John L. Pawlak **, David J.W. Lawton *** * Department of Chemical Engineering McMaster University, Hamilton, Ontario, Canada ** Xerox Corporation, Rochester, New York 14604, United States *** Xerox Research Center of Canada, Mississauga, Ontario L5K 2L1, Canada Abstract This paper reports on inline measurement techniques for the rheological behavior of aqueous polyester dispersion in batch reactor and twin screw extruder (TSE). Since the preparation of latex without hazardous solvent is a relatively new technique, very little has been reported to understand the kinetic aspects of the process for both batch reactor and TSE. A sudden viscosity drop is observed in a batch reactor whereas the viscosity tends to oscillate in TSE during the addition of water when surface tension is low enough. The viscosity changes during the addition of water are thought to be related to the morphological changes during the process since surfactant must be present else no change occurs. In this paper, different surfactant and NaOH concentrations have been studied for their influence on the viscosity so that emulsification may become a predictable process in a TSE. Introduction Solvent-free emulsification of polymers in an aqueous continuous process has rapidly progressed despite the fact that it has numerous challenges in obtaining a stable system without using hazardous solvent. In this system, water is added continuously into a highly viscous molten polymer (of the order of 10 +2-10 +3 Pa.s). The major issues include a tendency for poor mixing due to the extremely high viscous ratio of components (around 10 +6 ), and the low compatibility of the polymer and water phases. However, this process approach is not a novel idea, having been around since 1982[1]. There are a select few articles [2-4] that currently study the mechanism of this process as applied to a batch system or twin screw extruder. Measuring the rheological properties of the multiphase aqueous system can be quite challenging out of a necessity to keep the phase together in their mixed state. However, the rheological properties can still provide useful information about the morphology of this complex system. Commercial rheological instruments are not adequate for this type of multiphase system as they generally cannot provide the required pressure, temperature and shear constraints simultaneously. As a result, the dynamics of morphological changes in batch and/or extruder systems determined by commercial rheological instrumentation may be misrepresentative of the true material behaviour. Therefore, developing new inline methods for both batch and extruder systems are needed to understand the dynamics of rheology. The Metzner-Otto method [5-6] has proven to be a suitable approach to calculate the effective shear stress and shear rate for Newtonian and non-newtonian fluids in batch mixer systems. Conversely, the capillary tube die has been suitable developed to monitor viscosity on extruders, such as the measurement of polymer/co 2 systems [7]. Using the die at the end of the extruder is an appropriate method to understand the rheological behaviour of the SFEE system. However, the die needs to be flexible because of the broad range of viscosities in the system as water mixes into the polymer (around 0.1-200 Pa.s). Therefore, an orifice plate setup is introduced in this study to examine transient rheological behaviour of our solventfree emulsification process. The purpose of this paper is to demonstrate two inline methods to measure the dynamic rheological response of a polymer-water-surface active agent system as the components are mixed together in a batch reactor or twin screw extruder; the work studies the pre-inversion stage of the emulsification process. The batch study was disclosed in a recent article [8]. Materials Experimental A polyester resin was supplied by the Xerox Corporation. The resin was a high flow grade with an acid number of 16.2 mg KOH/g sample. An anionic surfactant, 4-dodecyl benzenesulfonic acid (SDBS) was purchased from Sigma-Aldrich. The sodium hydroxide (NaOH) was purchased from Caledon Laboratories Ltd., and was later grounded into a powder of comparable size to the polyester resin. Deionized water (>0.1 µs/cm 2 ) was used in the trials. SPE ANTEC Indianapolis 2016 / 710
Apparatuses Batch trials were done with a 2L Buchiglas BEP 280 laboratory pressure reactor system rated for 60 bars and fitted with an anchor type impeller modified with three turbine blades as seen in Figure 1(a). A grounded K-type thermocouple was fit into the reactor and a pressure transducer (PX302-500gv; Omegadyne) was installed to monitor the nitrogen gas head pressure. The reactor system was pressurized to a gas pressure of 1.72 MPa with nitrogen (99.999% purity; Air Liquide). The reactor was heated with 38L VWR circulating bath with a controller. Deionized water was fed into the reactor using an Optos pump from Eldex. 3D-open source computational software OpenFOAM 2.0.1 was used to solve the conservation of mass and momentum equations. The SIMPLE algorithm was used with a constitutive power law model (n= 0.96 obtained by a parallel plate and capillary rheometer for polyester containing NaOH and SDBS). The mesh of batch reactor geometry was developed using Gambit from Fluent Inc. The mesh was composed with 2.5x10 6 tetrahedral elements. Extrusion trials were done with a 40 L/D 27 mm Leistritz ZSE-HP co-rotating twin screw extruder (TSE) Dry materials were fed at 8 kg/h by a Brabender DDSR20 gravimetric feeder. Liquids were fed by an ISCO 260D syringe pump (Teledyne Technologies, Inc.). The screw design cannot be provided due to proprietary restrictions. A 2 mm orifice plate was added into the flow path of a 15 mm diameter tube for the die to increase the pressure difference. The aim was to detect the lower pressure drop along the die for high water content trials. Two pressure transducers (PT467E-300psi, Dynisco) were fit to the customized die. The customized tube die, shown in Figure 1(b), was simulated to convert the pressure difference data to viscosity by the same numerical software OpenFOAM. The flow domain of the die was meshed with 93,782 hexahedral elements by using Gambit. The SIMPLE algorithm was used again in this simulation. Batch Reactor Results and Discussion The batch reactor was used to study changes in torque during the addition of water while variables such as SDBS content, NaOH content, and impeller speed were adjusted. The processing and recipe conditions of batch reactor are summarized in Table 1. The reactor was initially charged with the dry ingredients and then heated. After the melt reached 140ºC (within 30 min. of heating), the batch reactor was pressurized to 1.72 MPa to keep the molten polymer and so-to-be-added water together in phase. After the impeller started, room temperature deionized water was added above the impeller at 20mL/min. The torque values were collected at the data acquisition sampling rate of 10 samples/sec. during the addition of water. The anchor type impeller with a three-turbine blade has a complicated geometry for determining the shear stress and shear rate. The purpose of the numerical study was to convert the torque data of the batch reactor into viscosity. A quasi-steady state was valid due to the low Reynolds number (Re<10-2 ) in such a polymeric system. The non-slip boundary condition was considered for simulation because a low shear rate exists in the system and no discontinuity was observed in the rheometers. At the vessel wall and its bottom surface, a null velocity was used while the impeller surface moved according to the local impeller speed. At the free surface, zero velocity was assumed in normal direction and zero shear stress was assumed for both normal and tangential directions. The torque of the simulated impeller was calculated using Eqn. 1 by OpenFOAM and integrated over the elements on the impeller: Γ=τAr (1) where Γ is the torque, τ is the shear stress, A is the surface area and r is the distance from the rotational axis. γ = 34.37ω (2) τ = K stress Γ n (3) K stress =6421e ΔT (4) where γ is the shear rate, ω is the impeller speed, τ is the shear stress and ΔT is the temperature difference. The empirical equations, listed as Eqn. 2-4, were derived from the simulated and experimental study for different shear rates and temperatures. Figure 2(a) indicates the shear stress distribution on the impeller. Figure 1. (a) Impeller dimensions. (b) Customized tube die dimensions. SPE ANTEC Indianapolis 2016 / 711
Figure 2. (a) Shear stress distribution on impeller of batch reactor. (b) Pressure drop over the die. A pressurized environment was required to keep the phases together, in order to better understand the rheological behaviour of this multiphase system at high temperatures. One of two main behaviours were observed in the batch reactor. Either the viscometer reported a gradual decline in torque (i.e. shear stress), or a rapid substantial drop in the torque, as seen in Figure 3. The sudden drop behaviour was never observed without water being added into the system. The inline viscosity measurements were done at 100 and 150 rpm during the water addition stage. It appeared that the sudden torque reduction was shear dependent. Therefore, unless otherwise stated, all conditions examined for transient behaviour were at 150 rpm for different chemical ingredients. The derivation of torque data (dγ/dt) was analyzed to quantify the sudden drop; the dγ/dt values for different contents and impeller speeds. Reduction in torque at 150 rpm was gradual without NaOH (dγ/dt=-0.0021) and neutralization ratio (NR) 43.7% with NaOH (dγ/dt=-0.0035). However, very sharp drops can be seen when the system has NR=131% (dγ/dt=-0.19); here the addition of salt groups to the polyester improved water incorporation. In addition, the sudden torque drop at NR=131% was shear dependent since it could not be observed at the lower speed of 100 rpm (dγ/dt=-0.0035). The sudden reduction in torque became more common among the trials that had SDBS surfactant, even at NR=43.7 %. dγ/dt became -0.191 and - 0.22 for NR: 43.7% and 131% respectively. Figure 3 indicated that the torque drop happened earlier when the system had SDBS surfactant. Lastly, the system change causing the sharp torque drop was less shear dependent with SDBS than the molten polyester-naoh without SDBS. It is believed that the rapid torque drop is related to morphological changes during the addition of water though in the absence of being able to sample from the reactor, this could not be confirmed. Figure 3. Viscosity changes during the water addition in batch reactor at 150 rpm. (a) NR=131%, 0% SDBS; (b) NR=131% -7.5 %w/w SDBS. Twin Screw Extruder (TSE) The desire was to translate the findings from the batch reactor over to the TSE where a continuous emulsification process was being studied. The TSE study probed changes in the pressure difference along the customized die during the addition of water. The conditions for extruder trials are summarized in Table 2. The barrel temperature profile was kept constant at 95ºC in every zone of extruder. The customized die, shown in Figure 1(b), was set to 95ºC. A NaOH aqueous solution was initially pumped into a third zone (Z3) of the extruder at 1 ml/min to avoid blocking the injector while melt filled the system. The extruder was started and set at 300 rpm. After 45 seconds from when polymer exited, the water-naoh flow rate was increased to 26.1-27.5 ml/min for various recipe conditions. The pressure difference data was collected continuously at the sampling rate of 100 samples/sec. over time. As with the batch mixer, the flow system of the die was complex and no analytical solution was apparent. Once again, we turned to the numerical software to assist in developing an empirical equation for relating the sensor readings (pressure in this case) to viscosity. The non-slip condition was assumed for the die simulation. At the die wall, the null velocity was used while the inlet of the die had set a flow rate of 8 kg/h. The pressure at the outlet of die was assumed to be zero. Different pressure drops between two pressure transducers were calculated for different viscosities by OpenFOAM. Eqn. 5 is the empirical relationship found to convert pressure drop into viscosity. Figure 2(b) indicated pressure drop over the customized die at a viscosity of 200 Pa.s. η=0.6222xδp-0.0024 (5) SPE ANTEC Indianapolis 2016 / 712
where η is viscosity and ΔP is pressure drop. Figures 4 (a) and (b) indicate that the viscosity fluctuates during the addition of water in TSE (though eventually stabilizing after about 300 s). The sudden torque drop observed in the batch reactor study, whereas the viscosity tends to oscillate in TSE. Figure 4 (a) indicates that the system viscosity fluctuates between 150 Pa.s and 400 Pa.s. Instead, Figure 4 (b) suggests that the viscosity fluctuations decrease between 120 Pa.s and 210 Pa.s. Less fluctuation with SDBS may be explained by the fact it helped the mixing of polyester and water and created more stable morphological state compare to the cases with no SDBS. In addition, it was observed that the extrudate sample became brown (indication of burning by NaOH) after the fluctuation stabilize (about 300 s.) without SDBS. The brown sample was never observed with SDBS in the TSE. The improved dispersion of water in the polyester in the presence of SDBS would facilitate end group conversion rather than create highly alkaline regions leading to burning. parallel plate and capillary rheometers. It has been shown that these dynamic techniques can be used to detect viscosity change which may be related to the morphogical state of an aqueous polyester dispersion in batch reactor and TSE. More stable morphologies of aqueous polyester were obtained when the surface tension was adequately in both batch reactor and TSE. Acknowledgements The authors would like to thank the Xerox Corporation and Xerox Centre of Canada for their research funding and technical advise on this project. In addition, special thanks go to the Society of Plastic Engineers Extrusion Division/Lew Erwin Scholarship. References 1. K. Abe, M. Tsuruoka, S. Kiriyama, and K. Nakamori, United States Patent 4,320,041 (1982) 2. G. Akay, J. Colloid Int Sci, 239, 342-357 (2001) 3. D. J.W Lawton, MASc Thesis, McMaster University (2013). 4. D. Song, W. Zhang, R. K. Gupta, and E. G. Melby, AIChE J, 57, 96-106 (2011). 5. A.B. Metzner and R.E. Otto, AIChE J., 3, 3 (1957). 6. P.A, Tanguy, F. Thibault and E. Brito De la Fuente, Can. J. Chem. Eng., 74, 220 (1996). 7. M. Lee, C.B. Park and C. Tzoganakis, Polym. Eng. Sci., 39, 99-109 (1999). 8. A. Goger, M.R. Thompson, J.L. Pawlak and D.J.W. Lawton, Ind. Eng. Chem. Res., 54, 5820 5829 (2015). Figure 4. Viscosity changes during the water addition in customized die at the end of twin screw extruder. (a) NR=100%, 0% SDBS; (b) NR=100%, 7.5 %w/w SDBS. Validation The numerical simulation results were validated whenever possible with a parallel plate and capillary rheometers, for the polyester as well as its mixtures with SDBS and/or NaOH. The difference between the calculated torque and measured torque from the batch reactor was less than 3% for polymer-surface active agent mixture. The pressure drop difference between calculated and measure value was less than 2% for the customized tube die. Conclusions The new inline rheological measurement techniques were developed with the validation of 3D numerical software OpenFOAM, as well as conventional SPE ANTEC Indianapolis 2016 / 713
Table 1. Operating conditions for the batch reactor study. Batch Reactor Type Content Neutralization Ratio Water Temperature Impeller Speed Water Flow Rate [-] [w/w %] [%] [ºC] [rpm] [ml/min] SDBS 0 0 25 100 20 Unicid 7.5 43.7 150 131 Table 2. Operating conditions for twin screw extruder. Type Content Neutralization Ratio Water Temperature Impeller Speed Water Flow Rate Twin Screw Extruder [-] [w/w %] [%] [ºC] [rpm] [ml/min] SDBS 0 0 25 300 20 7.5 100 SPE ANTEC Indianapolis 2016 / 714