Screw optimization of a co-rotating twin-screw extruder for a binary immiscible blend.
The predominant commercial method of blend preparation is melt blending in an extruder. The co-rotating twin-screw extruder (co-TSE) has outstanding distributive mixing efficiency and self-wiping characteristics (1). In general, the extruder for blending purposes must be capable of imposing an appropriate level of strain, and it should alternate the direction of the imposed shear stress (2).
Our objective in this study was to investigate systematically alternative screw configurations for immiscible blends prepared in a co-TSE. The residence time distribution and morphology evolution were of primary interest. The model blend was PA 6/PP 80:20 with no compatibilizer added.
Laminar mixing in real systems is a combination of distributive and dispersive mixing (3). In distributive mixing the material is divided and oriented by the shear strain. The moving fluid elements form parallel laminar planes and new area (1). The efficiency of distributive mixing depends on the number of the re-orientations (4, 5). Dispersive mixing, on the other hand, rearranges the fluid elements in random order and no laminar plane exists. Dispersive mixing depends on the flow characteristics, the melt viscosity, and elasticity (6).
Evaluation of Mixing Flow
Danckwerts (7, 8) introduced the residence time distribution (RTD) of a continuous mixer to represent the cross-flow characteristics, It has been shown (9) that RTD provides information primarily about mixing in the axial flow direction, and is not very useful for analyzing the laminar mixing of viscous fluids.
One easy way to determine RTD is to add a well-detectable tracer to the feed and measure it in the extruded fractions. FTIR provides a rapid and reproducible technique for the measurement. Many experimental studies have been made on the RTD of different types of extruders (10-21).
It has not been possible, however, to develop a general hydrodynamic flow model based on RTD. The flow in an extruder is usually characterized as a combination of plug flow and the flow in an ideal mixer (9, 10, 12, 18-23).
The Peclet number, [N.sub.Pe], is a measure of axial dispersion. It can also be defined as the inverse of the dispersion number, which is dependent on the eddy diffusion coefficient. In that case the eddy diffusion is considered as the source of the axial mixing. The Peclet number has been used to compare different screw geometries (12). Thus
[Mathematical Expression Omitted]
where [N.sub.d] is the dispersion number, [D.sub.e] the eddy diffusion coefficient, [N.sub.Pe] the Peclet number, u the flow velocity, and L the characteristic length.
Residence time dispersion is important for extruder performance. The dispersion can be evaluated by holdback value (H). It has also been shown by Noar and Shinnar (22) that the holdback is a measure of the relative dispersion (RD).
H = 1 - [integral of] e(t) dt between limits RT and 0 = RD/2 (2)
where H is holdback, RT the mean residence time, e(t) the RTD function, t time, and RD the relative dispersion. The value of the holdback lies within the range 0 [less than] H [less than] 1. The holdback value H = 0 describes plug flow, H = 0.368 the ideal mixer, and H = 1 excessive stagnancy and channeling (24). The holdback values can reflect dual behavior. High holdback values can describe good longitudinal mixing or they can reflect high stagnancy and channeling. The hold-back value cannot distinguish good macromixing from poor macromixing.
The mixing intensity function describes the mixing efficiency better than the holdback value. The mixing intensity function is defined as (22)
[Omega]([Theta]) = e([Theta])/i([Theta]) (3)
where [Omega]([Theta]) is the mixing intensity function, e([Theta]) the normalized RTD function, and i([Theta]) the internal age distribution. Good macromixing in the extruder is achieved when old fluid elements exit with greater probability than young ones. The slope of the curve [Omega]([Theta]) vs. [Theta] is an indicator of macromixing: The magnitude of the slope measures the extent, and the sign of the slope the quality, of the macromixing. Negative slope is evidence of channeling and stagnancy (22).
The screw configuration of a co-rotating twin-screw extruder is usually a combination of various mixing elements, i.e., kneading blocks or mixing gears, and conveying elements. The kneading blocks are used for melting the polymer, dispersing the additives, and mixing the melt. The degree of fill in the melting section has a marked effect on the mixing quality. The degree of fill in the kneading blocks can be adjusted by back-pressure elements, a barrel valve, or the staggering angle of the kneading element. The kneading elements may have positive, neutral, or negative conveying characteristics. Leakage flows caused by the high pressure zone in front of the head of the kneading block are very important to achieve high shearing forces and back-flow to the shearing zone (23).
Reversely conveying kneading blocks (RKB) create, through back-flow, one extra whirlpool in a fully filled screw (25). They also give a narrower residence time distribution and better distributive mixing efficiency (26). Experimentally it has been shown that the average residence time in the RKB and neutral kneading block (NKB) configurations is greater than that in the forward conveying kneading block (CKB) configuration (27).
The kneading type mixing element is preferred for polymer blends with viscosity ratio near unity. However, if the viscosity of the minor phase is much lower than that of the matrix, gear-type mixing elements are recommended (28, 29).
The commercial polymers used in this study are characterized in Table 1. Polyamide 6 grade, Ultramid B3S, was supplied by BASF AG, and isotactic homopolypropylene VB65 50B, by Neste Chemicals Oy. Both polymers were provided in the form of extruded pellets. No compatibilizer was added. The viscosity behavior of the components measured with a Gottfert Rheograph 2002 capillary viscosimeter is described in Fig. 1. The crossover of viscosities takes place at a shear rate of approximately 250 [s.sup.-1].
The mixing efficiency of seven different screw configurations for co-TSE was investigated. The screw configurations are summarized in Table 2 and Fig. 2. In addition to the different mixing elements added to [TABULAR DATA FOR TABLE 1 OMITTED] the screw, the effect of screen pack density (i.e., mesh number) was investigated with screw #1, and the effect of a barrel valve (i.e., degree of fill) with screws #2 and #7.
Blending and Sampling Positions
The blending was carried out in a Berstorff 25 x 39.5D co-rotating, closely intermeshing twin-screw extruder having the screw diameter ratio (D/d) of 1.45. Blending conditions are presented in Table 3. The blend composition was 80% PA 6 and 20% PP in all trials. Since screw speed, output, and melt temperature were constants as well, the only variable was the screw configuration.
Table 2. Tested Screw Configurations. Abbreviations: NKB: Neutral (90 [degrees]) Kneading Block; CKB: Conveying (+45 [degrees]) Kneading Block; RKB: Reversely Conveying Kneading Block (-22.5 [degrees]); and MG: Mixing Gear. See also Fig. 2.
Screen Barrel Screw # MIX1 MIX2 MIX3 Pack Valve
1A - - - - Open 1B - - - 100 MESH Open 1C - - - 250 MESH Open 2A NKB - - - Open 2B NKB - - - Closed 3 CKB - - - Open 4 CKB + RKB - - - Open 5 CKB + RKB NKB NKB - Open 6 CKB + RKB NKB MG - Open 7A CKB + RKB MG NKB - Open 7B CKB + RKB MG NKB - Closed
The sampling positions and sample designation are indicated in Fig. 2 by black arrows (12D, 22D, 32D, BS, BD, AD). The sample BS (before screen pack) was taken only when the screen pack was applied. The sampling procedure involved compression of the molten sample between two plates and an immediate quenching. The sample AD was in the form of an extruded strand, which was immediately cooled in a water bath after exiting from the die.
Residence Time and Residence Time Distribution
Residence time distributions were determined using potassiumthiocyanate, KSCN, as a tracer. The unit pulse of the tracer was fed into the feed hopper and samples were taken from the die. The samples taken at 5 s or 10 s time intervals were investigated by FTIR, by detecting the ratio of the absorbance peak of thiocyanate at 2056 [cm.sup.-1] and that of pure polymer blend at 841 [cm.sup.-1]. When the absorbance ratio was at maximum, also the concentration of the tracer was maximum. The normalized residence time distribution, e(t), was determined as follows:
e(t) = C/[summation of] C[Delta]t where 0 to [infinity] (4)
where e(t) is exit age distribution (i.e., RTD), C concentration, and t time. The mean residence time, RT, is the concentration weight average of the residence time
RT = [summation of] tC[Delta]t where 0 to [infinity]/[summation of] C[Delta]t where 0 to [infinity] (5)
The RTDs were converted to a more comparable form by introducing the reduced time (t/RT).
Particle Size Determination
The morphology of the fractured surfaces of the samples was investigated with a JEOL JXA-840A scanning electron microscope. All samples were fractured at liquid nitrogen temperature, and fractured surfaces were coated with a 15 nm layer of gold. At least 100 particles were included in each particle size determination. The average error in the particle size determinations was 0.5 to 1.0 [[micro]meter].
RESULTS AND DISCUSSION
Effect of Screw Configuration on the Residence Time
Table 4 lists the mean residence times and hold-back values of the various screw configurations. The shortest mean residence times were obtained with screws #1A and #3, which are purely conveying. The more mixing elements or elements causing back pressure that were added to the screw configuration, the longer was the mean residence time. The longest mean residence time was obtained with screw #5. The extended time evidently was due to the high back-flow rate.
The highest holdback value was obtained with purely conveying screw #1A, indicating possible stagnancy and channeling. The lowest holdback was obtained with screws #2 and #7A, which also exhibited narrow RTD. The holdback yields a crude estimation of the shape of the RTD, but is inadequate to characterize the extent and quality of macroscopic mixing.
The RTD for various screws are described in Fig. 3. Screws #1A and #3, exhibiting the most conveying capacity, yielded the broadest RTD. Screws #5, #6, and #7A yielded the narrowest distribution. No clear correlation was found between residence time and RTD.
Table 3. Blending Conditions For the Screw Configuration Tests. SEI is the Specific Mechanical Energy Input During the extrusion.
Screw Die [T.sub.m] Output Speed SEI Pressure Screw # ([degrees] C) (kg/h) (rpm) (kWh/kg) (bar)
1A 249 8.0 200 0.10 20 1B 252 8.0 200 0.10 30 1C 251 8.0 200 0.11 43 2A 252 8.0 200 0.12 22 2B 254 8.0 200 0.13 22 3 250 8.0 200 0.11 23 4 250 8.0 200 0.12 21 5 251 8.0 200 0.13 20 6 251 8.0 200 0.14 20 7A 251 8.0 200 0.14 20 7B 253 8.0 200 0.16 20 Table 4. Mean Residence Times (RT) Determined by Eq 5 and the Holdback Values (H) Calculated by Eq 2.
Screw # 1A 2 3 4 5 6 7A
RT (s) 67 69 62 72 95 86 83 H 0.37 0.23 0.33 0.36 0.34 0.31 0.24
The RTD and holdback value as such were found to be nonsensitive indicators of the total mixing, because only the axial mixing is expressed. They do not describe the distributive mixing in the mixing sections of the screw. Nevertheless, the shape of the RTD is of great importance for the information it gives of residence times in the various mixing sections. That information can be utilized in the mixing intensity function (Eq 3).
The mixing intensity functions for all screw configurations calculated by Eq 3 are depicted in Fig. 4. As can be seen, the screws with the greatest conveying capacity, #1A and #3, behaved like ideal mixers with low positive slope for the mixing intensity functions. This may be due to the high back-flow rate at the screw end. When additional kneading elements were added (screws #5, #6, #7A), the flow in the extruder was more like the plug flow (or the flow in ideal mixers in series), indicated by increasing positive slopes. These high slopes reflected better transverse mixing as well. Thus, every mixing section could be considered as an additional ideal mixer integrated to the screw.
A negative slope for the mixing intensity function is evidence of holdback. As indicated by the mixing intensity functions shown in Fig. 4, the screws #2, #3, #4, #5, and #6 exhibited a slight negative slope of in some part of the curve. No correlation. was found between the negative slopes and the holdback values presented in Table 4.
Particle Sizes of the Dispersed Phase and Morphology Evolution Along the Screw
Particle sizes of the dispersed phase of the blend at different positions along the screw are presented in Table 5 and Fig. 5. The sampling positions are explained in Fig. 2 and the screw configurations in Table 2. As can be seen, high shear stress was evidently important in the melting zone near the mixing section MIX1. The combination of CKB and RKB elements in the melting section (see screws #4, #5, #6, and #7) yielded the smallest particle size after the first mixing element (12D). This was due to a high degree of fill in the mixing section. The back-flow and mean residence time under high shear stress were increased as well.
Table 5. Particle Size of the Dispersed Phase at Different Sampling Positions.
Sampling Position Screw # 12D 22D 32D BS BD AD Number Average Particle Size ([[micro]meter])
1A 1200 1000 15 - 10 9.2 1B 1000 600 10 11 9 8.0 1C 1000 600 12 11 9 8.5 2A 250 40 11 - 9 6.6 2B 50 16 7 - 9 6.3 3 80 50 9 - 10 6.1 4 3O 15 9 - 7 6.1 5 15 15 10 - 9 5.9 6 16 8 6 - 8 5.9 7A 15 7 8 - 8 6.0 7B - - - - - 4.9
The importance of sufficient residence time in the mixing section was also indicated by the barrel valve tests. When the barrel valve was closed, the residence time and back-flow at the mixing section were increased and, as a result, the particle size diminished. The morphology and the efficiency of different kneading elements after the first mixing section and barrel valve (position 12D) are presented in Fig. 6.
The dispersion of the minor phase is dependent on the conveying capacity, and, as a consequence, on the degree of fill of the element. The NKB elements appeared to operate partially filled, as did the CKB elements. The mixing efficiencies of partially filled NKB and CKB differed only slightly. The combination of a MG and a NKB element in mixing sections MIX2 and MIX3 (screws #6 and #7) proved more efficient than two NKB kneading elements (screw #5) for phase particle size control of the minor phase. The order of the NKB element and the MG appeared to have little effect on the mixing efficiency and residence time.
The final particle size was not only controlled by the screw configuration but also by the back-flow at the screw end. Back-flow from the screw end-chamber to the screw section improved the dispersion of the minor phase. Slight evidence of coalescence was observed with screws #2B, #3, and #6 (see samples BD in Table 5) in the stagnant volume before the die plate. However, this was compensated for by the orientation and shearing effect at the die hole.
The effect of a screen pack (screw configurations #1B and #1C) on the final particle size was minor. The decrease was only 2 [[micro]meter] at maximum in the mean particle size when a screen pack was applied. Rather than improving the dispersion, the screen packs helped to homogenize the melt, promoting easier processing without pumping effect or other instabilities. The melt pressure at the die was clearly higher when a screen pack was applied.
Chemical Reactions in Blend Extrusion
Compatibilization of immiscible polymer blends is necessary if mechanical properties are to be achieved. Most compatibilizing agents induce a surface reaction between the matrix and dispersed phase. Usually these reactions are time dependent, which means that the formation of new surface area at an early stage of the mixing is advantageous. The analogy with reactive compounding is clear. The crosslinking reactions induced by peroxide, for example, are extremely difficult to control if the dispersion of peroxide in the matrix is poor or the residence time at the desired temperature is too short or too long.
If compatibilizing agent is to be added, then screws should produce high surface area of the dispersed phase at an early stage of the axial length of the extruder, facilitating the location of the compatibilizer at the interface and allowing maximum time for the preferred surface reaction to take place.
Long residence time in the molten stage because of an excessively long screw increases degradation of the polymer, and the properties of the material will be negatively affected. Furthermore, the investment and operation costs are increased when an unnecessary long compounding machine is installed, and energy consumption rises simultaneously. Thus, effective mixing is desirable for technical and economical reasons.
The effect of screw configuration of a closely inter-meshing co-TSE on the residence time and mixing efficiency was studied with uncompatibilized immiscible PA6/PP (80:20) blend.
As back pressure and number of mixing elements were increased, the mean residence time increased as well. The narrowest residence time distribution was achieved with screws exhibiting back-flow imposed by the RKB elements. There was no correlation between residence time and RTD.
The RTD and holdback value were found to be poor indicators of total mixing efficiency. However, the mixing intensity function derived from the shape of the RTD was of great assistance in evaluating the flow characteristics and mixing efficiency. As the number mixing elements was increased, better transverse mixing was achieved, indicated by increasing positive slope of the intensity function, changing the flow character of the extruder to be more plug-like. In short, it could be concluded that each mixing section acted like an ideal mixer applied to the screw.
We found that high shear stresses in the melting section were of great importance in achieving good dispersion in the immiscible blend. Long residence time and high fill ratio in the high shear melting zone also improved the dispersion.
The goodness of the dispersion was not only defined by the screw configuration but by the character of the downstream flow. We found that coalescence occurred if stagnant volume was present in the screw end-chamber. Finally, the orientation and shearing in the die hole determined the particle size of the dispersed phase.
The results of this study should also be relevant to systems in which a compatibilizer is added to the blend or where other chemical reactions are taking place possible during the extrusion. For example, the optimal downstream feeding positions and temperature profiles of the extruder can be determined more accurately on the basis of the present findings.
The authors wish to thank Neste Oy for assistance in the preparation of blends. The financial support of the Neste Foundation is greatly appreciated.
C = Concentration.
d = Inner screw diameter.
D = Outer screw diameter.
[D.sub.e] = Eddy diffusion coefficient. e(t), e([Theta]) = Exit age distribution = residence time distribution (area = 1).
H = Holdback value.
t([Theta]) = Internal age distribution.
L = Characteristic length.
[Mathematical Expression Omitted] = Number average molecular weight.
[Mathematical Expression Omitted] = Weight average molecular weight.
[N.sub.d] = Dispersion number.
[N.sub.Pe] = Peclet number.
RD = Relative dispersion.
RT = Mean residence time.
SEI = Specific (mechanical) energy input.
t = Time.
[T.sub.m] = Melt temperature.
u = Flow velocity.
[Theta] = Reduced time t/RT.
[Omega]([Theta]) = Mixing intensity function.
AD = After die.
BD = Before die.
BS = Before screen pack.
CKB = Conveying kneading block.
Co-TSE = Co-rotating twin-screw extruder.
FTIR = Fourier transform infrared spectroscopy.
KSCN = Potassiumthiocyanate.
LDPE = Low-density polyethylene.
NKB = Neutral kneading block.
PA 6 = Polyamide 6.
PP = Polypropylene.
RKB = Reversely conveying kneading block.
RTD = Residence time distribution.
1. C. J. Rauwendaal, Polym. Eng. Sci., 21, 1092 (1981).
2. K. Eise, J. Curry, and J. F. Nangeroni, Polym. Eng. Sci., 23, 642 (1983).
3. J. M. Ottino and R. Chella, Polym. Eng. Sci., 23, 357 (1983).
4. C. S. Tucker and R. J. Nichols, SPE ANTEC Tech. Papers, 33, 117 (1987).
5. M. H. Mack and T. F. Chapman, SPE ANTEC Tech. Papers, 33, 136 (1987).
6. C. J. Rauwendaal, Polymer Extrusion, Hanser Publ., New York (1986).
7. P. V. Danckwerts, Chem. Eng. Sci., 2, No. 1 (1953).
8. P. V. Danckwerts, Chem. Eng. Sci., 8, No. 93 (1958).
9. D. M. Bigg, Polym. Eng. Sci., 15, 684 (1975).
10. R. J. Nichols, J. C. Colba Jr., and P. K. Shete, "Experimental Study of Residence Time distribution in Counter-Rotating, Tangential Twin-Screw Extruders," AIChE Diamond Jubilee Meeting, Washington, D.C. (1983).
11. H. Potente, S. Schultheis, and K. P. Voss, Plastverarbeiter, 39, 149 (1988).
12. D. B. Todd, Polym. Eng. Sci., 15, 437 (1975).
13. G. Pinto and Z. Tadmor, Polym. Eng. Sci., 10, 279 (1970).
14. D. Wolf, N. Holin, and D. H. White, Polym. Eng. Sci., 26, 640 (1986).
15. V. S. Kao and G. R. Allison, Polym. Eng. Sci., 24, 645 (1984).
16. Z. Kemblowski and J. Sek, Polym. Eng. Sci., 21, 1194 (1981).
17. L. P. B. M. Janssen, R. W. Hollander, M. W. Spoor, and J. M. Smith, AIChE J., 25, 345 (1979).
18. D. Wolf and D. H. White, AIChE J., 22, 122 (1976).
19. H. Potente and H. Lappe, Plast. Rubber Process. Applic., 6, (1986), no. 2.
20. M. Ratzsch, Mackromol. Chem., Macromol. Syrup., 12, 165 (1987).
21. K. Motha, J. Seppala, and C. Bergstrom, Polym. Eng. Sci., 29, 1579 (1989).
22. P. Noar and R. Shinnar, I.E.C. Fundam., 2, 278 (1963).
23. D. H. Sebastian and R. Rakos, SPE ANTEC Tech. Papers, 36, 135 (1990).
24. J. A. Biesenberger and D. H. Sebastian, Principles of Polymerization Engineering, Chap. 3, "Mixing Effects," John Wiley & Sons, New York (1983).
25. A. P. Plochocki, S. S. Dagli, and M. H. Mack, Kunststoffe, 78, 254 (1988).
26. M. H. Mack, SPE ANTEC Tech. Papers, 35, 120 (1989).
27. A. D. Gotsis, Z. Ji, and D. M. Kalyon, SPE ANTEC Tech. Papers, 36, 139 (1990).
28. D. M. Kalyon and H. N. Sangani, Polym. Eng. Sci., 29, 1018 (1989).
29. D. M. Kalyon, C. Jacob, and P. Yaras, Plast. Rubber Compos. Process. Applic., 16, 193 (1991).
|Printer friendly Cite/link Email Feedback|
|Author:||Vainio, Tommi P.; Harlin, Ali; Seppala, Jukka V.|
|Publication:||Polymer Engineering and Science|
|Date:||Feb 15, 1995|
|Previous Article:||Characteristics of good stabilizers and methods of evaluation.|
|Next Article:||Comparison of microwave and thermal cure of an epoxy amine matrix.|