Online light scattering measurements: a method to assess morphology development of polymer blends in a twin-screw extruder.
Interest in twin-screw extrusion of polymer blends has been growing in recent years because of the relative ease of producing materials having new properties. As a consequence, many studies concerning the morphology of immiscible polymer blends have been carried out since the 1970s (1-5). Study of morphology evolution along the twin-screw (6, 7) is generally investigated by scanning electron microscopy (SEM) or transmission electronic microscopy (TEM). These methods give a direct picture of the morphology, but they are time-consuming as they require the polymer blend to be quenched rapidly. After that, each sample has to be prepared (curing, ultramicrotome cutting, ...) before observation by microscopy. But morphology can also be studied in the reciprocal space by using the light scattering principle. This technique has been widely used to characterize the morphology of polymer blends under simple shear flow (8-10). Recent papers (11-13) have demonstrated the use of this technique at the exit of an extruder.
This paper deals with this technique and the first results obtained with an original light scattering device that allows online morphological characterization of polymer blends. The instrument has been especially designed to be connected to high-pressure zones along the twin-screw extruder. Consequently, it can be used to study the influence of material and processing parameters. Furthermore, the instrument allows one to monitor the process and its stability.
2. LIGHT SCATTERING THEORY
When the size of a particle is comparable with the wavelength of the light and when the refractive indices of particle and its surrounding are different (Fig. 1). the light is scattered. Two theories describe physically the phenomenon, the Mie and the Rayleigh Debye theories.
2.1. Mie Theory for Spherical Diffuser
The theory of Mie (14, 15) gives an exact solution of the Maxwell equations governing the problem of the light scattered by an isotropic sphere of radius a into an isotropic medium. The light, of wavelength [lambda], is assumed to be monochromatic and plane. There is no restriction concerning the respective refractive indices of sphere, [n.sub.p], and surrounding medium, [n.sub.s]. The amplitude of the scattered light [I.sub.[parrall]([theta]) for a sphere located between two parallel polarizers is given by the following equation:
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (1)
where the coefficients [[blank].sup.e][B.sub.l] and [[blank].sup.m][B.sub.l] are built with Riccati and Bessel functions and depend on both refractive indices of the diffuser and the medium, on the number 2[pi]a/[lambda], and [P.sup.(l).sub.(cos[theta])] the first order Legendre functions of degree l.
2.2. Rayleigh Debye Theory
When the polarizability does not occur, the theory of Rayleigh Debye (16) gives a simpler solution of the Maxwell equations under the two following conditions:
\[n.sub.p]/[n.sub.s] - 1\ [much less than] 1 (2)
2[pi]a/[lambda] \[n.sup.p]/[n.sub.s] - 1\ [much less than] 1 (3)
Assumption 2 means that one can neglect the anomalous diffraction due to refraction occurring when the indices are too different. Restriction 3 assumes that no phase difference inside the particle occurs. The Rayleigh Debye theory takes advantage of these assumptions to describe scattering phenomena for nonspherical particles. Consequently, the intensity of the scattered light follows the equation:
I([theta]) = [I.sub.t] [k.sup.4] [cos.sup.2][theta] [[[([n.sub.p]/[n.sub.s]).sup.2] - 1].sup.2] [(V/4[pi]r).sup.2] [R.sup.2.sub.([theta], [phi])] (4)
where [I.sub.t] is the incident intensity, k the norm of the wave vector, V the volume of the particle and [R.sub.([theta], [phi])] the Rayleigh form factor (16).
2.3. Intensity of the Light Scattered by a Distribution of Particles
In the case of a highly dilute system, the light scattering intensity of a distribution of N particles corresponds to the sum of the intensities scattered by each particle. Using this additivity rule, we have developed an inverse method to obtain to the particle size distribution from an experimental light scattering pattern. The normalized intensity profile is then given by the following relations (17):
[I.sub.total]([theta]) = [summation over ([N.sub.class]/t=1)] [a.sub.i] [I.sub.t]([theta]) (5)
[I.sub.t]([theta]) = [I.sub.t]([theta])/[I.sub.t]([[theta].sub.o]) (6)
[a.sub.i] = [N.sub.t] [I.sub.t]([[theta].sub.o]/[summation over ([N.sub.class]/j=1])] [N.sub.j] [I.sub.j] ([[theta].sub.o]) (7)
[I.sub.i]([theta]) is the intensity profile for a particle of diameter [d.sub.i]. [N.sub.i] is the number of particles of diameter [d.sub.i], [N.sub.class] the number of classes corresponding to the different diameters and [I.sub.total] ([theta]) the normalized total light intensity.
For a given set of [N.sub.class] droplet diameters (0.2 [mu]m up to 10 [mu]m) and for a given experimental normalized intensity profile [I.sub.total] ([theta]), the method consists in computing each [N.sub.i] with a least square method and a nonnegativity constraint. The inverse method has been numerically tested. First, intensity profiles were calculated for different kinds of chosen distributions. The inverse method has then been used to recalculate these distributions and to compare them with the initial ones. The results show a good correlation between the initial and the re-calculated distributions.
Furthermore, Eq 4 shows that the intensity of light scattered by a large particle is much higher than that scattered by a smaller particle. As an example, a ratio of 2 for the radius results in a ratio of 64 for the scattered intensity at [theta] = [[theta].sup.[degrees]]. However, if one considers the normalized intensity I([theta])/I([[theta].sub.o]), small particles scatter light at wider angles (see also Fig. 6).
2.4. Multiple Light Scattering
For highly concentrated systems, multiple light scattering occurs. This means that the intensity scattered by a particle reaches its neighbors. Therefore, such systems scatter the light at wide angles. In this case, the two previous theories are not applicable anymore. Rusu (17) has proposed the following criteria to evaluate when multiple light scattering can be neglected:
[c.sub.vol] h/8a [less than or equal to] 0.2 (8)
where [c.sub.vol] is the volume fraction of spherical particles and h the thickness of the observed system.
As a consequence, the use of an inverse method will give smaller droplet sizes that one really gets.
3.1. The Light Scattering Device
The light scattering device has been especially designed and built for this study. It consists of three different parts (Fig. 2).
The light source: It is constituted of three He/Ne red laser beams of which the wavelength is 633 nm. Each laser can be individually translated and rotated in and around two directions to facilitate the alignment with the two other parts. The first laser is in the center of the optical bench, whereas the second and the third lasers are respectively at 35[degrees] and 70[degrees] in order to observe the scattered light at wide angles. In this study, only the centerline laser has been used.
The light scattering cell: This part is directly connected to the extruder by means of special connectors: therefore, a small flow of molten polymer is separated from the main flow inside the extruder to the light scattering device. The latter consists of a thermo-regulated barrel in which a 10-mm-diameter glass sphere is strictly aligned with the axis of the cell. Into the glass sphere a 0.8-mm hole is drilled to let the polymer blend flow through it. This is the heart of the device since the small volume of blend inside the sphere is illuminated by the laser beam and expected to scatter the light. A set of two distributors leads to a switch between the purge and the analysis mode (Fig. 3). In the former, the dead volume beneath the sphere is flushed away and fresh blend is therefore ready for analysis. The latter permits the fresh blend to flow through the sphere and lets evacuate it from the device. Furthermore, one can observe the scattered light when the polymer blend flows or not through the hole of the glass sphere (respectively dynamic or static mode). The ratio of flow through the scattering device to the main flow in the extruder is typically around 1/100 to 1/1000.
The optical bench: An aspheric condenser lens, an achromatic doublet lens, and a diaphragm are used to get back the scattered pattern on a semi-transparent plate. The pattern is recorded with a COD camera (Sony XC ST70). The bench can be translated and rotated in and around two directions for alignment purposes. Furthermore, the achromatic doublet lens and the diaphragm can be independently centered. The CCD camera is linked to a computer for data acquisition (Matrox MeteorII) and the pictures are processed with a commercial software (Matrox Inspector).
Two kinds of polymer blends have been used:
PS/PP blend: polystyrene is a Styron 678E from Dow and polypropylene a Stamylan P 48M10 from DSM. The PP is the minor phase whereas the PS is the matrix. They were chosen for their high refractive index contrast. As a consequence, even at low concentration we had sufficient turbidity to give adequate light scattering. At room temperature, the respective refractive indices of PS and PP are [n.sub.PS] = 1.6 and [n.sub.PP] = 1.49. Successive masterblends have been prepared in order to analyze a low concentrated blend of 99.2% PS and 0.8% PP in weight. First of all, a masterblend of 80% PS and 20% PP in weight is extruded at 200[degrees]C. A second step consisted in the preparation of a masterblend of 96%/4% in weight obtained at 200[degrees]C from the 80%/20% blend and PS. The blend of 96%/4% was analyzed at the same time. In a final step, we extruded and analyzed the 99.2%/0.8% blend obtained at 200[degrees]C from the 96%/4% and PS. The dependence of the viscosity on the shear rate was determined on a single-sc rew extruder equipped with a capillary die. Figure 4 shows the viscosities and the extrapolated viscosity ratio between the PS and the PP at 200[degrees]C. One can observe that in the range of shear rates from [10.sup.2] [s.sup.-1] to [10.sup.4] [s.sup.-1], the viscosity ratio [[eta].sub.PP]/[[eta].sub.PS] varies only between 0.7 and 1.1.
EBA/PA6 and REBA/PA6 blends: PA6 is an Ultramid B5 from BASF. EBA is a copolymer of ethylene and buthyl acrylate (Lotryl 17BA07 from Elf Atochem) while REBA is the EBA for which 4.6 molecules of maleic anhydrid (MAH) have been grafted per chain (Lotader AM3410 from Elf Atochem). MAH of the REBA can react with the terminal amine function of the PA6 and give rise to a reactive polymer blend. Both blends were extruded at 250[degrees]C.
3.3. Co-rotating Twin-Screw Extruder
The co-rotating twin-screw extruder (Werner & Pfleiderer ZSK30) has a length to diameter ratio of 40. Figure 5 shows the screw profile used. The optical cell is located at the point P4b at the end of the mixing zone. For all experiments, the feed throughput was 3 kg.[h.sup.-1] and screw rotational speed was set to 150 RPM.
3.4. Procedures for Light Scattering Experiments
Image processing: The data acquisition needs a statistical treatment to eliminate the noise resulting from variations in the scattered pattern. Therefore, a stream of 400 pictures (x, y patterns, see Figs. 2 and 3 for the x and y axis definition) was recorded at a frequency of 25 pictures per second. These 400 frames were averaged to get an image from which the variation of intensity as a function of the angle was extracted and compared. A statistical study has shown that one has a Gaussian distribution of the scattered intensity for each pixel of the CCD camera.
Run of on experiment: First, a diffraction grating film (140 grooves/mm) was used to correlate scattering angle with the coordinates of each pixel of the COD camera. Second, the polymer blend was extruded for at least half an hour to reach a steady state after some process parameters were changed. Then, the dead volume beneath the optical glass sphere is flushed away and the fresh blend was allowed to flow through the hole of the glass sphere for measurements. Then, the 400 patterns were recorded and processed in the dynamic or static mode.
A scanning electron microscope (JEOL JSM 5900 LV) operated at 20 kV was used to take micrographs of the blends. The samples were taken at the exit of the light scattering cell and rapidly quenched in water. After that, PS was selectively etched out using THF for PS/PP blends and PA6 was selectively etched out using formic acid for EBA/PA6 and REBA/PA6 blends.
4. RESULTS AND DISCUSSION
4.1. Characterization of the Morphology for Diluted Blend
Figure 6 shows experimental and theoretical light scattering profiles in the y direction for the 99.2%/ 0.8% PS/PP blend (see Figs. 2 and 3 for the y axis definition). The experimental y profile was obtained in the static mode. Several measurements, carried out for 2 hours, showed a very stable profile of the scattered light. Theoretical curves were obtained with the Mie model. No mono-disperse distribution was found to fit the experimental profile, whereas a bi-disperse distribution of particle size allows a fair fitting of the experimental curve. The distribution highlights a large number of small particles (90% of the total particle volume, which corresponds to 230 particles of 1.5-[mu]m diameter for only 1 particle of 4.5-[mu]m diameter). The respective intensity profiles for the 1.5-[mu]m and the 4.5-[mu]m diameter are also plotted to demonstrate the importance of each class of diameter.
Figure 7 shows a representative SEM micrograph for 99.2%/0.8% PS/PP blend. One observes that PP is dispersed in PS matrix. The micrograph shows that the droplet diameters are in the range from 0.2 [mu]m to 4 [mu]m, with a mean value around 0.8 [mu]m. This result is in agreement with the size distribution obtained from light scattering if one takes into account the precision of the experimental measurements.
4.2. Study Under Shear Flow
Figure 8 shows the x, y patterns (averages of 400 recorded frames) respectively in the static (Fig. 8a) and in the dynamic modes (Fig. 8b) for the 96%/4% PS/PP blend. One observes that the scattered light pattern for the dynamic mode is smoother than the one obtained for the static mode. The flow through the glass sphere acts as a filter for the signal by averaging the intensity during time. For the static mode, Fig. 8a shows an isotropic pattern, while Fig. 8b shows an elliptical pattern corresponding to an ellipsoidal deformation of the droplets inside the glass sphere.
Figure 9 shows the x and y intensity profiles obtained for the dynamic mode (flow through the glass sphere) and for the 96%/4% PS/PP blend. It should be noticed that the x profiles are available only up to an angle of 150. The anisotropy of the light scattering pattern shows that the particles are stretched under the flow during the dynamic mode. In this case, a throughput of 0.3 g.[min.sup.-1] was measured, which leads to a wall shear rate around 140 [s.sup.-1] (radius of the hole 0.4 mm and local power law index for the melt viscosity n = 0.35 (see Fig. 4). If one considers that shear flow leads to an ellipsoidal shape, Glatter and Hofer (18) have shown that the mean length of ellipsoidal particles can be estimated from the light scattering profile observed in the axial direction x of the flow. We have therefore fitted the experimental x scattering profile by a theoretical curve obtained with the inverse method using the Mie model. Calculations show a wide distribution of particle size, from 1.5 [mu]m to 5.5 [mu]m (see Fig. 9). It should be noticed that the size of 5.5 [mu]m, which is underestimated because of multiple light scattering, is nevertheless higher than the one obtained for the 99.2%/0.8% blend, which also highlights the particle stretching.
The flow through the capillary of the glass sphere has been modeled in order to estimate whether droplet breakup is possible in this flow section. A Carreau law has been chosen to characterize the rheological behavior of 96%/4% PS/PP blend. Velocity and shear rate fields have been calculated for a throughput of 0.3 g.[min.sup.-1], a capillary radius of 0.4 mm and a length of 5 mm. Figure 10 shows the local residence time and the critical time [t.sub.b] versus radial position in the capillary. Time [t.sub.b] corresponds to the time that a droplet needs to break up under flow corresponding to the critical capillary number [Ca.sub.crit] given by Grace (19) and MeiJer (20). Figure 10 shows that only droplets located between r = 0.27 mm and the capillary wall (r = 0.4 mm) are able to break up. Figure 11 shows minimum and maximum radii of the droplets that are broken for r = 0.27 mm up to 0.4 mm. The minimum radius is calculated from the critical capillary number given by Grace (19) and Meijer (20). The maximum rad ius is calculated for a droplet that breaks up after a time equal to the local residence time. Figure 11 shows that breakup does not occur at the center of the capillary (from r = 0 up to 0.27 mm), while it occurs for particle diameters from 0.5 [mu] up to 3 [mu]m (see the breakup zone in Fig. 11. This phenomenon should lead to a bidisperse distribution of the diameters near the wall of the capillary.
4.3. Nonreactive and Reactive Blends
Figure 12 shows the intensity profiles obtained for reactive blend (REBA/PA6 80%/20%) and its corresponding nonreactive blend (EBA/PA6 80%/20%). Such concentrated blends generate of course multiple light scattering. Nevertheless, one can have a correlation between the scattered light and the morphology; see Cielo (21) and Belanger (22). For EBA/PA6 blend, all measurements, which were carried out for more than 2 hours, have shown a high stability of the light scattering intensity profile. These results are in agreement with a previous work of Serra (23). As a matter of fact, Serra (23) has shown for a similar blend (EBA/Platamid 80%/20%) that there is no effect of processing parameters such as feed throughput and screw rotational speed. Nevertheless, one has observed high random fluctuations of the light scattering profiles in the reactive case. Figure 12 shows the two curves Min and Max corresponding to the envelope of all profiles obtained during 2 hours of measurements without variation of processing parame ters. These random fluctuations highlight, surprisingly, a non-constant morphology of the reactive REBA/PA6 blend. However, the level of the normalized intensity profile (Fig. 12) was always higher for the reactive blend than for the nonreactive one, which means that PA6 droplets are smaller for the reactive blend. This result was quite expected since the reaction at the interface of the two polymers decreases the surface tension, leading to smaller droplets.
Figure 13 shows the SEM pictures respectively for EBA/PA6 (Fig. 13a) and for REBA/PA6 (Fig. 13b) taken at the same scale, where the dark areas represent the PA6 phase. For each case, one observes that PA6 is dispersed in EBA or REBA matrix. The micrographs confirm, without any doubt, that the size of the droplets is much lower for the reactive blend.
We have developed a new apparatus for the purpose of following the morphology of polymer blends along an extruder by online light scattering measurements. The first tests have shown the capabilities of this device to assess the development of polymer blend morphology during their extrusion.
These preliminary results suggest:
* For dilute blends, deconvolution of the intensity profile gives access to the droplet size distribution and makes it possible to study the influence of processing parameters such as feed throughput, RPM, screw profile, temperature, ...
* The anisotropy of the light scattering pattern induced during the shear flow through the optical cell can be used to analyze the relaxation of the stretched droplets when the shear flow is suddenly stopped.
* The important difference in terms of light scattering intensity profile between reactive and nonreactive blends will provide deeper insight into how the morphology is affected by the functionality (number of reactive functions per chain) and the processing parameters.
Moreover, the use of the second and third laser will improve the sensitivity to polymer blends having very small particles by analysis at wide angles. Finally, this device could be probably used as a control process apparatus to follow the stability of the process and the quality of the product.
[FIGURE 1 OMITTED]
[FIGURE 2 OMITTED]
[FIGURE 4 OMITTED]
[FIGURE 6 OMITTED]
[FIGURE 9 OMITTED]
[FIGURE 10 OMITTED]
[FIGURE 11 OMITTED]
[FIGURE 12 OMITTED]
The authors wish to thank Christophe Melart and Patrice Simon for their help in running the experiments and designing the apparatus.
(1.) C. D. Han and T. C. Yu, Polym. Eng. Sci., 12, 81(1972).
(2.) B. D. Favis and D. Therrien, Polymer. 32, 1474 (1991).
(3.) N. Chapleau and B. D. Favis, J. Mat. Sci., 30, 142 (1995).
(4.) D. Bourry and B. D. Favis, Polymer, 39, 1851 (1998).
(5.) J. K. Lee and C. D. Han. Polymer, 40, 6277 (1999).
(6.) L. Delamare, Ph.D. Thesis, Ecole des Mines de Paris (1995).
(7.) J. K. Lee and C. D. Han, Polymer. 41, 1799 (2000).
(8.) S. Kim, J. W. Yu, and C. C. Han, Rev. Sci. Inst., 67, 3940 (1996).
(9.) N. G. Remediakis R. A. Weiss, and M. T. Shaw, Rubber Chemistry and Technology, 70, 71 (1997).
(10.) S. Kim, E. K. Hobbie. J. W. Yu, and C. C. Han, Macromolecules, 30, 8245 (1997).
(11.) E. K. Hobbie, K. B. Migler, C. C. Han, and E. J. Amis, Ad. In Polym. Tech., 17, 307 (1998).
(12.) S. Li, K. B. Migler, E. K. Hobbie, H. Kramer, C. C. Han, and E. J. Amis, J. Polym. Sci.: Part B: Polym. Phys., 35, 2935 (1997).
(13.) K. B. Migler, E. K. Hobble, and F. Qiao, Polym. Eng. Sci., 39, 2282 (1999).
(14.) H. C. van de Hulst, Light scattering by Small Particles, Wiley, New York (1964).
(15.) M. Born and E. Wolf, Principles of Optics, Pergamon Press, London (1975).
(16.) G. G. Fuller, Optical Rheometry of Complex Fluids, Oxford University Press, New York (1995).
(17.) Rusu, Ph.D. Thesis, Ecole des Mines de Pails (1997).
(18.) O. Glatter and M. Hofer, J. Colloid Interface Sci., 122, 484 (1988).
(19.) H. P. Grace, Chem. Eng. Comm., 14, 225 (1982).
(20.) H. E. H. Meijer and J. M. H. Janssen, in I. Manas-Zioczower and Z. Tadmor. eds., Mixing and Compounding--Theory and Practical Progress. Vol. 4 in Progress in Polymer Process Series, Hanser Verlag. Munich (1993).
(21.) P. Cielo, B. D. Favis, and X. Maldague. Polym. Eng. Sci, 27, 1601 (1987).
(22.) C. Belanger, P. Cielo, B. D. Favis. and W. I. Patterson, Polym. Eng. Sci. 30, 1094 (1990).
(23.) C. Serra, G. Schlatter, M. Bouquey. R. Muller, and J. Terrisse, the 17th Annual Polymer Processing Society Meeting, Montreal (May 2001).
|Printer friendly Cite/link Email Feedback|
|Author:||Schlatter, G.; Serra, C.; Bouquey, M.; Muller, R.; Terrisse, J.|
|Publication:||Polymer Engineering and Science|
|Date:||Oct 1, 2002|
|Previous Article:||Morphology and rheological properties of polypropylene/reactive elastomer blends.|
|Next Article:||Effect of dynamic vulcanization on the microstructure and performance of polyethylene terephthalate/elastomer blends.|