Continuous versus batch processing of recycled rubber and plastic blends.

Blending of recycled elastomers with plastics is a promising approach for developing new families of thermoplastic elastomers and rubber-toughened plastics (refs. 1-8). Most investigators (refs. 1-5, 7 and 8) have employed compatibilization methods and reactive processing techniques since simple blending with recycled elastomers generally retards mechanical properties (refs. 1, 2, 7-10). Problems arise, however, with the large variety of elastomeric compounds that are found in a typical recycled rubber stream. Since many of the blending approaches involve chemical reactions or are sensitive to temperature/shear history, scale-up from laboratory batch mixers to continuous processing operations has been challenging (ref. 11). Additionally, reactive processing schemes often require modification or substantial revisions when transitioning to twin-screw extrusion or when the recycled elastomer feed stock is changed.

Control of dispersed phase particle size is critical in obtaining adequate properties when blending for thermoplastic elastomers or rubber-toughened plastics (refs. 12-16). Optimum properties for dynamically vulcanized blends are generally observed with a rubber particle size in the range of one micron (refs. 15 and 16). This is extremely difficult to achieve using recycled elastomers since the cost of reducing crosslinked rubbers to this size range is prohibitive (refs. 1, 5, 17 and 18). Also, dispersion of the recycled rubber in the continuous phase is complicated by the fact that during processing the particles stretch, shear and recover their original shape. Small internal batch mixers, which can quickly grind glass marbles into finely divided powders, allow these problems to be addressed on a laboratory scale. For continuous processing on an industrial scale, co-rotating twin-screw extruders with intermeshing screw elements appear to provide the best, practical alternative (refs. 19 and 20).

The understanding and ability to model twin-screw extrusion operations has increased dramatically in recent years (refs. 19, 21-26). These investigations provide important benchmarks for efforts to scale-up multi-phase polymeric blends from laboratory batch processing. Especially noteworthy is the ability to program the screw geometry to obtain a desired residence-time distribution (refs. 21, 23 and 24), as well as the capability of predicting mixing efficacy of various element combinations (refs. 25 and 26). Both of these parameters are more important in reactive processing than traditional characterization values such as temperature, pressure and shear stress.

The purpose of this article is to report on an effort to scale-up blends of recycled rubber and plastics for use as thermoplastic elastomers or rubber-toughened plastics. Resultant blends are characterized with respect to rheological properties relevant to processing, phase morphology and mechanical properties. Development of specific formulations, reactive processing approaches and optimization of rheological and mechanical properties were previously reported (refs. 7, 8 and 27).

Experimental

Materials

Powdered EPDM elastomers (80 and 170 mesh) were provided by Rouse Rubber from recycled roofing compounds. Additionally, a comingled NR/SBR powder (80 mesh) from tire scraps was obtained from the same source. Particle size reduction was accomplished by Rouse using ambient temperature underwater processing. Reported particle sizes were converted from the mesh sizes provided by the processor. A density of 1.14 g/[cm.sup.3] for these rubbers was previously reported (ref. 8). The polypropylene used in the study came from Fina Oil & Chemical. It is identified by the manufacturer as PP-3622, with a melt flow index of 12 g/10 min., a number-average molecular weight of 29 kD, and a polydispersity index of 6.6. Characterization (ref. 8) of the polymer showed it to be 55% crystalline with a melting temperature of 173 [degrees] C. Its breaking and yield stresses were 28 and 32, respectively, with an elongational capability of 672% (ref. 8).

Chemical additives used in this study were t-butyl hydroperoxide from Aldrich Chemical. t-Butyl hydroperoxide was chosen based on its half life of initiation at processing temperature, 200 [degrees] C, and used as radical initiator during reactive EPDM/PP blending.

Blending

Batch mixing was performed in a torque rheometer. The mixer was preheated to the mixing temperature, 200 [degrees] C, until stable. This usually took approximately 30 to 45 minutes. The rotor speed was set at 30 rpm. Optimization of these parameters with respect to mechanical properties of EPDM/PP blends was accomplished using a six factor/two level design of experiment scheme (ref. 7). The rubber component was added when the PP was completely melted as indicated by a stabilized torque reading. In the case of EPDM/PP reactive blending, the rubber particles were allowed to imbibe the chemical additive at room temperature prior to addition to the mix. Blends prepared without the use of the chemical additive are referred to as simple blends throughout this article. All reported rubber contents are by weight.

Processing scale-up of the above materials was performed in a co-rotating intermeshing twin-screw extruder with a screw diameter of 32 mm and a length of 890 mm. The screws were programmed for a general shear profile. The barrel temperature profile for all blends was 174, 210, 228 and 230 [degrees] C, with a die temperature of 228 [degrees] C. A target output of 10kg/hr. was maintained using a baseline 250 rpm (starve feeding) with slight variations due to surging and feeder bridging. The screw speed was reduced to 30 rpm to match the residence time of the batch mixer, along with a corresponding reduction in hopper feed rate to maintain an equivalent channel fill fraction. Additionally, the temperature profile was also reduced to more closely approximate batch processing. The new temperature profile corresponding to the zones and die above are 170, 190, 200 and 210 [degrees] C, respectively. Reported residence times for both screw speeds represent measured time-of-first-appearance for a carbon-black tracer in the unfilled propylene. All blends were strand pelletized in line with the extruder.

Variable speed DC motors drove both primary and secondary feeders. The primary feeder was dedicated to PP and was located at the front most inlet port to the screw. The secondary feeder was used for all rubber particles and introduced this material into the melt stream through a second port located at the midpoint of the screw. A third port located approximately 200 mm on center from the die was left open for venting. Both feeders were calibrated by measuring the output versus feeder setting.

Specimen molding

Compression molding was performed using a heated press from a sandwich mold with a middle frame made from a 2 mm thick aluminum sheet. Rubber/plastic blends prepared by batch processing were placed between the platens, which were preheated to 225 [degrees] C. The blend was preheated for two minutes and then gradually compressed to a piston pressure of 30 MPa for an additional three minutes. Molded sheets were cooled to room temperature under a piston pressure of 100 MPa using a cooling press. The sheet was then removed and die-cut for the appropriate tests.

Injection molding of materials blended in the twin-screw extruder was performed on a 77 kN electric toggle injection-molding machine according to ASTM D638 to provide tensile test specimens. Specimens were molded at injection velocity of 64 mm/s. Processing conditions were determined using the short shot method.

Characterization and properties testing

Apparent viscosities [Eta] were measured with a capillary rheometer at 200 [degrees] C. A die with an L/D ratio of 10/1 was used in all cases. Each flow curve was generated from data collected at nine different shear rates ranging from 10 to 5,000 [s.sup.-1]. Data were reduced using a two-parameter power law equation:

1 [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

where k is the consistency index, n is the power law exponent, and [Gamma] and [[Gamma].sub.0] represent the shear rate and a reference shear rate of 1 [s.sup.-1], respectively.

Tensile specimens from compression molded sheets were prepared using a specially machined half scale ASTM D638 Die D, those from injection molding were tested as molded using a Die A specimen geometry. Tensile testing was performed on an Instron 6025 with attached computer operating system and a crosshead speed of 50mm/min. Strain was determined from the change in effective gauge length by optical methods.

Materials at various stages of processing were microtomed for optical image analysis. The image analysis system was comprised of a stereomicroscope, a digital microscope camera and a computer with Pentium III processor running Image J/1.18 software.

Results and discussion

Figure 1 summarizes the experimental pathways pursued for each of the processing approaches considered. The purpose of this outline is to show the sequence of processing/characterization/testing steps and illustrate differences in how the various blends were handled that might effect experimental interpretation. Beginning with the raw polymers, the internal batch mixer provided blends in nominal quantities of 60 g. This can be contrasted with the continuous, twin-screw extruder operations that produced approximately 200 kg. Rheological testing was conducted immediately after blending. Identical flow curves were generated for each blend/processing combination. The next step was to mold the tensile test specimens. A single sheet was compression molded for each batch mixer blend and specimens prepared by die cutting. Specimens obtained from injection molding of the blends were molded directly into the desired geometry. Image analysis, utilizing the smooth molded surfaces, was used to characterize the blend's morphology. Finally, tensile tests were conducted at equal crosshead speeds with the batch blends using half-scale specimens because of limited material availability.

[ILLUSTRATION OMITTED]

Rheological testing

Illustrative flow curves for the EPDM80/PP simple blends containing 20% recycled rubber are shown in figure 2. Strong pseudoplastic behavior was observed in all cases. Similar data were collected for all 18 simple blends/processing approaches considered. Data were reduced using equation 1. Resultant power law exponents n and consistency indices k are plotted as functions of rubber content in figure 3 to illustrate distinct behaviors within the EPDM80 blends (figure 3A) and their corresponding NR/SBR versions. (figure 3B).

[GRAPHS OMITTED]

Trends apparent in figure 3 appear almost identical regardless of whether recycled EPDM80 or NR/SBR is used. Processing conditions do, however, play an important role. This is especially true for the blends prepared by twin-screw extruder with a residence time of 52 s. As previously noted, this processing approach was developed by following the manufacturer's general processing guide. Figure 3 shows this approach results in a different blend as evident from lower n values and higher k values in all cases. Such behavior indicates poorer mixing under these conditions and was the reason the 120 s residence-time processing approach was developed. In addition to the longer residence time, this processing approach was also expected to generate materials with a greater resistance to flow due to the lower operating temperature. This feature could be expected to increase the shear stress experienced by the blend. Previous studies (refs. 7 and 8) on similar EPDM/PP blends found that polypropylene serves as the continuous phase over rubber concentrations ranging from 10 to 80%. This suggests that phase inversion does not play a role in interpreting the data in figure 3.

Nearly identical flow responses are observed once the twin-screw extruder operating conditions were modified to more closely approximate the time/temperature/shear stress history of the batch mixer. For these two blend/processing combinations, n decreases, while k increases as more rubber is added to the blend. The increase in k with crosslinked rubber content is consistent with the general theory for the rheology of suspensions, where resistance to flow of a composite increases exponentially with content of the dispersed phase. In this case, the carbon black filled, crosslinked elastomer acted as the dispersed solid while the PP melt served as the continuous phase. Nielsen and Landel (ref. 28) point out that hundreds of equations have been proposed to model this type of behavior dependent on the nature of the interaction between the two phases. For these blends, the viscosity rise is not as pronounced as would be expected from a rigid, solid filler, a feature that is believed attributable to the previously mentioned deformable nature of the rubber particles.

Figure 4 presents identically treated rheological data for two additional families of blend/processing approaches. The first, figure 4A, shows the effect of higher rubber concentrations using the EPDM80/PP blend. In this case, rubber fraction ranges from 40 to 50%, and can be compared with data in figure 3A. Again, the response of the blends prepared by internal batch mixer are inconsistent with those processed in the twin screw extruder with 52 s residence time. Figure 4B highlights the effect of rubber particle size on the rheological behavior of EPDM/PP blends. These materials were processed using the 170 mesh, or smaller particle size EPDM, and can also be directly compared with those previously discussed in figure 3B. Rubber particle size, at least at these concentrations and for these sizes, appears to have little or no impact on these rheological properties. A 120 s residence time twin screw extruder run was not accomplished on either of the blend families shown in figure 4.

[GRAPH OMITTED]

Morphological characterization

Image analysis was conducted to evaluate differences in the mixing morphology between blends prepared by internal batch mixing versus twin screw extrusion at equivalent (as determined from capillary rheometer data) processing conditions. Figure 5 presents photomicrographs obtained from NR/SBR simple blends processed using each approach. Visual inspection indicates that the rubber particles are better dispersed by batch mixing at the 10 and 20% concentrations. The superior degree of dispersion is less clear at 30%.

[ILLUSTRATIONS OMITTED]

More information concerning the efficacy of mixing can be obtained if the intensity of segregation is considered. The intensity of segregation is a measure of the degree-of-mixedness. It is defined as (refs. 29-30):

(2) [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

where [s.sup.2] is the variance of actual concentration from its mean value and [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] is the theoretical variance of a completely segregated mixture, or totally unmixed system. As I approaches zero, the mixture is considered homogenous, and is fully segregated near the other extreme of unity. Intensity measurements corresponding to the micrographs in figure 5 are 0.022, 0.037 and 0.056 for the batch mixed blend, and 0.077, 0.046 and 0.048 for blend prepared by twin-screw extrusion, at rubber concentrations of 10, 20 and 30%, respectively. These are consistent with the visual observations previously discussed and suggest that at low rubber concentrations, the internal batch mixer is superior. But as rubber content increases, the two mixing approaches converge.

[ILLUSTRATION OMITTED]

Tensile test results

Mechanical properties for the 18 simple blends/processing approaches are considered in figure 6. In general, tensile stress capability decreases almost linearly with increasing rubber content. These data are consistent with a volume rule of additivity whereby the higher strength polypropylene material is gradually diluted by a fractionally added lower strength recycled elastomer. Strength data range from a high near 25 MPa to a low just below 15 MPa for the NR/SBR blends, while their EPDM counterparts only reach a high of approximately 20 MPa. For the NR/SBR blends, ordering with respect to strength indicates batch mixing provides the best properties, followed closely by the 120 s residence time twin screw extrusion. The material blended by the 52 s approach displayed the weakest breaking stress capabilities. A less distinct trend is noted for the EPDM blends, where a crossover in all three curves complicates an ordering with respect to strength.

[GRAPH OMITTED]

Correspondingly, the tensile strain data generally increase with rubber fraction. This increase is relatively small and not as distinct for the NR/SBR blends, with all the observed elongational capabilities within the 20 to 50% range. Of the three processing approaches considered, twin screw extrusion with the 120 s residence time provides the highest values, while the internal batch mixer displays the lowest. More dramatic effects are noted for the EPDM blends. Again, the twin screw extruder with the 120 s residence time is superior, with values ranging from near 100 to near 300% for 10 and 30% EPDM powder, respectively. Internal batch mixing produces the least elongational capability, ranging from approximately 50 to near 150% for corresponding rubber fractions.

The mechanical properties of the two additional blend families, whose rheological properties were discussed in figure 4, are shown in figure 7. Figure 7A presents the EPDM-80/PP blends extended to rubber concentrations of 40 and 50%. These represent continuations of the trends noted in figure 6 with the twin screw extruder closely approximating the stress capabilities of the blends prepared by internal batch mixing. The primary difference is seen here again as a higher elongation at break for those blends processed via twin screw extrusion. Finally, data for the mechanical failure of rubber/ plastic blends utilizing EPDM170 are given in figure 7B. These show that the smaller particle size provides greater strain capabilities for blends processed in the internal batch mixer, but no concomitant change in stress capability. As previously noted, scale-up to twin screw extrusion at the slower rate was not accomplished on these last two families of blends.

[GRAPH OMITTED]

Conclusions

An experimental investigation has been conducted to scale-up a series of recycled rubber/polypropylene blends developed for internal batch mixing. Scale-up was accomplished using an intermeshing, co-rotating twin screw extruder. It was found that if the processing conditions are modified in an attempt to mimic parameters of the batch mixer, blends with similar rheological and mechanical properties can be produced. Image analysis revealed that even though agreement is excellent with respect to physical properties, the internal mixer still provides a superior degree of dispersion as quantified in the article by the intensity of segregation.

(This article is based on a paper given at the October, 2000 meeting of the Rubber Division)

References

(1.) M. Pittolo and R.P. Burford, Rubber Chem. Technol. 58, 97 (1985).

(2.) R.P. Burford and M. Pittolo, J. Mater. Sci. 21, 2308 (1986).

(3.) S. Al-Malaika and E.J. Amir, Polym. Deg. and Stab. 26, 31 (1989).

(4.) P. Rajalingam and W.E. Baker, Rubber Chem. Technol. 65, 908 (1992).

(5.) P. Rajalingam, J. Sharpe and W.E. Baker, Rubber Chem. Technol. 66, 664 (1993).

(6.) J. McKirahan, P. Liu and M. Brillhart, "Proceedings of the Annual Technical Conference of the Society of Plastics Engineers," 3110 (1996).

(7.) H.S. Liu, C.P. Richard, J.L. Mead and R.G. Stacer, "Proceedings of the Annual Technical Conference of the Society of Plastics Engineers," 2884 (2000).

(8.) H.S. Liu, J.L. Mead and R.G. Stacer, paper 61 presented at a meeting of the Rubber Division, American Chemical Society, April 4-6, 2000,

(9.) A. Accetta and J.M. Vergnaud, Rubber Chem. Technol. 55, 961 (1982).

(10.) M.J. Myhre and D.A. MacKillop, Rubber World, 42, May 1996.

(11.) A. Chidambaram and K. Min, "Proceedings of the Annual Technical Conference of the Society of Plastics Engineers," 2927 (1994).

(12.) A.Y. Coran and R. Patel, Rubber Chem. Technol 53, 141 (1980).

(13.) F.C. Stehiing, T. Huff, C.S. Speed and G. Wissler, J. Appl. Polym. Sci. 26, 2693 (1981).

(14.) B.Z. Jang, D.R. Uhlmann and J.B. Vander Sande, J. Appl. Polym. Sci. 30, 2485 (1985).

(15.) A.Y. Coran, in "Handbook of Elastomers - New Development and Technology," A.K. Bhowmick and H.L. Stephens, Eds., Dekker, New York, 1987.

(16.) A.Y. Coran and R.P. Patel, in "Thermoplastic Elastomers, 2nd Edition," G. Holden, N.R. Legge, R.P. Quirk and H.E. Schroeder, Eds., Hanser/Gardner, Cincinnati, 1996.

(17.) R.A. Swor, L.W. Jensen and M. Budzol, Rubber Chem. Technol. 53, 1215 (1980).

(18.) K. Oliphant and W.E. Baker, Polym. Eng. Sci., 33, 166 (1993).

(19.) C. Rauwendaal, "Polymer Extrusion, 3rd Edition," Hanser/Gardner, Cincinnati, 1994.

(20.) H.U. Stiegenthaler and T. Stropoli, "Proceedings of the Annual Technical Conference of the Society of Plastics Engineers," 69 (1996).

(21.) W. Michaeli and A. Grefenstein, Adv. Polym. Tech. 14, 263 (1995).

(22.) W. Michaeli and A. Grefenstein, Polym. Eng. Sci. 35, 1485 (1995).

(23.) G.E. Gasner, D. Bigo, C. Marks, F. Magnus and C. Kiehl, Polym. Eng. Sci. 39, 286 (1999).

(24.) R. Hettema, J. Van Tol and L.P.B. Janssen, Polym. Eng. Sci. 39, 1628 (1999).

(25.) M. Yoshinaga, S. Katsuki, M. Miyazaki, L. Liu, S.-I. Kihara and K. Funatsu, Polym. Eng. Sci. 40, 168 (2000).

(26.) H. Potente and M. Bastian, Polym. Eng. Sci. 40, 727 (2000).

(27.) H.S. Liu, J.L. Mead and R.G. Stacer, "Proceedings of the 32nd International SAMPE Conference, Boston" in press (2000).

(28.) L.E. Nielsen and R.F. Landel, "Mechanical Properties of Polymers and Composites, 2nd Edition," Dekker, New York, 1994.

(29.) S. Middleman, "Fundamentals of Polymer Processing," McGraw-Hill, New York, 1977.

(30.) D.G. Baird and D.I. Collias, "Polymer Processing Principles and Design," Wiley, New York, 1998.

H.S. Liu, J.L. Mead, R.G. Stacer, University of Massachusetts Lowell