Recycled Poly(ethylene terephthalate) Chain Extension by a Reactive Extrusion Process

FIRAS AWAJA and FUGEN DAVER* RMIT University School of Aerospace, Mechanical and Manufacturing Engineering PO Box 71, Bundoora, Victoria 3083, Australia EDWARD KOSIOR Visy R&D 13 Reo Crescent Campbellfield, Victoria 3061, Australia
A commercial-scale reactive extrusion processing system for recycled poly(ethylene terephthalate) (PET) flakes with an added chain extender, pyromellitic dianhydride (PMDA), was investigated. The PMDA concentration was varied with the intention of reaching a higher recycled PET intrinsic viscosity ([ ]). The effect of changing the extruder residence time on the systems stability and the recycled PET [ ] was also investigated. Reactive extruded PET with a PMDA concentration up to 0.3 wt% was found to have a higher [ ] and lower carboxyl content than recycled PET processed in a normal extrusion system. A shift in [ ] of about 0.18 dl/g was obtained with a 0.3 wt% PMDA concentration. A PMDA concentration above 0.3 wt% produced chemical, thermal and hydrodynamic instability in the system, causing crosslinking reactions and gel formation. The reactive extrusion system was stable at low residence time (45 s) and moderate (0.15 wt%) PMDA concentration; however, using 0.2 wt% PMDA produced higher reactive extruded recycled PET [ ] with lower carboxyl content than other PMDA concentration levels examined. Residence times higher than 45 s produced higher reactive extruded recycled PET [ ]. Reactive extruded recycled PET was also tested for mechanical properties. Polym. Eng. Sci. 44:15791587, 2004.
2004 Society of Plastics Engineers.

INTRODUCTION

ost-consumer PET undergoes a reduction in intrinsic viscosity, [ ], when recycled in a normal extrusion system. The occurrence of thermal and hydrolytic degradation reactions during recycled PET melt processing is responsible for the reduction in [ ] or molar mass of the PET. The presence of water and polyvinyl chloride (PVC) in the recycled PET flakes produces PET chain scission during normal extrusion. At processing temperature (280C), hydrolysis reactions occur between water and PET, resulting in shorter chains with carboxyl and hydroxyl end groups. The thermal cleavage

* To whom correspondence should be addressed. E-mail: fugen.daver@rmit.edu.au 2004 Society of Plastics Engineers Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/pen.20155

of the PET ester bond results in PET chains with carboxyl and vinyl ester end groups. Recently, intensive drying to remove moisture, and vacuum degassing processing, were introduced by Erema in their plasticrecycling systems to minimize the effect of these reactions, resulting in higher PET [ ] in comparison with normal extruded PET (1). Solid state processing has also been reported to achieve higher PET [ ] (2, 3). However, this process is considered to be slow and expensive. In this study, chain extension by reactive extrusion was chosen to overcome the reduction of [ ] for several reasons. It is less expensive than solid state processing and easier to apply in an existing normal extrusion system, and because of the proven success of the chain extension process with virgin and recycled PET (3). In order to raise the [ ] of the PET to a higher desired level, reactive blending of virgin PET with chain extenders was investigated by Inata and Matsumura (4 9).
1579

POLYMER ENGINEERING AND SCIENCE, AUGUST 2004, Vol. 44, No. 8

Firas Awaja, Fugen Daver, and Edward Kosior Their efforts concentrated on evaluating potential chain extenders. They showed that additive type di- or polyfunctional chain extenders were the preferred chain extenders because of their high reaction rate without generating by-products. PMDA has previously been reported as an efficient chain extender or branching agent (10, 11). It is thermally stable, produces no side products on reaction with PET, and is tetra-functional, commercially available, and economical. Khemani (10) showed that when 0.2 to 0.3 wt% PMDA was used with virgin PET, a significant increase in melt strength was obtained. Because of these results, PMDA was the selected chain extender in this study; its concentration was chosen to be one of the process variables. The influence of variables such as the chain extender concentration and the reaction time on the virgin and recycled PET chain extension process has been investigated (4 17). Chain extender concentration has been the most important parameter examined in reactive blending process research in recent years (14, 15, 18, 19). The theoretical amount of chain extender required was calculated according to the chain extension reaction stoichiometry. Chain extender in amounts higher than theoretically needed leads to crosslinking reactions, resulting in gel formation. Akkapeddi (20) investigated a reactive extrusion system of virgin PET with added chain extenders. He obtained lower carboxyl content using different types of chain extenders. Different models have been introduced to describe the reactive extrusion system. Janssen (21) introduced an interaction chart that presented the parameters affecting reactive extrusion. These parameters and conditions influencing the reactive extrusion process and its stability have been described by many researchers (21 24). Vergnes et al. (25) developed a model and software to characterize the flow in a co-rotating twinscrew extruder. Thermal, hydrodynamic and chemical reactions were found to be the main cause of instability in the reactive extrusion process. Controlling the causes of these instabilities and then minimizing them is the key to successful reactive extrusion. It is well known that in general, an increase in reaction conversion can be expected as a result of an increase in residence time (22). In this study, the residence time was considered as the process variable, and its effect on the process stability and product properties was examined. So far, there have been no reported studies describing a recycled PET reactive extrusion system implemented on a commercial-size extruder. This study investigates the reactive extrusion of recycled PET with a PMDA chain extender, in order to produce reactive extruded recycled PET material with high intrinsic viscosity. EXPERIMENTAL PROCEDURE Reactive extrusion experiments were performed using a commercial PET co-rotating twin-screw extruder at Visy Plastics, Melbourne facilities. Table 1 shows the normal extrusion operating conditions. PMDA was obtained from Nippon Shokubai Co. Ltd. with a purity of 99.7%, and recycled PET having an [ ] of 0.75 dl/g was obtained from Visy Plastics. Flakes of recycled PET were subjected to intensive drying at 170C before being fed to the extruder. The flakes were dried by blowing desiccated air (dew point 20C) at 170C through them for 4 hours. The average moisture content after drying was about 500 ppm. The reason for the relatively high moisture level of the recycled flakes is the limited drying time. Less moisture content could be achieved using a drying time longer than 4 hours; however, the commercial PET recycling industry limits the drying time of the flakes for economic reasons. The relatively high moisture content in the flakes is compensated by the efficient vacuum system attached to the extruder. The PET (flakes) used in the experiments had an average content of 400 ppm of PVC. In order to determine the amount of chain extender to be used in the reactive extrusion process, the theoretical amount of chain extender required was calculated using the chain extension stoichiometry equation. Equation 1 shows the theoretical amount of chain extender to be added in the recycled PET (5). W 11>Nm 2 * G *CVh *10
4

(1)

where W is the weight percentage of chain extender, N m is number of moles of recycled PET reacted, and G is the molecular weight of chain extender, equal to 218 g/mole in this study. CVh is the hydroxyl content of recycled PET, which is calculated from the Berkowitz equation, Mn 2,000,000/(CV CVh ) where CV is the carboxyl content; in this study it was measured to be equal to 52 eq./ton, Mn for Post Consumer PET flakes was calculated as 33,630 g/mole according to the Mark-Houwink equation. The possible combinations of PMDA- PET reactions are shown in Table 2. Reactive Extrusion Process A commercial-size twin-screw co-rotating extruder was used for the reactive extrusion process. The extruder was attached to a pelletizer and a crystallizer,

Table 1. Normal Extrusion Conditions. Flow rate Flakes moisture Melt temperature Backpressure Temperatures 500 kg/h 500 ppm 290C 1.7 MPa Adapter 2 Adapter 1 Zone 1 to 8 Berringer 0.1 MPa 90 rpm 210 rpm Vacuum 1 Vacuum 2

279C 278C 279C 279C

Melt pressure Feeder screw rotation speed Extruder screw rotation speed Pressure

10.3 kPa 9.6 kPa

1580

POLYMER ENGINEERING AND SCIENCE, AUGUST 2004, Vol. 44, No. 8

Recycled PET Chain Extension

Table 2. The Amount of Chain Extender Needed for 1 Mole of PMDA To React With 2, 3 or 4 Moles of PET. PMDA (Moles) 1 1 1 PET (Moles) 2 3 4 Chain Extender wt% 0.31 0.20 0.15

and was operated under a vacuum so that all light components resulting from the recycled PET degradation reactions were removed continuously. Experiments were conducted at 0.05, 0.08, 0.1, 0.15, 0.2, 0.25 and 0.3 wt% PMDA concentrations at a constant residence time of 112 seconds, corresponding to the normal production speed of 500 kg/h. Another set of experiments were conducted at a constant PMDA concentration of 0.15 wt% and at residence times of 112, 90, 64, 56, 50 and 45 seconds corresponding to a extruder production speed of 200, 250, 350, 400, 450 and 500 kg/h respectively. During the experiments, the extruder screw rotation speed was maintained at a constant speed for each test run and PMDA was fed to the extruder as a powder via a volumetric feeder. Characterization of Reactive Extruded Samples Reactive extruded samples were characterized and compared with samples taken after a normal extrusion process. Several tests were conducted as described below. The change in molar mass of the reactive extruded samples, which could be indicated by the change in [ ], is one of the important outcomes of reactive extrusion. [ ] tests have been performed with a solution viscometer (PAAR AMV200). A mixture of 60/40 (w/w) of phenol-tetrachloroethane at 120C was used to dissolve the PET. Solutions of PET were tested at a PET concentration of 0.5 wt% at 25C. All reactive extruded recycled PET samples were tested for [ ]. Virgin and recycled PET were also tested for comparative purposes. The degree of crosslinking (the percentage of insoluble gel) test was performed in order to obtain an understanding of PMDA-PET reaction mechanism. All reactive extrusion experimental samples were tested; recycled and virgin PET were tested also for comparison. Approximately 250 mg of PET pellets were dissolved in a mixture of phenol and tetrachloroethane (60/40 w/w) at 120C for one hour. The solution was then filtered and the insoluble gel was separated and washed with acetone. The gel material was then dried under vacuum until a constant weight was reached (15). Carboxyl end-group analysis is important to illustrate the conversion rate of the chain extension reaction in the reactive extrusion process. Determination of carboxyl end groups in PET was done according to the Pohl method (26). The PET samples were finely ground in a mill to about 20 mesh screen. A small amount 0.1 to 0.2 g of
POLYMER ENGINEERING AND SCIENCE, AUGUST 2004, Vol. 44, No. 8

ground PET was transferred to a small analyzing flask, to which 5 ml of benzyl alcohol was added using a 5 ml pipette. The flask was then transferred to a hot oil bath (210 1C) for 1.8 minutes and stirred using a magnetic stirrer. When the flask was in the bath for the required length of time, it was removed and dipped immediately into a beaker of cool water (20C) to quench the sample for 6 to 7 seconds. After quenching, the sample was immediately poured into a 50 ml beaker containing 10 ml of chloroform. The chloroform acts as a dispersant and prevents thick gel formation. A rinse of 5 ml of benzyl alcohol was then added to wash the walls of the flask. The flask containing the rinse was put into a 210C bath for 80 5 seconds, and then its contents were carefully added back to the rest of the sample. Two drops of an indicator solution (phenolic red) were added to the mixture and the mixture was titrated with 0.1 N sodium hydroxide dissolved in benzyl alcohol. A 10 l injection needle capable of 2 l drops was used for the titration. The titration was carried to the first recognized overall pink color that remained for 10 seconds. The results were expressed by eq./106g. Calibration was made for PET degradation during solution. The reactive extruded samples along with virgin and recycled PET were injection molded to produce specimens for impact and tensile strength tests. The reactive extruded, recycled, and virgin PET samples, in the form of pellets, were dried for 3 hours at 170C and 0 mbar in a vacuum oven before injection molding. The injection molding machine was a Battenfeld BA 350/75 PLUS, and the operating conditions are shown in Table 3. PET specimens produced after injection molding were conditioned for at least a week at 20C and were then subjected to an impact strength test performed according to ASTM D256. The machine used was a Ceast (RESIL); the test parameters and specimen dimensions are shown in Table 4. The tensile properties test was conducted according to Australian Standard AS 1145. All samples were conditioned for at least a week before testing. The instrument used was a Zwick 2010. A test speed was 50 mm/ min. A type 1 test specimen was used with an overall length of 150 mm, width at end of 20 mm, length of narrow parallel portion of 60 mm, width of narrow parallel portion of 10 mm, radius of 60 mm and initial distance between grips of 115 mm.
Table 3. Injection Molding Operating Conditions. Operating Temperatures Mold temperature Nozzle temperature Barrel temperature (zone 1) Barrel temperature (zone 2) Barrel temperature (zone 3) Injection Section Injection time Shot size 7.22 s 53.5 mm 40C 300C 295C 295C 295C

1581

Firas Awaja, Fugen Daver, and Edward Kosior

Table 4. Impact Strength Test Operating Conditions and Specimen Dimensions. Specimen type 3 Notch type Dimensions A Radius Angle Hammer Impact velocity Dissipation energy 63.5 12.7 3.17 mm B 10.16 0.05 mm 1.00 0. 05 mm 45 1 2.8 J 3.46 m/s 0.005 J

RESULTS AND DISCUSSION Reactive Extruded RecycledPET Processing Reactive extrusion is a process with the potential for chemical, thermal and hydrodynamic instabilities to occur. The system has many parameters, all of which can affect the reactive extrusion process. Performing a stable operation was required in order to achieve a successful reactive extrusion. To accomplish this, chemical, thermal and hydrodynamic reactions need to be controlled. During tests, the thermal and hydrodynamic instabilities were controlled by keeping the barrel temperature between 282C and 285C and the screw rotation speeds at 75, 92, 154, 174, 188 and 216 rpm, corresponding to residence times of 112, 90, 64, 56, 50 and 45 s respectively. At a PMDA concentration between 0.25 and 0.3 wt%, an increase in the extruder die pressure was observed. A hazy thicker material coming from the edge of the extruder die was indicative of chemical instabilities. The chain extension reaction conversion increased at stagnant regions because of extended residence times. It was noticed that the process was stable when the screw rotation speed was increased by 10 rpm even though the die pressure was still high. The higher die pressure did not appear to further disturb any other parameters. At 0.35 wt% PMDA concentration, chemical and hydrodynamic instabilities were detected; the extruder struggled to overcome the increase in melt viscosity as indicated by the increase in die pressure. Thick, crosslinked PET was obtained from the extruder. The viscosity was dramatically increased in the fully filled extruder section that intensifies crosslinking reactions. As the viscosity increased, it led to an increase in the die resistance, which in turn increased the residence time of the reactive extruded PET, leading to process instabilities. During these instabilities, the extruder suffered few blockages in the flow. The screw rotation speed was increased by 40 rpm in an attempt to overcome the blockage. In a few minutes after feeding PMDA to the extrusion system, it was not possible to overcome the blockage and the material obtained remained gel-like. The above experience shows that the extruder die resistance indicated by die pressure is another important parameter in the reactive extrusion process, as was also shown in a previous study (23). According to Janssens interaction diagram for reactive extrusion (21), the die resistance is affected by flow rate and viscosity. It was expected that when the feeding of PMDA commenced, a
1582 Fig. 1. The extruder die pressure response readings to the chain extension process at different PMDA concentrations.

sudden increase in melt viscosity would take place as a result of the fast reaction between PMDA and PET. Preliminary trials showed that stable die pressure in the first few minutes of the reactive extrusion process was a very good indication of a stable operation, as the majority of the die pressure fluctuations happen when the feeding of the chain extender is initiated. Steady die pressure, in other words, a uniform distribution of the melt viscosity in the extruder, is preferred. Figures 1 and 2 show the extruder die pressure readings during the first few minutes of selected tests with respect to variable PMDA concentrations and residence times, respectively. The extruder die pressure initially increases with the start of PMDA feeding. This increase, which corresponds with the commencement of chain extension reaction, immediately introduces higher-viscosity material, resulting in an instant die resistance. The die pressure increased by a value of 21 bar when the 0.3 wt% PMDA was added to the extrusion system, as shown in Fig. 1. Figure 2 illustrates

Fig. 2. The extruder die pressure response readings to the chain extension process at a constant PMDA concentration of 0.15 wt% for 112, 64 and 45 s residence times and at a PMDA concentration of 0 and 0.15 wt% for 45 s residence time. POLYMER ENGINEERING AND SCIENCE, AUGUST 2004, Vol. 44, No. 8

Recycled PET Chain Extension the effect of 0.15 wt% PMDA on die pressure at a residence time of 45 s, in comparison to no PMDA addition. It also compares the effect of residence time on the die pressure at a constant PMDA concentration of 0.15 wt%. At constant PMDA concentration of 0.15 wt%, the die pressure increased by 7 bar when the residence time decreased from 112 to 45 s, owing to increased extruder production speed. Another important aspect of the die pressure is the process safety limitation. It is considered hazardous to have die pressure readings above 40 bar. Many operating problems occurred when the die pressure was at this level and above. At that level of die resistance, it is believed that the reacted recycled PET became so viscous near the die that it was impossible for the incoming PET to force that material out of the extruder. As a result, the material is forced out through the vacuum lines, causing a disturbance in the extruder as well as being deleterious to the vacuum system. It is seen from Fig. 1 that operating with a PMDA concentration higher than 0.3 wt% would be likely to produce a die pressure that is above the safety limit. With 0.15 wt% PMDA concentration, there were no noticeable pressure fluctuations when operating at different residence times. As shown in Fig. 2, the reactive extruder was always kept within the safety limits for all the residence times studied. The Reaction Mechanism Between PMDA and Recycled PET The reaction between PMDA and recycled PET can be categorized into three possible types: (a) blocking reaction, which means a PMDA molecule joins one PET molecule, (b) coupling reaction, which comes after the blocking reaction when a PMDA molecule joins two molecules of PET linearly, and (c) branching reactions, in which PMDA molecules join more than 2 PET molecules and/or more than one PMDA molecule joins more than two PET molecules, which leads to crosslinking and later to gel formation. Blocking and coupling reaction rates are important determinants of increasing molar mass or [ ] of the recycled PET inside the extruder. The results of the crosslinking test showed that the amount of gel formation was less than 4%. Branching reactions at such low PMDA concentrations did not reach the point of extensive cross-linking. The expected chain extension reaction mechanism, where the PET hydroxyl end group attacks the anhydride functional group of PMDA, will lead to coupling between the PET and the PMDA, thus forming two carboxyl groups. The resulting carboxyl groups will later react with PET and introduce more coupling and/or branching reactions and produce a molecule of water in each stage of the reaction, as presented in Fig. 3. PMDA is a tetra-functional material; one molecule of PMDA can join four molecules of PET. All blocking reactions (joining one molecule of PMDA with one molecule of PET) produce one free carboxyl group, Fig. 3a. In the case of two PET molecules reacting with PMDA, as shown in Figs. 3b and 8c, there is a possibility of proPOLYMER ENGINEERING AND SCIENCE, AUGUST 2004, Vol. 44, No. 8

Fig. 3. PMDA reactions with PET.

ducing two carboxyl end groups with each coupling, as in Fig. 3b, or a molecule of water, as in Fig. 3c. Branching reactions take place when three PET molecules are connected to PMDA; one carboxyl group and one water molecule are generated as well, as revealed in Fig. 3d. The typical reaction would be a complete connection of the four PMDA functional groups with four PET molecules, generating one water molecule, as shown in Fig. 3e. Other cases include heavy branching and crosslinking reactions, where two or more PMDA molecules linked together to form a network, shown in Fig. 4. These carboxyl groups resulting from the incomplete extension of the four functional groups of the PMDA molecules will lead to an increase of the carboxyl end groups in the mixture. In addition, the water molecules generated will initiate further hydrolytic reactions, generating even more free carboxyl ends. The PMDA concentration and the reaction time are the main factors that determine the blocking, coupling and branching reactions, and thus the carboxyl content. Impurities in the recycled PET, such as PVC and moisture, will catalyze the degradation reaction, and increase its rate (3, 13), thus increasing carboxyl content. The carboxyl content of recycled PET has a significant negative effect on the materials physical properties and hydrolytic stability. The increase of carboxyl groups, which are generated during extrusion, is very significant. Starting from 8.5 eq./ton carboxyl content of
1583

Firas Awaja, Fugen Daver, and Edward Kosior

Fig. 4. PMDA extensive branching reactions with PET.

feeding flakes, 52 eq./ton resulted in the produced material after a normal extrusion operation. The introduction of PMDA chain extender reduces both hydroxyl groups (by reacting directly) and carboxyl content by limiting the degradation reactions effect. Figure 5 shows the hydroxyl content in response to varying the PMDA concentration. The hydroxyl content of reactive extruded recycled PET is calculated from the Berkowitz equation, where M n refers to an apparent number average molecular weight since the Mark-Houwink equation is applicable for linear polymers. Branched PET shows considerable deviation from a linear relationship between [ ] and M n based on the Mark-Houwink equation (7, 27, 28). The calculation of the hydroxyl group content in PET was based on two assumptions. The first assumption is that PET chains with hydroxyl group ends are not produced by the chain extension reaction; the other assumption is that PMDA reacts only with the PET chains hydroxyl groups. Hydroxyl content decreases rapidly with increasing PMDA concentration, and this is indicative of the reaction progress of PMDA with PC-PET. Figure 6 shows the calculated hydroxyl content vs. residence time. The hydroxyl content decreases with increasing residence time at constant

Fig. 6. Carboxyl content vs. residence time at PMDA concentration of 0.15 wt%.

PMDA concentration of 0.15 wt%. A higher residence time for the PMDA/PET reaction leads to a higher number of hydroxyl groups coupled with PMDA. The chain extension process depends on PMDA concentration more than the residence time. The reduction in hydroxyl content at different PMDA concentrations and constant residence time is double the reduction in hydroxyl content at different extruder residence times and constant PMDA concentration. The [ ] versus the PMDA concentration relationship can be divided into three stages, as displayed in Fig. 7. The first stage lies between 0 and 0.1 wt% PMDA concentration, where an increase of PMDA concentration leads to a steep increase of [ ]. At PMDA concentration lower than the concentration needed theoretically for complete PET-PMDA reaction, 0.15 wt% in the case of 4 moles of PET reacting with 1 mole of PMDA, PET short chains with functional end groups are more available

Fig. 5. Carboxyl content vs. PMDA concentration at a residence time of 45 s. 1584

Fig. 7. The effect of PMDA concentration on the [ ] of the reactive extruded recycled PET at a residence time of 45 s. POLYMER ENGINEERING AND SCIENCE, AUGUST 2004, Vol. 44, No. 8

Recycled PET Chain Extension for each individual PMDA molecule to react with. PMDA molecules are more likely to react tri-functionally and/ or tetra-functionally with PET short chains. Coupling and branching reactions are active; however, it is believed that the branching reactions dominate over coupling reactions. Branching reactions will develop higher molecular weight, increasing the [ ] of PET rapidly. At this stage, the chain extension reactions minimize the effect of degradation reactions but do not overcome it completely. The second stage lies between 0.1 and 0.2 wt% PMDA concentration. [ ] is still increasing with increasing PMDA concentration but at a steady rate. At this stage, where PMDA concentration is at a level closer to the theoretical amount needed for complete PET-PMDA reaction, both branching and coupling are active and in equilibrium with each other. Compared with the previous stage, a larger number of PMDA molecules react with degraded PET chains, and branching and coupling reactions increase PET viscosity at a steady rate. The last stage lies between 0.2 and 0.3 wt% PMDA concentration. PMDA-PET reactions start to form crosslinking, and the molecular weight starts to rise exponentially. Branching and coupling reactions coexist in this stage; however, the branching of PET increases with increasing PMDA concentration. The chain extension reactions overcome the degradation reaction, as indicated by the higher [ ]s obtained than that of the original feedstock. Reactive extruded PET with [ ] above 0.73 dl/g was obtained using a PMDA concentration at 0.2 wt% and residence times above 45 s during reactive extrusion. In normal extrusion, using the same PET flakes ([ ] of 0.75 dl/g) produced a PET material with [ ] of only 0.67 dl/g. Figure 8 shows that at different residence times with a constant PMDA concentration of 0.15 wt%, [ ] increases linearly with increasing residence time. The longer residence time leads to more reaction conversion between PMDA and PET resulting in an increase in viscosity. By implementing a reactive extrusion process consisting of PET flakes (having a [ ] of 0.75 dl/g) and PMDA, a shift between 0.01 and 0.18 dl/g (depending on the PMDA level) in reactive extruded recycled PET [ ] was obtained. As expected, the extruder die pressure increases with increasing PMDA concentration when the residence time is kept constant and the relationship is linear, as presented in Fig. 5. Increasing the PMDA concentration in the reactive extrusion process increases the melt viscosity. Flow becomes more difficult for the reactive extruded recycled PET, and it applies more pressure to the extruder die. In response to this, the extruder die resistance increases. By extrapolating the linear fit in Fig. 9, it can be seen that the die pressure would exceed the safety limit of 40 bar at about 0.35 wt% PMDA. At a constant PMDA concentration of 0.15 wt%, the relationship between die pressure and residence time is linear, as shown in Fig. 10. The die pressure decreases

Fig. 8. The effect of residence time on the [ ] of the reactive extruded recycled PET at constant PMDA concentration of 0.15 wt%.

with increasing residence time because of lower extrusion production speeds, and it is within the safety limit even when operating at a very low residence times corresponding to maximum extrusion production speed. Figures 7 to 10 summarize the various influences of PMDA concentration, residence time and die pressure on [ ], and from them Fig. 11 is derived. The linear relation between die pressure and [ ] shown in Fig. 11 is very useful for operating the reactive extrusion process. It makes it easy to predict the [ ] of the PET in the reactive extruder using the die pressure reading at a constant residence time of 45 s. Mechanical Properties of Reactive Extruded Recycled PET It was important to investigate the effect of the PMDA chain extension process on the mechanical properties of recycled PET.

Fig. 9. The effect of PMDA chain extender concentration on the extruder die pressure at a constant residence time of 45 s.

POLYMER ENGINEERING AND SCIENCE, AUGUST 2004, Vol. 44, No. 8

1585

Firas Awaja, Fugen Daver, and Edward Kosior

Fig. 10. The effect of residence time on the extruder die pressure at constant PMDA concentration of 0.15 wt%.

Fig. 12. Elasticity modulus vs. PMDA concentration at residence time of 45 s. Virgin PET has an elasticity modulus of 1756 MPa.

After injection molding, the virgin PET specimens were transparent, while the majority of recycled and the reactive extruded PET specimens were light yellow in color. The impact strength of reactive extruded recycled PET materials is 311 j/m. No significant differences were noticed between the different reactive extruded recycled PET samples. In comparison with the impact strength of the virgin PET sample (virgin PET had impact strength of 41.7 j/m), the recycled and reactive extruded recycled PET samples were more brittle than the virgin PET samples. Reactive extrusion does not significantly change the elasticity modulus of PET. Figure 12 shows that for the investigated PMDA concentrations, the reactive extruded recycled PET has a slightly higher elasticity modulus than the recycled PET sample. Figure 13 shows that higher residence times produce a lower value of elasticity modulus of the investigated samples. Increasing residence time increases the chain

extension reaction conversion, producing higher molecular weight, heavily branched and entangled chains. Heavily branched and entangled chains have reduced flexibility, leading to a reduction in elasticity (29). A residence time of 45 s with a PMDA concentration of 0.15 wt% resulted in reactive extruded PET with an elasticity modulus higher than that of virgin PET. The elasticity modulus of virgin PET was 1756 MPa, measured on the same instrument. Varying PMDA concentrations and residence times did not significantly change the upper yield stress of the samples. Reactive extruded recycled PET samples have an upper stress value of 58 1 MPa. Compared with the mechanical properties of virgin PET, reactive extruded recycled PET has better tensile strength and slightly lower impact strength, while recycled PET shows lower tensile and impact strength than virgin PET.

Fig. 11. The die pressure vs. [ ] for a residence time of 45 s derived from different PMDA concentration. 1586

Fig. 13. Elasticity-modulus vs. residence time at a constant PMDA concentration of 0.15 wt%. POLYMER ENGINEERING AND SCIENCE, AUGUST 2004, Vol. 44, No. 8

Recycled PET Chain Extension CONCLUSION PMDA has proved to be an effective chain extender in an industrial-scale reactive extrusion system for recycled PET. Reactive extruded recycled PET with higher [ ] and lower carboxyl content than in non-chain extended recycled PET was obtained. With appropriate adjustment to the extruder temperatures and the screw rotation speed, the reactive extrusion system was successfully implemented using an existing normal extrusion system. During this study, it was found that the chain extension reactive extrusion process depends strongly on chain extender concentration. The range of residence times studied showed a lesser effect on the product [ ], the product hydroxyl content and the process stability. Because of the dominating factor of PMDA concentration in the reactive extrusion process, it is critical to determine the most appropriate concentration in order to produce reactive extruded recycled PET with a particular viscosity. ACKNOWLEDGMENTS The first author acknowledges the Advanced Engineering Centre for Manufacturing (AECM) and Visy Plastics for a scholarship. The authors thank Prof. Dr. Werner Mormann of University of Siegen, Germany, for valuable discussions and Ms. Juliette Milbank for editing the paper. REFERENCES
1. EREMA, Plastic recycling systems, The PET plant technology (2002); www.erema.at/e_pet_tech.htm. 2. H. Al Ghatta, S. Cobror, and T. Severini, Polymers for Advanced Technology, 8, 161 (1996). 3. J. Scheirs, Polymer Recycling, Science, Technology and Application, Wiley series in polymer science, John Wiley and Sons Ltd. (1998). 4. H. Inata and S. Matsumura, J. Appl. Polym. Sci., 30, 3325 (1985). 5. H. Inata and S. Matsumura, J. Appl. Polym. Sci., 32, 5193 (1986). 6. H. Inata and S. Matsumura, J. Appl. Polym. Sci., 32, 4581 (1986). 7. H. Inata and S. Matsumura, J. Appl. Polym. Sci., 33, 3069 (1987). 8. H. Inata and S. Matsumura, J. Appl. Polym. Sci., 34, 2609 (1987). 9. H. Inata and S. Matsumura, J. Appl. Polym. Sci., 34, 2769 (1987). 10. K. C. Khemani, SPE ANTEC, 1934 (1998). 11. L. Incarnato, P. Scarfato, L. Di Maio, and D. Acierno, Polymer, 41, 6825 (2000). 12. S. M. Aharoni, D. E. Forbes, W. B. Hammond, D. M. Hindenlang, F. Mares, and R. D. Sedgwick, J. Polym. Sci. Part A: Polymer Chemistry, 24, 1281 (1986). 13. N. Cardi, R. Po, P. R. Giannotta, E. Occhiello, F. Garbassi, and G. Messina, J. Appl. Polym. Sci., 50, 1501 (1993). 14. D. N. Bikiaris and G. P. Karayannidis, J. Polym. Sci. Part A: Polymer Chemistry, 33, 1705 (1995). 15. D. N. Bikiaris and G. P. Karayannidis, J. Polym. Sci. Part A: Polymer Chemistry, 34, 1337 (1996). 16. A. A. Haralabakopoulos and D. Tsiourvas, J. Appl. Polym. Sci., 71, 2121 (1999). 17. G. P. Karayannidis and E. A. Psalida, J. Appl. Polym. Sci., 77, 2206 (2000). 18. S. M. Hong, S. S. Hwang, S. Yongsok, I. J. Chung, and U. Kim, Polym. Eng. Sci., 37, 3 (1997). 19. S. J. Chang and F. C. Chang, Polym. Eng. Sci., 38, 1471 (1998). 20. M. K. Akkapeddi and J. Gervasi, Polym. Prep., ACS Division of Polymer Chemistry Meeting, 29, 567, Toronto (1988). 21. L. P. B. M. Janssen, Polym. Eng. Sci., 38(12), 2010 (1998). 22. K. J. Ganzeveld and L. P. B. M. Janssen, Canad. J. Chem. Eng., 71, 411 (1993). 23. M. J. H. Bulters and P. H. M. Elemans, SPE ANTEC, 313 (2001). 24. M. H. Kim and C. G. Gogos, SPE ANTEC, 145 (2001). 25. B. Vergnes, G. Della Valle, and C. Delamare, Polym. Eng. Sci., 38, 11 (1998). 26. A. H. Pohl, Anal. Chem., 26, 1614 (1954). 27. S. Japon, L. Boogh, Y. Leterrier, and J. E. Manson, Polymer, 41, 5809 (2000). 28. L. Incarnato, P. Scarfato, L. Di Maio, and D. Acierno, Polymer, 41, 6825 (2000). 29. N. Torres, J. J. Robin, and B. Boutevin, Euro. Polym. J., 36, 2075 (2000).

POLYMER ENGINEERING AND SCIENCE, AUGUST 2004, Vol. 44, No. 8

1587