Assessing changes on poly(ethylene terephthalate) properties after recycling: Mechanical recycling in laboratory versus postconsumer recycled material

Materials Chemistry and Physics 147 (2014) 884e894

Contents lists available at ScienceDirect

Materials Chemistry and Physics journal homepage: www.elsevier.com/locate/matchemphys

Assessing changes on poly(ethylene terephthalate) properties after recycling: Mechanical recycling in laboratory versus postconsumer recycled material
pez a, Ana Isabel Ares Pernas a, Mª Jose
Abad Lo
pez a, María del Mar Castro Lo a, b a
pez Vilarin ~ o , Mª Victoria Gonza
lez Rodríguez a, * , J.M. Lo Aurora Lasagabaster Latorre
ns Tecnolo
gicas (CIT), Departamento de Física, Escuela Universitaria Polit
~ a, Grupo de Polímeros, Centro de Investigacio ecnica, Universidade de A Corun Campus de Ferrol, 15403 Ferrol, Spain

nica I, Facultad de Optica
n 118, 28037 Madrid, Departamento de Química Orga y Optometría, Universidad Complutense de Madrid (UCM), Arcos de Jalo Spain a

b

h i g h l i g h t s

g r a p h i c a l a b s t r a c t


Combination of multiple techniques to characterize the effects of recycling in PET.
Cleavage of ester bonds reduced viscosity, Mw, toughness in mechanical recycled PET.
Virgin, mechanical recycled and commercial recycled PET differ in crystal populations.
Cyclic oligomers [GTc]n and [GTc]n-G increase from the fourth extrusion cycle onwards.

a r t i c l e i n f o

a b s t r a c t

Article history: Received 21 March 2014 Received in revised form 20 May 2014 Accepted 15 June 2014 Available online 4 July 2014

Keeping rheological, mechanical and thermal properties of virgin poly(ethylene terephthalate), PET, is necessary to assure the quality of second-market applications. A comparative study of these properties has been undertaken in virgin, mechanical recycled and commercial recycled PET samples. Viscoelastic characterization was carried out by rheological measurements. Mechanical properties were estimated by tensile and Charpy impact strength tests. Thermal properties and crystallinity were evaluated by differential scanning calorimetry and a deconvolution procedure was applied to study the population of the different crystals. Molecular conformational changes related to crystallinity values were studied by FTIR spectroscopy. Variations in average molecular weight were predicted from

Keywords: Polymers Ageing

Abbreviations: acU, Charpy impact strength; CI, carboxyl index; DEG, diethylene glycol units; DHB, 2,5-dihydroxybenzoic acid; sB, the stress at the break point; sy, the stress at the yield point; E, Young modulus; ESI, electrospray ionization; εB, the strain at the break point; εy, strain at the yield point; G(t), stress relaxation modulus; G0 , storage modulus; G00 , loss modulus; [GTc]n-G, cyclic oligomer with a glycol unit; H-[GTL]n-GA, linear oligomer with a glycol-aldehyde unit; LVE, linear viscoelastic region; Me: MeOH, Methanol; Mn, number average molecular weight; MS, mass spectrometry; MW, molecular weight; MWD, molecular weight distribution; Mw, weight average molecular weight; h*, complex viscosity; PET, poly (ethylene terephthalate); PET_N1 to PET_N5, reprocessed PET samples (being N from 1 to 5 the number of reprocessing cycles suffered by the material); PET-vg, virgin bottle-grade poly (ethylene terephthalate); RPET1 and RPET2, two commercial recycled PET samples; T-[GTc]n, cyclic specie bearing one extra terephthalic unit; Tc, crystallization temperature; TFA, trifluoroacetic acid; Tg, glass transition temperature; Tm, melting temperature; TTS, timeetemperature superposition; %T, fraction of trans conformer; a, crystallinity degree; DHc, crystallization enthalpy; DHcc, cold crystallization enthalpy; DHm, melting enthalpy; n(CH), aliphatic CeH stretching vibrational mode; nC(]O)eOe, carboxyl stretching vibrational mode; n(OH), hydroxyl stretching vibrational mode; u(CH2), CH2 wagging vibrational mode; (u), frequency. * Corresponding author. Tel.: þ34 981 337 400 3051, þ34 981 337 400 3485; fax: þ34 981 337 416.
pez), [email protected] (A.I. Ares Pernas), [email protected] (M.J. Abad Lo
pez), [email protected] (A.L. Latorre), [email protected] E-mail addresses: [email protected] (M.M.C. Lo
pez Vilarin ~ o), [email protected] (M.V. Gonza
lez Rodríguez). (J.M. Lo http://dx.doi.org/10.1016/j.matchemphys.2014.06.034 0254-0584/© 2014 Elsevier B.V. All rights reserved.


pez et al. / Materials Chemistry and Physics 147 (2014) 884e894 M.M.C. Lo Mechanical properties Thermal properties Chemical techniques

885

rheology. Besides, the presence-absence of linear and cyclic oligomeric species was measured by mass spectrometry techniques, as MALDI-TOF. Mechanical recycled PET undergoes a significant decline in rheological, mechanical and thermal properties upon increasing the number of reprocessing steps. This is due to the cleavage of the ester bonds with reduction in molar mass and raise in cyclic oligomeric species, in particular [GTc]n and [GTc]nG type. Chain shortening plus enrichment in trans conformers favour the crystallization process which occurs earlier and faster with modification in crystal populations. Additional physicochemical steps are necessary to preserve the main benefits of PET. © 2014 Elsevier B.V. All rights reserved.

1. Introduction Due to the good barrier properties of blow-moulded containers made with poly(ethylene terephthalate) (PET), it has become a very important commercial plastic used in food packaging, films and carbonated soft drink bottles. PET presents good thermal and mechanical properties and is also used in reinforced plastics [1]. PET recycling process has received considerable attention as a result of societal pressure to reduce environmental pollution and to improve waste management. PET products have a slow rate of natural decomposition; therefore, recycling processes are the best way to economically reduce PET waste. Two major processes have been applied in order to recycle post-consumer PET, chemical and mechanical recycling. Although, both have been extensively studied, the disadvantage of chemical recycling is its high cost. Conversely, mechanical recycling is a relatively simple process that has received considerable attention as it is a successful route in terms of energy saving and emission of gases contributing to global warming. Mechanical recycling normally consists of contamination removal by sorting and washing, drying and melting processing [1,2]. Keeping high molar mass is necessary to assure the quality of second-market applications of recycled polymer. However, the exposure to several reprocessing cycles can modify the molecular structure and molecular weight (MW) of PET materials causing changes in their thermal, rheological and mechanical properties and thus, turning them useless for main commercial applications. Irreversible macromolecular changes generated by different chemical processes take place during PET mechanical recycling process. The main chemical reactions are chain scissions that predominate in “well-oxygenated” zones, but other minor modifications such as chain coupling can also take place in “poorly oxygenated” zones [2e4]. As result of chain-scission, ester-scission or auto-catalysed hydrolytic reactions, among others, the formation and disappearance of linear and cyclic oligomeric species have been identified in PET degradations studies [5]. Matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS) has become a valuable technique to study the chemical degradation of PET [6], allowing a faster identification of the characteristic degradation products. Application of the MALDI technique opened new vistas in polymer degradation analysis since it allows desorption and ionisation of intact polymer molecules and their identification in well resolved mass spectra, especially in the lower molecular mass range (<10,000 Da) [7,8]. In addition, it has been reported that small amounts of cyclic low molar mass oligomers present in PET samples come from PET polymerization process as it is usual for polymers prepared by condensation polymerization [6,9]. In the same way, the relation between the percentage of individual cyclic oligomers formed in the polymer melt and the percentage of cyclic oligomers initially found in commercial PET has been reported [10].

With all this in mind, the oligomeric degradation products have been used to investigate the chemical degradation of mechanical recycled PET. Although the application of MALDI-TOF MS to analyse PET degradation is not new [4e6], the combination of this method with a wide range of characterization techniques in order to correlate the chemical degradation with declining in physical properties, critical for industrial processes, has seldom been published. In the current work, several numbers of reprocessing cycles have been used to simulate the mechanical recycling of PET. The polymer degradation state for each reprocessing cycle has been characterized by measuring its rheological, thermal and mechanical properties and by FTIR spectroscopy. Furthermore, number average molecular weight (Mn) and weight average molecular weight (Mw) of the different reprocessed samples have been calculated from data obtained in the dynamic rheological tests. Finally, linear and cyclic oligomeric degradation species (100e10,000 Da) have been monitored by liquid chromatography coupled to mass spectroscopy (MS) and MALDI-TOF-MS. In the same way, two postconsumer recycled PET have been studied and compared with the mechanically recycled PET samples. 2. Experimental 2.1. Materials Samples of virgin bottle-degree poly (ethylene terephthalate) (Seda PET SP04) were supplied by Catalana de Polimers S.A., Grup LaSeda (Barcelona, Spain) in the form of pellets (PET-vg). The Seda Pet SP04 consists of opaque semicrystalline granules with an Intrinsic Viscosity value (IV) of 0.8 ± 0.02 dL g1 which corresponds approximately to 125 repeat units per molecule and an approximate MW of 2.53$104 g mol1. It has a melting point (Tm) is 245e250  C and a service temperature range between 20 and þ70  C. Besides, two commercial recycled PET (RPET1 and ~ a, Spain). RPET2) were provided from Nosoplas S.L. (A Corun Methanol (MeOH) and dichloromethane HPLC-gradient for instrumental analysis were supplied by Merck (Darmstadt, Germany). Formic acid 98e100% puriss p.a. was from SigmaeAldrich (Stockholm, Sweden). Water was purified using a Milli-Q Ultrapure water-purification system (Millipore, Bedford, MA, USA). MALDITOF MS matrix compound 2,5-dihydroxybenzoic acid (DHB) and trifluoroacetic acid 99% (TFA) were purchased from SigmaeAldrich. 2.2. Polymer samples manufacture: reprocessing simulation To simulate the mechanical recycling of PET, multiple reprocessing up to five times was performed using a corotating twinscrew extruder (DSE20, Brabender, GmbH & Co, Duisburg, Germany). Five reprocessing protocol was selected since the extrusion parameters were difficult to settle after the 5th step due to the low polymer viscosity; in addition, the resulting polymer strand was

886


pez et al. / Materials Chemistry and Physics 147 (2014) 884e894 M.M.C. Lo

hard to handle owing to very poor mechanical properties. A length/ diameter ratio of 40, operating at a speed of 20 rpm with a barrel temperature of 270  C was used. Previously, the pellets were dried in a Heraeus Vacutherm oven (Heraeus SA (Madrid, Spain)) at 150  C for 8 h. The different reprocessed PETs were labelled PET_N1 to PET_N5 (being N from 1 to 5 the number of reprocessing cycles suffered by the material). After each reprocessing step, some pellets were injection moulded into standard specimens (thickness: 3.96 ± 0.10 mm, width: 9.98 ± 0.07 mm, length: 79.6 ± 0.49 mm) using an injection machine (Battenfeld Plus 350, BattenfeldeWittmann, Kottingbrunn Austria), with an injection pressure of 1200 bar, a barrel temperature of 270  C and an injection rate of 35 cm3 s1. For rheological tests, another pellet fraction was shaped into discs by compression moulding in a hot plate press (IQAP-LAP PL-15, IQAP-LAP, Barcelona), at 270  C, applying a pressure of 40 bars for 2 min. 2.3. Rheological measurements Viscoelastic characterization was performed using a controlled strain rheometer (ARES, TA Instruments, and New Castle (USA)), with parallel-plate geometry (25 mm diameter, 1 mm gap) at 270  C. The complex viscosity (h*), storage modulus (G0 ) and loss modulus (G00 ) were measured as a function of frequency (u). Tests were performed in nitrogen atmosphere. The rheological tests were carried out in the linear viscoelastic region (LVE). This LVE region was determined previously from a strain sweep test. The u sweep measurements were set up in the u range 1  101 to 102 rad s1. In an effort to both map the flow behaviour of PET over wide ranges of u and temperature and to obtain molecular weight of reprocessed samples, the TimeeTemperature Superposition (TTS) principle was used to generate master curves for each polymer. Dynamic tests were performed at individual temperatures, ranges from 250  C to 290  C ranging of ten in ten in order to construct TTS diagram at reference temperature of 270  C. Shift factors were obtained for constructing TTS master curves. Information about the MW of the samples before and after reprocessing was achieved using the ARES rheometer software (Rheology Advantage Data Analysis) according to the method discussed in Section 3.1.1. 2.4. Mechanical properties 2.4.1. Tensile tests Tensile tests were performed at a crosshead speed of 10 mm min1 using a universal testing machine (Instron 5566, Instron, Norwood, USA) according to ISO 527. The Young modulus (E), the stress (sy) and strain (εy) at the yield point, and the stress (sB) and the strain (εB) at the break point were measured from stressestrain curves. At least five specimens from each blend were tested to obtain the average values of the mechanical parameters and their corresponding standard deviations. 2.4.2. Impact test Charpy impact strength (acU) was evaluated with an InstroneWolpert PW5 impact pendulum with a potential energy of 7.5 J following ISO 179 standard. At least ten unnotched specimens of each PET sample were tested to calculate the average value of the impact resistance and its standard deviation. The measurements were performed at room temperature. 2.5. Thermal properties by DSC Thermal measurements were carried out with a differential scanning calorimeter (DSC-7, PerkineElmer) under nitrogen

atmosphere. The samples of 6e10 mg were heated from 10  C to 280  C at a heating rate of 10  C min1 and held in the molten state for 3 min to erase their thermal history. After being cooled at 10  C min1, the samples were reheated to 280  C at 10  C min1. The glass transition temperature (Tg), crystallization temperature (Tc), crystallization enthalpy (DHc), melting temperature (Tm) and melting enthalpy (DHm) were obtained from the cooling scan and the second heating scan. Moreover, the crystallinity degree (a) was calculated according to Eq. (1), with DH0 ¼ 140 J g1 for the melting enthalpy of 100% crystalline PET [11]. In this way, in order to evaluate the original crystalline state of PET, the a value was calculated by subtracting the cold crystallization enthalpy (DHcc) measured on the heating scan.

a¼

ðDHm  DHcc Þ  100 DH0

(1)

2.6. FT-IR spectroscopy The ATR-FTIR analyses were performed in a Bruker Vector 22 in the spectral range of 4000e400 cm1. For comparison purposes, the pellets were subjected to the same thermal treatment performed for DSC experiments, then placed on a thermostated MK II Golden Gate Diamond 45 ATR accessory and compressed onto the ATR crystal with the Sapphire Anvil (10,531). The spectra were the results of 100 coadded interferograms at a spectral resolution of 4 cm1. Peak height measurements and integral absorbance were performed with the spectral analysis software (Opus 5.5). Results are the average of at least 6 spectra measured in different samples. The area (A) or intensity (I) of the 1410 cm1 band (rings CH in plane bending and CeC stretching), which has already proved its suitability as internal reference band as it is not conformationally sensitive, is used for normalization in order to compensate the variations in thickness [12,13]. 2.7. Extraction of PET oligomers by microwave procedure A Milestone microwave laboratory system ETHOS TC (Sorisde, Italy) equipped with 10-vessel position carousel was employed to extract PET oligomers. The instrument is temperature controlled. 1 g of PET samples cut into small pieces were accurately weighted and extracted by microwave energy under the following conditions: 30 mL of dichloromethane extraction solvent, 2 min of heating time, 13 min of extraction time, and 60  C of temperature. After extraction, vessels were allowed to cool to ambient temperature. The liquid phase was filtered through an AcrodiscR PTFE CR 13 mm, 0.2 mm filters (Waters, Mildford, MA, USA) and analysed by means of HPLC-PDA-QqQ. 2.8. HPLC-PDA-QqQ equipment and operating conditions An Agilent 1200 Series Rapid Resolution LC system (Agilent Technologies, Waldbronn, Germany) equipped with an on-line degasser, a binary pump delivery system, a high performance SL autosampler, a thermostated column department and on-line coupled to a mass spectrometer detector was used for analysis. Samples were filtered through a 0.2 mm AcrodiscR PTFE CR (Waters) and injected in Zorbax SB-C18 (50  2.1 mm, 1.8 mm) column (Agilent Technologies). A two solvent gradient elution was performed, with flow rate of 0.3 mL min1 and injection volume of 3 mL. Mobile phase was composed by watere0.1% formic acid (A) and MeOH (B) under the following gradient elution profile: starting at 30% of B was linearly increased to 100% B in 20 min and maintained for 5 min. It was then brought back to the initial conditions.


pez et al. / Materials Chemistry and Physics 147 (2014) 884e894 M.M.C. Lo

887

The column effluent was directly introduced into the Agilent Technology DAD SL G1315C system (ranged from 190 to 500 nm). The mass spectrometer Agilent 6410 Triple Quad LC/MS (Agilent Technologies) was linearly coupled to the PDA. It operated in a positive ionization mode. Ions were formed using electrospray ionization (ESI). The following ESI source parameters were used: temperature of the drying gas (N2) was set to 350  C and flowed at 10 mL min1. The nebulizing pressure (N2) was maintained at 35 psi. Capillary voltage was set at 5.5 kV. Integration and data elaboration were performed using Agilent MassHunter Workstation software, version B03.00 (Agilent-Technology, Santa Clara, USA). The full mass scan range was m/z 100e2000 (1 s/scan). Mass spectral data and retention time were used for peak identification. 2.9. MALDI-TOF: sample preparation and analysis About 0.05 g of virgin, commercial-recycled PET and reprocessed PET samples were dissolved in 5 mL of TFA. A portion of 20 mL of the polymer solution was mixed with 20 mL of the DHB matrix prepared in CHCl3-TFA (5:2). Then, 1 mL of the mixture was dropped onto a stainless steel MALDI-TOF MS plate and allowed to dry at ambient temperature before insertion into the instrument. Separation of the oligomers was also done by precipitating the polymer by successively adding 4 mL of chloroform and 4 mL of acetone to a portion of 2 mL of polymer solution in TFA under continuous stirring. The PET oligomer solution was separated from the precipitated polymer by filtration. 20 mL of the oligomer solution was then mixed with 20 mL of DHB matrix. 1 mL was dropped onto the plate and allowed to dry before analysis. MALDI-TOF-MS experiments were conducted by means of a Voyager DE STR mass spectrometer (Applied Biosystems Inc., Foster City, CA) equipped with a nitrogen laser (337 nm). All spectra were obtained in the linear, positive ion mode, using an acceleration voltage of 15 kV, grid voltage of 90 and delay time set to 200 ns. Spectra were acquired in m/z range from 450 to 10,000 with low mass gate set at m/z ¼ 450. To obtain best spectral resolution, the laser intensity was set at medium to high levels. External calibration was performed using standards of angiotensin-1, ACTH (1e17), ACTH (18e39) and insulin (bovine). For each sample, 100 spectra from 6 different points on the sample spot and by triplicate were collected and combined (100 spectra  6 points  3 ¼ 1800 spectra). All spectra were processed and calculated using Sata Explorer ver. 4 program (Applied Biosystems Inc.).

Fig. 1. Complex viscosity versus frequency of PET for different extrusion cycles.

modulus (G0 ) reduction is especially evident at higher frequencies. In fact, reprocessed material is less elastic than PET-vg and recycled commercial PETs. 3.1.1. Molecular weight from rheological data It is well known that the rheological properties of polymer melts depend strongly on the molecular structure, especially MW and MWD. In recent years, there has been considerable interest in this area regarding the prediction of MW and MWD from rheological data. Since there are many viable methods of determining MW and MWD of polymers, such as gel permeation chromatography [14e16] intrinsic viscosity measurements [17,18] or light scattering, it is important to note the specific advantages provided by the application of the rheological MW determination method. Many polymers of commercial importance, such as Teflon, polyethylene and PP, are only slightly soluble in common solvents at room temperature. The use of rheological methods to assess their MW avoids this problem, although the calculation of Mn, Mw and MWD is not as accurate as with GPC. Anyway, several methods have been used for

3. Results and discussion 3.1. Rheological results The h* versus u for PET after different reprocessing cycles is illustrated in Fig. 1. Overall, Newtonian behaviour is shown by all samples. Initially, differences between PET-vg and the two recycled commercial PET samples (RPET1 and RPET2) are not revealed. On the other hand, all reprocessed PET samples show lower viscosity than the original one and the value of h* dramatically decreases with increasing cycles at constant u. Although a stabilization phase of viscosity values is observed between PET_N2 and PET_N3, this behaviour is related to the thermal decomposition of main PET structural irregularities, such as diethylene glycol units (DEG) that is frequently added to decrease the crystallization rate and improve ductility, processability and clarity of PET [6]. The global performance of studied samples is in agreement with the recycling PET behaviour described in a previous study [2]. G00 is higher than G0 during all the u range, showing a typical viscoelastic liquid behaviour. The moduli decrease with extrusion cycle, indicating loss of material properties (Fig. 2). The elastic

Fig. 2. Moduli as a function of frequency for neat PET and PET after different extrusion cycles.

888


pez et al. / Materials Chemistry and Physics 147 (2014) 884e894 M.M.C. Lo

prediction MW from rheological data and some software has also been developed. William H. Tuminello and N. Cudre-Mauroux showed that MW can be determined from linear viscoelastic melt properties, such as the storage modulus, G0 (u) or the stress relaxation modulus G(t) and from viscosity vs. shear rate [19e22]. To fit the rheological data, the adjustment parameters summarized in Fig. 3 were used. Both number average molecular weight (Mn) and weight average molecular weight (Mw) of the PET samples were estimated as function of reprocessing cycle. First of all, the estimated Mn and Mw values for PET-vg (2.53$104 and 2.70$104 g mol1 m, respectively) are in accordance with the molecular weight provided in the supplier technical data sheet. Besides, the reprocessed samples present pronounced extrusion cycle dependence. Therefore, reductions in Mn and Mw higher than 60% are observed compared with the material without processing (Fig. 3), due to chain scissions as the main thermal degradation mechanism. Nevertheless, a different behaviour is shown between the values of the first (PET_N1) and second (PET_N2) reprocessed samples. A small decrease in Mn and greater increase in Mw are displayed between both reprocessing cycles, presumably due to a balance between chain scissions and chain branching, according to a previous study of recycling PET by extrusion [2]. Conversely, no important differences are revealed between data of PET-vg and commercial recycled PETs (RPET1 and RPET2) for Mn (2.60$104 and 2.90$104 g mol1, respectively) and Mw (2.90$104 and 3.20$104 g mol1, respectively) values. This is probably related to a chain extension process where poly-functional low MW material is reacted with PET carboxyl and/or hydroxyl end groups to rejoin the broken chains [1]. 3.2. Mechanical properties The tensile parameters and the Charpy impact strength (acU) are reported in Table 1 for all samples. Only virgin PET and the polymer after the first reprocessing step display plastic deformation with a well-defined yield point in the stressestrain curves. Then, upon increasing the number of extrusions steps, the specimens break before the elastic limit and the strain at break point (εB) undergoes a dramatic drop (from 35% after 1st cycle to 0.7% after 5th cycle). In addition, the variation trend is more complicated for the tensile strength (sB). The value for PET_N1 is comparable to PET-vg, whereas the sB for PET_N2 nearly duplicates those values. Conversely, there is a clear drop from PET_N3 onwards. Most of

Fig. 3. Number average molecular weight (Mn) and weight average molecular weight (Mw) of PET after different reprocessing cycles.

these changes could be related with the variations in the molecular weight values (Mn and Mw) and with the reduction in elastic modulus (G0 ) measured from the rheological tests. In the latter experiments, molecular weight reductions are particularly significant from the third extrusion cycle onwards proving that PET chains have suffered scissions during reprocessing. Similar results were previously reported by other authors [3]. Besides, the Young modulus (E), after an initial decrease for PET_N1 and PET_N2 increases after the third cycle and then it remains constant within experimental error. Concerning the impact tests, material toughness drops dramatically with the numbers of reprocessing steps. Both the above mentioned chain shortening and the raise in the crystallization degree, discussed in the following section, explain the increasing polymer stiffness (E increase) and fragility (εB and acU decrease). Conversely, the striking high value of sB for PET_N2 or even the initial drop in Young modulus may be related to the increase observed in its Mw, mentioned in the preceding section, which was tentatively ascribed to a balance between chain scissions and chain branching. In addition, the tensile parameters obtained for commercial recycled PET (RPET1 and RPET2) are comparable to PET-vg except for the higher standard deviations in the tensile strength (sB) and deformation at the break point (εB). The latter parameter has not been reported for RPET1 and RPET2 because the deviation obtained in the experiment was excessive. The random break in these specimens may be indicative of polymer chain heterogeneity produced during the recycling process and/or to the presence of contaminants [1]. Moreover the commercial recycled PETs do not break during the impact tests. In summary, it is clear that reprocessing by extrusion produces severe chain degradation in PET which is evident from both tensile and impact parameters. The negative effects in the mechanical properties are visible from the first extrusion process. Conversely, as stated in the rheological study, the industrial processes that recycle the post-consumer PET usually incorporate chain extension reactions or solid state polymerizations to overcome the decrease in its molecular mass. These processes could explain that commercial recycled PETs are able to maintain some of the tensile mechanical properties (E, sy, εy, sB) of PET-vg or even improve the impact strength [1]. 3.3. Thermal properties 3.3.1. Melting and crystallization behaviour DSC tests were performed in order to identify the effects of reprocessing steps on crystallization and melting behaviour of PET. The cooling curve of virgin PET shows a very wide crystallization peak. This indicates a very slow crystallization rate characteristic of a resin with high molar mass (that is required for bottle-grade with high transparency). The remaining samples have a well-defined exotherm with a sharp crystallization peak during the cooling scan (not shown). During the second heating scan (Fig. 4), the DSC curve of virgin PET presents an exotherm (Tcc ¼ 163  C and DHcc ¼ 14.2 J g1) associated with cold crystallization stage previous to melting. It could be attributed to morphological and structural re-organization of less perfect crystals formed during cooling. Its enthalpy was subtracted to calculate the crystallization degree (a) of virgin PET. In the reprocessed materials, the cold crystallization peak does not appear but their melting peak presents a shoulder (typical of thermo-mechanical degradation) [6], more pronounced as the number of reprocessing cycles increases. Table 2 summarizes the DSC data obtained from crystallization and subsequent melting scan (second heating run). The Tc of


pez et al. / Materials Chemistry and Physics 147 (2014) 884e894 M.M.C. Lo

889

Table 1 The tensile parameters and the Charpy impact strength (acU) for all the PET samples assayed. Samples

E (MPa)

PET-vg PET_N1 PET_N2 PET_N3 PET_N4 PET_N5 RPET1 RPET2

1405 1103 1247 1783 1750 1729 1381 1423

± ± ± ± ± ± ± ±

εy (%) 40 108 61 40 17 26 22 16

5.8 7.4 e e e e 5.8 5.3

± 0.2 ± 0.4

± 0.1 ± 0.1

sB (MPa)

εB (%)

sY (MPa)

23.7 ± 0.2 22.4 ± 0.9 50 ± 4 29 ± 2 18 ± 4 10 ± 1 33 ± 12 38 ± 10

42 ± 4 35 ± 2 5.7 ± 0.5 2.6 ± 0.3 1.6 ± 0.3 0.7 ± 0.1 e e

55.5 53.2 e e e e 55.3 52.5

reprocessed materials shifts to higher temperatures, indicating that the crystallization process occurs earlier and faster for samples subjected to increasing number of extrusion cycles. There is a substantially increase of the crystallization enthalpy for the recycled samples compared to virgin PET, but no dependence with the number of reprocessing steps is found. During the second heating scans, the Tg of reprocessed samples shows a clear decrease from the first to the second cycle and then, slight oscillations around 78  C. Furthermore, Tm increases as a function of the number of reprocessing cycles as well as the intensity of the new peak at lower temperature (the latter appears as a shoulder in PET_N1 and PET_N2 and as an intense peak from PET_N3 onwards) (Fig. 4). Multiple melting peaks have already been observed in PET during DSC thermal analysis by other authors [1,4]. Furthermore, it is interesting to note that the increase in

acU (kJ m2)

± 0.4 ± 0.6

135 ± 32 71 ± 35 25 ± 12 16 ± 8 6±1 6±2 NB NB

± 0.9 ± 0.5

melting enthalpy and a depend on the number of reprocessing cycles, up to third extrusion cycle and then, they remain constant within experimental error. This trend is reflected in the tensile modulus (E) from the third cycle onwards. Probably, the chain shortening caused by the thermomechanical degradation supported the crystallization during the cooling, increasing the crystalline degree and consequently, the polymer stiffness. For commercial recycled PET, whereas Tg values are similar to the corresponding PET-vg value, the melting behaviour is alike the mechanical reprocessed samples after several cycles, with two neat peaks at approximately 244.4 and 236.7  C. In addition, an increase in the crystallization rate and crystallinity is observed compared with the virgin PET. The corresponding values are similar to the polymer after one or two reprocessing steps. 3.3.2. Multiple melting peaks Previous researchers, as Badía et al. [4], suggested that the multiple melting peaks, observed during the DSC heating scan, are attributable to the distribution of crystals with different lamellar thickness and to the melting of different crystal structures. In order to clarify how the sequential reprocessing cycles affect to the population of the different crystals, a deconvolution procedure has been applied to the endothermic peak of the second heating scan. Thus, a Gaussian function has been used to fit the area under the endotherm in three or four different peaks and the fit was considered valid when R2  0.9. For the reprocessed PET, the endotherms are composed of three peaks, centred roughly at 220  C, 235  C and 240  C for all the samples. The crystal populations have been labelled as 1, 2 and 3 with increasing temperature. Lu et al. proved that PET suffers two different processes during its crystallization [23]. The different melting peaks are related to this behaviour. The population 1 could be identified as the smallest lamellae produced by secondary crystallization or inter-lamellar crystals developed after the primary step of crystallization is complete. The population 2 would be formed during the primary crystallization and the crystals of population 3 could be due to recrystallization of crystals Table 2 DSC data for virgin PET, reprocessed PET samples and commercial recycled PET samples. Sample

PET-vg PET_N1 PET_N2 PET_N3 PET_N4 PET_N5 RPET1 RPET2 Fig. 4. DSC thermograms for second heating scans.

Cooling scan

Second heating scan

TC ( C)

DHC (J g1)

Tg ( C)

Tm ( C)

DHm (J g1)

a (%)

158.2 172.8 191.0 194.2 197.5 201.3 191.6 188.0

29.5 47.6 50.9 57.5 60.4 52.6 53.3 44.2

79.9 79.4 78.7 77.6 78.7 77.3 80.6 80.3

238.6 240.1 (225.8) 240.6 (231.3) 241.1 (231.3) 241.3 (236.4) 243.1 (237.0) 244.5 (236.8) 244.3 (236.3)

35.1 41.0 43.2 48.9 49.2 48.7 45.8 41.0

14.9 29.3 30.9 34.9 35.1 34.8 32.7 29.3

890


pez et al. / Materials Chemistry and Physics 147 (2014) 884e894 M.M.C. Lo

of populations 1 and 2. This is consistent with the fact that PET-vg displays only crystals of population 3 after the cold crystallization process. Fig. 5 shows the evolution of the relative contribution of each crystal population to the total melting peak area (partial crystallinity %). This parameter was calculated from the partial melting enthalpy of each population divided by the whole area under the melting peak in percentage. The data show that the relative contribution of population 2 progressively increases, mainly at the expense of population 3, whereas the relative contribution of population 1 remains practically constant for all extrusion recycled PETs. The population 2 corresponds with the area under the shoulder of the melting peak, which intensity increases with the number of recycling steps. Regarding commercial recycled PET, the deconvolution of the melting peaks leads to four populations of crystallites. In addition to the populations 1, 2 and 3 previously commented, there is a new population (labelled as population 0) related to a lower Tm (about 205  C) and only detected in these samples. Probably, these crystallites were produced during the secondary crystallization and are due to heterogeneities in the crystallization processes such as additives, chain extenders or contaminants. It seems clear that the process to recycle the post-consumer PET encourages heterogeneous crystallization with crystals of different sizes and melting enthalpy. 3.4. FT-IR spectroscopy FTIR is an important analytical method that allows evaluating chemical and conformational changes during the recycling process. Possible changes in the infrared spectrum are followed in the regions corresponding to the subsequent vibrational modes: n(OH), n(CH), u(CH2) and the complex lower wave number region containing the ester bands and ring modes [24]. Besides decreasing molar mass, chain scission during PET recycling often generates polymer radicals with hydroxyl and carboxyl end groups and that carboxyl end-groups act as catalyst to

promote further degradation. The index of carboxyl end-groups has been determined as the ratio of the peak areas at 3430 cm1 and 1410 cm1 (reference peak for normalizing) [25,26]. This ratio remains unchanged within experimental error upon increasing the number of extrusion cycles (0.11 ± 0.01). On the contrary, there is a dramatic change in the shape, number and intensity of the bands of the n(CH) region (3000e2800 cm1) (Fig. 6 (I)) during the degradation process. The intensity of the band at 2924 cm1 decreases upon raising the number of extrusion cycles whereas a new band appears at 2907 cm1; the relative intensity of the band at 2854 cm1 with respect to the band at 2960 cm1 also diminishes. The profile of these bands for the commercial recycled samples, RPET1 and RPET2, is similar to PET_N1. These changes may be attributed either to a chemical change in the system due to chain scission, to conformational changes or both. Decreasing the polymer chain length causes variations in the ratio of the number of terminal “ethylene glycol” groups relative to the number of bulk polymer “ethylene glycol” groups which alters the molar extinction coefficient of the n(CH) bands as a result of the change of electronic environment [24]. More to the point, as FTIR spectroscopy is highly sensitive to structural modifications, the conformational changes during degradation can also be observed. One of the most important characteristics of the molecular conformation of PET concerns the existence of trans and gauche rotational conformers for the ethylene glycol moiety. Both types of conformers are found in the amorphous phase, but only the trans conformer is present in the crystalline phase. The marker bands attributed to ethylene glycol moiety in the trans conformation fall at 1470 cm1 (CH2 bending), 1340 cm1 (CH2 wagging), 1118 cm1 (OeCH2 and ring CeC stretching, ring CH in plane bending), 970 cm1 (OeCH2 and C(] O)eO stretching), and 845 cm1 (various bending modes of the benzene ring). On the contrary, the bands at 1370 cm1 (CH2 wagging), 1044 cm1 (CeO stretching), and 898 cm1 (CH2 rocking) are attributed to the gauche/amorphous conformation of PET [24,27,28]. The fraction of trans conformers (%T) was calculated taking into account the integral absorbance of the bands at 1340 cm-1 and 1370 cm1 (A ¼ based on peak areas), respectively assigned to trans and gauche conformers, by Eq. (2) [29,30]:

 %T ¼

Fig. 5. Evolution of relative partial crystallinities as function of the number of extrusion cycles.

 A1340 $100 A1340 þ 6:6A1370

(2)

The conformational composition has also been followed from the trans bands at 1470 cm1 and 845 cm1, the gauche band at 1044 cm1 and the doublet near 1090 cm1 (CeO stretching and other vibrations). In fact, the component at about 1090 cm1 has been attributed to gauche and trans conformers or to the amorphous conformation, whereas the component at 1118 cm1 has been assigned to trans conformers in crystalline PET. For complementary qualitative assessment of the trans/(gauche þ trans) ratio, the I1118/I1090 ratio (I ¼ based on peak heights) was also calculated (Table 3) [28]. It can be inferred from Fig. 6(II) and the data displayed in Table 3 that upon increasing the number of extrusion cycles, the trans conformers were enriched either by selective hydrolysis and removal of the gauche conformers, or by gauche / trans conversion. Furthermore, a raise in trans conformers, similar to the values obtained for the first extrusion cycles, is confirmed for commercial recycled PETs. Although not directly correlated with the % of crystallinity, a (%) (Table 2), the fact that the intensity of the FTIR band markers related with the trans conformers increase whereas the intensity of the band markers associated with the gauche conformers decrease with thermal degradation corroborate the trend


pez et al. / Materials Chemistry and Physics 147 (2014) 884e894 M.M.C. Lo

891

Fig. 6. FTIR-ATR film spectra in the 3600e2700 cm1 range (I) and 1800e1100 cm1 range (II) of: PET-vg (a); PET_N5 (b) and RPET1. Inset: enlarged view of the 1500e1325 cm1 spectral range.

observed by DSC. These conclusions are in accordance with other workers results which found enrichment in trans conformers as a result of degradation and, hence, an increase of the overall degree of crystallinity and/or in conformational changes leading to a more ordered state of the polymer chains [24,28]. 3.5. Mass spectrometry The presence-absence of degradation products, linear and cyclic oligomeric species, have been studied in the virgin PET, commercial recycled PET and reprocessed PET samples by two methods: extraction with microwave energy and subsequently analysis by liquid chromatography (Sections 2.7 and 2.8) and solubilisation of PET samples and analysis by MALDI-TOF-MS (Section 2.9). 3.5.1. Microwave-assisted extraction of PET oligomers for HPLCPDA-QqQ PET degradation has been usually explained by the formationdisappearance of linear and cyclic oligomeric species such as [GTL]n and [GTc]n and their derivatives [5,6], representing G and T a glycol and terephthalate units, respectively. L and c standing for lineal and cyclic, respectively; and being n the number of repeating units forming the oligomer. Therefore, cyclic and lineal oligomeric species [PET]n (100e2000 Da) have been identified based on the ions (m/z) labelled as [MþH]þ and a pseudo-molecular ion [MþNa]þ. As n increased [MþNa]þ was the predominant adduct detected. The main oligomeric species [PET]n found are listed in Table 4 with their corresponding m/z (n ranged from 1 to 8). PET-vg and RPET1 only show detectable levels of the cyclic oligomers [GTc]n and [GTc]n-G with n ¼ 2e3 while all the reported Table 3 Area absorbance ratios (A) or intensity ratios (I) calculated from the spectra of ATR pellets subjected to the same thermal treatment than in DSC experiments. Sample

I1470/I1410

PET-vg PET_N1 PET_N2 PET_N3 PET_N4 PET_N5 RPET1 RPET2

0.29 0.30 0.34 0.35 0.36 0.44 0.40 0.35

± ± ± ± ± ± ± ±

0.003 0.01 0.01 0.02 0.03 0.03 0.05 0.03

T (%) 44 49 52 54 58 58 48 47

± ± ± ± ± ± ± ±

I1118/I1090 1 1 1 7 1 2 4 3

0.66 0.73 0.69 0.76 0.73 0.78 0.73 0.71

± ± ± ± ± ± ± ±

0.01 0.04 0.03 0.02 0.03 0.02 0.07 0.04

A1044/A1410 0.23 0.23 0.19 0.18 0.20 0.12 0.16 0.19

± ± ± ± ± ± ± ±

0.01 0.01 0.01 0.01 0.02 0.03 0.02 0.02

A845/A1410 0.25 0.26 0.28 0.29 0.32 0.28 0.27 0.27

± ± ± ± ± ± ± ±

0.01 0.02 0.01 0.01 0.02 0.02 0.02 0.01

species (Table 4) have been observed in reprocessed PET samples. The species corresponding to n ¼ 3 are predominant among all detected oligomers. Besides, cyclic oligomers are detected in higher relative abundance than linear ones. The formation of cyclic oligomers [GTc]n and [GTc]n-G as main species have been related to successive extrusion cycles [5,6]. In addition, small quantities of two other species, also described as degradation products in reprocessed PET degradation, have been detected. A cyclic specie bearing one extra terephthalic unit (T[GTc]n) and a linear oligomer with a glycol-aldehyde unit (H-[GTL]nGA). The formation of the latter is assigned to degradation of DEG units [5,6]. Nevertheless, both kind of molecules were only clearly observed for species of low molecular weight (n < 4). Thus, multiple reprocessing steps lead to modifications in the oligomeric distribution in the studied PET samples. Very high levels of the low-molecular mass oligomer ([GTc]3) have been found in reprocessed PET sample (PET_N5) compared to PET-vg, as previously reported [6,9], and to commercial recycled PET (RPET1). 3.5.2. MALDI-TOF-MS MALDI-TOF MS measurements have been carried out in the mass range from 450 to 10,000 Da, in which it exhibits the highest mass resolution as well as the best conditions for the formation of molecular ions, favouring the study of the degradation reactions that take place during PET processing. These conditions allow the detection and identification of [PET]n oligomers with n ranging from 3 to above 20, decreasing its relative abundance as n is increased, in compliance with HPLC-PDA-QqQ analysis. Fig. 7 shows the MALDI-TOF MS spectra of virgin PET, commercial recycled PET and reprocessed PET samples (for better visualization, only the m/z 570e760 region is displayed). The results show that the relative abundance of [PET]n oligomers decreases as n is increased. As in the microwave extraction method, cyclic oligomers [GTc]n and [GTc]n-G detected as protonated molecules [MþH]þ are the main species found by MALDI-TOF MS. Nonetheless, the presence of linear oligomers with different structures can be more clearly seen. Besides the cited H-[GTL]n-GA, the oligomer TFA-[GTL]n-OH is also observed. The latter is related to the typical reaction in PET thermo-mechanical degradation process, resulting from acetylation between the hydroxyl end groups and the TFA when NaTFA is used as catalyser in MALDI experiments [5,6].


pez et al. / Materials Chemistry and Physics 147 (2014) 884e894 M.M.C. Lo

892

Table 4 Oligomeric (n ¼ 1e8) species assigned for virgin PET, commercial recycled and reprocessed PET by MW/HPLC-PDA-QqQ. m/z

[PET]*1

[PET]2

[PET]3

[PET]4

Structure

[M þ H]þ

[M þ Na]þ

149 193 237 341 Nd 385 429 533 475 577 621 725 667 769 813 Nd 859

Nd Nd 259 Nd Nd 407 451 Nd Nd 599 643 Nd Nd 791 835 Nd Nd

[Tc] [GTc]n [GTc]n-G T-[GTc]n H-[GTL]n-GA [GTc]n [GTc]n-G T-[GTc]n H-[GTL]n-GA [GTc]n [GTc]n-G T-[GTc]n H-[GTL]n-GA [GTc]n [GTc]n-G T-[GTc]n H-[GTL]n-GA

m/z [M þ H]þ

[M þ Na]þ

961 1005 Nd Nd 1153 1197 Nd Nd 1345 1389 Nd Nd Nd Nd Nd Nd

983 1027 Nd Nd 1175 1219 Nd Nd 1367 1411 Nd Nd 1559 1603 Nd Nd

[PET]5

[PET]6

[PET]7

[PET]8

*[PET]1…[PET]8 ¼ structures with repeating PET units from one to eight. Nd: not detected.

The evolution of the relative ion abundances, taking into account intensities and areas for calculations, has been compared (Fig. 8). The relative abundance of the [GTc]n considerably increases with further reprocessing cycles. Similarly, the average intensity of the cyclic PET oligomer with an ether linkage in the backbone ([GTc]n-G) slightly raised as result of the reprocessing steps. The

presence and increase of both cyclic oligomers in PET samples can be explained by different mechanisms [5,6,31]. The slight decrease shown for minor cyclic oligomeric adducts T-[GTc]n with an extra terephthalate unit might be guided by the loss of terephthalic units, thus interacting with the formation of [GTc]n species [5,31].


pez et al. / Materials Chemistry and Physics 147 (2014) 884e894 M.M.C. Lo

893

Fig. 7. MALDI-TOF MS spectra in the m/z 570e760 range of PET-vg compared to commercial recycled PET (RPET1, RPET2) and reprocessed PET (PET_N1 to PET_N5).

Linear species, H-[GTL]n-GA, and TFA-[GTL]n-OH, remain constant during reprocessing within experimental error, suggesting that they barely interact in the general degradation mechanism of PET. As can be observed in Fig. 8, commercial recycled PET (RPET1 and RPET2) show similar PET oligomeric distribution to virgin PET, although smaller levels of the main oligomers are observed in the former.

From another point of view, the percentage of the major cyclic oligomers, [GTc]n and [GTc]n-G, with different n values in the range 3e20 was estimated in reprocessed samples in relation to virgin PET (results not shown). The relative percentage of each and every one of these oligomers remain constant or slightly increase during the three first reprocessed samples, decrease for the fourth one whereas significantly increase for the fifth reprocessed sample, specially the oligomer with n ¼ 20.

894


pez et al. / Materials Chemistry and Physics 147 (2014) 884e894 M.M.C. Lo

to maintain some of the tensile mechanical properties (E, sy, εy, sB) of PET-vg or even improve their impact strength despite the increase in trans conformers and crystallinity. These results highlight the need for additional physicochemical steps intended to preserve molecular weight and viscosity parameters and thus meet the requirements demanded of packaging intended for the conservation and transport of foodstuffs. Acknowledgements This study was supported by the Xunta de Galicia Govern (Autonomous Community Government)-FEDER under Program of Consolidation and structuring competitive research units 2011e2013 (CN2011/008) and project 10TAL172002PR. References Fig. 8. Normalized ions abundance changes of the main detected oligomers of PET-vg, RPET1, RPET2 versus reprocessed PET (PET_N1 to PET_N5).

Finally, the prevalence of cyclic oligomers in comparison with linear species explains the fact that the carboxyl index, estimated by FTIR, remains unchanged within experimental error. 4. Conclusions Multiple reprocessing by means of successive extrusion cycles was used to simulate degradation effects on PET samples under thermo-mechanical recycling. Their effects were compared with non reprocessed samples (virgin PET) and with commercial recycled PET. Reductions in viscosity, moduli (G0 and G00 ), Mn, Mw and toughness are observed for all reprocessed PET samples in comparison with the material without processing. These properties dramatically drop upon increasing the number of reprocessing steps. This behaviour is related to the cleavage of the ester bonds of PET, being chain scission the main ageing mechanism during reprocessing cycles according to MS analyses. The latter techniques have shown the raise of cyclic oligomeric species as PET degradation proceeds. In particular, [GTc]n-G and the most abundant [GTc]n increase with the number of reprocessing cycles. These data are in accordance with the consistency of the carboxyl index observed by FTIR. Furthermore, the crystallization process occurs earlier and faster for samples subjected to extrusion cycles and modifications in crystal population are observed: three different crystal populations were found (labelled as 1, 2 and 3 with increasing temperature) in extruded samples; by contrast, PET-vg displays only crystals of population 3. Probably, the chain shortening caused by the thermomechanical degradation plus the enrichment in trans conformers proved by FTIR, boosted the crystallization during the cooling, increasing the crystalline degree and consequently, the polymer stiffness and fragility. From another point of view, no important differences are revealed between PET-vg and commercial recycled PETs (RPET1 and RPET2) for h*, Mn, Mw values and cyclic oligomeric levels. Moreover, the industrial recycling processes lead to materials able

[1] F. Awaja, D. Pavel, Eur. Polym. J. 41 (2005) 1453e1477. [2] L.K. Nait-Ali, X. Colin, A. Bergeret, Polym. Degrad. Stabil. 96 (2011) 236e246.
, M.A. de Paoli, J. Appl. Polym. Sci. 80 (2001) 20e25. [3] M.A. Silva Spinace [4] J.D. Badia, F. Vilaplana, S. Karlsson, A. Rives-Greus, Polym. Test. 28 (2009) 169e175. €mberg, A. Ribes-Greus, S. Karlsson, Anal. Chim. Acta 692 [5] J.D. Badía, E. Stro (2011) 85e95. ~o, M.F. Franco, Y.E. Corilo, M.N. Eberlin, M.A.S. Spinace
, M.A. De Paoli, [6] W. Roma Polym. Degrad. Stabil. 94 (2009) 1849e1859. [7] S. Weidner, G. Kuehn, B. Werthmann, H. Schroeder, U. Just, R. Borowski, R. Decker, B. Schwarz, I. Schmuecking, I. Seifert, Polym. Sci. Part A Polym. Chem. 35 (1997) 2183e2192. [8] F. Samperi, C. Puglisi, R. Alicata, G. Montaudo, Polym. Degrad. Stabil. 83 (2004) 3e10. [9] K.A. Barnes, A.P. Damant, J.R. Startin, L. Castle, J. Chromatogr. A 712 (1995) 191e199. [10] V. Vermylen, P. Lodefier, J. Devaux, R. Legras, W.A. Mac donald, R. Rozenberg, E. de Hoffmann, J. Polym. Sci. Pol. Chem. 38 (2000) 416e422. [11] B. Wunderlich (Ed.), Macromolecular Physics, vol. 1, A Press, New York and London, 1973, p. 389. vremont, A. Ajji, M.M. Dumoulin, Appl. Spectrosc. 48 (1994) [12] K.C. Cole, J. Gue 1513e1521. [13] C. Sammon, J. Yarwood, N. Everall, Polymer 41 (2000) 2521e2534. [14] M. Mutsuga, Y. Kawamura, K. Tanamoto, Food Addit. Contam. A 25 (2008) 1283e1290. [15] H. Tsuji, Y. Ikada, Polym. Degrad. Stabil. 67 (2000) 179e189. [16] S.K. Saha, H. Tsuji, Polym. Degrad. Stabil. 91 (2006) 1665e1673. [17] K. Xu, A. Kozluca, E.B. Denkbas, E. Piskin, J. Appl. Polym. Sci. 59 (1996) 561e563. [18] Z. Su, Q. Li, Y. Liu, G.H. Hu, Ch Wu, J. Polym. Sci. Pol. Phys. 47 (2009) 1971e1980.
-Mauroux, Polym. Eng. Sci. 31 (1991) 1496e1507. [19] W.H. Tuminello, N. Cudre [20] J.J. Cooper-White, M.E. Mackay, J. Polym. Sci. Pol. Phys. 37 (1999) 1803e1814. [21] H.J. Lehermeier, J.R. Dorgan, Polym. Eng. Sci. 41 (2001) 2172e2184.
pez, [22] S. Dopico-García, A. Ares-Pernas, J. Otero-Canabal, M. Castro-Lo
pez-Vilarin ~ o, V. Gonza
lez-Rodríguez, M.J. Abad Lo
pez, Polym. Adv. J.M. Lo Technol. 24 (2013) 723e731. [23] X.F. Lu, J.N. Hay, Polymer 42 (2001) 9423e9431. [24] C. Sammon, J. Yarwood, N. Everall, Polym. Degrad. Stabil. 67 (2000) 149e158. [25] G.J. M Fechine, M.S. Rabello, R.M. Souto Maior, L.H. Catalani, Polymer 45 (2004) 2303e2308. [26] H. Fashandi, A. Zadhoush, M. Haghighat, Polym. Eng. Sci. 48 (2008) 949e956. iæ, J. Jelene iæ, e-Polymers 089 [27] L.K. Krehula, Z. Hrnjak-Murgiæ, B. Andrie (2008) 1e12. [28] I. Donelli, P. Taddei, Ph. F. Smet, D. Poelman, V.A. Nierstrasz, G. Freddi, Biotechnol. Bioeng. 103 (2009) 845e856. [29] M. Drobota, M. Aflori, V. Barboiu, Dig. J. Nanomater. Bios. 5 (2010) 35e42. [30] Y. Zhang, J. Zhang, Y. Lu, Y.D. Yan, D. Shen, Macromolecules 37 (2004) 2532e2537. [31] I. Goodman, B.F. Nesbitt, Polymer 1 (1960) 384e396.