Influence of the addition of glycerol-derived polymers on the properties of post-consumer recycled PET

In the present work, the influence of glycerol prepolymers as potential chain extenders in the reactive mixture with post-consumer recycled PET (PET-PCR) was studied, namely: poly(glycerol citrate) (PGC), poly(glycerol succinate) (PGSu) and poly(glycerol citrate-co-succinate) (PGCSu). Mixtures of recycled PET with 2 wt% of glycerol prepolymers or Joncryl ADR®4638, a commercial chain extender, were processed in an internal mixer at 265 °C for 6 min and 60 rpm. The compositions were characterized by viscosimetry, carboxylic end groups (CEG) analysis using the Pohl method, crosslinking degree as the percentage of insoluble gel, Fourier transform infrared spectroscopy (FTIR), thermogravimetric analysis (TGA) and differential scanning calorimetry (DSC). The addition of glycerol prepolymers increased the fluidity in the post-consumer PET resin, consequently, decreasing its viscosity by 33% for PETPCR/PGC, 24% for PET-PCR/PGSu, 44% for PET-PCR/PGCSu and 17% PET-PCR/Joncryl; the molecular weight in the order of 42% for PET-PCR/PGC, 31% for PET-PCR/PGSu, 55% for PETPCR/PGCSu and 23% for PET-PCR/Joncryl and increasing the number of carboxylic end groups. Although they did not show extensive chain crosslinking, the samples were fragile and brittle, thus making the tensile test unfeasible. Thermogravimetry showed that the polyesters were thermally stable up to approximately 350 °C and the sample that presented the lowest thermal stability was the PETPCR/PGCSu corroborating to the highest reduction in the molecular weight recorded. The DSC results showed an almost constant melting and glass transition temperature among the samples, a decrease of 8 °C in the crystallization temperature and an increase in the degree of crystallinity relative to neat PET-PCR, which corroborates to the intrinsic viscosity results. These results indicated that at the concentration and processing conditions used, the potential additives to chain extenders had an opposite effect to the desired one, that is, they reduced the molecular weight of recycled PET and the same effect occurred for the commercial chain extender. However, the additive that showed the greatest potential for use as chain extender was PGSu, which presented results similar to the commercial extender Joncryl ADR®4638, composed of multifunctional epoxy groups.


Introduction
Poly(ethylene terephthalate) (PET) is one of the most industrially used polymers because of its excellent balance between production cost and optical, mechanical and thermal properties [1].Due to the diversity of applications, PET weight fraction in municipal solid wastes (MSW) has been continuously increasing [1].Therefore, in order to avoid MSW accumulation there is an urgent need to recycle these materials through, e.g., mechanical recycling.However, during mechanical recycling changes in the final properties of the recycled polymer occur as a consequence of hydrolysis and thermomechanical degradation that cause a decrease in molecular weight [2].In this sense, studies have been 372 Page 2 of 13 carried out aiming to recycle PET with the addition of substances such as chain extenders, capable of improving PET final properties [3][4][5][6][7].In general, chain extenders are di-or multifunctional chemical compounds intended to recover molecular weight and physical properties of degraded thermoplastics through reactions between PET chain ends [8].
The most commonly chain extenders used in the literature for PET are organic chemical compounds, based on dianhydrides, oxazolines, diepoxides, isocyanates and phosphates [5,[9][10][11][12].However, the formation of excess acid groups during the reaction of pyromellitic dianhydride (PMDA) with PET, among the dianhydrides, limits polymer stability, and the high costs and toxicity associated with most organic phosphites also limit extended-chain PET in applications such as in food packaging [13,14].
Glycerol-based polymers can be synthesized through the polycondensation reaction of glycerol with various dior triacids, such as succinic, citric, sebacic acid, resulting in prepolymers with controllable molecular weight and functionality [15][16][17][18].In addition to adding value to glycerol, one of the main by-products of biodiesel, these polymers can be used in food packaging industry because they are derived from nontoxic, renewable and biodegradable sources [15].Moreover, glycerol-based polymers can be potentially used as chain extenders since their hydroxyl and carboxylic end groups are capable of reacting with PET chains [19].
In this work, poly(glycerol succinate) and poly(glycerol citrate) prepolymers were synthesized by bulk polycondensation polymerization of succinic and citric acids [20].Succinic acid is a precursor used in many important industrial applications, including biodegradable plastics [21,22]; and citric acid has a highly functional, abundant, and low-cost moiety, whose applications range from food to pharmaceutical and cosmetics industries [23].As chain extenders are used at levels of up to 2 wt% [8,14,[24][25][26], the proposed extenders were added at this concentration in post-consumer recycled (PCR) PET in order to assess their potential use comparing with a commercially available one (Joncryl ADR ® 4368).

Synthesis of glycerol prepolymers
Glycerol prepolymers was synthesized via polycondensation reaction [20] of glycerol, and citric and succinic acids.The reaction temperature was maintained at 120 °C, under constant stirring and an atmosphere of nitrogen and 0.01 wt% of catalyst.Each reaction was conducted using an appropriate molar ratio of reagents, in order to keep the [OH]:[COOH] ratio at 1:1 as can be seen in Table 1.
Reaction was followed by calculating the acid number (AN) for according to Carothers gelation theory (Eq. 1) [28], where Pc is the critical polymerization extension value for polymer, that is, the amount of acids that interact with glycerol for the expected polymerization to occur and fav, the average functionality of the polymer.The calculation of AN (Eq.2) consists of determining the amount of mass in mg of potassium hydroxide (KOH) needed to neutralize one gram of acid.Polymerization was stopped when the calculated acid number was close to the theoretical critical acid number of each polymer, for the formation of Fig. 1 General structure of the chain extender, Joncryl ADR ® 4368, where R1-R5 are H, CH 3 , higher alkyl group, or a combination of; R6 corresponds to an alkyl group, while X, Y and Z are each between 1-20 [27] a three-dimensional network, thus giving rise to prepolymers that were further used as chain extenders.

Melt-processing of PET-PCR with glycerol prepolymers
Prior to processing, post-consumer recycled PET (PET-PCR) was placed in a vacuum oven (Solab SL 104) for 6 h at 160 °C and kept under vacuum in a desiccator for at least 21 h before processing due to the high hygroscopicity of PET.Then, the PET-PCR without (PET-PCR-0%) with 2 wt% of each glycerol prepolymer were processed in an internal mixer coupled to the Haake torque rheometer (Rheomix 3000QC from PolyLab QC) using roller rotors to obtain a filling factor (f) of 70% of the equipment's chamber (Eq.3).
where ρ corresponds to the density of the material and fVf to the free volume of the processing chamber of the internal mixer.
Mixing was carried out for 6 min at 60 rpm and 265 °C.Initially, only PET was added to the mixing chamber, after 1 min of processing, the mixing chamber was opened and the glycerol prepolymers added, without interrupting the processing.Another formulation of PET-PCR with 2 wt% of multifunctional epoxy-based extender additive composed of three classes of monomers: styrenics, methyl acrylates and glycidyl acrylates (Joncryl ADR ® -4368 supplied by BASF-Brazil) was prepared under the same processing conditions.The description of melt-processed samples is shown in Table 2.
PET molecular weight behavior for each sample was evaluated from the torque and mechanical energy measurements (Fig. 2).The same chamber opening procedure after 1 min was used for PET-PCR-0%. (1)

Intrinsic viscosity ( [] ) and average viscosimetric molar mass ( Mv)
The intrinsic viscosity test was performed in accordance with ASTM D4603:2018.The samples were solubilized in phenol/1,1,2,2-tetrachloroethane (60/40 wt%) at a concentration of 0.5 g.dL −1 and at a solution temperature of 30 °C in duplicate.The intrinsic viscosity[ ] was then calculated by measuring relative viscosity using the Bilmeyer equation: where η is the intrinsic viscosity; r is the relative viscosity; and C is polymer concentration.

Determination of carboxylic end groups (CEG)
The carboxyl content was determined according to the Pohl method [29] by titrating a standard solution of NaOH (0.1 N) in a solution of 0.15 g PET in 5 mL of benzyl alcohol and 10 mL chloroform using phenol red as an indicator.The concentration of carboxylic end groups in eq/g was calculate by Eq. ( 6) below: where V1 is the volume of standard basic solution consumed in the blank titration of the benzyl alcohol and chloroform solution; V2 is the volume of standard basic solution consumed in the PET solution titration; N is the normality of the standard basic solution; and w is the polymer weight.From this value, 1.6 eq/10 6 g must be deducted to eliminate the degradation that occurred during the titration process.

Determination of gel content
The degree of crosslinking defined as the percentage of insoluble gel was performed in duplicate to understand the reaction mechanisms.Approximately 250 mg of samples were dissolved in 25 mL of phenol/tetrachloroethane (60/40 wt%) at 120 °C for 1 h.The solution was then filtered, washed with acetone, and allowed to dry to a constant weight.The gel content was determined by the Eq. ( 7) below [30].
where wf is final weight of samples after filtration, washing and drying and wi is the initial weight of the filter paper after filtration, washing and drying of the neat PET-PCR-0% solution to discount the amount of polymer present in the polymeric solution that was precipitated by acetone rinsing and retained in the filter.

Fourier transform infrared spectrophotometry (FTIR)
Samples in the form of pellets and films were characterized in the attenuated total reflectance (ATR) mode between 600 and 4000 cm −1 using 64 scans at 4 cm −1 resolution on a model IRAffinity-1-Shimadzu FTIR spectrometer.

Thermogravimetric analysis (TGA)
The thermal stability of PET with and without addition of glycerol polymers was evaluated using thermogravimetric analysis.The conditions used were heating rate of 10 °C.min −1 , nitrogen flow of 50 mL.min−1 and temperature range from 25 to 600 °C.

Differential scanning calorimetry (DSC)
DSC analysis was performed on Shimadzu DSC-60 equipment, with a heating/cooling rate of 10 °C.min −1 , in a temperature range between 25 and 300 °C under nitrogen flow of 50mL.min−1 .After the first heating, to erase the thermal history of melt-processing, samples were kept at 300 °C for 3 min and then cooled to 30 °C, kept at that temperature for 3 min and then heated again to 300 °C.The degree of crystallinity was then determined from the second heating by means of Eq. ( 8): where ΔHm is the melting enthalpy of samples (obtained from the area of the crystalline melting peak of the DSC thermogram); ΔH 100% is the melting enthalpy of PET 100% crystalline, being 140 J.g −1 [31]; and w is the PET weight fraction.

Synthesis of prepolymers
The polycondensation reaction of poly(glycerol citrate), poly(glycerol succinate) and poly(glycerol citrate-co-succinate) was followed according to Carothers theory [28] and stopped when the calculated acid number was close to the theoretical critical acid number of each one (Table 3), preventing crosslinking from occurring inside the reactor.Prepolymers had color and final texture depending on the acid used: transparent and fluid for PGC, white and pasty for PGSu and white and viscous for PGCSu (Fig. 3), whose structural forms present large amounts of hydroxyl and carboxylic groups with potential to be used as chain extenders to PET-PCR [20].

Melt-processing of PET-PCR compositions
The torque curves as a function of processing time are shown in Fig. 4. It was possible to observe an increase in the torque values during the feed, due to the friction and the plastic deformation of pellets, followed by an asymptotic decay.As the polymer starts to melt, viscosity decreases and so does the torque.Friction and deformation are reduced until the ( 8) polymer is completely melted.The torque stabilizes up to the total processing time of 6 min and there was little change in torque (Fig. 4a) after incorporating the additives.
Since the torque can be directly related to melt viscosity and, consequently, to polymer molecular weight [4], it was expected that the addition of glycerol-based prepolymers, would increase and/or stabilize torque in the compositions due to the great amount of -OH and -COOH groups capable of reacting by condensation with the ends group of PET chains and thus, inhibit the reduction of PET molecular weight during reprocessing.However, in this work, the curve profile showed a slight increase and then continued the decay similar to the other curves.Differences in the homogenization rate of the chain-extender agents in the melt PET may be one of the reasons for the torque curve oscillations, as a function of mixing time in this region.The torque decay of the additive formulations relative to that PET-PCR-0% is possibly because these additives are acting as lubricants at 2 wt% and hindering the chemical chain extension reaction to take place [32].Another hypothesis is that the mixing time used (6 min) was not enough to promote the reaction between the commercial chain extender and PET.
Regarding the mechanical energy (Fig. 4c), it was observed that the incorporation of glycerol polymers was not able to increase the energy needed to move the viscous material inside the processing chamber and consequently its viscosity.Since mechanical energy can be used as a parameter to evaluate the viscosity of molten polymeric materials, the rate of mechanical energy dissipation in the processing chamber is directly proportional to the melt viscosity [33].

Intrinsic viscosity [η] and average viscosimetric molecular weight Mv
The intrinsic viscosity [η] results obtained according to Eq. ( 5) for PET-PCR-0% and PET-PCR with the incorporation of 2% of glycerol prepolymers and commercial additive are shown in Table 4. Results were transformed into viscosimetric average molecular weight ( Mv ) by applying Eq. ( 3), which relates the intrinsic viscosity of a polymeric solution to the molecular weight of the polymer.
. A decrease in the intrinsic viscosity and, consequently, in the molecular weight of the samples after processing and incorporation of the additives greater than for PET-PCR-0% was noticed.Furthermore, according to the intrinsic viscosity value of recycled PET, declared by the supplier company (0.78 dL.g −1 ), it was analyzed that there was a 42% reduction in the intrinsic viscosity of PET-PCR-0% after its processing in the internal chamber of the torque rheometer.The decrease in molecular weight is due to chain scission, which is the main mechanism of thermo-oxidative, mechanical and hydrolysis degradation of PET in the melt state [12].A similar behavior was observed by Romão et al. [34] that processed bottle grade PET in an extruder using temperatures of 220, 260, 275, 280 and 280 °C for 5 min along the five zones, and obtained a decrease in the intrinsic viscosity from 0.80 to 0.43 dL.g −1 and in the molecular weight from 46,700 to 17,700 g.mol −1 .Furthermore, they [34] observed that the intensity of this decrease increased as a function of the mixing time, since after processing for 15 min, the intrinsic viscosity dropped to 0.26 dL.g −1 and the molar mass to 8,600 g.mol −1 .
The same authors [35] processed post-consumer PET obtained from flakes of carbonated beverage bottles in an internal mixer at 250 °C, 50 rpm for 5 min and again a significant decrease in the intrinsic viscosity was observed, i.e., from 0.76 dL.g −1 to 0.59 dL.g −1 .This behavior is expected for PET subjected to thermomechanical processing of drying step is not synchronized to melt processing and contact with room atmosphere is unavoidable.Therefore, all evaluations of this study consider the PET-PCR-0% processed under the same conditions of all additivated PET-PCR samples.
The intrinsic viscosity reduction of the additivated PET sample in relation to PET-PCR dried and processed under the same conditions was of the order of 33% for PET-PCR/PGC, 24% for PET-PCR/PGSu, 44% for the PET-PCR/PGCSu copolymer and 17% for the commercial additive PET-PCR/Joncryl.For average viscosimetric molecular weight, this decrease was in the order of 42% for PET-PCR/PGC, 31% for PET-PCR/ PGSu, 55% for PET-PCR/PGCSu and 23% for PET-PCR/ Joncryl.These results indicated that at the concentration and processing conditions used, the chain-extending additives had an opposite effect to the desired one, i.e., they accelerated the reduction of PET molecular weight.
Regarding the composition containing citric acid based prepolymer (PGC), despite the highest amount of reactive sites, the presence of bulky -OH e -COOH groups in the chemical structure of the citric acid near to the reaction sites, probably, caused steric hindrance and slowed down the reaction [36].As a result, the PET-PCR sample containing PGSu presented the best performance among the glycerol prepolymers chain extenders used were similar to the commercial chain extender, Joncryl ADR ® 4368.
It is worth noting that the synthesis reaction of glycerol prepolymers was interrupted before reaching the acid number (AN) which corresponds to the critical extent of the polymer for crosslinking (Pc) of each polymer according to Carothers' theory of gelation [28].At that point, the chains tend to present themselves in a linear structure.However, for AN valuers lower than Pc, a higher extent of the reaction (pre-polymeric form) with the formation of ramifications, without increasing the size of the chain takes place [20].Thus, the linearity of the prepolymers chains may have interfered in the melt viscosity of PET and in the effectiveness of the reaction between the components of the mixture [32].The short mixing time chosen was based on the fact that glycerol-based polymers have an initial temperature of thermal decomposition (~ 300 °C) [16,20] close to the processing temperature of PET (265 °C) and that a long processing time could cause degradation.
The Joncryl additive presented an unexpected result in relation to other studies [37,38], that reports an increase in PET molecular weight between 20 and 68% with the addition of up to 1.5 wt%.However, the composition used in this work (2 wt%) showed different results.Probably, the processing time used in this work (6 min) was not enough to promote the extension reaction between Joncryl and PET, as some studies [37,38] used different mixing times.Tavares et al. [38] used 30 min of processing with insertion of the additive after 7.5 min and Duarte et al. [37] used 20 min of processing with additive insertion after 10 min.Both [37,38] obtained effective PET chain extension with Joncryl.

Determination of carboxylic end groups (CEG)
It has been reported in the basic polymer literature [39] that the number of carboxyl end groups decreases with increasing molecular weight of PET.Therefore, it can be concluded that the decrease in the amount of carboxyl groups can be interpreted as an efficient chain-extension reaction between recycled PET and chain-extender additive [40].However, in this work there was an increase in carboxylic end groups, corroborating with the intrinsic viscosity results (Table 4).This increase was 64.58%, 83.77%, 77.53% and 46.25% respectively for PET-PCR/ PGC, PET-PCR/PGSu, PET-PCR/PGCSu and PET-PCR/ Joncryl, with respect to PET-PCR-0%, probably because of thermomechanical and/or hydrolysis degradation during processing.This degradation leads to PET chain scission and, therefore, to a reduction in the length of the polymer chain.As a result, the number of chain end increases and, also, the number of carboxylic groups [5,41,42].The concentration of additive and/or the reaction time used were possibility not appropriate for the chain extension reaction to effectively occur, as previously discussed.

Determination of gel content
To understand the reaction mechanism of the samples, the crosslinking degree was calculated for each composition (Table 4).Results show that the additive samples have a small fraction of gel formation indicating that the reactive mixing process can theoretically result in the formation of lightly cross-linked structures in addition to the chain scission reaction [12].
The literature explains that the chain extender and reaction time in amounts greater than theoretically necessary produce chemical, thermal and rheological instability in polymers, causing crosslinking reactions and gel formation and consequently negative effects on the mechanical property of elongation at break [10,30,42].As an example, there is the study carried out by Awaja et al. [41] using pyromellitic dianhydride (PMDA), as a chain extender additive in recycled PET by reactive extrusion (zone temperature 279 °C for 112 s) which did not achieve extensive crosslinking and results in a gel formation of less than 4% at concentrations up to 0.3 wt%.However, using 0.35 wt% of PMDA, there were chemical and hydrodynamic instabilities detected with the increase in melt viscosity that produced thick and crosslinked PET.
In this work, the reactions in the used concentrations of additives (2 wt%) did not reach the point of extensive crosslinking of the samples.Conversely, they showed a certain fragility, probably, due to the decrease in intrinsic viscosity caused by the chain scission promoted during the processing conditions and the additives concentration used.

Fourier transform infrared spectrophotometry (FTIR)
FTIR spectra of post-consumer recycled PET resin, before (PET-PCR) and after processing in the internal mixer (PET-PCR-0%) and PET-PCR-0%, prepared by hot pressing (260 °C for 1 min) are illustrated in Fig. 5a.The spectra of the compositions of PET-PCR with glycerol prepolymer after processing and of their respective films prepared by hot pressing (260 °C for 1 min) are illustrated in Fig. 5b-c.
The spectra in Fig. 5a show the characteristic bands of PET.The bands around 2962 cm −1 and 1717 cm −1 are attributed respectively to the CH 2 group of the aliphatic chain and to the carbonyl group (C = O).The vibration at 1407 cm −1 is attributed to the C-H bonds of the aromatic ring.The bands between 1240 − 1100 cm −1 and 725 and 848 cm −1 are attributed, respectively, to the ester (-COO-) groups and C-H vibrations associated with the aromatic rings [43].Evaluating the spectra of the samples after the addition of the possible chain extenders (Fig. 5b and c), it was possible to observe that apparently there were no new peaks and that in all of them occurred in the characteristic ester bands (1245 − 1090 cm −1 ).

Thermal analysis
The thermal decomposition of PET samples with and without chain extenders was investigated by thermogravimetric analysis (TGA) as shown in Fig. 6.All samples were thermally stable up to approximately 350 °C, indicating volatile levels below the sensitivity of the equipment.A reduction between 2 and 24 °C in the initial temperature of thermal decomposition (T onset ) of the samples of PET-PCR with additives in relation to the PET-PCR-0% was observed, indicating that the additives interfered in the thermal stability of the PET-PCR.The composition with the copolymer (PET-PCR/PGCSu) shifted the thermal event (T onset ) to a temperature of 352 °C, a difference of 24 °C in relation to the PET-PCR-0% that presented the highest thermal stability (376 °C).This behavior is probably due to the lower thermal stability of glycerol polymers [16,20] relative to virgin PET [44].The residue at 600 °C was practically the same for all samples: 15.5% for PET-PCR, 13.5% for PET-PCR/PGC, 13.9% for PET-PCR/PGSu, 12.7% for PET-PCR/PGCSu and 15.2% for PET-PCR/Joncryl.These variations may be related to the amount of PET used in the formulations and the absence of residues from the decomposition of glycerol polymers.
The DSC curves of PET-PCR samples with and without a chain extender are shown in Fig. 7. Literature describes the melting temperature of virgin PET samples between 250 and 265 °C [45].Little oscillation was observed at this temperature, according to the results presented in Table 5.However, the broadening of the melting peaks for the samples containing glycerol polymers was probably due to the formation of crystals of varying sizes that ended up melting at a wider temperature range, interfering with the degree of crystallinity of the samples [46,47].The observation of a single peak for melting temperatures (Tm) (Fig. 7) is interesting, since two peaks of Tm are commonly observed as a consequence of thermomechanical degradation [35,40].The glass transition temperature (Tg) was also practically constant among samples.The crystallization temperature (Tc) of the PET-PCR samples with glycerol prepolymers changed systematically to lower temperature ranges (~ 8 °C) in relation to PET-PCR-0% and the degree of crystallinity (Xc) increased: 57.23% for PET-PCR/PGC, 48.97% for  5.These results corroborate those of intrinsic viscosity and molar mass reduction (Table 4), as smaller chains have greater mobility, facilitating the packaging of macromolecules in a crystalline structure and leading to the chemic crystallization process [46,48]. .

Conclusions
It was possible to synthesize the glycerol-derived prepolymers by measuring the acid number of each one and thus using them as potential chain extenders for recycled PET.The results indicated that at the concentration and processing conditions used, the chain-extender additives had an opposite effect to the intended one, that is, they did not inhibit the thermo-mechanical, thermo-oxidative and hydrolysis degradation of PET during melt processing.Despite the composition containing citric acid (PGC) having the highest number of reactive, it was the sample containing succinic acid (PGSu) that presented the best results and behaved like the commercial extender Joncryl.Therefore, this work will serve as a basis for future research, in order to study the concentration range and another possible compositions of glycerol prepolymers, and, if necessary, modify the processing time.

Fig. 6
Fig. 6 TGA and DTG curves and T onset (inset) of PET-PCR-0% and PET-PCR with the incorporation of 2% of glycerol polymers and commercial additive

Fig. 7
Fig. 7 DSC graphs of PET-PCR-0% and PET-PCR with the incorporation of 2% of glycerol polymers and commercial additive at second heating cycle (a) and enlarged graphs at glass transition (b), crystallization (c) and melting temperatures (d), respectively

Table 1
Stoichiometric compositions and proportions of reagents in the reaction of polymers and glycerol copolymer a Values are based on stoichiometric amounts of hydroxyl and carboxyl groups [OH]=[COOH]

Table 2
Labeling of PET-PCR with and without chain extenders used in this work

Table 3
Theoretical and calculated critical acid number of each prepolymer

Table 4
Data of the intrinsic viscosity, viscosimetric average molecular weight, concentration of carboxylic end groups and the degree of crosslinking of the melt-processed PET samples η Intrinsic viscosity, Mv Viscosimetric average molecular weight, Equivalents/10 6 g Concentration of carboxylic end groups

Table 5
Data obtained from the DSC analysis of PET-PCR-0% and PET-PCR samples with the incorporation of 2% of glycerol polymers and commercial additive Tg Glass transition temperature, Tc Crystallization temperature, TmMelting temperature, ΔH f Enthalpy of meting, Xc Degree of crystallinity