Sustainable polymer reclamation: recycling poly(ethylene terephthalate) glycol (PETG) for 3D printing applications

Due to their versatile properties and wide-ranging applications across various industries, including manufacturing, polymers are indispensable for today’s society. However, polymer-based products significantly impact the environment since many are single-used plastics and require a long time to degrade naturally. A method to attenuate end-of-life polymers’ ill effects is recycling them to bring them again into the production cycle, from grave to cradle. This investigation involves recycling PETG sheets used in face shield production during the COVID-19 outbreak to fabricate 3D printing filaments for FFF. We assessed poly(ethylene terephthalate) glycol (PETG) processability to up to five recycling cycles and obtained filaments with properties adequate for 3D printing. Rheological, thermal, morphological, and mechanical characterization were analyzed to verify the effect of the number of processing cycles on the properties of the polymer. The recycling cycles originated a decrease in viscosity and elasticity, and the gain in molecular mobility resulted, relatively, in solids with a higher degree of crystallinity and prints with more elliptical depositions. The mechanical properties of printed parts fabricated of recycled material were comparable to those from commercial filament, especially after three extrusion cycles. Both extrusion and additive manufacturing processes successfully recycle material into filaments and printed parts, indicating that the proposed methodology is a promising alternative to bring value back to polymers from solid waste.


Introduction
Additive manufacturing (AM), popularly known as 3D printing, is defined by ISO/ASTM 52900:2015 as joining materials layer upon layer to build an object from a 3D computer model (ISO, ASTM 2021).The democratization and increased accessibility of the public to the AM process have resulted in the emergence of low-cost equipment and materials, thus enabling new technological opportunities and continuous expansion of this field, including for the consumer market.Even under economically unfavorable conditions brought on by the COVID-19 pandemic, according to Wohler's 2024 report, there has been an 11.1% growth in independent AM service providers worldwide, resulting in a revenue Page 2 of 14 Seno Flores et al. J Mater. Sci: Mater Eng. (2024) 19:16 of approximately $ 20.035 billion in 2024 (Wohlers et al. 2020).
In particular, the fused filament fabrication (FFF) process has had considerable relevance to the popularization of AM (Volpato et al. 2007;Wang et al. 2017).The advent of low-cost 3D printers has culminated in the increase of popular and domestic applications, with FFF, an AM technique based on material extrusion, being the highlight for appearing in the most economical and easyto-operate equipment, playing a relevant role in the popularization of this technology (Volpato et al. 2007;Wang et al. 2017;Gao et al. 2021).
This extraordinary interest in AM can be attributed to the versatility of the process, the possibility of material savings compared to subtractive manufacturing methods, and the ability to produce complex geometries (Volpato et al. 2007;Gibson et al. 2015).Aspects like these make AM suitable for aerospace, automotive, architecture, artistic, educational, and medical industries, and even startups use it (Volpato et al. 2007;Wang et al. 2017;Gibson et al. 2015).The materials commonly used in the FFF process are acrylonitrile butadiene styrene (ABS), poly(lactic acid) (PLA), poly(ethylene terephthalate) glycol (PETG), polycarbonate, polyphenylene sulfide, polyamides, polymer blends, and composites with varied matrices (Wong and Hernandez 2012; Goh et al. 2020).These polymers must have adequate viscosity, between a minimum value to allow extrusion and a maximum value sufficient to provide structural support (Volpato et al. 2007; N. Turner et al. 2014).Many of these polymers are not biodegradable, requiring a long time for their degradation to occur through natural agents.These materials have low densities, occupying large volumes in landfills, i.e., in 2022, 400.3 million tons of polymer resins were manufactured worldwide (PlasticsEurope 2024).In Brazil, the production of resins reached 6.7 million tons, and the demand for processed plastics was 7.1 million tons, while only 1.0 million tons of post-consumer waste were recycled (Brazilian Plastics Industry Association (ABIPLAST) 2022).The high consumption, incorrect disposal, and low recycling efforts are responsible for negative social, environmental, and economic impacts; therefore, recycling has an important role since it allows the reuse of plastic residues.
Recent studies show that recycling polyesters, such as poly(ethylene terephthalate) PET and poly(ethylene terephthalate) glycol (PETG), is a viable alternative in environmental and economic terms.This helps mitigate the consequences of the high consumption of virgin resins and aims at more sustainable production processes.In commercial terms, PETG is more widely used than PET for FFF filaments since it presents lower crystallinity, greater toughness and flexibility, a wider processing window, and greater transparency than PET.Besides, it maintains its properties when subjected to heat treatment.In addition, PETG is considered food-safe in both amorphous and semi-crystalline conditions due to its chemical resistance (Latko-Durałek et al. 2019;Quental 2004;Chen et al. 2015a).
Franciszczak et al. (Franciszczak et al. 2018) reused PETG as a matrix to manufacture composites processed by injection molding.They found that the flow rate increased, and the injection process was only possible after including the chain extender Joncryl 4368S.The application of the chain extender allowed the use of more degraded PETG scrap with small amounts of residual impurities, providing a low-cost method for material and reprocessing.In the same study, the researchers developed a blend of recycled PETG with PET fibers, which showed tensile and flexural mechanical properties comparable to composites with 20 to 40% by weight of talc-reinforced polypropylene.Dealing with blends with recycled PETG, Latko-Durałek et al. (Latko-Durałek et al. 2019) formulated a blend of PETG and Polystyrene, both recycled and virgin PETG.The results show that when comparing virgin PETG with a blend of 30% recycled PETG and polystyrene, the blend had a 42% increase in tensile strength, 15% in modulus of elasticity, and 27% in elongation at rupture.
In addition to its application in the injection molding process, recycled PETG has shown itself capable of producing filaments for FFF.Vidakis et al. (Vidakis et al. 2021) submitted PETG to 6 times extrusion recycling cycles.Analyzing the mechanical properties, they observed that the tensile strength and modulus of elasticity reached their maximum values in the third and fourth processing cycles.The thermogravimetry (TGA) analyses presented by the authors indicated that the maximum operating temperature of the samples manufactured by 3D printing, approximately 380 ºC, is suitable for engineering thermoplastic materials.In addition, the thermal analysis revealed that there was no significant degradation until the fifth recycling cycle.Kováčová et al. (Kováčová et al. 2020) evaluated the mixture of recycled and virgin PETG with added fillers as an alternative for applying the recycled polymer for FFF.Using this mixture of PETG with expanded graphite and carbon fiber (5-10 wt%), they defined that the recycled filament with fillers had no significant differences in mechanical and thermal properties compared to virgin PETG (Kováčová et al. 2020).Nikam et al. (Nikam et al. 2023) investigated the feasibility of producing sustainable filaments for 3D printing (3DP) by recycling PET bottles.They obtained filaments with a consistent diameter and mechanical properties comparable to polymers such as polylactic acid (PLA).The results demonstrated that recycling PET bottles is a viable and environmentally friendly alternative, effectively reducing plastic waste and promoting sustainability in additive manufacturing.Furthermore, Nguyen et al. (Nguyen et al. 2024) studied the effects of printing parameters on the tensile strength and thermal stability of recycled PET and PETG, optimizing parameters using the design of experiments (DOE) method to obtain high-quality prints and concluded that these optimal conditions are promising for sustainable applications in various industries, contributing to the advancement of the circular economy.
Manufacturers, researchers, and consumers claim that PETG filament combines the best properties found in the filaments of the most consumed materials, PLA and ABS, being considered easy to print like PLA and at the same time having good mechanical strength, heat resistance, and flexibility like ABS. (Nikam et al. 2023).Therefore, to produce FFF filaments, a recycled raw material commonly employed in low-value applications, can undergo a transformation process to create a product of significantly increased value.Then, this study aims to investigate the recycling of PETG from face shields used during the COVID-19 pandemic to manufacture 3D printing filaments.Furthermore, we addressed the technical and economic feasibility of multiple PETG recycling cycles, evaluating its rheological, thermal, morphological, and mechanical properties employed to compare its properties to commercial filament.This approach offers a potentially sustainable solution for the reuse of plastic waste, in line with current environmental and economic sustainability imperatives.

Materials and processing
PETG, in the form of fragments, was obtained by cutting face shields used to reduce transmission during the COVID-19 outbreak.The PETG parts of these personal protective equipment were extruded using grade Cadence Copolyester GS2 material from EASTMAN.For comparative purposes, commercial PETG filament for AM was purchased from the manufacturer Sethi3D (Brazil), with a 1.75 ± 0.05 mm diameter.The PETGs studied have a cis/trans ratio of 30/70.
The PETG sheets used to produce the fragments had a protective film (liner) that kept the material free from dirt, thus eliminating the need for an additional washing step.Only the protective film was removed before grinding, and after removing it, the PETG fragments were ground into flakes in a knife mill (Wittmann).The resulting flakes were dried in an air circulation oven at 70 °C for at least 16 h.This process ensured that the PETG presented suitable flowing properties to be ready for extrusion.Figure 1 shows a schematic of the processing steps to obtain PETG filaments from PETG fragments.
The dried flakes were fed into a B&P Process Equipment and Systems MT19TC co-rotational twin-screw extruder (D = 19 mm, L/D = 25) with a water-cooling system.The diameter of the filaments was monitored through measurement with a digital caliper and controlled through the Fig. 1 Schematic illustration of the processing method adopted.*must be colored speed of the puller rollers aiming to fabricate 1.75 mm filaments.The extrusion process was carried out from one to five times to simulate recycling cycles, and before being fed back into the extruder, the filaments were processed in a pelletizing machine and dried in an oven using the same conditions described before.Filaments that passed through different reprocessing (REP) cycles were obtained; hence, the samples were named 1-5 REP.The extrusion conditions were kept constant for 1-5 REP, screw speed of 60 RPM, 5 heating zones (210, 220, 225, 230, and 235 ºC), and only slight modifications were made at die temperature (from 235 ºC in the first 2 cycles to 230 ºC for the others).Figure 2 shows the PETG filaments produced in the laboratory and the commercial filament, both used to print specimens.
The printer used was the Sethi3D S2 equipment, with a closed cabin and a brass hotend with a diameter of 0.4 mm.The printing parameters are described in Table 1.The printing conditions were chosen based on the properties of PETG and the need to guarantee highquality parts for the tests.The printing conditions were adapted from literature and based on PETG properties to ensure high-quality samples for testing (Kumaresan et al. 2023).A printing temperature of 245 ºC and bed temperature of 80 ºC were chosen to maintain optimal material viscosity, minimize warping, and ensure adequate adhesion of the initial layer.Additionally, using a 0.2-mm layer height and a print speed of 50 mm/s strikes a favorable balance between printing quality, low warping, adequate printing time, and reduced risk of failures.The rectilinear fill pattern, combined with a raster angle of ± 45º, maximizes mechanical strength and evenly distributes stresses along XY planes.Lastly, the XYZ orientation guarantees geometric consistency according to ASTM D638 Type I specifications, and employing 100% infill ensures solid and precise samples for mechanical testing.The G-Code for manufacturing the specimens was obtained using the Ultimaker Cura ® slicing software, and the specimens were manufactured according to the dimensions of ASTM D638 Type I (American Society For Testing And Materials 2014).
During the slicing of each specimen, eight points along the filament length were measured to correct filament diameter, and two cross-section measures were collected orthogonal to each other to evaluate filament circularity since slight diameter variations can influence printing accuracy.

Fourier-transform infrared spectroscopy
Post-consumer PETG fragments were submitted to Fourier-transform infrared spectroscopy (FTIR) for  qualitative analysis to determine and validate the chemical structure of the investigated polymer as PETG by identifying chemical bonds in the obtained spectra.
The sample used was a piece of PETG fragments with a square shape of 10 cm.The ThermoScientific equipment, model Nexus 4700 FTIR, with Attenuated Total Reflectance mode, was used for this test.The spectrum was taken with 32 scans, from 3500 to 500 cm −1 and 2 cm −1 of resolution.

Rheology analysis
Before the rheological tests, unprocessed PETG, filaments from 1 to 5REP, and the commercial were manually cut and dried in an air circulation oven at 70 °C for at least 16 h.Tests in steady state at low shear rates were conducted in an inert nitrogen environment using TA Instruments' AR-G2 controlled tension equipment.Parallel plates of 25 mm were used, with a gap of 1 mm.The temperature was 245 °C, equal to the printing temperature, and the shear rate range was from 0.01 to 100 s −1 .For rates between 300 and 30,000 s −1 , an INSTRON 4467 capillary rheometer was used, with a capillary of 0.762 mm in diameter and 26.162 mm in length.The shear rate on the inner wall of the capillary was determined using the Rabinowitsch correction for non-Newtonian fluids, according to Eq. ( 1): Where γw is the corrected shear rate, γmeasured is the apparent viscosity rate at the die wall, n is the power law index and b = 1/n.
In the AR-G2 equipment, rheological tests were also carried out in the oscillatory regime under the same conditions described for the permanent regime, except that the angular frequency ranged from 1 to 500 rad/s.In the same rheometer, the creep behavior of the materials was evaluated through a test under constant shear stress of 1000 Pa applied for 300 s while the deformation was measured.The materials' elastic recovery γ r was assessed by completely removing the shear stress and measuring the strain for 300 s.The γ r was determined by Eq. 2: (1) Where γ total is the strain measured after 300 s of stress application, and γ final is the strain measured after 300 s from unloading.

Differential scanning calorimetry
Differential scanning calorimetry (DSC) tests were carried out to evaluate the thermal transitions and the degree of crystallinity X c of the specimens in TA Instruments' Q2000 equipment.Differential scanning calorimetry (DSC) tests were conducted to assess the specimens' thermal transitions and degree of crystallinity (X c ) using TA Instruments' Q2000 equipment, which has a temperature accuracy of ± 0.1 ºC.The samples from 3D-printed parts using filaments from 1 to 5REP and commercial were cut, including the three regions shown in Fig. 3, i.e., a thin layer of the specimen's middle cross-section, with similar fractions of the base, middle, and top of the printed part.
This technique is relevant in this study since differences in heat flux can result in different values of X c depending on the height analyzed since it usually employs a heated bed during printing.The X c and thermal properties of the printed parts were evaluated by heating them from 20 to 300 °C at a rate of 10 °C/min.Equation (3) was considered for the determination of X c : Where ΔH m is the enthalpy of fusion, ΔH cc is the enthalpy of cold crystallization, and ΔH m 0 is the enthalpy of fusion of a theoretical sample of 100% crystalline PET, considered 140 J/g (Spinacé and Paoli 2001;Wunderlich 2003).

Scanning electron microscopy
The morphology of the contact surfaces between the adjacent depositions of the different layers and the dimensions of the material deposited in the AM process was studied by scanning electron microscopy (SEM) images.For this analysis, 3D-printed specimens with 1REP to 5REP and commercial filaments were subjected (2) to cryogenic fracture via submersion in liquid nitrogen for 15 min before fracture.The fractured surfaces were coated with gold and evaluated in a Philips scanning electron microscope, model XL-30 FEG.The interior of each sample was assessed using ImageJ software, and it was used to identify the perimeter of 45 deposited and obtain the measurements of interest: a larger dimension referring to the width of the ellipse (Y axis) and a smaller one referring to the height (Z axis).

Tensile testing
Tensile tests were performed on five specimens of each type of printed part (1-5 REP and commercial).The samples were previously conditioned under controlled temperature and humidity for at least 24 h according to the standard recommendation ASTM D638 Type I and tested using an INSTRON universal testing machine, model 5569, with a 5-kN load cell and a 5-mm/min test speed (American Society For Testing And Materials 2014).
For each group of specimens (1-5 REP and commercial filaments), the data referring to tensile strength σ max and modulus of elasticity E were submitted to the Tukey test to demonstrate the significant differences through the grouping classification.

Chemical investigation
The spectrum obtained from the FTIR analysis for the post-processed PETG is shown in Fig. 4. The absorption band at 1712 cm −1 is typical for polyesters and corresponds to the stretching of the C = O of the ester group (Chen et al. 2015a;Nicolino et al. 2020;Paszkiewicz et al. 2017).The less intense bands at 2930 cm −1 and 2855 cm −1 are attributed to the asymmetric and symmetric stretching of aliphatic C-H, respectively, present in the methylene group of PETG (Shirali et al. 2014;Chen et al. 2016;Wu et al. 2004).Regarding the 1,4-cyclohexanedimethanol terephthalate (CT) units of PETG, the peaks 1456, 1262/1246, and 957 cm −1 correspond to the C-H bending of the methylene group, the stretching of C(= O)-O, and the stretching of C-H of the cyclohexane ring, respectively (Chen et al. 2015a(Chen et al. , 2015b;;Lee et al. 2000).The bands with values of 1409 and 873 cm −1 are attributed to the aromatic ring's in-plane and out-ofplane C-H flexion, respectively (Chen et al. 2016).The spectrum at 1099 cm −1 is attributed to the symmetrical stretching of C-O (Chen et al. 2015b).Finally, the band at 727 cm −1 comes from the out-of-plane C-H flexion and C = O flexion (Chen et al. 2015b).The peaks were identified based on literature studies that performed the same analysis for PETG.In this way, it was possible to determine that the material of the flakes to be recycled is a PETG without any other polymer contaminations.

Rheological properties
Figure 5 shows rheological curves obtained in steadystate tests, and the power factor (n).The materials presented typical shear-thinning behavior, as Newtonian plateaus are observed at low shear rates and Fig. 4 FTIR spectrum of post-processed PETG residue pseudoplastic behavior at high shear rates.At low shear rates, the result obtained for the Flakes-Form PETG sample, before reprocessing, behaves similarly to the commercial filament PETG sample.Still, over the profile at low shear rates, there is a tendency for viscosity to drop with the number of processing cycles, and this effect can be attributed to the reduction in molar mass.The high temperatures and shear rates inherent to the processing cycles promote the thermomechanical degradation of the polymer, causing scissions in the polymeric chains (Cruz et al. 2017).Therefore, greater molecular mobility and less occurrence of entanglements are expected, justifying the greater fluidity of the most processed materials, which are still over the profile at low shear rates.
The curve obtained for high shear rates can be compared to the Power Law model, described by Eq. ( 4): In Eq. ( 4), η represents melt viscosity (Pa.s), m denotes consistency, γ ̇ denotes shear rate (s −1 ), and n is the power law index and indicates material pseudoplasticity.Logarithmically relating Eq. (4) to experimental data allows the determination of the n − 1 term from the slope in the shear thinning region and log m from its intersection with the vertical axis.
For the commercial filament, the relatively high value of n and the less steep curve in the shear thinning region, as seen in Fig. 5, indicate that this material has a less pseudoplastic behavior than the others. (4) This comparison with recycled material reveals the viscosity variation in the studied shear rate range.The pseudoplasticity behavior of the polymer increases with its molar mass distribution; therefore, the commercial filament is suggested to be composed of material with a narrower molar mass distribution than the recycled materials (Spinacé and Paoli 2001;Cruz et al. 2017).
The materials studied have similar viscosity in the shear-thinning region.This condition contributes to the fact that, during the AM process, where the material is subjected to shear rates between 1.000 and 1.600 s −1 along the 3D printer nozzle, there are no significant differences in flow that could interfere with the melt flow rate and the deposit stability (Sanchez et al. 2019).However, the AM process by material extrusion is influenced by the viscous and elastic components of the polymer.Therefore, the oscillatory rheology test results are presented in Fig. 6.For each measured frequency, the values of G′′ were higher than G′, confirming that the PETG acts as viscous liquids under the conditions tested (Cruz et al. 2017).
The G′ and G′′ curves did not cross during the angular frequency analysis, and the crossover must occur at higher frequencies, which suggests that the materials have a low relaxation time (Cruz et al. 2017).It is then assumed that the shear stresses these materials suffer at the exit of the printer nozzle are quickly relaxed and do not affect the deposition process.However, at the exit, the polymer still experiences elongational stress during deposition, and the material undergoes temperature reduction.Its viscosity may be high enough to impair Comparing the different samples studied, it is observed that G′ decreases the greater the number of processing steps PETG undergoes.This phenomenon can be attributed to the thermomechanical degradation of the material, which results in a reduction in entanglements and, consequently, in the elastic behavior (Cruz et al. 2017).This result reinforces what was observed through tests in a steady state.The commercial filament did not undergo reprocessing and showed a more pronounced elastic behavior (higher G′) than the other samples.With the increase in processing steps, a decrease in G′′ is also observed for all materials.
During the AM processing, the polymers underwent high shear rates, and the materials that present lower G′ tend to have lower accuracy.Due to the lower viscosity and G′ of the 1-5 REP conditions, the printing conditions needed to be fine-tuned to ensure printing fidelity without some problems associated with PET printing, i.e., stringing and warping.
Finally, the creep and elastic recovery curves of the samples studied are presented in Supplementary Figure S1.It was observed that the samples that underwent more processing cycles showed higher values of γ total , indicating prominent viscous behavior.This deformation suggests that the processing cycles caused degradation to the materials, reducing the size of the polymeric chains and thus reducing entanglements and contributing to molecular mobility.These results agree with the findings obtained by analyzing the G′ and G′′

Degree of crystallinity and thermal properties
Figure 7 presents the DSC test curves for the first heating of the additively manufactured specimens.Table 2 shows the values of X c and thermal properties obtained through the DSC heating curves, where T g is the glass transition temperature, T cc is the cold crystallization temperature, and T m is the melting temperature.While the commercial PETG printed part showed no evidence of crystallization, maintaining an amorphous structure, the printed parts from the other materials showed cold crystallization and melting events.
There is a tendency to increase X c with the number of processing cycles.This tendency can be attributed to the degradation suffered by PETG, as verified during the rheological tests.The reduction in PETG molar mass favors molecular mobility and structural ordering.Due to the sample collection method, it is reasonable to consider that the X c values obtained are average values, considering top, middle, and bottom layers, and not specific values for a particular region of the printed part, which may be subjected to a different thermal flow than another region.
The significant difference in ΔH m and ΔH cc values between 1 to 5 REP suggests crystallization during printing and induced cold crystallization during the DSC test.This behavior, potentially advantageous, could be a result of the molten filament fabrication process.The high cooling rate after filament deposition leads to a low crystallization rate, which could contribute to the maintenance of the defined computational modeling, good dimensional stability, and low displacement of the deposited layers.
The T g did not suffer significant variations, possibly because the mobility of the amorphous fraction, characteristic of the material that influences this thermal property, is not so restricted in the case of materials with relatively low X c .T cc , on the other hand, tends to decrease with the increase in the number of processing cycles, indicating that less energy is required for crystallization to occur.On the other hand, T m tends to increase because, as discussed, the processing cycles favor the crystallization of PETG.
Another factor observed during crystal melting is a double or bimodal fusion peak for the 4REP and 5REP specimens, with a lower temperature (T m1 ) and a higher temperature (T m2 ).Increasing the number of reprocessing leads to the appearance of the bimodal peak starting at the 4th processing step, representing the formation of less perfect crystalline arrangements at lower temperatures.
A few studies using semicrystalline PETG have reported this bimodal melting event.Badía et al. (Badía et al. 2009) simulated PET recycling, subjecting the polymer to five extrusion steps.They pointed out that the bimodal melting event and its peak are more comprehensive, which becomes wider as the number of reprocessing increases, suggesting the relationship between the greater range of thickness of the lamellae of the crystalline population and the function of repeated processing (Badía et al. 2009).In this study, PETG suffered degradation by chain scission, and the higher the number of processing steps, the higher the density of chain ends, which may act as impurities during the PETG crystallization process  and contribute to the formation of less perfect crystals, as indicated by the bimodal melting event and the observation of a T m1 peak, both representing less perfect crystal melting and, it can be inferred that the specimens made with the 4 REP and 5 REP filaments have crystallites with two predominant lamellae thicknesses.

Morphological characterization
Figure 8 shows SEM micrographs of samples after cryogenic preparation.Despite the predominant elliptical geometry, even with a 100% filling, voids are observed, and this effect is less intense for the specimen manufactured with commercial filament.This behavior can be explained by the variations observed in the diameter of the filaments manufactured in the laboratory.The irregular filament diameter can generate slight flow variation during polymer feed into the printer and contribute to significant variations in the shape of the deposits of the specimens manufactured with recycled filaments.Due to high adhesion (deformity), measuring the dimensions of a minimum acceptable number of deposits for the specimen manufactured additively with the filament extruded five times (5 REP) was impossible.In all the samples printed, no warping was observed, even using 100% infill, which is a condition that improved this phenomenon.
Table 3 shows the measurements at two perpendicular points on the filaments, taken every 3 min during the additive manufacturing process, and the average deposit dimensions of the printed specimens measured using ImageJ software.The increasing number of processing cycles contributes to a higher average of the dimension of the Y axis of the ellipse, ranging from 0.33 mm in the specimen recycled once to 0.43 mm in the specimen recycled four times.The increase in the width (Y-axis) of the ellipse can be attributed to the reduction of both the viscous and elastic components of PETG with increasing reprocessing numbers, as observed in the rheological results.Thus, in the molten state, the polymer has a decrease in its ability to sustain subsequent layers (more significant deformation of the circular shape) and an increase in molecular mobility, both of which are associated with a reduction in the viscoelasticity of the polymer, and this can compromise the dimensional stability of the part and generate variation from the part designed in the 3D model.
The increase in width provides a better condition for the molecular chains to diffuse between the layers if there are temperature and time conditions.The adhesion between the layers is the limiting factor for the mechanical performance of additively manufactured parts, and the higher the quality of this adhesion, the better the mechanical properties, which will be discussed hereafter.

Mechanical properties
Figure 9 shows the averages of ultimate tensile strength (σ max ) and Elastic Modulus (E).Table 4 presents the mean values of these properties and their respective groups obtained from the Tukey analysis.
The number of processing cycles for both properties directly impacts the mechanical properties.Regarding the σ max , the Tukey test proved the interference of the number of processing cycles and pointed out four distinct groups.The samples that presented the best performances were those manufactured with the commercial and 3 REP filaments, belonging to groups A and AB, respectively.The 1 REP, 2 REP, and 4 REP samples showed intermediate performance, all belonging to group B. The sample with the worst performance was the 5 REP, belonging to group C. The mechanical properties of the recycled material used in this work tend to increase with the number of recycling steps, reaching maximum performance after three processing cycles.
The processing cycles to which PETG has been subjected influenced its rheological properties and X c and, therefore, should originate variations in σ max and E. It was observed in the rheological, and DSC analyzes an expressive indication of molar mass reduction according to the increase in the number of processing cycles imposed on the PETG, a phenomenon inferred through the observation of the decrease in viscosity as a function of the shear rate in the Newtonian plateau (low shear rates), the reduction of G′ and G′′ in the dynamic rheological tests, and the increase of X c of the printed specimens.
Therefore, it is suggested that the main characteristics of the material, which influence the mechanical properties of the printed material, are the viscous flow (molecular mobility) and the crystallization of the polymer, both favored by the processing steps.Furthermore, polymer crystallization simultaneously with the process of diffusion of the polymeric chains between deposits (inter and intralayer), causes the molecules to diffuse to the surface of the crystals.When incorporated in the crystalline region, polymer molecule segments have reduced mobility due to the high intermolecular interaction forces.Thus, the increase in σ max from the first to the third processing cycle can be attributed to the rise in molecular diffusion due to the degradation of the material.However, from the fourth processing cycle onwards, the polymer chain's diffusion and entanglement cannot withstand applied loads, which justifies the reduced σ max of 4 REP and 5 REP samples.
The modulus of elasticity (E) tends to increase due to the increase in reprocessing, with the highest average being the specimens manufactured with the 5 REP filaments (Group A).Property E is related to the material's stiffness, and its increase may be associated with the degradation of the polymer and the tendency for high X c .In addition, the greater the contact area between deposits from different layers, the increase in bond strength between deposits (inter and intralayer) also requires adequate molecular diffusion, depending on temperature and time conditions.The SEM micrographs indicate that a greater number of processing cycles contributes to reducing the material viscosity and the filling of voids, evidenced by the larger dimension width (Y axis) of deposits and difficulty in identifying the shape of deposits due to the fusion between them.This behavior contributes to the adhesion between layers due to the greater contact area and molecular diffusion, and the 5 REP sample, which probably underwent the most degradation and reduction in molar mass, presented the highest surface area for filament bonding (Levenhagen and Dadmun 2017;Ko et al. 2019).
Additionally, during DSC tests, it was found that the X c achieved low values between 11.5 and 18.5%.The difference in X c of prints made with 1-3 REP materials is slight and within the DSC technique's standard deviation.For 4 REP and 5 REP, there was a tendency towards an increase in X c , which may have caused a reduction in σ max since it reduces the mobility of the polymer chains, hindering the diffusion between the deposits and decreasing the bond strength between layers and consequently its modulus of elasticity.Regarding the specimens manufactured with the commercial PETG filament, the average E appears to be like those manufactured with the 3 REP filaments belonging to group B, and a similar tendency was observed by Vidakis et al. (Vidakis et al. 2021), where it was observed that the modulus of elasticity reached their maximum value at the 3rd and 4th processing cycles, respectively.
The printed parts manufactured with the 3 REP material presented the best relationship between mobility and resistance the polymeric chains offer, and it is achieved when the polymeric chains have sufficient mobility to favor diffusion between the deposits but can withstand the stresses imposed on the part.This behavior confirms the fulfillment of one of the objectives of this study: to reuse a material that would possibly be discarded in a landfill to produce filaments destined for AM based on material extrusion with satisfactory characteristics for this application, that is, ease of material deposition, besides dimensional quality and good mechanical properties.

Conclusions
The primary objectives of this study were to investigate the feasibility of recycling PETG from face shields used during the COVID-19 pandemic into 3D printing filaments and to evaluate the material's properties after multiple recycling cycles.The results demonstrate that extrusion and additive manufacturing can effectively repurpose PETG into high-quality filaments and printed components, confirming that PETG fragments are a viable feedstock for such processes.Rheological tests revealed that increasing the number of extrusion cycles reduces the elastic behavior of PETG due to thermomechanical degradation, which decreases molar mass and viscosity while increasing molecular mobility.This degradation accounts for the observed decreases in storage modulus (G') and loss modulus (G′′) and the changes in T cc , X c , and T m in DSC tests.SEM micrographs indicated more elliptical fill lines in printed parts, reflecting increased mobility and decreased viscosity due to slight degradation, however without traces of warping.Despite these effects, PETG exhibited increased stiffness (E) and comparable maximum tensile strength (σ max ) to commercial PETG, particularly after three extrusion cycles.The findings suggest that PETG can be successfully recycled multiple times through thermomechanical processes without significantly losing mechanical properties or processability in FFF 3D printing.This method offers a sustainable alternative for transforming PETG waste into valuable products.However, precise control over filament diameter is crucial to avoid irregular geometries in printed parts, as observed in this investigation.In conclusion, this study demonstrates that recycling PETG from COVID-19 face shields into 3D printing filaments is technically feasible and economically viable.The findings validate PETG as a robust alternative feedstock for highvalue applications, aligning with the initial aim to explore sustainable solutions for plastic waste reuse in advanced manufacturing.

Fig. 3
Fig. 3 Cross-section strip taken from the tensile specimens (a) and the final sample for DSC testing (b).*must be colored

Fig. 5
Fig. 5 Viscosity as a function of shear rate for the PETG samples (a) and power law indexes (b).*must be colored

Fig. 7
Fig. 7 DSC curves of the printed samples for the second heat must be colored

Fig. 8
Fig. 8 SEM micrographs of samples with the yellow markings on the perimeters of the deposited a 1 REP, b 2 REP, c 3 REP, d 4 REP, e 5 REP, and f commercial.*must be colored

Table 1
Printing conditionsa Printing speed stands for the first layer, perimeters, and infill speeds

Table 2
Thermal properties of printed parts obtained

Table 3
Average dimensions of the ellipses Fig. 9 Average values of ultimate tensile strength (σ max ) and modulus of elasticity (E).*must be colored

Table 4
Average values of σ max and E and their respective groups were obtained in the Tukey test