Post-consumer poly (ethylene terephthalate) (rPET) from beverage bottles was used. Phosphoric acid (85% wt./mL), titanium isopropoxide (97%), ethanol (99%), ethylamine (70%), pyromellitic acid (96%) and zinc acetate were purchased from Sigma-Aldrich. Silver nitrate was obtained from Isofar. Two commercial dyes (red and violet) were purchased from Guarany Ind. e Com. Ltda and used as received.

The synthesis and chemical modification of titanium phosphate have already been described previously [30] but will be briefly mentioned as follows. Under vigorous stirring, titanium isopropoxide was added dropwise into the phosphoric acid at a 1:8 molar ratio. Sequentially, deionized water was added, and the system was heated to 120˚C and maintained at that temperature for 24 hours. Thereafter, the medium was washed until the pH reached 6 and then dried in an oven at 90˚C. TiP modification was sequentially conducted with ethylamine, pyromellitic acid, zinc acetate and silver nitrate. Intercalation was performed by using ethanol and water as solvents, according to the solubility of each precursor. In each step, the final product was labeled as TiP/EA (ethylamine), TiP/EA/AP (ethylamine/pyromellitic acid), TiP/EA/AP/Zn (ethylamine/pyromellitic acid/zinc) and TiP/EA/AP/Zn/Ag (ethylamine/pyromellitic acid/zinc/silver).

Colorless, transparent, and multilayer bottles were collected, washed, dried, cut into flakes and ground. A portion was preserved as a blank. The remaining flakes (10 × 5 × 0.20 mm) were extruded in a twin-screw extruder Teck Trill DCT-20, at 300 rpm, within aprocessing window from 90˚C to 250˚C, resulting in pellets. To improve the filler dispersion, a masterbatch was prepared using the initial flakes combined with fillers (pristine TiP and modified - TiPEAAPZnAg), resulting in a compound with an rPET/filler ratio of 95/5 wt.%. The processing was conducted in a Haake Rheocord 9000 mixing chambe, at 240˚C, 50 rpm, for 5 minutes. The material was cut into pellets. The composite was prepared by mixing processed rPET with masterbatch to attain a final composition with 1 wt.% of filler. The processing was using a Haake model Rheocord 9000 single-screw extruder at 30 rpm, with 4 heating zones (processing window of 210˚C - 240˚C), generating filament. For aging, polarized optical microscopy, dynamic-mechanical analysis and contact angle evaluations, film specimens (25 × 15 × 0.16 mm) were prepared using a Marconi press model MA 098/A, at 250˚C, under 6 tons of pressure, for 6 minutes.

Raman spectroscopy was performed using a DXR Raman Microscope equipped with a laser at a wavelength of 532 nm and a 50x lens, in the range of 200 - 4000 cm⁻¹. The vibrational modes of the samples were evaluated.

¹HNMRTD relaxation time distribution (T2H) was obtained using the Ostroff-Waugh (OW4) pulse sequence. The analysis was performed with a recycle time of 2 s and 30˚C. The results were expressed as distribution domain curves.

The rheology measurements were conducted using a TA rheometer, model AR-2000, with parallel plate geometry of 25 mm diameter, at 270˚C, under an inert atmosphere and oscillatory regime. A strain sweep test was performed at a frequency of 1 Hz to determine the region of linear viscoelasticity. The storage and loss moduli, as well as the complex viscosity, were assessed.

The film was exposed to UV lamps emitting a wavelength of 365 nm for 120 hours in a UV darkroom model SP-204. Before and after aging, the sample surface underwent FTIR analysis, using a Perkin-Elmer Frontier MIR/FIR equipment, at a spectral range of 4000 - 400 cm⁻¹, by attenuated total reflectance (ATR), using 60 scans and resolution of 4 cm⁻¹. To monitor stabilization, the carbonyl index was determined considering the carbonyl band at 1723 cm⁻¹, characteristic of the rPET ester group, and the reference band at 2962 cm⁻¹. The average of four replicates was considered.

POM was conducted using an Olympus SZH10 microscope, coupled with an HDCE-50B camera and LinKam hot stage. The film specimen was prepared by pressing the sample between glass plates under the action of heating. Before the crystallization process, photographs were taken with and without polarized light. After that, the sample was heated to 260˚C at 10˚C/min, held for 2 minutes to erase the thermal history, cooled to 200˚C and allowed to crystallize for 3 hours. Finally, the sample was cooled to room temperature and photographs were taken again with and without polarized light.

DMA analysis was carried out using a TA Instruments Q800 instrument in tension film mode with a rectangular specimen measuring 9 × 3.2 × 1 mm. The analysis was conducted from 25˚C to 200˚C, at a heating rate of 3˚C min⁻¹ and frequency of 1 Hz. Storage modulus (E’), loss modulus (E”) and tangent delta (tan δ) were determined.

The contact angle was measured using OCA 15 EC equipment. Water was the test liquid with a dosing rate of 0.5 μL/s. The contact angle was recorded, and its value was calculated by a computational program. The result represents the average of three determinations.

Before infrared analysis, 5 grams of each dye were dissolved in 50 mL of acetone and allowed to stand for 24 hours. After that, the solution was filtered, and the colored liquid was subjected to FTIR evaluation. The infrared analysis was conducted in a Perkin-Elmer Frontier model equipment, from 4000 - 500 cm⁻¹, with 60 scans and a resolution of 4 cm⁻¹. The resultant spectrum was compared with those in the equipment’s spectrum collection.

The rPET and composites were dyed. Approximately 1.3 g of each sample was packed into a polyamide bag. On a hot plate, a three-neck flask equipped with a reflux apparatus was filled with 300 mL of distilled water and a packed sample. The mixture was heated to 60˚C under magnetic stirring at 175 rpm. The device was maintained for 10 minutes. After that, 8 grams of dye were added. The mixture was kept for an additional 10 minutes. The temperature was gradually raised to 100˚C and the mixture was kept at this temperature for 40 minutes. The system was allowed to cool down to ambient temperature. Finally, the packed sample was removed, washed with distilled water, and dried in an oven until a constant weight was achieved. The degree of fading was evaluated according to AATCC 16-2004 using an SDL-Atlas Ci 3000 + Xenon Fade-Ometer with a wavelength of 420 nm, at 1 rpm, and relative humidity of 31.6%, using a gray scale for color assessment.

Figure 3 presents the distribution domain curves of the samples. The rPET showed a rigid domain at 0.031 ms and a mobile domain at 0.167 ms. In rPET/TiP, both domains were observed to be displaced along the time axis. The lower domain increased the rigidity while the higher domain increased molecular mobility. The rPET/TiPEAAPZnAg revealed distribution domain curves very similar to those of rPET, but the slight narrowing at domain curves could indicate better affinity between polymer and filler. Mazur et al. studied the effect of commercial dimethyl dihydrogenated tallow ammonium cations-modifying montmorillonite (Cloisite® 20A, C20A) at a content of 5 - 10 wt.% in blends of poly (butylene succinate) (PBS)/poly (lactic acid) (PLA) [44] . Although with a high C20A content, the proton spin-lattice relaxation time (T₁) of the composite indicated a slight variation of the blend’s molecular relaxation. Using time domain low field NMR, Jing et al. investigated the molecular mobility of poly (vinyl butyral) (PVB) plasticized with triethylene glycol bis (2-ethylhexanoate) (25 wt.% of TEG-EH) [45] . The evaluation of T₂ relaxation time showed a considerable increase in PVB molecular dynamics, ascribed to the plasticizing effect. The limited influence of the fillers on the molecular mobility of rPET composites could be attributed to ethylamine (a low-chain amine), which restricted the expansion of TiP interlayers, and to the low filler content (only 1 wt.%) used in the composite preparation.

Figure 4 represents the evolution of storage modulus (G’), loss modulus (G”) and complex viscosity (η*), respectively. G’ showed different behaviors above and below 2 × 10¹ rad/s. Below this threshold, the curves of rPET and rPET/TiPEAAPZnAg are practically superimposed. The curve of rPET/TiP showed a lower G’ value. Above that limit, the curves showed a trend, indicating the rearrangement of fillers in the rPET/TiP composite promoted by the oscillatory work during the test. For G”, all curves are superimposed indicating no apparent influence of composite processing and filler addition on the degradation of rPET chains. For all curves, complex viscosity (η*) exhibited pseudoplastic behavior. Below 2 × 10˚ rad/s, the rPET and rPET/TiPEAAPZnAg curves were nearly superimposed. The rPET/TiP curve displayed a lower value of η* due to the arrangement of filler in the composite. Above 2 × 10˚ rad/s, the curves tended to overlap. Table 1 shows the behavior of G”/G’ at different frequencies. The values of G”/G’ converge at both frequencies with the profiles of G’, G”, and η*. At 10˚ rad/s, for rPET and rPET/TiPEAAPZnAg, the G”/G’ values were very close, while for rPET/TiP, G”/G’ was too high due to the reduction of G’. At 10² rad/s, all G”/G’ values were similar. Chowreddy et al. studied recycled PET/carbon nanotubes composites [46] . They observed an increase in G’ and η* with increasing filler concentration. It was deduced that the filler acted by hindering the flow of the PET and, therefore, increasing the complex viscosity. In another article, Chowreddy et al. prepared a nanocomposite of recycled PET/clay [47] . At a concentration of 1%, little variation of G’, G’’ and η* was registered. According to Litchfield and Baird, the aggregation of the filler’s particles increases the distortion of the polymer flow lines, causing huge effects on the rheological properties [48] . Exfoliated or well-dispersed nanoparticles exhibit different rheological behavior compared to their agglomerated counterparts. Herein, the materials exhibited different rheologic behavior at low and high frequencies. At low frequencies, G’ and viscosity curves behaved according to the dispersion and distribution of each filler in the composites. At high frequencies, the rheologic behavior tended to be similar.

Figure 5 displays the carbonyl index measurement. Although variations could be detected, they were within the experimental error, indicating stabilization in the applied conditions. Debska et al. investigated the effect of accelerated aging (UV radiation of 340 nm, 63˚C, simulating rain) on the physical and mechanical properties of composites based on plastic waste including recycled PET from beverage bottles [49] . They reported that until 500 hours of exposure, no color change or remarkable mechanical properties were achieved. Pires et al. investigated the action of nano zinc oxide (nZnO, 0 - 3 wt.%) as a shield for ultraviolet radiation of recycled PET/polycarbonate (rPET/PC, 80/20 wt./wt.%) blend [50] . The PET and PC carbonyl indexes suggested that the UV damage was attenuated in the sample containing the highest content of nZnO. This result could suggest that the presence of TiP and TiPEAAPZnAg did not induce notable changes in rPET chains during processing.

Figure 6 shows POM images before and after crystallization, with and without polarized light. The sample film (sandwiched between sheets of glass) was heated to 260˚C, left for 1 minute to erase the thermal history, cooled up to 200˚C and held for 180 minutes. After that, the sample was cooled until 25˚C. For all samples, before and after the application of the crystallization protocol, the optical images seem to be similar. Viora et al. studied the crystallization of virgin and recycled PET [51] . POM images revealed variations in organizational spherulitic crystals between both PETs. For recycled PET, the final diameters of the spherulites were on the order of magnitude of 4 μm or even less. The result was associated with factors such as contaminants and shorter chain lengths, which limited the growth due to numerous active crystallization sites. In the study of the action of carbonate salts in the rPET crystallization kinetics, Xanthos et al. reported that PET has a low crystallization rate (10 µ/min) and nuclei formation when compared to polyethylene (5000 µ/min) [52] . Calcagno et al. investigated the effect of four types of montmorillonite—(2-5 wt.%, Cloisite: Na, 15 A (cationic modifier C18), 30 B (cationic modifier C16) and 10A (cationic modifier C14)—in nanocomposites based on PET [53] . They detected higher crystallization rates and smaller spherulites’ size for nanocomposites than for pure PET. Through the POM analysis, no variation in nucleation and crystal size of rPET was observed. It was believed that ethylamine (a low chain amine) acted as a small intercalator for TiP interlayers, affecting the polymer/filler interaction to some extent. Also, the amount of filler (only 1 wt.%) could have contributed. This result corroborated the time domain NMR analysis.

Firstly, some considerations about the chemical nature of dyes are provided. Figure 9 shows the infrared spectra of red and violet dyes. According to the equipment’s spectrum collection, the red one presents similarity to the coumarin family and the blue one to fluorescein. The lightfastness test is appropriate for evaluating textile color retention (degree of fading). The grey scale of the lightfastness is rated between 1 - 5. Low grade (1) means that higher degree of fading while on the contrary, a higher value of lightfastness (5) represents excellent color retention [56] . Figure 10 shows the samples after incorporation of red and violet dyes. Table 2 shows the lightfastness measurement of rPET and composites. For all samples, there was no variation in the degree of fading for both types of dyes. Gashti et al. reported that polymer substrate, physical state and concentration of dye and the presence of UV absorbers are factors which could affect the lightfastness of dyes [57] . Summarizing, it was possible to infer that the color retention could be attributed to some intermolecular interaction among polymer, filler, and dye.