Development of a continuous PET depolymerization process as a basis for a back-to-monomer recycling method

1.1 Chemical recycling pathways

Pure and slightly colored PET waste can be recycled thermomechanically, but the recycling of multilayered and highly colored complex PET waste is still a major challenge in waste processing [7].

Back-to-monomer recycling (BMR) offers, in contrast to pyrolysis, an opportunity to overcome this issue by producing the corresponding monomers through depolymerization rather than the production of bulk petrochemicals [8].

BMR is possible via different reaction pathways displayed in Figure 1: hydrolysis, methanolysis, glycolysis, or aminolysis, and ammonolysis [9]. The hydrolysis of PET is either conducted in water using high temperatures, high pressure, and catalysts or by strong acids and bases [10,11,12,13]. Water acts as a weak nucleophile and breaks the ester bonds to form TA and EG [14]. Alternatively, strong bases, such as sodium hydroxide, can be used to attack the ester linkages to form the corresponding monosodium terephthalate salt and disodium terephthalate salt (DST) and EG [15]. The monomer TA is recovered through the treatment of DST with a strong acid, such as hypochloric or sulfuric acid [7,15]. Several processes are based on alkaline hydrolysis. However, sodium hydroxide concentrations of up to 20 wt%, high temperatures of up to 250°C, and pressures between 1.4 and 2 MPa are mostly used for the depolymerization of PET [16,17].

Figure 1

Different methods of PET chemical recycling paths, adapted from ref. [29].

During methanolysis, which has been thoroughly investigated by several groups, a transesterification reaction occurs, and dimethylterephthalate (DMT) and EG are produced [18,19,20,21,22]. This transesterification is usually catalyzed by sodium carbonate or zinc acetate [23,24]. This approach to PET recycling suffers from the formation of a mixture of several glycols, alcohols, and phthalic acid derivatives [22]. Downstream efforts to isolate and purify DMT diminish the economic viability of this process. However, methanolysis and glycolysis can be performed economically when pure PET waste is used as feed material, such as production waste [22]. During glycolysis, EG is used as a solvent and breaks the ester bonds of the polymer at high temperatures. This reaction is usually catalyzed by a transesterification catalyst, such as sodium carbonate, cobalt acetate, lead acetate, manganese acetate, and zinc acetate [20,25,26]. The reaction products include bis(hydroxyethyl)terephthalate (BHET), EG, and varying amounts of oligomers. This process benefits from the use of EG as a solvent, which is also produced during the process. The reaction rate of the depolymerization reaction has been found to be directly proportional to the polymer surface area [20]. This indicates that either a smaller polymer particle size or more vigorous mixing can lead to a more effective reaction. The most important disadvantages of methanolysis and glycolysis are the use of catalysts, high pressure and high temperatures during the process, and the resulting oligomers, which are difficult to separate from the products [22].

During aminolysis, PET is depolymerized with amine solutions, such as methylamine, allylamine, hydrazine, or ethanolamine [27]. During the process, bis(2-hydroxyethylene)terephthalamide (BHETA) is formed [27]. Alternatively, PET can be depolymerized by ammonolysis, where PET is treated with an aqueous ammonia solution to form terephthalamide [28].

A more detailed description of the advantages and disadvantages of the different approaches has been presented in several review articles [23,30].

In contrast to the known processes in the literature, the depolymerization is conducted in this study as a continuous process at ambient pressure, with short residence time, without catalyst, and at moderate temperatures.

1.2 Mechanism of PET depolymerization

Regarding the PET depolymerization mechanism, several research groups have investigated the reaction behavior of the solid/liquid interface between PET and the reaction solution. Yoshioka et al. investigated PET degradation in nitric and sulfuric acid and proposed a reaction mechanism for initial material degradation [31,32]. It is assumed that the reaction occurs on the surface of the material via a modified shrinking-core mechanism, where pores and cracks are formed during the reaction.

López-Fonseca et al. investigated the depolymerization mechanism of PET in aqueous media in the presence of a catalyst. They observed a lamellar reaction behavior on the external surface of the PET particles [33]. By contrast, Oku et al. reported that the reaction of PET with sodium hydroxide in EG appears to be a two-stage process with a fast initial reaction step and a slow second stage. They proposed that reaction products, such as DST, are formed on the surface of the PET particle and limit the rate of reaction [15].

Even though different reagents were used in the aforementioned studies, the results suggest that mechanical stress and constant renewal of the PET particle surfaces lead to a higher reaction rate.

1.3 Chemical depolymerization as a continuous process

While laboratory investigations on reaction pathways and kinetics are typically conducted in batch experiments, a transfer to an industrial application will benefit from a continuous operation. Given the enormous quantities of the PET recycling market, developing a continuous process seems mandatory from both an economic and ecological perspective. Initial studies have shown that the chemical recycling of PET has to overcome economic issues. In their study, Aguado et al. demonstrated that, from an economic perspective, wastewater management has a large impact on the feasibility of the recycling process in the alkaline hydrolysis of PET waste [7].

The processes described earlier show a reaction time ranging between 6 min and 70 h, typically achieved in batch processes [16]. Continuous production processes offer many advantages, especially with respect to constant product quality, automated operability and control, energy, and resource efficiency, as well as scalability. As shown for polymerization, batch processes have several disadvantages, such as high operating costs due to product changeover, including intensive cleaning procedures and the risk of poor batch-to-batch consistency. Furthermore, batch processes are difficult to scale up due to the limited capability of heat transfer, which requires low reaction rates and results in longer batch times [34].

In the past, few attempts have been made to implement a continuous process with short reaction times for the recycling PET scrap material. Ostrowski described the continuous depolymerization of polyester at atmospheric pressure in EG at 230°C. The reaction was performed in a fed-batch reactor and was operated continuously [35]. The residence time was 1.5–3.6 h, and the resulting product was a mixture of various oligomers. Bergmann et al. reported a method to glycolyze PET waste with EG in a twin-screw extruder to produce oligomers with short chain length. The average residence time of the reaction mixture varied between 1 and 5 min. However, these investigations focused on online monitoring of the reaction rather than the recycling of the PET material [36].

The first experimental realization of continuous depolymerization of PET scrap material with sodium hydroxide to form EG and disodium terephthalate was described by Benzaria et al. in the patent US5545746 [37]. The reaction was performed in a kneading extruder with an approximately 6 min residence time at 120–160°C. They observed a depolymerization degree above 60% and could recover the EG formed directly during the depolymerization. Even greater depolymerization degrees >95% were achieved when a second downstream process step was introduced, where the raw product was tempered at 80–130°C for 30 min [37]. This approach presents a distinct improvement over the typically conducted batch reactions because of the rather short reaction times, the low temperatures, and solvent-free conditions [37]. The authors hypothesized that the mechanical stress in the kneading extruder accelerated the depolymerization reaction. The investigated process concept presented here follows a similar approach to the depolymerizing PET with a strong base to produce TA and EG. The authors developed a novel concept for process intensification for PET depolymerization. A twin-screw extruder was used to achieve more rigorous mechanical stress. This led to a high degree of depolymerization, with a residence time below 1 min. The “hydrolysis” route, as displayed in Figure 1, benefits from the fact that the resulting TA can be used without further processing for the production of PET in contrast to DMT and therefore functions as a “drop-in” building block for the current PET production [38].

The global increase in plastic production and the consequently generated PET waste have led to the need for new approaches for plastic waste recycling since the state-of-the-art technology is not capable of closing the material loop. In the past few decades, many studies have described the depolymerization of PET and the regain of the monomers TA and EG. In this study, the development of a continuous process to solve the issues associated with the various approaches for the chemical recycling of PET is proposed.

2.1 Process concept and experimental setup for the continuous depolymerization of PET

2.1.1 Process concept

In this approach, the alkaline hydrolysis of polyesters is investigated as a method to recover TA and EG from PET scrap material in a continuous process. The depolymerization of PET is conducted in a three-phase system of solid PET, the solvent EG, which is produced during depolymerization, and the corresponding base solid sodium hydroxide. The underlying chemical reaction is displayed in Figure 2.

Figure 2

Reaction scheme of the PET depolymerization.

For compounding applications where mechanical stress and mixing are key requirements, twin-screw extruders are frequently used [39]. Therefore, such a device was selected to ensure high mechanical stress on the starting materials, as well as distinct mixing.

2.1.2 Experimental setup

The depolymerization experiments were conducted on a Leistritz, Nuremberg, ZSE 27 MAXX twin-screw extruder with a k-tron doser supplied by Coperion GmbH, Stuttgart. The screws had a diameter of 28.3 mm and an L/D ratio of 60. The twin screws consisted of 82 segments, which could be selected as mixing, holding, or kneading elements to create high shear stress and residence times. The screw design provided several reaction zones where the reaction mixture was intensely blended and put under mechanical stress with kneading elements. The PET material was dispensed in the main shaft using a gravimetric doser by k-tron (K-CL-SFS-KQX4), as displayed in Figure 3. Solid sodium hydroxide was fed through a side feed LSB27 in cylinder 4 and dispensed with a gravimetric doser (K-SFS-I 240 N). The chemicals as well as the extruder screw were flushed with N₂. Water was dispensed in cylinder 12 with a gravimetric pump (K-CL-SFS-P).

Figure 3

Configuration of the reactive extrusion process.

2.2 Reagents and PET samples

2.3 Experimental procedure and parameters

2.4 Data extraction and analytics

3.1 Depolymerization of PET (light blue)

With the setup displayed in Figure 3, the influence of rotational speed on the degree of depolymerization was investigated. The rotational speed varied from 200 to 400 min⁻¹. At each operating point, the residence time was measured, and a sample was taken to determine the degree of depolymerization. As depicted in Figure 4, a depolymerization degree above 97% was achieved in all experiments, with the highest degree of depolymerization observed at a rotational speed of 300 min⁻¹. At a rotational speed less than 200 min⁻¹, the feed rate exceeded the extruders’ conveying capability and thus the torque limit of the device.

Figure 4

Influence of rotational speed on the degree of depolymerization of b-PET and p-PET calculated according to Eq. 4.

The characteristic residence times in the extruder are displayed in Figure 5 and Table 1. All residence times were reduced with the increasing rotational speed. Moreover, the plot shows that the residence time distribution broadens with a reduced rotational speed. These results are in line with the assumption of plug flow in an extruder, as suggested by Shon et al. [47]. The rotational speed of 200 min⁻¹ is taken as a reference, with a degree of depolymerization of 97.3% at a residence time at color maximum of t _max ≈ 58 s.

Figure 5

Influence of rotational speed on characteristic residence times of the extrudate for b-PET.

According to Oku et al., the PET depolymerization reaction is a two-stage process in which the lower rate constant represents the coverage of the surface with disodium terephthalate [15]. The results presented here suggest that better mixing and higher shear stress on the material can compensate for a lower residence time in the twin-screw extruder.

In consideration of the results displayed in Figure 5, residence time cannot be the only factor influencing the degree of depolymerization. The residence time t _max was reduced by 38% when the rotational speed was doubled from 200 to 400 min⁻¹, but the degree of depolymerization was reduced by only 0.3%. High shear stress and mechanical work on the polymer from the extruder screws led to higher depolymerization of the PET material at higher rotational speeds through faster renewal of the PET reaction surface. In addition, the intense mixing of the extruder led to a constant renewal of the reaction mixture in the proximity of the surface of the PET particles, resulting in a very high PET turnover at a residence time of 36 s at 400 min⁻¹.

3.2 Depolymerization of PE/PET material

In addition to light blue PET flakes, a PET composite material consisting of PET and PE was used as a feed material for the depolymerization process. The results, displayed in Figure 4, clearly show that the depolymerization of PET is not affected in great measure and that high yields of TA are obtained even though the PET material is covered on one side with a quasi-inert layer and the residence time in the extruder is very short. This might be due to a delamination process of the p-PET flakes through mixing and high shear stress in the extruder.

According to Table 1, the residence time t _max of the p-PET material is 6 s lower at a rotational speed of 200 min⁻¹ and 4 s lower at a rotational speed of 400 min⁻¹, which might be due to the content of the inert PE material in p-PET (see Figure 6).

Figure 6

Influence of rotational speed on the characteristic residence times of the extrudate for p-PET.

3.3 Comparison of the residence times of b-PET and p-PET

To compare the residence times of the two PET materials, the width of the residence time was calculated by the time of maximum coloration (see Eq. 2). These residence time coefficients are shown in Figure 7. For all experiments, the normalized residence time for p-PET was larger than for b-PET. At a rotational speed of 400 min⁻¹, the coefficient of the p-PET was 28% higher than the coefficient of the b-PET. During the depolymerization of the p-PET material, the PE flakes were maintained, resulting in a broader residence time distribution.

Figure 7

Residence time coefficients of b-PET and p-PET.

3.4 Characterization of depolymerization products

The extruder output was treated as described in Section 2.4.1 to receive TA and unreacted PE and PET residues. Figure 8 displays the absorbance of the TA samples from 330 to 350 nm. It is clearly recognizable that the postproduction waste material p-PET results in lower absorption values, indicating a higher TA quality. The absorbance values of TA from b-PET are higher, presumably due to the content of blue particles and the fact that b-PET is a postconsumer material.

Figure 8

UV/Vis-spectra of the obtained TA in DMF (50 g/L) solution at a wavelength of 330–350 nm.

Furthermore, a higher rotary speed resulted in a higher absorbance, which indicated the formation of undesired colored contaminants at high rotary speeds.

As a representative example, Figure 9 and Table 2 present the results of the NMR analysis of p-PET at a rotary speed 200 min⁻¹. The ¹H-NMR-spectra in Figure 9 show that a mixture of the isomers TA and IPA has been isolated. No EG residues or CH₂ groups are present within the detection limit of approximately 1% of the NMR spectra. The depolymerization of PET to TA/IPA and EG seems to be complete since no oligomers, such as corresponding CH₂-groups, have been observed. Since recycled bottle flakes have been used as a feed material, an IPA content of 1–5 (w/w)% was expected. Usually, IPA is added during the production of PET to influence its crystallinity and mechanical properties [48].

Figure 9

¹H-NMR-Spectra of the obtained TA/IPA mixture from p-PET material.

3.5 Comparison of recycled TA vs standard commercial TA

All TA samples were characterized by IR spectroscopy, and the obtained spectra were compared with the commercially available TA standard (see Figure 10). Obviously, the IR spectrum of the recovered TA matches very closely with the corresponding spectrum of the commercial standard. At 3000–2900 cm⁻¹, the C–H phenyl stretching band was observed, and at 1668 cm⁻¹, the intense C═O-group stretching band indicated the presence of a carbonyl group. At 1422 and 878 cm⁻¹, OH-group bands were visible, as described by Téllez et al. [49]. The absorption peak at 783 cm⁻¹ proves the paraposition of the carboxyl groups in the obtained terephthalic acid [45].

Figure 10

FT-IR-spectra of the obtained terephthalic acid in comparison with the terephthalic acid standard.

Therefore, it was concluded that the obtained product was mainly monomer terephthalic acid with a small amount of isophthalic acid. In further experiments, the suitability of terephthalic acid for polymerization with EG to PET will be tested.

3.6 Characterization of the depolymerization residue

The residues of Experiments 1 and 2 were analyzed by IR spectroscopy, and four intense bands were detected (see Figure 11). Between 3000 and 2800 cm⁻¹, the absorption bands of the CH₂-groups were observed at 1470 cm⁻¹, and the deformation band of CH₂-groups and the CH₂ rocking band were observed at 715 cm⁻¹. These four characteristic intense bands show that the residue mainly consists of PE [50,51]. In comparison, the FTIR-spectrum of the starting material p-PET shows characteristic peaks for PET, such as the C–H phenyl stretching band at 3000 and 2900 cm⁻¹ and the carbonyl-group band at 1713 cm⁻¹. Polyolefinic constituents of PET-containing waste appear inert under the given PET depolymerization conditions and may be separated after PET depolymerization for further processing.

Figure 11

FT-IR-spectra of the residue of the depolymerized p-PET starting material.

3.7 Classification of the results in the context of a concept for a PET-recycling process

Based on the presented results for the successful continuous depolymerization of bilayer p-PET waste, a process concept for a complete PET-recycling process may be derived. The high degree of depolymerization in the extruder within a few minutes thus provides a good starting point for the recovery of TA. The concept of recycling technology consists of seven process steps for the production of TA from waste PET material (see the process block diagram in Figure 12).

Figure 12

Block diagram of the continuous back-to-monomer recycling process.

In the first step, depolymerization is conducted in a twin-screw extruder as a continuous reactor with an average residence time in the order of minutes. Twin-screw extruders are widespread machines in the plastic industry. Due to the modular design, extruders offer a wide range of options for especially adapting the thermal and mechanical conditions for processing the respective material.

In the next process step, disodium terephthalate in the extrudate is completely dissolved in water. The subsequent filtration and optional discoloration step, depending on the feed material, purify the solution from nonreacted and undesired compounds. In the following precipitation step, the TA is precipitated through the addition of a strong acid. The suspension is filtered, and the resulting TA is washed, recrystallized, and dried. EG can be recovered during the depolymerization process or following the purification of the filtrate.

With this process, it is possible to recover TA and EG from hardly recyclable PET waste and to close the raw material cycle. This innovative technological approach thus represents a promising possibility for the creation of a circular economy for PET waste. In future studies, this technological development should be complemented with economic and ecological investigations of the presented recycling technology.