Solid waste contamination of non-degradable PET products, such as bottles, textile fibers, and films (Fig. 1a), has raised widespread environmental concerns15. Physical recycling technologies, e.g., melt extrusion, often decrease the molecular weight of PET and downgrade product quality, failing to establish a closed recycling loop (Fig. 1b). This is exemplified by the existing recycling chain, starting with bottles, fibers and wood plastic composites but ultimately leading to unusable trash16-18. To address this, chemical upcycling emerges as the most promising approach to achieve high-value, close-loop recycling of waste PET19,20. This approach primarily relies on reversible chain-exchange reactions (Fig. 1c) based on ester bonds in PET using small molecular amines, water, or alcohols9,21. These reactions encompass amino-ester exchange (AEE) and hydroxyl-ester transesterification (HET), with the latter being well-established for PET synthesis. Recently, the concept of carboxyl-ester transesterification (CET) has been introduced for polycondensation synthesis of polyesters and shows potential for PET conversion in the presence of carboxylic acids22-24. Nevertheless, these exchange reactions, along with a side reaction generally known as ester-ester transesterification (EET)25,26, can result in uncontrolled breakdown of the PET polymer chain due to an unbalanced ratio of diol/diacid units27-29. While this chain breakdown allows for PET depolymerization into its monomers30,31, e.g., terephthalic acid (TPA), ethylene glycol (EG), and their derivatives, from which high value-added plastics, particularly biodegradable polymers, can be resynthesized1,32,33, it multiplies the procedure and suffers from low yield and selectivity of monomers due to the difficulty in extracting monomers from the multimers produced34,35. Consequently, the complete utilization of PET remains a challenge. An appealing alternative is to achieve the direct polymer-to-polymer conversion from waste PET to biodegradable polymers; however, identifying an efficient yet closed-loop pathway presents a significant obstacle.
We noted that the reverse processes of the HET, CET, and EET hold the potential to propagate the molecular chain of product25, provided that the small molecules can be artificially eliminated from the reaction system. Accordingly, we proposed integrating the depolymerization and repolymerization of PET into a single procedure by combining these three transesterifications, enabling direct polymer-to-polymer conversion of waste PET into biodegradable plastics (Fig. 1d). To achieve this, we devised two parallel routes where waste PET was simultaneously reacted with commercially available aliphatic diacid and diols (denoted as X-monomers, Fig. 1e), e.g., EG, 1,4-butanediol (BDO), as well as succinic acid (SA), and adipic acid (AA), through the identical process employed in PET synthesis9. In the first stage, we employed the forward forms of HET, EET, and CET to insert X-monomer units into the PET chain, yielding hydroxyl- or carboxyl-terminated prepolymers. In the second stage , we applied the reverse HET or CET to propagate the molecular chain of the prepolymers under high temperature (260-280 °C) and reduced pressure (<100 Pa). This stage removed the excess end-group units, thereby achieving the required stoichiometric diacid/diol ratio and affording high-molecular-weight (HMW) biodegradable products named “PEXT”.
We observed a significant decrease in the intrinsic viscosity ([η], 0.80 dL/g) and viscosity-average molecular weight (Mη, 37.0 kDa) of bottle-derived PET upon reaction with SA/1.1 equiv. EG (Fig. 2a) or EG/1.1 equiv. SA (Fig. 2b), resulting in the poly(ethylene succinate-co-ethylene terephthalate) (PEST) prepolymers with [η] of 0.07-0.47 dL/g and Mη of 1.2-18.0 kDa. Despite this, the ratios of PET unit/poly(ethylene succinate) (PES) unit, marked as n(PET):n(PES), closely matched the starting feed ratios ranging from 8:2 to 2:8 (see Supplementary Table 1 and Figs. 1-28). Encouragingly, applying reverse HET or CET to remove the corresponding excess EG or SA led to a substantial increase in [η] (0.57-1.07 dL/g) and Mη (23.3-54.9 kDa) for the PEST products. Moreover, by maintaining the diol/diacid ratio at 1.1:1 (Supplementary Tables 2 and 3), substituting SA with AA or EG with BDO also yielded HMW poly(ethylene adipate-co-ethylene terephthalate) (PEAT, 27.0-50.6 kDa) and poly(butylene succinate-co-ethylene terephthalate) (PEBST, 28.6-67.7 kDa) products following a brief oligomerization stage (3.1-11.2 kDa). Notably, the values of n(PET):n(PEA) or n(PET):n(PBS) in the products remained consistent with the initial feed ratio (see Supplementary Figs. 29-56). Comparatively, the products obtained by melt-mixing PET and the forward HET-synthesized PES polymer experienced a considerable reduction in Mη (5.1-25.9 kDa, Fig. 2c), with n(PET):n(PES) comparable to those of the feed (Supplementary Table 4 and Figs. 57-65). This finding proves that conventional physical blending suffices for chain depolymerization due to the occurrence of EET25. In contrast, repolymerization of the blended products using reverse HET or CET effectively improved their Mη to a high range of 27.8-64.4 kDa (Fig. 2c, Supplementary Table 4 and Figs. 66-72).
We further verified the insertion and propagation of the molecular chain through detailed analyses on the product composition. To exemplify, in the case of PEST using excess EG and n(PET):n(PES) of 5:5 (Fig. 2d), the content of the TPA-EG-SA segment equaled that of the TPA-EG-TPA and SA-EG-SA segments. The total diol:diacid values (>1) revealed the hydroxyl-terminated structure of both the prepolymer and product (see also Supplementary Figs.1-14). This is different in the case of excess SA, where TPA-EG-SA in the prepolymer and product were half of the total amount of other two segments, but the ratios of (SA-EG-SA)/(TPA-EG-TPA) and total diol/diacid were both less than 1, indicating the chain propagation by removing SA resulted in the reduction of SA units and the formation of carboxyl-terminated products (reconfirmed in Supplementary Figs. 15-28). These results aligned with the NMR spectra of PEAT and PEBST (Supplementary Figs. 29-56). In comparison, the TPA-EG-SA segment in the PET/PES blend only accounted for 62% of the total amount of TPA-EG-TPA and SA-EG-SA segments in Fig. 2d. Nevertheless, following repolymerization, these three types of segments in the product regained similar levels as the corresponding cases of excess EG or SA (also refer to Supplementary Figs. 57-72). Moreover, attenuated total reflectance-Fourier transform infrared (ATR-FTIR) spectra disclosed heightened hydroxyl peaks (3400-3500 cm-1) in PET, PES, and their blend products compared to other PEXT products (Supplementary Figs. 73-79), suggesting that the decrease in molecular weight of the blends could be caused by the absorbed moisture from the air36,37. MALDI-TOF-MS also confirmed the uniform structure of the prepolymer obtained using excess EG or SA, with repetition periods of 192.17 and 48.04 (Fig. 2e and Supplementary Figs. 80-83), respectively assigned to the m/z of PET repeating unit and the difference in m/z between the PET and PES repeating units. The blends exhibited periods dominated by PES repeat units (144.13 and 96.09) in addition to the aforementioned periods; however, these periods vanished after repolymerization, indicating the formation of a homogeneous molecular structure (Supplementary Figs. 84-85).
To assess the mechanical suitability of PEST products for practical applications, we tested their tensile properties across a range of n(PET):n(PES) from 8:2 to 2:8 (Fig. 3a and 3b, along with Supplementary Tables 5-7). The results indicated that an increase in the proportion of PES units led to a significant decrease in the average tensile strength of PEST obtained using excess EG or SA, ranging from 32.9-34.9 MPa to 0.20-0.33 MPa. However, their elongation at break demonstrated a notable peak, reaching values of 2442.2-3419.2% at n(PET):n(PES) of 5:5 or 4:6, compared to minimum values of 11.3-12.1%. Interestingly, the blended products showed relatively stable tensile strength (1.97-5.35 MPa) and elongation at break (2.2-28.5%) across different n(PET):n(PES) ratios. By contrast, the tensile properties of the repolymerized product after blending resembled those obtained by directly using excess EG or SA. Specifically, the tensile strength decreased from 37.7 MPa to 0.6 MPa and increased to 8.6 MPa when n(PET):n(PES) reached 2:8; meanwhile, the maximum elongation at break of 3054.4% was observed at n(PET):n(PES) of 4:6. Comprehensively, PEST achieved the highest toughness at n(PET):n(PES) of 6:4 with an average value ranging from 192.48 to 270.79 MJ m-3 (Fig. 3c and Supplementary Fig. 86), surpassing those of most elastic materials including commercial low-density polyethylene (LDPE, 62.55 MJ m-3) used for packaging38-45.
In terms of thermal performance of PEXT, we conducted differential scanning calorimetry (DSC), dynamic mechanical analysis (DMA), and thermogravimetric analysis (TGA) to determine their glass transition temperature (Tg), melting point (Tm), and thermal decomposition temperature (Td). Comparing Tg values from DSC and DMA (Supplementary Figs. 87-121 and Tables 8-10), we noted a lower Tg for PEST using DSC. Tg values exhibited a nearly linear decline with the logarithm of n(PET):n(PES) (Fig. 3d). Notably, when utilizing 1.1 equiv. EG, the reduction in Tg of PEST was significantly higher than those using 1.1 equiv. SA and blending after polymerization. As anticipated, the blended products displayed two distinct Tg corresponding to those of PET and PES polymers, irrespective of n(PET):n(PES), implying challenging in achieving molecular-level homogeneity. Similarly, DSC analysis indicated reduced Tm for PEST with decreasing n(PET):n(PES) (Fig. 3e). Intriguingly, the n(PET):n(PES) of 1:1 or higher made Tm detection difficult, indicating a glassy state of PEST under these conditions. For the blended products, two separate Tm corresponding to PET and PES were observed. Likewise, we investigated the thermal decomposition of the samples (Supplementary Figs. 122-127) and identified the temperature at 10% mass loss (Td,10%, Fig. 3f). Generally, the Td,10% of PEST with excess EG or SA decreased linearly with the logarithm of n(PET):n(PES). The thermal stability of repolymerized PEST after blending was slightly lower than that of the former two. The Td,10% of the blended product remained consistently around 382.19 °C, which falls between those of pure PET (410.2 °C) and PES (361.2 °C, Supplementary Table 11).
Leveraging the tunable thermal properties, we discovered that PEST with a specific composition exhibits exceptional shape memory capability (Fig. 3g) due to distinct molecular regions with dominant hard PET and soft PES segments46. As an example at n(PET):n(PES) of 4:6, PES segments become reorientable under external stress within the temperature range of Tg to Tm, allowing reshaping and retaining of the new shape at room temperature (25 °C). Reheating beyond Tg caused PEST revert to its original molecular configuration, thereby recovering its initial shape. Additionally, by adapting preparation conditions, PEST showcased immense potential for diverse applications, e.g., transparent flexible films, opaque rigid plastic parts, high-stretch materials, and packaging bags (Fig. 3h).
By constructing the molecular composition and end-group types of PEST, we sought to regulate its biodegradation cycle within an optimal range. To validate this, we conducted composting degradation experiments spanning 84 days (Supplementary Figs. 128-132). By monitoring the molecular weight of the degradation products at intervals and analyzing the process using first-order reaction kinetics (see composting degradation characterization in Supplementary Information), we calculated their degradation rate (k, Supplementary Figs. 133-162 and Tables 12-15). The k values of PESTs (4.6×10-4-51.9×10-4 h-1) aligned with the mainstream biodegradable plastic–poly(butylene adipate-co-terephthalate) (PBAT, 9.9×10-4-24.7×10-4 h-1) (ref.47). Accordingly, the respective degradation half-life was determined (t1/2, Fig. 4a). Notably, hydroxyl-terminated PEST products had approximately 10%-90% longer t1/2 (10.7-62.9 days) than carboxyl-terminated products (5.6-56.8 days) across n(PET):n(PES) from 8:2 to 2:8, showing a faster degradation rate due to the catalytic acceleration effect induced by the acidity of the carboxyl groups. Similar to thermal performance, t1/2 correlated with the logarithm of n(PET):n(PES), reinforcing the predictability of PEST degradation performance. Given the inherent biodegradability of PES24, the t1/2 of PET/PES blends varied in the range of 33.4-68.1 days. However, the repolymerization of PEST after blending accelerated the degradation rate, yielding t1/2 (17.6-56.1 days) comparable to the case of 1.1 equiv. EG. Composting degradation of virgin PET and PES polymers was compared (Fig. 4b). Strikingly, PET displays an extremely high t1/2 of 7019.5 days (19.2 years) compared to PES (27.6 days), implying its degradation over several centuries in a natural environment48. These results highlight the attainability of complete biodegradation for PEXT within a desired timeframe.
In short, we have highlighted that PEXT possesses programmable comprehensive properties through precise structural control. On this basis, we proceeded to demonstrate the feasibility of closed-loop recycling for PEXT by regenerating it into specialized functional components by incorporating PET or X-monomers, analogous to PET conversion (Fig. 4c). Furthermore, the biodegradability of PEXT enables its microorganism decomposition similar to biomass and active participation in the natural carbon cycle, thus presenting immense economic and environmentally friendly prospects.