Emma
McCrea
,
Peter
Goodrich
,
John D.
Holbrey
and
Małgorzata
Swadźba-Kwaśny
*
The QUILL Research Centre, School of Chemistry and Chemical Engineering, Queen's University Belfast, Belfast BT9 5AG, UK. E-mail: m.swadzba-kwasny@qub.ac.uk
First published on 31st July 2025
Methanolysis of polyethylene terephthalate (PET) to dimethyl terephthalate (DMT), carried out in a microwave reactor, was catalysed by an inexpensive and recyclable non-stoichiometric protic ionic liquid, formulated from sulfuric acid and triethylamine. The influence of the catalyst composition (excess of acid or base), reaction temperature and time, as well as methanol excess, on the conversion of PET and the yield of DMT, was investigated. Under optimised conditions (3 h, 180 °C), waste PET from milled plastic bottles was depolymerised, reaching 100% PET conversion and 98% isolated yield of DMT. Pure DMT was separated through recrystallisation directly from the reaction mixture. Preliminary experiments with carpet waste (dyed mixed polymer waste, without milling) gave results on par with those achieved for PET bottles, with 100% PET conversion and 97% of DMT (isolated yield).
Sustainability spotlightThis work addresses the need for an economically viable PET depolymerisation process, catalysed by an inexpensive and reusable protic ionic liquid. It handles contaminated waste (dyed carpets) without preprocessing and enables closed-loop recycling by producing a virgin-quality PET monomer, dimethyl terephthalate. The process contributes to truly circular PET recycling (SDG12 – sustainable consumption and production). Addressing plastic pollution promotes cleaner waters (SDG14 – life below water) and decreases landfill waste (SDG11 – sustainable cities and communities). |
Although mechanical recycling plays a crucial role in extending the life cycle of plastics, it progressively degrades the quality of the material. In contrast, chemical recycling offers a more sustainable alternative by depolymerising PET into its original monomers. These building blocks can then be re-polymerised into high-quality PET, equivalent to its virgin counterpart.
Conventionally, PET is synthesised via polycondensation reactions between terephthalic acid (TPA) and ethylene glycol (EG) through direct esterification (Scheme 1a), or between dimethyl terephthalate (DMT) and EG via transesterification (Scheme 1b). Consequently, PET recycling can be achieved by reversing these reactions. Specifically, the ester bond in PET can be cleaved through methods such as methanolysis, glycolysis, hydrolysis, pyrolysis, aminolysis, or ammonolysis.4,5 Among these, glycolysis is the most prevalent, producing bis(2-hydroxyethyl) terephthalate (BHET), a key intermediate for PET synthesis (Scheme 2a). Despite its high energy demand and complex purification steps, glycolysis remains attractive due to the direct reusability of BHET.6–13 Hydrolysis, in contrast, decomposes PET into TPA and EG (Scheme 2b) under acidic, basic, or neutral conditions. However, this method requires corrosion-resistant equipment, increasing capital expenditures and limiting its commercial appeal.14–18 In comparison, methanolysis yielding DMT and EG (Scheme 2c) offers advantages in terms of easier product separation and purification.
![]() | ||
Scheme 1 Polycondensation of TPA and ethylene glycol via direct esterification (a); polycondensation of dimethyl terephthalate and ethylene glycol via transesterification (b). |
![]() | ||
Scheme 2 Glycolysis of PET to BHET (a); hydrolysis of PET to TPA (b); methanolysis of PET to DMT (c). |
Historically, methanolysis was first reported in the 1960s by Gruschke and co-workers, who developed a two-stage catalyst-free process involving PET melting (265–285 °C) followed by reaction with excess hot methanol under high pressure (30–40 atm) to produce pure DMT.19 Building on this, Eastman Kodak patented a process in the 1990s using zinc acetate as a catalyst in either batch or continuous flow systems, operating with methanol or a methanol/glycol mixture at a PET:
methanol molar ratio of 1
:
7–1
:
20.20 Today, commercial methanolysis processes typically use PET flakes at 180–280 °C and 2–4 MPa.21 However, these harsh conditions increase side reactions, including PET degradation to acetaldehyde, formaldehyde, and CO2, as well as the partial ester exchange with methanol. Moreover, decomposition of EG to glyoxal and formic acid further compromises process efficiency, driving the search for milder and more efficient alternatives.
One such alternative is methanolysis using supercritical methanol, which offers near-complete PET conversion and a 95% DMT yield. Nonetheless, the extreme conditions required (260–270 °C, 9–11 MPa) result in high energy demands and elevated carbon emissions, limiting its commercial viability.22,23
To improve efficiency, catalytic methanolysis is commonly employed, typically with basic or acidic catalysts. In both cases, the ester bond is broken via nucleophilic substitution, where methanol's –OCH3 group attacks the carbonyl carbon of the ester linkage, forming EG and DMT (Scheme 2c). Basic catalysts enhance this process by activating methanol for stronger nucleophilic attack, while Brønsted acids protonate the ester oxygen to increase electrophilicity.24 Acidic catalysis often results in higher selectivity for DMT and better compatibility with contaminated PET waste.25,44
Recent patents reflect this trend. A 2021 patent granted to Eastman describes a catalytic process for PET methanolysis, using sodium carbonate, magnesium methoxide, 1,8-diazabicyclo[5.4.0]undec-7-ene (DBU) and triazabicyclodecene (TBC) as chosen catalysts.26 Other methods rely on basic, heterogeneous catalysts, such as sodium silicate,27 potassium carbonate,28 metal oxides and metal acetates, e.g. ZnO or Zn(OAc)2, as well as metal hydroxides.29 Although these catalysts accelerate the reaction, increase the conversion and the yield of DMT, they also have disadvantages. They lack selectivity to DMT, increasing the rate of side product formation alongside the targeted reaction, necessitating additional purification steps. Furthermore, metal catalysts cause fouling and leach into the product, resulting in discolouration of the recycled plastic.13,29–31
Muangmeesri et al. introduced an organocatalytic methanolysis process employing triethylamine to selectively depolymerise PET into high-purity dimethyl terephthalate (DMT) and ethylene glycol (EG) under milder conditions (200 °C, 2 h) with a reduced solvent-to-substrate ratio. Notably, this method allows efficient catalyst and solvent recirculation, minimising waste and energy input. Extending its versatility, the approach is applicable to other polyesters and polycotton textiles, yielding recoverable cotton suitable for viscose fiber production. However, when comparing product yield from polyester to polycotton material, the yield of DMT drops from 88% to 56% indicating that there is a reduced reactivity to the system when using mixed feedstock systems. A comprehensive life cycle assessment (LCA) indicates a net reduction of 1.88 kg CO2-equivalent emissions per kg of PET recycled, outperforming incineration and conventional recycling processes. This work establishes a scalable, low-impact strategy for closed-loop recycling of PET and blended textile waste.32
In response to these limitations, ionic liquid-based approaches take advantage of their low vapour pressure and the ability to solubilise a broad range of polymers. For instance, lanthanide-containing ionic liquids (ILs) were used for the glycolysis of PET, showing cooperativity between imidazolium and phosphonium ILs with basic (chloride and acetate) anions, and lanthanide salts of the general formula LnCl3·xH2O, where Ln = La, Sm, Gd, Y, or Tm (x = 6 or 7). This was evidenced by the increase in BHET yield compared to that of the ionic liquid without the lanthanide salt.12 Poly(ionic liquid)s synthesised from 1-vinyl-3-ethylimidazole acetate and acrylic acid zinc salts formed Pil-Zn2+ and achieved 100% PET conversion and 89% yield of DMT at relatively low temperature (170 °C). Pil-Zn2+ facilitates the activation of the ester bond, making the PET chain more susceptible to nucleophilic attack, which allows for milder conditions and minimises potential side reactions.33
Organocatalytic approaches have also demonstrated potential. For example, 1,3-dimethylimidazolium-2-carboxylate, [C1mim-CO2], and 1,3-dimethylimidazolium acetate, [C1mim][OAc], were compared in a study of PET glycolysis. Basicity was found to be the deciding factor, and [C1mim-CO2] was more effective, as it works through N-heterocyclic carbene, which is a very strong base.8
Similarly, basic deep eutectic solvents (DESs), such as a combination of 1,5-diazabicyclo [4.3.0]-5-nonene (DBN) and phenol, also gave excellent results. The strong hydrogen-bonding network between the DES and the PET activates the ester bond and increases the nucleophilicity of methanol. The interactions stabilise the transition state and improve the kinetics, leading to rapid depolymerisation under milder conditions and lower energy barriers (100% conversion and 95.3% DMT yield at 130 °C in 1 h).34
In the context of hydrolysis, sulfonic acid-functionalised Brønsted acidic ILs have been used for the hydrolysis of PET using 1-(3-propylsulfonic)-3-methylimidazolium chloride, [PSMIM]Cl. This catalyst was more effective than the benchmark (sulfuric acid), as the IL is understood to have greater hydrogen bonding interactions with PET, facilitating the depolymerisation (94% yield of TPA with 100% conversion at 210 °C over 24 h).35
A representative comparison of PET alcoholysis and hydrolysis processes is shown in Table 1, comparing reaction conditions and results.
Catalyst | Solvent | Temperature (°C) | Pressure (atm) | Time (h) | Conversion of PET (%) | Yield of DMT (%) | Ref. |
---|---|---|---|---|---|---|---|
None | Methanol | 265–285 °C | 30–40 | 99 | 99 | 19 | |
Zn(OAc)2 | Methanol for batch | 240–260 | 24–40 | 2–6 | 95 | 85–90 | 20 |
Methanol and alkylene glycols | |||||||
[Bmim][BF4] | Supercritical ethanol | 240 | 64 | 0.75 | 98 | 97 (diethylterephthalate) | 22 |
DBU | Methanol | 140 | 1–15 | 4 | 90 | 75 | 23 |
TBD | Methanol | 140 | 1–15 | 4 | 88 | 70 | 23 |
NaOMe | Methanol | 110 | 1–15 | 4 | 81 | 62 | 23 |
Na2CO2 | Methanol | 140 | 1–15 | 4 | 99 | 77 | 23 |
Mg(OMe)2 | Methanol | 180 | 1–15 | 4 | 99 | 80 | 23 |
MgO-modified NaY zeolite | Methanol | 220 | 1 | 3 | 95 | 85 | 24 |
N222 | Methanol | 180 | 1 | 4 | 99 | 88 | 25 |
Zn(OAc)2 | Methanol | 200 | 1 | 1 | 99 | 98 | 26 |
[C4mim]Cl with CeCl3 | Ethylene glycol | 200 | Autogenous | 4 | 95 | 85 (BHET) | 12 |
[C1mim-CO2] | Ethylene glycol | 185 | 1 | 3 | 100 | 60 (BHET) | 8 |
DBN and phenol | Methanol | 130 | 1 | 1 | 100 | 95 | 29 |
[PSMIM]Cl | Water | 210 | Autogenous | 24 | 100 | 94 (TPA) | 30 |
Al[OCH(CH3)2]3 | Methanol | 200 | 20–40 | 2 | 96 | 64 | 23 |
Pil-Zn2+ | Methanol | 170 | Autogenous | 1 | 100 | 90 | 33 |
N222 | Methanol, toluene | 200 | Autogenous | 2 | 100 | 88 | 32 |
While there are many IL-based systems that offer good yields and conversions under mild conditions, the key consideration is the high cost. Superbases such as DBU or sulfonic acid-functionalised ILs, although highly efficient, are significantly more expensive than conventional basic catalysts or thermal methods, thereby impacting the economic viability of their large-scale implementation.
Considering this, the present work investigates an economical alternative: a non-stoichiometric protic ionic liquid as a catalyst for PET methanolysis. Our group has previously explored protic ILs based on sulfuric acid and inexpensive amines, formulated with excess H2SO4.36 These ILs demonstrated superior catalytic performance in esterification and Beckmann rearrangement reactions due to their favourable phase behaviour.37,38 Building on this foundation, we now assess the efficacy of these ILs across various compositional ranges, as either acid- or base-rich systems for catalysing the methanolysis of PET.
Polyethylene terephthalate (PET) was sourced from commercially available water bottles, washed with detergent, and dried before use. Carpet samples (dyed PET with latex backing) were sourced from a commercial carpet supplier.
Subsequently, the hot reaction mixture, including the unreacted solid PET, was transferred to a centrifuge vial (15 ml) and centrifuged using an Eppendorf Centrifuge 5702 (4000 rpm, 3 min). Unreacted PET was separated by vacuum filtration. Note: The reaction mixture must remain above 70 °C to ensure that DMT does not precipitate out before the vacuum filtration of unreacted PET. Separated PET was subsequently dried under vacuum (40 °C, 0.5 h) to constant mass, which was recorded.
The filtrate was transferred to a round-bottomed flask and placed in an acetone-dry ice bath (−78 °C, 1 h) to crystallise the product. The white, needle-like crystals were separated by vacuum filtration, washed with cold methanol (3 × 100 ml), and dried in air to constant mass, which was recorded.
Representative analysis of isolated DMT are shown in Fig. S2 (1H NMR spectrum), Fig. S3 (GC-MS chromatogram), Fig. S4 and S5 (TGA measured curves) and Fig. S6 (FT-IR spectra).
Conversion of PET was determined using eqn (1), where mPET ini is the mass of PET used for the reaction and mPET fin is the mass of PET after the reaction.
![]() | (1) |
Yield of DMT was calculated using eqn (2), where nDMT exp is the quantity of DMT measured or isolated (in moles) and nDMT theor is the theoretical quantity of DMT assuming complete conversion of PET.
![]() | (2) |
Selectivity of DMT was calculated using eqn (3), where nDMT iso is the quantity of DMT measured or isolated and ntotal is the total quantity of all measured or isolated products.
![]() | (3) |
A sample (1 μL) was injected into an Agilent 8890 GC system, equipped with a 5977B GC/MSD detector and an HP-5MS UI (60–325 °C) 30 m × 250 μm × 0.25 μm column, using a post-column splitter with He carrier gas and split tubing (2.3 m × 150 μm internal diameter) to the MSD and to the front detector FID with an MSD transfer line temperature of 280 °C under the following conditions: helium (1.0 ml min−1) as the carrier gas and injection with a 10:
1 split ratio at 250 °C.
The oven temperature program was 50 °C (2 min hold), ramped at 10 °C min−1 to 250 °C (5 min hold), followed by a post-run at 300 °C (2 min). Mass spectrometric detection was performed in electron ionisation (EI) mode at 70 eV with a scan range of 50–500 m/z. The ion source and quadrupole temperatures were 230 °C and 150 °C, respectively. Data were acquired in full-scan mode for the identification and quantification of methanolysis products. DMT, EG, TPA, BHET, mono(2-hydroxyethyl) terephthalate (MHET), and monomethyl terephthalate (MMT) were quantified using five-point calibration curves (Fig. S8–S12).
The acceptor numbers of the three ionic liquid compositions were determined via31P NMR spectroscopy. All sample preparation was conducted under an inert atmosphere in a glovebox. Each ionic liquid (1.00 g) was weighed into a 10 cm3 vial containing a PTFE-coated magnetic stir bar. Triethylphosphine oxide (TEPO, 1 mol%) was then added as the probe molecule for acceptor number measurement. Following the addition, the vials were sealed and stirred overnight to ensure complete dissolution of TEPO. After equilibration, the mixtures were transferred to 5 mm borosilicate NMR tubes containing sealed capillaries of deuterated dimethyl sulfoxide as a field lock. NMR tubes were sealed with parafilm and removed from the glovebox prior to analysis. 31P NMR spectra were recorded on a Bruker AVANCE 400 MHz spectrometer at 60 °C. Chemical shifts were referenced externally to 85% H3PO4 in water. For comparison, TEPO solutions in hexane (1 mol%) were prepared using an identical procedure.
The acidic ionic liquid [HN222][HSO4·H2SO4] was excluded from further study due to a rapid increase in pressure above the safety threshold, likely caused by dehydration of methanol to dimethyl ether. Interestingly, this issue was not observed with neat H2SO4, demonstrating how modifying a Brønsted acid into an acidic ionic liquid can significantly alter the reaction pathway.
Among the remaining catalysts, the basic ionic liquid [HN222·N222][HSO4] consistently delivered the highest DMT yields. Its improved performance is attributed to increased basicity and reduced acidity. In contrast, the more acidic [HN222][HSO4] exhibited lower activity. Under acidic conditions, the reaction proceeds via protonation of the ester carbonyl, followed by methanol attack. However, high acidity reduces methanol nucleophilicity and increases viscosity and ionic strength, which limits PET interaction and slows reaction kinetics (Fig. 1).
In the base-rich [HN222·N222][HSO4], triethylamine may partially deprotonate methanol to generate methoxide in situ, which acts as a stronger nucleophile. This promotes PET bond cleavage via a tetrahedral intermediate. While neat triethylamine can initiate this pathway, its volatility, lack of ionic structure, and poor PET solubility limit its practical application. Combining it with sulfuric acid to form an ionic liquid provides thermal and chemical stability, a polar ionic medium for better PET interaction, and reduced volatility.
The acidity of the PILs was quantified via the Gutmann Acceptor Number: [HN222][HSO4·H2SO4] had the highest value (118.46 AN), followed by [HN222][HSO4] (107.93 AN) and [HN222·N222][HSO4] (20.62 AN). Lower acidity correlated with higher DMT yields, supporting a mechanistic distinction between acid- and base-driven catalysis. These structural and mechanistic factors together explain both the superior performance of the excess base formulation and the enhanced activity of the ionic liquid compared to triethylamine alone.
At low methanol excess (1:
30), reactions catalysed by H2SO4 and N222 showed high PET conversion (100% and 80%, respectively), but low selectivity to DMT (14% and 12%). The equimolar PIL [HN222][HSO4] had low activity (20% PET conversion, <10% DMT yield), while the basic PIL [HN222·N222][HSO4] gave moderate conversion (65%) and slightly better DMT yield (18%). Across all catalysts, DMT selectivity was poor, with most PET converted to TPA, as indicated by GC-MS (Fig. S3). This highlights the challenge of driving the reaction towards DMT formation at lower methanol concentrations.
To improve DMT selectivity, reactions were repeated at a higher methanol excess (1:
75). This increased PET conversion and DMT yields across all catalysts. N222 gave full PET conversion, while others reached 80%. DMT yields improved significantly, ranging from 40–70%, with the highest selectivity (57–86%) observed for [HN222·N222][HSO4].
This improvement is consistent with Le Chatelier's principle: increasing the concentration of methanol shifts the equilibrium towards DMT and EG, limiting repolymerisation. Excess methanol also helps suppress MHET formation by competing with EG and water as nucleophiles. Although higher methanol loading increases downstream recovery costs, it clearly enhances product selectivity and conversion efficiency under these conditions (Fig. 2).41
In this work, PET methanolysis was studied within the temperature range of 100–180 °C (Fig. 5), in a reaction catalysed by [HN222·N222][HSO4], which exhibits the highest selectivity towards DMT. At 100 °C, no products were detected after 1 h. At 120 °C and 140 °C, both conversion and DMT yield remained under 10%. Only at 160 °C, PET conversion reached 60%, and DMT yield reached 54%, with very good selectivity (90%). Further increase in temperature to 180 °C gave satisfactory conversion (85%) and DMT yield (70%), but selectivity to DMT dropped to 81%. A reaction at 200 °C failed due to the generated pressure exceeding the safety threshold of the microwave. As such, a reaction temperature of 180 °C was selected as offering sufficiently high conversion within 1 h, without significantly compromising selectivity.
At the lower ratio (1:
30), the yield of DMT was analysed by qNMR spectroscopy in the post-reaction mixture and compared to the isolated yield of the crystallised product (Fig. 4). It was possible to quantitatively separate pure DMT in the form of colourless crystals. The isolated yield of DMT reached a plateau at 30% after 2 h, despite a steady rise in conversion up to 6 h.
At the higher methanol excess (1:
75), it was impossible to quantify isolated yields in a reproducible manner, because increased volume of methanol with the constraint of fixed volume of the microwave reactor resulted in small absolute quantities of the product. Instead, the reaction progress was followed by GC-MS, enabling quantitative analysis of both DMT and the side products (Fig. 4). At higher methanol excess, conversion of PET reached 80% within 1 h and 100% after 3 h. The main product was DMT, reaching 98% yield after 3 h. For longer reaction times, the yield of DMT dropped due to hydrolysis to TPA, transesterification to MHET and several other side reactions. It is known that extended reaction times result in the degradation of EG to acetaldehydes, glycolic acid or formic acid.44 Furthermore, it allows the formation of oligomers that are not detected by GC-MS, as they precipitate out during the derivatisation of the GC-MS sample.
In conclusion, while it was not feasible to reach satisfactory yields of DMT using the low methanol excess, it was possible to reach full conversion of PET to DMT at high methanol excess, under reaction conditions on par with the literature (Table 1), but using an ionic liquid catalyst that is as cheap as acetone.36
Using a high methanol ratio offers sustained catalytic performance, cleaner product recovery, and minimal interference from side-product formation. In contrast, the low methanol excess was used to ensure a more rigorous test of catalyst robustness, revealing the limitations imposed by increasing concentrations of EG and other side products, which were extracted by ethyl acetate (Fig. S29 and S30, 1H and 13C NMR spectra of the organic layer). These may reduce isolated DMT yields through transesterification or other secondary reactions. In conclusion, using higher methanol excess is recommended in practice to maximise catalyst lifetime and reduce waste processing burdens in future scale-up applications.
The key test for the [HN222·N222][HSO4] catalyst was its robustness against more challenging PET waste, in the form of a carpet sample containing dyed PET fibres with a latex backing (Fig. 7). Methanolysis of carpet samples was carried out by cutting them into small pieces using scissors and placing them on a balance. The latex backing was removed from the PET strands to weigh the quality of PET. The PET was combined with the latex backing and placed in the reaction vessel. The reaction was carried out under optimised conditions (180 °C, 3 h, high excess of methanol). The isolated yield was calculated based on the starting moles of PET and measured using GC-MS.
![]() | ||
Fig. 7 Visual representation of the PET carpet and the end results. Unreacted latex was separated from the DMT product. |
Methanolysis of carpet waste under the conditions optimised for PET bottles gave 100% conversion of the PET fibres and a 97% isolated yield of DMT. It has been determined that the presence of dyes and the latex backing did not inhibit the depolymerisation reaction or reduce product yield, which is an important advantage for practical applications in chemical recycling. The latex backing remained intact post-reaction and was easily separated (Fig. 7), in contrast to many other ionic liquids that degrade latex,45 which would complicate the separation and purification of DMT. GC-MS analysis showed no detectable dyes in the spectrum, and the purity of the DMT was determined by NMR using 1,3,5-trimethoxybenzene as an internal standard, resulting in a purity of 98% (Fig. S31 and S32) While some dye accumulation in [HN222·N222][HSO4] may occur over time and require periodic removal, detailed analysis of dye contamination and its long-term effects was beyond the scope of this study. Nevertheless, catalytic performance appeared unaffected, indicating the ionic liquid catalyst's tolerance to such additives. Unlike the findings reported by Muangmeesri et al., where yield decreased significantly upon introducing a mixed feedstock,32 the use of ionic liquids in this study offers the advantage of maintaining higher yields despite feedstock complexity.
The supplementary information provides 13C, 1H, 31P, and qNMR spectra, GC-MS chromatograms with calibration curves, TGA analyses, and a mass balance table supporting the findings of this study. See DOI: https://doi.org/10.1039/d5su00316d.
This journal is © The Royal Society of Chemistry 2025 |