Marco
Rollo
a,
Massimo A. G.
Perini
a,
Alessandro
Sanzone
b,
Lorenzo
Polastri
b,
Matteo
Tiecco
c,
Alejandro
Torregrosa-Chinillach
d,
Elisa
Martinelli
*a and
Gianluca
Ciancaleoni
*a
aDepartment of Chemistry and Industrial Chemistry, University of Pisa, Via Giuseppe Moruzzi 13, I-56124 Pisa, Italy. E-mail: elisa.martinelli@unipi.it; gianluca.ciancaleoni@unipi.it
bgr3n SA, Via Probello 19, 6963 Lugano, Switzerland
cChemistry Interdisciplinary Project (ChIP), School of Pharmacy, University of Camerino, Via Madonna delle Carceri, 62032 Camerino, MC, Italy
dDepartment of Organic Chemistry, Faculty of Sciences, Institute of Organic Synthesis (ISO), University of Alicante, Apdo. 99, 03080 Alicante, Spain
First published on 21st November 2023
Chemical recycling offers a convenient solution for the disposal of plastic items made of polyethylene terephthalate (PET); however, there is still much room for improvement in terms of integration into the current waste treatment cycles. Recently, deep eutectic solvents (DESs) have exhibited interesting properties in PET glycolysis and hydrolysis, in some cases under mild conditions. In particular, we recently reported good results with Lewis/Brønsted acidic DESs (LBDESs) containing iron(III) chloride and sulfonic acids. However, the choice of weaker acids, such as acetic acid, is more cost effective and sustainable, with an associated reduced risk of corrosion and improved safety. In this study, we demonstrate that a simple post-reaction procedure significantly enhances the yield of terephthalic acid (TA) using FeCl3·6H2O/acetic acid (molar ratio 1:1) LBDES from 4% (literature value) to 54% under the same experimental conditions. Furthermore, we investigate the effect of chloride salts as additives and microwave irradiation on the reaction, achieving quantitative conversion and a high yield of TA in 10 minutes at 180 °C.
Sustainability spotlightModern society needs to move from the current model of produce-use-discard to a circular economy, where no new polymers are produced and all the plastic is recycled. This is a very challenging task given that the quality of the recycled polymer is often negatively impacted by thermo-mechanical recycling processes. However, chemical recycling processes, such as depolymerization and repolymerization, can produce recycled polymers that are similar to virgin materials. Our goal is to develop new solvents for the depolymerization of polyethylene terephthalate, allowing the use of milder conditions with respect to existing industrial processes, in agreement with the following UN sustainable development goals: SDG 11 (sustainable cities and communities) and SDG 12 (responsible consumption and production). |
Tertiary recycling is still in its infancy; polymers, such as polyethylene,4 require harsh conditions5 and precious metal catalysts,6 and there is generally no selectivity.7 For condensation polymers, the potential is higher; indeed, research8–13 and industrial protocols already exist.2
In particular, polyethylene terephthalate (PET) is one of the preferred polymers for depolymerization studies; this research field is very active, and original ideas are continuously emerging.14–17 Generally, the depolymerization process of PET leads to bis(2-hydroxyethyl)-terephthalate (BHET) if ethylene glycol (EG) is the nucleophile,18 and to dimethyl terephthalate using methanol and terephthalic acid (TA) with water. In most cases, harsh conditions are necessary (T > 160 °C, many hours, sometimes high pressure) although exceptions can be found in recent studies.19–22
On the industrial side, many processes are used,23 including the pyrolysis of unsorted or sorted waste plastics and solvolysis, demonstrating widespread interest in the depolymerization process.
We recently contributed to this field by showing that the synergy between Lewis and Brønsted acids24–26 can be extremely beneficial in PET hydrolysis.27 Moreover, DESs formed by FeCl3·6H2O and organic sulfonic or carboxylic acids (mixed Lewis/Brønsted acidic DES, LBDESs) can efficiently hydrolyze PET under mild conditions (100 °C, 1 atm). The best performing LBDES contains methanesulfonic acid, but it is desirable to use different acids that are more environmentally friendly, cheaper, and easier to dispose of.
Acetic acid seems to be the ideal choice because it is cheaper and less corrosive than sulfonic acid (but still, significantly corrosive at high temperature28) and is generally synthesized by the carbonylation of methanol but can also be produced from lignocellulosic biomass via a bioconversion process.29,30 Acetic acid also has another advantage because it has a boiling point (Tb) of 118 °C, which is lower than that of MSA (Tb = 167 °C)27 and significantly different from that of EG (Tb = 197 °C), making the fractional distillation of the mixture to extract EG easier. However, our preliminary data showed that the liquid formed by FeCl3·6H2O/acetic acid (molar ratio 1:1, system LBDES1) is significantly less active than the FeCl3·6H2O/methanesulfonic acid (MSA) 1:1 LBDES, with PET conversion of around 20% and a terephthalic acid yield of around 4% in 30 minutes (0.3 g of PET in 4 g of LBDES, 100 °C, 1 atm).27
In this paper, we thoroughly characterize LBDES1, and we show that the performance of the system LBDES1 can be markedly improved from 4% of TA yield at 100 °C and 30 min to 56% under the same reaction conditions using a simple post-reaction treatment. We also demonstrate that the use of CaCl2 as an additive increases the yield up to 69% in 30 min, leading to quantitative conversion after 60 min of reaction at 100 °C. However, its use is not convenient because of the scale-up of the process.
The same reaction was also carried out by microwave (MW) irradiation,10,20,31 reducing the time to 10 min at 180 °C and greatly improving the PET/solvent ratio. Surprisingly, the results revealed that the beneficial effect of MW is that it is a fast and efficient heating strategy, without any significant specific effect.
(1) |
(2) |
The theoretical melting temperatures were determined from the theoretical curves by considering the activity coefficients γi = 1. The eutectic points were determined as the minimum in the experimental curves, and they were compared to the theoretical ones. The experimental γi values were determined using eqn (3) with the experimentally observed melting temperatures:
(3) |
Conversion is calculated using the following equation:
(4) |
The yield of the purified TA is calculated using the following equation:
(5) |
(6) |
In addition to E, the energy economy coefficient (ε) is defined as
Finally, the energy impact factor (ξ) is the ratio between E and ε.
Additionally, in this case, the experimental Tm is significantly lower (ΔTm around 40 K) than that estimated by considering a simple eutectic without any specific interaction between the components. Coherently, the activity coefficients of the components are much lower than 1 in all compositions. Another consideration comes from the theoretical/experimental comparison: theoretical equations indicate the eutectic point at xacetic acid at around 0.9, while the experimental trend indicates the eutectic point at 0.5 (1:1 molar ratio). Interestingly, for the FeCl3·6H2O/methanesulfonic acid (MSA) mixture, the theoretical and experimental eutectic points are close to each other.27
To complete the LBDES characterization, the trends in the density and conductivity of LBDES1 with temperature were measured (Fig. S1, ESI†). This leads to an activation energy of conduction (Eσ) of 34 kJ mol−1. For comparison, the system FeCl3·6H2O/p-toluenesulfonic acid·H2O has a Eσ of 23 kJ mol−1.27 The difference is likely attributed to the lower acidity of acetic acid with respect to p-toluenesulfonic acid. A stronger acid generates more H3O+ ions, which are smaller and diffuse very efficiently (possibly also through the Grotthuss mechanism32). The other ions possibly present in LBDES1, apart from acetate, are FeCl2(H2O)4+ and Cl−, which are the ions expected from the crystal structure of FeCl3·6H2O.33
A mild post-reaction treatment maintaining all the recovered solid fractions (TA, oligomers and unreacted PET) in contact with 20 mL of NaOH 0.5 M for 12 h at room temperature under magnetic stirring was found to improve the conversion up to 54%. This suggests that the less crystalline polymer on the surface of the flakes was hydrolyzed and converted to Na2TA.21,34 The surface of the PET flakes was much smoother after this treatment (Fig. 2b). The white powder was also completely dissolved, with a consequent increase in yield to 56%. The percentages reported are the average of three reactions with deviations around ±5%. Fig. 2a shows that after 12 h of post-reaction treatment, the yield did not increase sensibly.
The fact that the conversion and yield after the post-reaction treatment were similar implies that with this procedure, all the oligomers were hydrolyzed to Na2TA. It is noteworthy that keeping post-consumer 0.3 g of PET in 10 mL of NaOH 0.5 M for 12–18 h at room temperature did not lead to a measurable yield of Na2TA, therefore confirming that the hydrolysis reaction occurs in the LBDES.
For stronger acids, such as sulphonic ones, the conversion is so high that the improvement in the work-up is not relevant, but for weaker ones, such as acetic acid, it becomes crucial. The purity of obtained TA is verified by 1H (see Fig. 3), 13C NMR and IR spectroscopies (Fig. S3 and S4†). These impurities are mainly isomers of TA already present in the original PET blend, such as isophthalic acid (the singlet and triplet peaks are around 8.4 and 7.6 ppm, respectively, around 0.5%).
The speciation of chloride salts in DES is unclear because many equilibria could be active. For example, in the presence of a higher chloride concentration, FeCl2(H2O)4+ (the cation present in the solid state structure of FeCl6·6H2O33) could be converted into FeCl3(H2O)3 or FeCl4(H2O)2−. Even without these equilibria, chloride salts can enter the hydrogen bond network of the DES, interacting with acetic acid (CH3COOH⋯Cl−) or with water coordinated to iron. Hydrated AlCl3 has also more chances to interact owing to the water coordinated to aluminum, which can interact with bond hydrogen bond donors (such as –COO) or acceptors (such as Cl− or –COH). In all the cases, such interactions are insufficient to reach a high concentration of chloride salts.
In all the cases, we added from 1 to 4 equivalents of chloride salt with respect to the PET (0.3 g of PET = 1.60 mmol of monomeric units) based on the salt solubility in LBDES1. The results are summarized in Table 1. In all the cases, the liquid formed by FeCl3·6H2O and acetic acid (1:1) was used as solvent under the same experimental conditions (0.3 g of PET in 4 g of LBDES1, 100 °C, 30 min). The blank reaction (no additives added) gave a PET conversion of 54%.
The presence of 2 equivalents of NaCl did not affect the PET conversion, which was basically the same as that of the blank (57%). Additionally, in this case (and in all the following ones), all the conversion values are the average of three independent measurements, with a standard deviation of around 5%. Increasing the number of salt equivalents led to undissolved material in the flask and was not investigated.
However, 2 equivalents of CaCl2 (0.17 equivalents with respect to the iron) boosted the conversion up to 69%, but a further increase in the concentration led to a loss of performance. Such an effect, even if significant, is not drastic, but it deserves further investigation.
The physical properties of the liquid FeCl3·6H2O/CaCl2/acetic acid (1:0.17:1, LBDES2) are slightly different from those of LBDES1. The density is higher (Table S2†), and more importantly, the conductivity is higher (Table S3†). This denotes that the solution contains more ions, demonstrating that CaCl2 dissociates (partially, at least) in its ions. More detailedly, Eσ is reduced to 27 kJ mol−1 (Fig. S2†), confirming that the ions are less paired together and, consequently, more mobile. It is also interesting to note that the melting point of LBDES2 is practically the same as that of LBDES1 (Table S4†).
Using LBDES2 as a solvent, conversion became quantitative after 60 min (Fig. 4), while yield reached 94% after 75 min. This result indicates a significant reduction in the reaction time, which is reported to be 180 min for a quantitative yield for the same DES but without CaCl2.27
Fig. 4 Trend of PET conversion and TA yield with time (0.3 g of PET in 4 g of LBDES2, 100 °C, conventional heating). Solid lines serve only to guide the eye. |
Finally, the addition of AlCl3·6H2O surprisingly did not affect the PET conversion (Table 1), as the latter did not change significantly by increasing the number of equivalents if we considered an experimental uncertainty of ±5%.
Proposing a single hypothesis to explain all these effects is not straightforward. In the absence of any additive, the catalytic effect of LBDES1 can be explained by a synergy between the Lewis and the Brønsted acids present in the DES. In particular, a likely mechanism is illustrated in Scheme 1: the Lewis acid can coordinate the carbonyl moiety, whereas the proton of the Brønsted acid can attack the ester oxygen.
Scheme 1 Proposed mechanism of double activation of the ester group by the Lewis and Brønsted acids. |
This double attack leads to a double activation of the ester group, making the ester carbon more prone to nucleophilic attack from water. This hypothesis can also explain the double role of water: it is necessary to carry on the hydrolysis reaction, but it also acts as a reaction inhibitor (see below and ref. 27). A trimolecular adduct is favored only at very high concentrations, and any dilution is detrimental to the reaction rate.
Under this assumption, it is possible to rationalize the role of the additives. We know from conductivity measurements that calcium chloride dissociates, at least partially, in its constituting ions, and it is also known that calcium ions strongly interact with water (calcium chloride is routinely used as a drying agent). Similarly, calcium coordinates some water molecules, favoring the reaction because it removes the inhibitor. If the calcium concentration exceeds its optimal value, too much water is sequestrated, and the hydrolysis is less efficient. However, sodium is known to be less effective in binding water because of its reduced charge density. This implies that sodium does not significantly reduce the amount of free water and that the conversion remains practically unaltered. Furthermore, NaCl has solubility issues in LBDES1 that limit its use at 2 equivalents.
Aluminum chloride brings its water molecules, increasing or maintaining stable inhibitor concentrations, depending on the number of water molecules that remain coordinated. It is expected to be a good Lewis acid; therefore, it is also possible that [AlClx(H2O)y]3−x replaces [FeClx(H2O)y]3−x for the activation of the carbonyl moiety (Scheme 1), leading to an unaltered mechanism and a similar PET conversion. Unfortunately, in this case, the solubility became an issue beyond 4 equivalents, not allowing further investigation. In addition, mixing AlCl3·6H2O and acetic acid did not lead to a stable liquid phase in the molar ratios 1:1, 1:2 and 2:1.
First, we verified the ability of LBDES1 to absorb MWs by irradiating a sample of 8 g with a constant power of 100 W for 30 s (Fig. 5). Under these conditions, the sample increased its temperature to 53 K.
Therefore, the specific heat capacity of LBDES1 can be calculated using the following equation:
P × t = Q = mcΔT | (7) |
The experimental results using MW irradiation are listed in Table 2. Run 1 was performed under experimental conditions similar to conventional heating (100 °C, 35 min, no water added, PET/solvent ratio = 0.075), showing slightly higher conversion and yield values, probably because the target temperature is reached faster. As expected, adding 10 eq. of water (w) with respect to iron(III) was strongly detrimental to the reaction, even after 60 min, both under MW (run 2) and conventional heating (run 3).
Run | m (g) | T (°C) | t (min) | w (eq.) | PET conversion (%) | TA yield (%) |
---|---|---|---|---|---|---|
a Conventional heating. b Standard deviation calculated on 3 replicates: ±0.8. c Standard deviation calculated on 3 replicates: ±1.0. d Reaction heated by MW using a SiC reactor. | ||||||
1 | 1.125 | 100 | 35 | 0 | 74 | 73 |
2 | 1.125 | 100 | 60 | 10 | 0 | 0 |
3a | 1.125 | 100 | 60 | 10 | 0 | 0 |
4 | 1.125 | 140 | 10 | 0 | 100 | 87 |
5 | 1.125 | 140 | 35 | 10 | 30b | 26c |
6 | 1.125 | 140 | 60 | 20 | 4 | 2 |
7 | 1.125 | 180 | 10 | 0 | 100 | 92 |
8 | 1.125 | 180 | 10 | 10 | 100 | 91 |
9d | 1.125 | 180 | 10 | 10 | 100 | 88 |
10 | 1.125 | 180 | 35 | 20 | 100 | 96 |
11 | 4.000 | 180 | 10 | 10 | 100 | 95 |
12 | 6.000 | 180 | 10 | 10 | 100 | 94 |
By increasing T to 140 °C, the reaction was quantitative in 10 min (run 4), with a good yield (87%). Moreover, in this case, the addition of water led to a loss of conversion even when a longer time was used (run 5). Increasing the water addition (w) to 20 eq. (run 6) led to complete inactivity of the system, even with a longer reaction time (60 min).
At 180 °C, the reaction was always quantitative regardless of the amount of water added (from 0 to 20 eq., runs 7, 8 and 10), with TA yields always higher than 90%. As a confirmation that MW does not have a specific effect other than heating, run 9 was replicated with a SiC reactor, which efficiently absorbs MW.38 Comparable results were obtained, and the conversion and yield were 100% and 88% (run 9), respectively.
It is important to note that even at 180 °C and in the presence of ethylene glycol (coming from the depolymerization of PET), no trace of BHET was found in the 1H NMR spectrum (Fig. S6, ESI†). Moreover, BHET is soluble in basic water, but after some hours, it is hydrolyzed to Na2TA and EG.
Because the reaction was completed even in the presence of dilution, the amount of PET was increased to 4 g in run 11 and then up to 6 g in run 12 (ratio PET/DES = 0.4, PET/(DES + H2O) = 0.26) by maintaining the same amount of water (10 eq.) and DES (15 g) as well as the temperature (180 °C) and time (10 min) as in run 8. In both cases, the conversion was quantitative, and a high yield of pure TA was obtained. The result is particularly good compared with examples of acidic hydrolysis in the literature, in which the ratio PET/solvent typically varies from 0.09 to 0.4, reaction times from 140 to 300 min and temperatures from 98 to 140 °C.13
The experimental conditions of run 10 (Table 2) were used to test the solvent recycling. The reaction crude was filtered through a Whatman© filter paper, and the liquid fraction was reused without any treatment. At the end of the first cycle, the yield in TA was quite low (32%, Fig. 6) because TA was partially soluble in the LBDES. The yield increased in the second cycle by up to 82% because the DES solvent was saturated in TA. Finally, in the third cycle, the yield was found to exceed 100% (139%) because the TA lost in the first two cycles recovered after the complete work-up procedure. Globally, the conversion was always around 100%, and the yield over the three cycles was 78%. Furthermore, as the solid fraction and the filter remained impregnated with the solvent, some of the latter were lost during the filtration process.
MW irradiation can be compared to traditional heating through the environmental factor (E) defined as the mass of waste for the mass of the product. It should always be coupled with other factors, such as the energy economy (ε) and the combination of E and ε (environmental energy impact, ξ) (Table 3).13
Run | E | ε (°C−1 min−1) | ξ (°C min) |
---|---|---|---|
1 | 15.44 | 2.08 × 10−4 | 7.42 × 104 |
2 | — | — | — |
3 | — | — | — |
4 | 12.99 | 6.18 × 10−4 | 2.10 × 104 |
5 | 67.07 | 5.29 × 10−5 | 1.27 × 106 |
6 | 1119.14 | 2.50 × 10−6 | 4.48 × 108 |
7 | 12.16 | 5.13 × 10−4 | 2.37 × 104 |
8 | 19.09 | 5.06 × 10−4 | 3.78 × 104 |
9 | 19.74 | 4.89 × 10−4 | 4.04 × 104 |
10 | 24.40 | 5.35 × 10−4 | 4.56 × 104 |
11 | 5.15 | 5.27 × 10−4 | 9.79 × 103 |
12 | 3.46 | 5.23 × 10−4 | 6.61 × 103 |
The increase in T and decrease in t led to an improvement in the parameters (E and ξ should be minimized; ε should be maximized), and the strategy of diluting the system is particularly winning because it allows to processing of more PET concurrently. The environmental parameters are comparable to those obtained in our previous contribution27 using a much stronger acid, such as MSA (E = 2.88, ε = 4.07 × 10−5, ξ = 7.27 × 104). As depicted in Table 4, our method is compared with other methods proposed in recent literature, especially focusing on DES- and iron-based protocols.13
Method | E | ε (°C−1 min−1) | ξ (°C min) | Ref. |
---|---|---|---|---|
FeCl3·6H2O/acetic acid/MW | 3.46 | 5.23 × 10−4 | 6.61 × 103 | This work |
FeCl3·6H2O/MSA | 2.88 | 4.07 × 10−5 | 7.07 × 104 | 27 |
NaOH/MW | 1.70 | 4.47 × 10−4 | 3.8 × 103 | 39 |
H2SO4(aq.) | 8.87 | 2.22 × 10−5 | 3.99 × 105 | 40 |
Urea/ZnCl2 | 0.407 | 1.63 × 10−4 | 2.5 × 103 | 41 |
K2CO3/EG | 1.33 | 4.07 × 10−5 | 3.3 × 104 | 42 |
FeCl3/2-ethylhexanol | 1.18 | 1.21 × 10−5 | 9.75 × 104 | 43 |
FeCl3/ChCl/2-ethylhexanol | 0.57 | 2.65 × 10−5 | 2.15 × 104 | 43 |
To compare the sustainability of state-of-the-art PET management,23 a low temperature (100 °C) or a short time (10 min) when MWs are used makes our proposed technology competitive with other processes currently used for pyrolysis (T between 350 and 600 °C), glycolysis (T between 180 and 240 °C) or methanolysis (T between 180 and 280 °C, pressure between 20 and 40 atm).2
Thus, the improvement in LBDES1 efficiency was achieved by adding a simple post-reaction procedure, specifically by keeping unreacted PET, dispersed oligomers and the crude TA for 12 h in contact with an aqueous solution of NaOH (0.5 M) at room temperature under stirring. This step significantly improved both PET conversion and yield in TA, thus making all the depolymerization processes more effective. Furthermore, the effect of chloride salts as additives was investigated, and it was found that cheap and non-toxic CaCl2 positively affected the solvent performance, thus increasing the conversion up to 69%. This effect is explained by the known ability of calcium ions to coordinate water molecules, whose excess acts as an inhibitor. Moreover, the increase in conversion and yield is not so marked as to justify the use of an additive, especially because of a scale-up of the process.
Results obtained for the same reaction carried out under microwave (MW) irradiation demonstrated that if the temperature was increased to 180 °C, the reaction was quantitative and very high TA yields were obtained at low reaction time (10 minutes), even in the presence of 10 equivalents of excess water. The presence of the latter is an important goal because it allows processing in a single reaction to a significantly higher amount of PET (PET/solvent ratio = 0.26 in mass).
Under these conditions, the environmental parameters were comparable with those of much stronger acids, with an evident improvement in corrosion, costs, eco-sustainability and the ease of disposing of exhaust solvent.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d3su00205e |
This journal is © The Royal Society of Chemistry 2024 |