An Oxalate-Bridged Binuclear Iron(III) Ionic Liquid for the Highly Efficient Glycolysis of Polyethylene Terephthalate under Microwave Irradiation

Supporting Information for this article is given via a link at the end of the document. ABSTRACT: A new oxalate-bridged binuclear iron(III) ionic liquid combined with an imidazolium based cation, (dimim) 2 [Fe 2 Cl 4 (μ-ox)], was synthesized and characterized by a wide range of techniques. This halometallate ionic liquid was active catalysing the depolymerization by glycolysis of polyethylene terephthalate (PET), under conventional and microwave-assisted heating conditions. Both methodologies were very selective towards the production of bis(2-hydroxyethyl)terephthalate (BHET). The employment of microwave heating proved beneficial in terms of time and energy saving when compared to the use of thermal heating. Indeed, dielectric spectroscopy studies revealed that the binuclear iron-containing ionic liquid exhibits an excellent heating response under an electromagnetic field. The catalyst provided quantitative conversions to BHET in the glycolysis of post-consumer PET bottles in only 3 h through microwave heating, as compared to 80% conversion after 24 h under conventional heating. A new ionic liquid based on an imidazolium cation and an iron oxalate anion for the efficient glycolysis of polyethylene terephthalate (PET) under microwave and thermal heating conditions is reported here. The dielectric properties of the catalyst lead to a considerable reduction in reaction time when microwave irradiation is employed. This work provides valuable insights for the development of highly active, stable and recyclable catalysts for the efficient depolymerization of PET into highly pure monomers.


INTRODUCTION
The annual production of plastics is around 311 million metric tons. [1] Poly(ethylene terephthalate) (PET) is one of the most important plastics, since this versatile polymer is non-toxic, odourless, tasteless and exhibits excellent physical and chemical properties, such as thermal stability, mechanical and performance under conventional and microwave-assisted heating conditions and compare both methodologies. To the best of our knowledge this is the first case which describes the synthesis and catalytic application of an oxalate-bridged paramagnetic ionic liquid.
ILs, being in essence molten salts with a melting point below 100 ºC, generally exhibit very strong microwave heating properties. [17] The interaction of ILs with a microwave field is underpinned by their dielectric properties. The dielectric properties of a material are described by the value of the complex permittivity (*), [18] which consists of a real component ('), for convenience often termed dielectric constant ('), and an imaginary component termed dielectric loss (''). ' determines the amount of energy that can be stored by a material in the presence of an electromagnetic field, whereas '' denotes the amount of the stored energy that can be converted into heat. Both ' and '' are functions of temperature and frequency. In the present work, we have studied the dielectric properties of (dimim) 2 [Fe 2 Cl 4 (μ-ox)] in order to ascertain its response under microwave frequencies and determine the contribution of the iron-containing IL in the context of its heating behavior during the glycolysis process.

Scheme 1. Formation of (dimim)2[Fe2Cl4(μ-ox)] (3).
Due to the paramagnetic nature of the sample, nuclear magnetic resonance (NMR) is not a suitable spectroscopic tool to characterize iron-containing ionic liquids. Moreover, in contrast with our previous work about PET depolymerisation by imidazolium-based halometallate complexes, [13] we were not able to obtain single crystals suitable for X-ray diffraction analysis. Therefore, (dimim)2[Fe2Cl4(μ-ox)] was fully analysed by the use of a wide variety of techniques, including elemental analysis (EA), thermogravimetric analysis (TGA), inductively coupled plasma optical emission spectrometry (ICP-OES), differential scanning calorimetry (DSC), Fourier-Transform Infrared (ATR FT-IR) and Raman spectroscopies, magnetic measurements, and dielectric spectroscopy studies.
ICP gave an iron content of ca. 20.2%, which is in good agreement with the expected value (20.1%), whereas TGA showed that the reproducibility is high (ca. 19.4%, see Figure S1, Supporting Information). In addition, EA was consistent with the calculated contents of C, N and H (for further details see Experimental Section). DSC ( Figure S2, Supporting Information) did not show solid-solid (s-s) phase transitions upon cooling the complex from above the melting point (74 ºC), which starts to decompose at about 250 ºC ( Figure S1). The iron complex 3 was investigated in detail by Raman and IR spectroscopies, since these techniques are extremely useful for studying the bonding and type of imidazolium compounds [19] and the metal coordination environments of halometallates species. [20] Figure 2 displays the non-polarized Raman and FT-IR spectra of 3 in the range of 150-3900 cm -1 at 300 K. The vibrational assignments proposed herein are based on the comparison of spectroscopic behavior of the "free" oxalate anion [21] with the data described in the literature for iron oxalate complexes [22] and those published in our previous study of (dimim)[FeCl4] complex (Table S1, Supporting Information). [23] The tetrachloridoferrate (III) ion in tetrahedral (Td) coordination exhibits vibrational spectra characterized in the low-frequency region by four Raman-active vibration modes. [22] According to the proposed structural formula for the metal complex [Fe2Cl4(μ-ox)] 2-, which will display an identical structure in the plane as that depicted in Figure   1 in Td symmetry, such an anion presents four Fe-Cl Raman-active vibration modes; two of them are attributed to the bending modes [s = 111 and as = 137 cm -1 ], while the others are related to the stretching modes [ s = 332 and  as = 376 cm -1 ] (Table S1). These bands are shifted to higher energies as a result of the substitution of chlorine by oxygen atoms in Td symmetry. An estimation of vibration bending and stretching modes for the Fe-O bond can be performed on the basis of the Raman shifts corresponding to the Fe-Cl bond. [24] The Raman frequency can be described approximately by the equation (1) [21,22,25] Furthermore, the vibrational spectra of 3 feature the characteristic symmetric and antisymmetric C=O stretching bands in the vicinity of 1302 and 1605 cm -l (Table S1 and figure 2a). Hence, the observed fingerprint for the oxalate-bridged binuclear iron(III) complex is in good agreement with the reported literature. [22,25] We have also compared the Raman and IR spectra of 3 to those of (dimim) [FeCl4] in the region where the bands corresponding to imidazolium moiety appears, [23] thus confirming the presence of this cation in the Fe-based IL. Both IR and the Raman spectra show three bands at 600-1550 cm -1 assigned to CH3 (CH3-N and H-C-H stretching absorption, figure 2a). The Raman spectrum also displays two weak bands at 3110 and 3164 cm -1 recognized as H−C−H asymmetric stretch of CH3 ( Figure 2b). [19,23]   In addition, the magnetic properties of 3 were determined over the temperature range 2-300 K (for further details see Section 3, Supporting Information). The plots for the inverse of the magnetic susceptibility (1/m) and the mT product versus T at 1 kOe are shown in Figure 3. The mT value at R.T. (6.24 emuKmol -1 Oe -1 ) is lower than expected for a dinuclear species containing non-interacting Fe 3+ (S = 5/2) ions (8.754 emuKmol -1 Oe -1 ), which suggests the presence of antiferromagnetic exchange interactions at R.T. across the oxalate bridges. [26] This value decreased to 0.54 emuKmol -1 Oe -1 at 2 K and reaches an inflexion point at ca. 20 K. The inverse of the magnetic susceptibility data can be fitted to a Curie-Weiss law (red line in figure 3) with paramagnetic Curie temperatures (p) close to 44 K and an effective paramagnetic moment eff = 10.68 B/molecule. The latter is consistent with the value expected for two Fe 3+ ions with high spin d 5 configuration (eff = 5.92 B/Fe ion) and is in a good agreement with that found for other Fe(III)-based halometallate ions. [27,28] In addition, the negative temperature between both curves at Tirr = 14 K. This behaviour could be attributed to a 3D spin canting structure and is similar to that observed for analogous oxalate bridged iron(III) complexes of formulas [29] and {(MeNH3)2[Fe2Cl4(μ-ox)]·2.5H2O}n. [30] On the basis of these studies, we conclude that {(dimim)2[Fe2Cl4(μ-ox)]·H2O}n is the most likely structure for 3. Finally, and as a novelty with respect to our previous study, [13] a series of dielectric spectroscopy studies at 2470 MHz with different temperatures were undertaken in order to evaluate the heating propensity of 3 under microwave irradiation compared to that of the metallic precursor FeCl3. The dielectric characterization was carried out employing the cavity perturbation technique. [31] The dependence of both ' and '' with temperature are presented in Figure 4. The dielectric properties of FeCl 3 do not display any significant variation with temperature and remain fairly stable throughout the measurement process. Such behavior is typical of a salt below its melting point. [31d] On the other hand, both ' and '' of 3 exhibit a substantial increase with temperature above its melting point, demonstrating a thermally activated conductivity mechanism, typical for an IL. [32] These results suggest that the imidazoliumbased oxalate-bridged binuclear iron(III) IL will show superior heating behavior under an electromagnetic field at 2450 MHz, with a heating rate substantially higher than that exhibited by FeCl3 once the temperature of the reaction reaches ca. 80 o C. This observation implies a potentially good catalytic performance of 3 during a microwave-assisted glycolysis process.

Catalytic glycolysis of PET with (dimim)2[Fe2Cl4(μ-ox)] (3)
The catalytic activity of 3 was studied in the depolymerisation of PET to BHET by glycolysis under conventional heating in ethylene glycol (EG). Two sources of PET were employed: pellets (Goodfellow Inc. (3-5 mm)) (PET-1) and used PET sourced from water bottles obtained from a local market and cut to an approximate size of 6 mm (PET-2). Firstly, in this work we have performed a thorough analysis of the optimal reaction conditions. Entry [cat] (mol) Mass of consumed PET (%) BHET selectivity (%) [b] BHET conversion (%) [ [a] Reagents and conditions: 170 ºC, 24 h, PET-1 (250 mg), EG (1.5 mL). [b] BHET amount determined by 1 H NMR spectroscopy as the average of two runs, employing 1-phenyl-1,2-ethanediol as internal standard. [33] Mass of consumed PET and conversion to BHET product increase with the reaction time, but with a slight decrease in selectivity towards BHET ( Figure 5). This is attributable to the existing equilibrium between monomers and oligomers, which leads to the repolymerization of BHET product generated in the glycolysis process as the time progresses.
[12c,f] Finally, we studied the influence of temperature on the catalytic reaction ( Table 2). As expected, the degradation of PET is favoured by an increase in the reaction temperature (entries 1-4), since the higher the temperature is, the higher the mass of consumed PET and conversion to BHET product, obtaining 96% consumption and 60% conversion to BHET at 170 ºC (entry 4). In contrast, the selectivity of BHET product reached a maximum of 94% at 150 ºC but then decays with the rise of temperature due to the re-polymerisation of BHET monomer into dimers and oligomers. Therefore, the optimal reaction conditions found under conventional heating are 1.5 mL of EG, 170 ºC and 24 h of reaction time, resulting in a 60% conversion. Interestingly, the glycolysis of PET-2 (waste bottles) instead of PET-1 (pellets) was performed with complete consumption of all initial PET, which was transformed into BHET, dimers and oligomers (Table 2, entry 5). Higher selectivity and conversion towards BHET product was obtained. This can be explained by the higher surface area from PET-2. BHET selectivity (%) [b] BHET conversion (%) [b] [b] BHET amount determined by 1 H NMR spectroscopy as the average of two runs, employing 1-phenyl-1,2-ethanediol as internal standard.
In addition, several control experiments were performed under thermal heating to determine the implication of 3 in the catalytic reaction (Table 3). Firstly, no reaction was observed without metalcontaining IL (entry 1) or when glycerol was employed instead of EG (entry 2), indicating that both catalyst and EG are involved in the process. In addition, the catalytic activity of (dimim)Cl and the metal precursor FeCl3 were lower than that of the Fe-based ionic liquid under the optimized reaction conditions (entries 3 and 4). These experiments reveal that both the cation ((dimim) + ) and the anion ([Fe2Cl4(μ-ox)] 2-) play an important role in the degradation of PET, and demonstrates a synergistic effect of the two fragments. Entry Catalyst Mass of consumed PET (%) BHET selectivity (%) [b] BHET conversion (%) [ increase. [34] First, we performed a set of experiments with PET pellets at 160 ºC and 170 ºC ( Table 4).
The system required 5 W to maintain 160 ºC as reaction temperature, giving 13% conversion and 70% selectivity towards BHET in 3 h (entry 1). An increase in the reaction time up to 6 h provided an improvement in the results, obtaining the BHET product with higher selectivity and conversion (80 and 32%, respectively, entry 2). Interestingly, a significant enhancement of activity was observed at 170 ºC (entries 3-5), reaching 72% consumption of PET and 52% conversion to BHET in 3 h (entry 5), whereas the system only needed 6 W to maintain the desired temperature. On the contrary, the depolymerisation of PET catalysed by FeCl3 required 10 W, yielding 94% selectivity and 17% conversion towards BHET product (entry 6). In good agreement with dielectric spectroscopy studies, the halometallate complex acts as main heating source in such a way that the catalyst absorbs more efficiently the microwave radiation and thus needs less energy than the simple metal salt to maintain the heating at the required temperature.
Subsequently, a set of catalytic experiments were performed at a constant microwave power of 100 W and 170 ºC as maximum reaction temperature (entries 7-9). The catalytic activity increased with higher irradiation power. In fact, the process required less time (2 h, entry 9) to show comparable performance to that obtained at a lower irradiation power of 6 W (3 h, entry 5). Logically, an increase in the specific surface area resulted in an increase in activity and the use of ground PET led to a 65% conversion and 68% selectivity in 30 min, while the consumption of PET was 95% (entry 8). BHET selectivity (%) [b] BHET conversion (%) [b]  Commercial PET cut from bottles (PET-2) showed a higher mass consumption of PET and conversion and selectivity towards BHET ( Table 5). The glycolysis of PET at 160 ºC proceeded with 96% consumption in 3 h (entry 1), while the conversion and selectivity were 54% and 56%, respectively.
Quantitative consumption of PET and slightly higher conversion and selectivity towards BHET product were observed when increasing the reaction time up to 6 h (entry 2). Most importantly, we noted a significant increase in the activity at 170 ºC (entries 3-5), providing quantitative formation of BHET in 3 h (entry 7). Good results were also obtained fixing the microwave power at 100 W and the temperature at 170ºC (entries 6 and 7). Indeed, nearly quantitative consumption of PET was reached in 30 min, whereas the conversion and selectivity to BHET were higher than 70%. The catalyst reported herein displays better catalytic performance under microwave heating than those shown by iron-containing ILs described in our previous work. [13] The importance of microwave methodology in terms of efficiency and reduction in reaction time is evidenced by the comparison with the use of thermal heating, where quantitative consumption of PET was observed after 24 h of reaction time. BHET selectivity (%) [b] BHET conversion (%) [ [b] BHET amount determined by 1 H NMR spectroscopy as the average of two runs, employing 1-phenyl-1,2-ethanediol as internal standard.
In addition, gram-scale reactions with 2.5 g of discarded commercial PET bottles (PET-2) were demonstrated under both thermal and microwave heating conditions (Table 6). Through conventional heating, complete consumption of PET, and 78% conversion and selectivity towards BHET were observed (entry 1). The use of microwave methodology also provided quantitative consumption of PET, and 89% yield and selectivity towards BHET (entry 2). These experiments prove the potential application of the described catalytic system for large-scale glycolysis of PET. BHET selectivity (%) [b] BHET conversion (%) [ [a] Reagents and conditions: PET-2 (2.5 g), 3 (0.9 mmol), EG (15 mL), 170 ºC. [b] BHET amount determined by 1 H NMR spectroscopy as the average of two runs, employing 1-phenyl-1,2-ethanediol as internal standard.
Finally, the catalyst exhibited a limited potential for recyclability. We carried out several experiments to investigate the recyclability of 3. The binuclear iron complex could be recovered and used in successive cycles by precipitation in water. However, a significant decrease in the activity was observed in the second cycle ( Figure S4, Supporting Information). We have characterized the iron-containing IL after the catalytic reaction by IR and TGA. No stretching bands corresponding to oxalate were found in the IR spectrum of the recovered catalyst, which shows that the binuclear iron complex is decomposed ( Figure S5). There is also a shift in the N-CH3 band of the imidazolium moiety. This is supported by TGA, since the mass of the final residue is higher than that found in the TGA of the non-used catalyst ( Figure S6). In addition, the TGA curve exhibits several weight loss due to new decomposition temperatures. Consequently, these analyses suggest that the iron catalyst has decomposed during the purification process.

Characterization of BHET Product
The BHET product was isolated by crystallization in water at 4 °C (see Experimental Section for more details) and characterized by DSC, ATR FT-IR, TGA, NMR, electrospray ionization mass spectrometry (ESI-MS), powder X-Ray diffraction (PXRD), and gas chromatography mass spectrometry (GC-MS).
PXRD data ( Figure S7, Supporting Information) confirmed unit cell parameters in agreement with literature reports (Table S2, Supporting Information), [35] with no unmatched diffraction peaks. The polymer (Inset of figure S7) shows the characteristic diffraction pattern of PET with a semicrystalline structure. [36] 1 H and 13 C NMR spectra confirmed the purity of the product ( Figure S8 and S9, Supporting Information). [37] A melting point of 109.7 ºC was confirmed by DSC, which is in good agreement with that published for BHET ( Figure S10, Supporting Information). [34,38] Similarly, TGA gave a weight loss of ca. 27% at 210 ºC, due to the thermal decomposition of BHET product, and another weight loss of about 61% at 422 ºC corresponding to the thermal decomposition of PET repolymerized during the heating performed in the TGA analysis ( Figure S11, Supporting Information). [12c,37,39] In addition, FT-IR spectrum is consistent with literature reports ( Figure S12, Supporting Information). [37] Finally, the purity of the BHET synthesized was further confirmed by GC-MS and ESI-MS (Figures S13 and S14, Supporting Information), which demonstrated that highly pure BHET product was obtained. Indeed, no signals attributable to other compounds were detected.

CONCLUSION
In summary, a new oxalate-bridged binuclear iron(III) ionic liquid based on imidazolium cation, , has been synthesized, fully characterized and successfully employed as catalyst for the glycolysis of PET, under thermal and microwave-assisted heating conditions. Such an oxalate complex was selected for its dielectric properties, which involve a higher heating rate under an electromagnetic field compared to that of FeCl3 and thus a potentially better catalytic performance during a microwave-assisted glycolysis process. Quantitative consumption of PET and high conversions to BHET were obtained through the two approaches, but much shorter reaction times and energy consumption were observed when microwave-assisted heating conditions were employed. Indeed, the complex provided higher conversion to BHET product through microwave conditions in remarkably less time than that required under conventional heating (>99% in 3 h vs 80% in 24 h for post-consumer PET).
In conclusion, we have devised a novel oxalate-bridged halometallate ILs with very exciting dielectric properties, which has evidenced its suitability as an active catalyst for degradation of PET, in particular under microwave heating conditions.

General
Reagents were purchased from Sigma Aldrich and purified when required following literature protocols. [40] Chemical shifts of 1 H and 13 C NMR are reported in ppm, employing the solvent as internal standard. Signals are quoted as s (singlet), t (triplet).

Catalytic depolymerisation of PET
General procedure for thermal heating: Protocol adapted from reference 13. Typically, 250 mg of PET (commercial pellets or PET waste), 1.5 mL of EG and the corresponding amount of catalyst were placed in a 10 mL round bottom flask. The mixture was heated under reflux to the selected temperature and maintained during the reaction time, and afterwards the solution was allowed to cool down to R.T.
Then, the reaction mixture was combined with 100 mL of distilled water. The unreacted polymer was separated and dried at 80 °C, and the mixture was filtered out. The water was evaporated under vacuum (72 mbar) at 40 °C. 100 mg of the remaining solution was dissolved in DMSO-d6 and analyzed by 1 H and 13 C NMR using 1-phenyl-1,2-ethanediol (0.1 mmol) as internal standard. BHET was crystallized at 4 °C by addition of 7 mL of distilled water to the residue.
General procedure for microwave-assisted heating: Protocol adapted from reference 13.
These experiments were performed in a CEM microwave reactor. Typically, 250 mg of PET (pellets or waste PET) 1.5 mL of EG and the corresponding amount of catalyst were placed in a special reaction vial. The vial was inserted in the microwave reactor. Then, the microwave was programmed to maintain a constant temperature with either fixed or automatically controlled heating power. After the reaction was finished, the vial was allowed to cool to R.T. and a similar procedure to that previously described was employed.
The consumption of PET is calculated as follows: Mass of consumed PET (%) = 0 − 1 0 × 100 (Eq. 2) where W0 corresponds to the initial weight of PET and W1 corresponds to the weight of unreacted PET. In addition, the conversion and selectivity towards BHET product are defined by equations (3) and (4)

CONFLICTS OF INTEREST
There are no conflicts of interest to declare.