Synthesis and Verification of Biobased Terephthalic Acid from Furfural

Exploiting biomass as an alternative to petrochemicals for the production of commodity plastics is vitally important if we are to become a more sustainable society. Here, we report a synthetic route for the production of terephthalic acid (TPA), the monomer of the widely used thermoplastic polymer poly(ethylene terephthalate) (PET), from the biomass-derived starting material furfural. Biobased furfural was oxidised and dehydrated to give maleic anhydride, which was further reacted with biobased furan to give its Diels-Alder (DA) adduct. The dehydration of the DA adduct gave phthalic anhydride, which was converted via phthalic acid and dipotassium phthalate to TPA. The biobased carbon content of the TPA was measured by accelerator mass spectroscopy and the TPA was found to be made of 100% biobased carbon.

carbon derived from biomass [27][28][29] . It is determined by measuring the value of the 14 C/ 12 C ratio in the material, and is based on the assumption that this ratio for petroleum-derived (or 'ancient') carbon is 0, while the ratio for biobased (or 'modern') carbon is 1 3 10 212 . In our work, this ratio was measured using accelerator mass spectroscopy (AMS) according to guidelines listed in ISO 16620-2. The details of the measurement procedure have been previously reported in the literature 28 .

Results and Discussion
Oxidation of furfural to fumaric acid and maleic acid. The oxidisation of furfural with NaClO 4 as an oxidant, and V 2 O 5 as a catalyst, gave a mixture of fumaric acid and maleic acid in 58% yield, lower than the 72% yield reported in the literature 30 . This lower yield could be caused by vigorous oxidation, which is difficult to control. The oxidation of furfural to maleic acid and fumaric acid has been performed by other means with improved yields and better selectivity for the products, either maleic acid or fumaric acid, elsewhere in the literature 31,32 . However, the oxidation with NaClO 4 is, for our purposes, a more practical laboratory process, and, therefore, we adopted this traditional oxidation method in this study.
Dehydration of fumaric acid and maleic acid to maleic anhydride. The dehydration of the mixture of maleic acid and fumaric acid to maleic anhydride was performed using P 2 O 5 as a dehydration agent. As the ratio of maleic acid and fumaric acid, determined by 1 H NMR, was 157, the yield of maleic anhydride could potentially be less than 30%. However, once maleic acid was dehydrated to give maleic anhydride, phosphoric acid, produced by the reaction of P 2 O 5 and water, could isomerise fumaric acid to maleic acid 33 . Consequently, the mixture of maleic acid and fumaric acid was quantitatively converted to maleic anhydride.
Diels-Alder (DA) reaction of anhydrous maleic acid and furan to the exo-DA adduct. The synthesis of furan from furfural could have been demonstrated in this study. However, commercially available furan is a biobased chemical and has been verified as such in previous work 13 . The biobased carbon contents of the furan and furfural used in this study are shown below. Therefore, although we have not performed the process ourselves, for the purposes of this work, commercially available furan is assumed to be biobased.
The DA reaction of maleic anhydride and furan readily gave the DA adduct. At the beginning of the reaction, the regioselectivity of DA cyclisation is for the endo-adduct due to its kinetic stability. However, the DA reaction is reversible and, after some time, the product is converted to the more thermally stable exo-DA adduct [34][35][36] . Consequently, the reaction was carried out for 12 h at room temperature, yielding the DA adduct in almost quantitative yield. Melting point analysis showed the m.p. of the product to be 127-129uC, corresponding to the exo adduct.
Dehydration of the exo-DA adduct to phthalic anhydride. The oxabicyclo moiety in the exo-DA adduct is readily dehydrated with an acid. We attempted to dehydrate the exo-DA adduct using sulfonic acid, phosphoric acid, and P 2 O 5 , but the yield and purity of the phthalic anhydride produced were not high enough to isolate it. However, a more effective protocol for the dehydration of the exo-DA adduct has been recently developed in which it is treated with a mixture of trifluoromethane sulfonic acid and acetic anhydride 37 . We employed this new method and obtained phthalic anhydride in 84% yield.
Hydrolysis of phthalic anhydride to dipotassium phthalate. Phthalic acid was readily hydrolysed with aqueous potassium hydroxide to give dipotassium phthalate quantitatively.
Transfer reaction and acidification of dipotassium phthalate to TPA. Half a century ago, a transfer reaction known as the Henkel method, which converts dipotassium phthalate to dipotassium terephthalate at high temperature (above 400uC) with CdI 2 as a catalyst, was the most common industrial process for the production of TPA 38,39 . The development of an alternative method involving the oxidation of p-xylene to TPA led to the Henkel method losing its competitive advantage and falling out of favour. Consequently, it is rarely used industrially. However, in this study, we adopted the Henkel method to convert biobased phthalate to biobased TPA, as we found it to be a practical method for obtaining biobased TPA from furfural. The reaction was carried out with CdI 2 at 420uC, and the resulting mixture was acidified to give biobased TPA. At 44%, the yield of TPA obtained in this study is not sufficient. However, this figure is obtained at the milligram scale, but the process was optimised industrially, so, therefore, it is reasonable to expect that the yield would increase for industrial production. Additionally, the Henkel method is proven as an industrial process, and, therefore, the commercial viability of this synthetic route from furfural to TPA is already established.
Biobased carbon content. The biobased carbon contents of the reagents and products are summarised in Table 1. The synthesis of fully biobased TPA is verified by the fact that the values of furfural, furan, and TPA were almost 100%. Thus, we can conclude that both the starting materials and product are fully biobased chemicals.
On the other hand, the values for furfural and furan reported previously were 100.8 and 105.0% 11,13 and slightly higher than those measured in this study. These values were obtained in 2010, while those in this study were obtained in 2014. Since the lot numbers of furfural and furan used in this study are different from those used in the previous study, the actual value of the 14 C/ 12 C ratio could be slightly different. This difference could be explained by the difference in the definition of biobased carbon content between ISO 16620-2 and ASTM D6866 and the manufacturing year of furfural and furan. In principle, the percentage of modern carbon (pMC) calculated from the 14 C/ 12 C concentration ratios, is the biobased carbon content. However, the pMC for biomass produced by fixation of CO 2 in the atmosphere by photosynthesis was 108-110% in 2002 [27][28][29] . The pMC is possibly slightly higher than 100% because of the continuing but diminishing effects of nuclear testing in the atmosphere in the 1950s, during which large amounts of 14 C were emitted into the atmosphere. Because the 14 C in all the samples is referenced to a ''prebomb'' standard, i.e., modern carbon-based oxalic acid radiocarbon [Standard reference material (SRM) 4990c, National Institute of Standards, USA], all pMC values must be multiplied by a cofactor to correct for the bomb carbon and to obtain the true biobased carbon content of the sample. In our previous study, the biobased carbon contents were calculated using a strong cofactor of 0.93, that being the old value based on ASTM D6866 (2008) because the furfural and furan used were purchased before 2009. Nevertheless, the reason why the value of furan was above 105% is that it was produced before 2008. In this study, as the reagents used were purchased in 2013, the biobased carbon contents were calculated using the new value for the weak cofactor of 0.95 based on ISO 16620-2. These indicate that the nuclear testing effect on the old reagents was strong and the biobased carbon content was above 100%, even though the strong cofactor 0.93 was used, while the effect on the new reagents was weaker, giving a biobased carbon content of almost 100%. Therefore, the biobased carbon contents of furan and furfural in this study were slightly different from the values measured in the previous study.
The precise method for the measurement of biobased carbon content is detailed in ISO 16620-2 and ASTM D6866 and is an industrially indispensable verification procedure 29,40 . It is important, not just to prevent mistakes by researchers, but also to detect whether supposedly biobased materials have undergone some contamination from, or carbon exchange with, petrochemical sources such as other reaction reagents or non-biobased solvents. For example, in the case of the Henkel method, the transfer reaction could involve the incorporation of a carbonyl carbon from carbon dioxide, produced as a by-product of a petrochemical process 38,41 . In addition, biobased carbon content measurement is also an invaluable method for identifying materials mistakenly or falsely supplied as biobased. Therefore, we propose that the measurement of biobased carbon content should be necessary when biobased chemicals are used, especially when the products can be synthesised from commercially available petroleum-derived starting materials or involve the use of non-biobased reagents or solvents.
In summary. We successfully synthesised biobased TPA from furfural and furan using viable and proven organic synthetic procedures. Furthermore, the biobased carbon content of the TPA that we synthesised confirmed that it is a truly biobased product. Using furfural as a single resource is a novel and interesting concept, since furfural can be produced from inedible cellulosic biomass.
The aim of this study was to propose a viable synthetic route from furfural alone to TPA, and we have succeeded in this. It is our hope that more research, conducted by both ourselves and, perhaps, other groups, will optimise this process so that it may be industrialised.
We finally propose that the measurement of biobased carbon content is indispensable as a verification method in the research area of biobased synthesis.

Methods
Materials. Furan, sodium chlorate, vanadium pentoxide, phosphorus pentoxide, methane sulfonic acid, acetic anhydride, cadmium iodide, toluene, and diethyl ether were purchased from Wako Pure Chemical Industries (Osaka, Japan). Furfural, potassium hydroxide, and hydrochloric acid were purchased from Kanto Kagaku Co., Inc (Tokyo, Japan). Phosphorus pentoxide was purchased from Kishida Chemical Co., Ltd (Osaka, Japan). Furfural, trifluoromethane sulfonic acid, and acetic anhydride were used after distillation under reduced pressure. All other chemicals were of reagent grade and used without further purification.
Instrumentation. 1 H NMR spectra were recorded on a 400 MHz NMR spectrometer (JNM-ECX400; JEOL, Tokyo, Japan) using deuterated chloroform or deuterated dimethyl sulfoxide as a solvent, and tetramethylsilane as an internal standard.
Measurement of biobased carbon content 28 . Measurements of the ratios of the three carbon isotopes ( 14 C, 13 C, and 12 C) using AMS were performed at the Institute of Accelerator Analysis Ltd (IAA) (Fukushima, Japan) using a 3-MV tandem accelerator (National Electrostatics Co., Middleton, WI, USA, Pelletron 9SDH-2). The pMC was calculated from the 14 C/ 12 C concentration ratios for the sample. The biobased carbon content was determined from the ratio of 14 C/ 12 C concentrations of the sample according to ISO 16620-2. D 14 C is the isotope differential ratio of 14 C between the sample and reference material. Reference materials were also analysed using AMS. The biobased carbon content was calculated as follows: Oxidation of furfural to fumaric acid and maleic acid 30 . Furfural (36 g, 367 mmol) was carefully added dropwise to a solution of sodium chlorate (80 g, 751 mmol) and vanadium pentoxide (360 mg, 1.98 mmol) in water (10 mL) at 90uC over 3 h, and the mixture was stirred at 80uC for a further 10 h. The mixture was allowed to stand at room temperature for 11 h, affording a white crystalline precipitate. The precipitate was filtered and dried to give 42.3 g (58%) of a mixture of fumaric acid and maleic acid as white crystals. 1  Dehydration of exo-DA adduct to phthalic anhydride 37 . The exo-DA adduct (1.00 g, 6.02 mmol) was added to a mixture of methane sulfonic acid (10.0 mL, 154 mmol) and acetic anhydride (2.0 mL, 21 mmol) under a N 2 atmosphere at 0uC. The reaction mixture was allowed to warm to room temperature and stirred for 2 h. It was then heated to 80uC and stirred for 4 h. After cooling to room temperature, the reaction mixture was extracted with toluene (3 3 20 mL). The combined toluene extract was washed with saturated sodium hydrogen carbonate solution and saturated sodium chloride solution before being dried over anhydrous sodium sulfonate. After filtration, the organic layer was evaporated in vacuo to give 746 mg (84%) of phthalic anhydride as white crystals. 1  Hydrolysis to phthalic anhydride to dipotassium phthalate 43 . Phthalic anhydride (500 mg, 3.38 mmol) was dispersed in a solution of potassium hydroxide (1.00 g, 17.8 mmol) in water (10 mL). The reaction mixture was allowed to stir at 80uC for 2 h. The resulting solution was poured into ethanol (150 mL) affording a white precipitate which was filtered to give 802 mg (98%) of dipotassium phthalate as white crystals. 1  Conversion and acidification of dipotassium phthalate to TPA 43 . A mixture of dipotassium phthalate (800 mg, 3.30 mmol) and cadmium iodide (40 mg, 0.11 mmol) was ground in a pestle and mortar. The mixture was placed in a sealing tube and sealed in vacuo. After heating to 420uC for 2 h, the resulting solid was placed into hot water (30 mL). The dispersion was refluxed for 10 min and the remaining precipitate was removed by filtration. The filtrate was acidified with 1 M HCl and the resulting solid was filtered off to give 242 mg (44%) of a white solid. 1