Highly Efficient Conversion of Fructose to Furan Compounds in Ethanol Using Sulfonated Polymers with Solvent Moieties to Inhibit Product Degradation

TL;DR

A new sulfonated polymer catalyst efficiently converts fructose to furan compounds in ethanol while minimizing product degradation.

Contribution

A novel sulfonated polymer with solvent moieties enhances fructose conversion efficiency and product stability.

Findings

The 1.5VP/0.64SPSS/0.37DVB catalyst achieved an 81.9% EMF yield and 92.7% total furan yield.

The catalyst showed high HMF yield in pure tetrahydrofuran.

Improved product stability was attributed to the polymer's solvent moieties.

Abstract

The catalytic dehydration of fructose to 5-ethoxymethylfurfural (EMF) in ethanol provides a promising approach for low-carbon chemical production. However, current catalytic systems generally suffer from a trade-off between reaction efficiency and product selectivity. Herein, we show that incorporating solvent moieties to sulfonated polymer enables the highly efficient conversion of fructose to furan compounds in ethanol via restraining product degradation. The co-polymerization of N-vinyl-2-pyrrolidinone, with divinylbenzene (DVB) and sodium p-styrene sulfonate (SPSS) gave 1.5VP/0.64SPSS/0.37DVB that has slightly lower acid contents and inferior pore structure than the co-polymer of DVB and SPSS. The 1.5VP/0.64SPSS/0.37DVB catalyst exhibited maximal EMF yield of 81.9% with a total furan yield of 92.7%, Which is remarkably higher than previous reports. Moreover, the 1.5VP/0.64SPSS/0.37DVB catalyst gave a high HMF yield in pure tetrahydrofuran. The superior performance was attributed to the improved stability of the product. Our findings will instruct the design of active and selective catalysts to facilitate the production of biomass-derived products.

Figures

Full text

1. Introduction

The exploration of renewable lignocellulosic biomass resources to progressively displace fossil resources is an important approach to achieve carbon neutrality and sustainable development [1,2,3]. As the main constituents of lignocellulosic biomass, inedible carbohydrates including cellulose and hemicelluloses provide abundant feedstocks for biorefinery [4,5,6,7]. The most outstanding difference between biomass-derived carbohydrates and fossil-derived feedstocks is the remarkably higher oxygen contents. Theoretically speaking, carbohydrates are more suitable than fossil-derived feedstocks for the production of oxygen-containing fine chemical and fuels if oxygen can be reduced via hydrodeoxygenation or dehydration [8]. The conversion of biomass feedstocks to target products via hydrodeoxygenation generally requires harsh reaction conditions (high temperature, high pressure) and consumes large amount of hydrogen (H_2_), resulting in unfavorable input-output ratios. Comparatively, the conversion of carbohydrates to furan compounds via the facile dehydration process is more likely to achieve high input-output ratios due to the high potential to reduce energy consumption and production costs [9]. Thus, diverse catalytic systems and reaction routes have been investigated for this kind of conversion [8,9].

As the pivot of biorefinery, synthesizing 5-hydroxymethylfurfural (HMF) from hexoses and their glycans has been attempted for more than one centenary [10]. In this process, the fundamental reaction mechanisms, pathways and the corresponding catalyst design strategy for HMF production have been clearly established [11,12,13,14]. High HMF yield have been achieved in some elaborately designed catalytic systems under certain conditions [15,16,17,18,19,20,21,22]. However, achieving high yield usually requires low substrate loading and high-boiling point solvents, such as ionic liquids, deep eutectic solvents and dimethyl sulfoxide (DMSO) even with the most tractable fructose as feedstock, virtually resulting in low production efficiency [8,9]. The increase in substrate loading, and the subtle change in catalytic systems and reaction parameters all lead to the decline in HMF yield and selectivity due to its poor stability [23,24]. 5-(chloromethyl)furfural and 5-(bromomethyl) furfural from carbohydrates have been studied as alternatives to circumvent the awkward product degradation problem, in view of their relatively easier synthesis and better stability than HMF [25,26,27]. Nevertheless, the consumption of concentrated hydrochloric acid, dichloromethane or dichloroethane as both solvent and reactant, as well as the treatment of Cl-containing wastes would be an intractable obstacle for possible industrial production.

Another HMF alternative is 5-ethoxymethylfurfural (EMF) which can be synthesized from hexoses in ethanol, a cheap and green solvent with a low boiling point [28]. The production of EMF does not consume an external hydrogen source, and the caloric value (8.7 kWh/L) of EMF is comparable to gasoline (8.8 kWh/L) and 42.6% higher than ethanol (6.1 kWh/L) [29]. The application of EMF as a fuel additive could significantly reduce the emission of fine particles and SO_2_ from engines [30]. In addition, EMF is more stable than HMF, which shows great potential to replace the latter as a feedstock to synthesize other valuable products [31,32]. With Brønsted acid as the catalyst, the etherification of HMF with ethanol could afford the highest EMF yield of 98% [33]. However, the one-pot conversion of fructose to HMF of high yield is very challenging. The use of acidic ionic liquid 1-butyl-3-methylimidazolium hydrogen sulfate (abbreviated to [Bmim][HSO_4_] [34] or [C_4_mim][HSO_4_] [35] in different references) as both a homogeneous catalyst and reaction medium gave the highest EMF yield around 79%, while most of the heterogeneous catalysts gave maximal EMF yields around 67% [31,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52] with three exceptive yields of 72.2%, 74.6% and 71% obtained with MOP-SO_3_H [36], SNC-chitin-2 [40] and 25wt%TPA/OAC [42], respectively. On the whole, the conversion of fructose in ethanol has larger potential for practical production due to the advantage of the reaction medium, but current catalytic systems generally suffer from lower product yield in relative to fructose dehydration to HMF in high-boiling point solvents [20,53,54,55,56]. Therefore, it is imperative to establish effective catalytic systems to boost the EMF yield.

In this study, we report a sulfonated polymer with solvent moieties via the facile co-polymerization process to achieve the highly efficient conversion of fructose to furan compounds in ethanol via restraining product degradation. A series of characterizations showed that the co-polymer (1.5VP/0.64SPSS/0.37DVB) of N-vinyl-2-pyrrolidinone, divinylbenzene (DVB) and sodium p-styrene sulfonate (SPSS) have slightly lower acid contents and inferior pore structure than the co-polymer of DVB and SPSS. As one of the most representative solid acid materials, commercial Amberlyst-15 was used as the benchmarking for material characterization and performance evaluation. The 1.5VP/0.64SPSS/0.37DVB catalyst gave maximal EMF yield and total HMF/EMF yield in ethanol, superior to previously reported catalytic systems, and also gave high HMF yield in pure tetrahydrofuran (THF), as a consequence of the improved stability of furan compounds under reaction conditions. These findings may guide the design of active and selective catalysts to facilitate the production of biomass-derived products.

2. Results and Discussion

2.1. Preparation and Characterizations of Sulfonated Polymers

Scanning electron microscopy (SEM) and transmission electron microscopy (TEM) were used to observe the morphology of different materials. The SEM images (Figure 1) showed that 3.2SPSS/2DVB comprises large amounts of diminutive particles of several nanometers with a coarse surface and affluent pores, whereas 1.5VP/0.64SPSS/0.37DVB has large particle sizes, a smooth surface and few pores. The TEM images (Figure 2) further showed that 3.2SPSS/2DVB has plenty of wormhole-like mesopores with wide pore distribution, while 1.5VP/0.64SPSS/0.37DVB (Figure 3) exhibits a smooth surface without notable pores. The highly porous structure of 3.2SPSS/2DVB is in accordance with previous reports [57]. The disappearance of the porous structure in 1.5VP/0.64SPSS/0.37DVB suggests that introduction of vinylpyrrolidone remarkably reinforce the crosslinking of polymers. In addition, high angle annular dark-field (STEM-HAADF) images and the corresponding energy dispersive X-ray spectroscopy (EDS) mapping images demonstrated that C, O, N and S elements are homogeneously distributed on the polymer.

The N_2_ adsorption–desorption analysis was carried out to further analyze the textural properties of different materials (Figure 4). The 3.2SPSS/2DVB material showed typical type-IV isotherms, confirming the formation of a mesoporous structure. The BET specific surface area of 3.2SPSS/2DVB was 339.7 m^2^ g^−1^ with a relativly high pore volume (0.62 cm^3^ g^−1^) and average pore diameter of 7.7 nm, as is in accordance with previous reports [57]. Compared with 3.2SPSS/2DVB, 1.5VP/0.64SPSS/0.37DVB exhibited a remarkably lower BET specific surface area and pore volume (Table 1), as is consistent with the SEM and TEM results. These results indicated that the co-polymerization of N-vinyl-2-pyrrolidinone with DVB and SPSS could greatly reinforce the crosslinking of polymeric networks to generate large particles with smooth surfaces and few pores.

Element analysis was carried out to investigate the C, N and S contents of different materials. Among the three kinds of materials, commercial Amberlyst-15 exhibited the highest S content (Table 2). Both C and H contents in 1.5VP/0.64SPSS/0.37DVB were comparable to those in 3.2SPSS/2DVB, while the S content in the former was slightly lower than the latter. The N content in 1.5VP/0.64SPSS/0.37DVB was measured to be 1.4 wt%. In view of N content in N-vinyl-2-pyrrolidinone, the content of pyrrolidinone groups in 1.5VP/0.64SPSS/0.37DVB was estimated to be around 11.2 wt%. Although the incorporation rate is lower than the initial ratio in precursors, this analysis proved that considerable amount of N-vinyl-2-pyrrolidinone has been successfully incorporated into the resultant 1.5VP/0.64SPSS/0.37DVB material.

The thermal stability of the three kinds of material was analyzed with thermogravimetric analysis. The TG curves (Figure 4c) of all the samples show weight loss in three temperature ranges, corresponding to the removal of adsorbed water (30–180 °C) disintegration of sulfonic groups (240–460 °C) and complete decomposition of the polymeric framework (>460 °C), respectively [57]. The weight retention rates of both 1.5VP/0.64SPSS/0.37DVB and 3.2SPSS/2DVB are obviously larger than that of Amberlyst-15 in the second temperature range, indicating that the highly cross-linked structure triggered by DVB might enhance the stability of sulfonic groups to some extent.

As shown in the FTIR spectra (Figure 4d), all the three samples exhibit the bands (1032, 1124, 1232 cm^−1^) due to the stretching vibration of S-O and O=S=O bonds, affirming the successful incorporation of SO_3_H groups into the polymeric framework [36]. As is different from 3.2SPSS/2DVB, 1.5VP/0.64SPSS/0.37DVB show an additional band at around 1651 cm^−1^ that is assigned to C=O groups [58], confirming the incorporation of pyrrolidone into 1.5VP/0.64SPSS/0.37DVB. It should be pointed out that the similar band of the C=O group in Amberlyst-15 is mainly resulted from the COOH generated by concentrated H_2_SO_4_ induced oxidation during the sulfonation process. In addition, the large bands at 3435 and 2928 cm^−1^ are associated with OH groups from both surface groups and adsorbed water. The intensity of the OH groups decreases in the following order: Amberlyst-15 > 1.5VP/0.64SPSS/0.37DVB > 3.2SPSS/2DVB, suggesting that both 1.5VP/0.64SPSS/0.37DVB and 3.2SPSS/2DVB are more hydrophobic than Amberlyst-15. Moreover, X-ray photoelectron spectroscopy (XPS) analysis further confirms the existence of C, O, N and S elements (Figure 5). The high-resolution O 1s XPS spectra exhibited three peaks of C=O (531.5 eV), SO_3_H group (532.9 eV) and C–O (534.1 eV), respectively [59], while the high-resolution S 2p XPS spectra showed S 2p peaks [60]. These results confirm the existence of SO_3_H groups and N-containing species in 1.5VP/0.64SPSS/0.37DVB. In addition, the NH_3_ temperature programmed desorption (NH_3_-TPD) curve (Figure S2) could also demonstrate the presence of acidic sites in 1.5VP/0.64SPSS/0.37DVB and 3.2SPSS/2DVB. Nevertheless, quantitative comparison of the acid amounts of different materials with NH_3_-TPD analysis was not conducted in view of their decomposition at temperatures higher than 300 °C.

2.2. Catalytic Tests for Fructose Conversion to EMF

The catalytic performances of 1.5VP/0.64SPSS/0.37DVB and 3.2SPSS/2DVB were studied for fructose conversion to EMF under different conditions. When the reaction was performed with relatively high fructose loading (4.5 wt%) at a reaction temperature between 110 and 130 °C, both 1.5VP/0.64SPSS/0.37DVB and 3.2SPSS/2DVB could completely convert fructose within 60 min (Figure 6a). The 1.5VP/0.64SPSS/0.37DVB and 3.2SPSS/2DVB also exhibited comparable total yields of furan compounds but with different product distributions. At 110 °C, 3.2SPSS/2DVB gave a maximal EMF yield of 64.8% in 90 min (Figure 6b). In contrast, 1.5VP/0.64SPSS/0.37DVB gave an EMF yield of 49.5% even with a prolonged reaction time. Under this temperature, the total HMF/EMF yield was around 74%. When the reaction temperature was improved to 120 °C (Figure 6c), the maximal EMF yields over 1.5VP/0.64SPSS/0.37DVB and 3.2SPSS/2DVB were raised to 62.3% and 67.5%, respectively. Meanwhile, the total HMF/EMF yield (78.1%) over 1.5VP/0.64SPSS/0.37DVB was obviously higher than that (74.4%) over 3.2SPSS/2DVB. When the reaction temperature was improved to 130 °C (Figure 6d), the maximal EMF yields over 1.5VP/0.64SPSS/0.37DVB and 3.2SPSS/2DVB were raised to 69.5% and 69.3%. The total HMF/EMF yield over 1.5VP/0.64SPSS/0.37DVB was further raised to 79.1%. Overall, the reactions over 3.2SPSS/2DVB were faster than that over 1.5VP/0.64SPSS/0.37DVB, probably due to the higher amount of acid sites and larger specific surface area. However, neither the EMF yield nor the total HMF/EMF yield was limited by the compromise between the low reaction rate at relatively low reaction temperature and aggravated side-reactions at elevated temperatures. At relatively high temperature, 1.5VP/0.64SPSS/0.37DVB exhibited some advantages over 3.2SPSS/2DVB in achieving a high total yield. These results indicated that the incorporation of NMP-associated groups could inhibit the degradation of target products to promote selective conversion.

To further improve the yield, we also investigate the influence of other parameters, including initial fructose loading and catalyst loading. At relatively low fructose loading (1 wt% of fructose), 1.5VP/0.64SPSS/0.37DVB exhibited more obvious advantages over 3.2SPSS/2DVB for achieving higher total HMF/EMF yields (86.5% vs. 80.5%) probably due to the inhibition of product degradation (Figure 7a). Under constant reaction conditions (130 °C, 40 min, 1 wt% of fructose), the use of an appropriate amount of catalyst (15 mg) could further improve the EMF yield to 81.9% with total HMF/EMF yields as high as 92.7% in the medium of pure ethanol (Figure 7b), as has seldom been achieved in previously reported catalytic systems. In contrast, no HMF and EMF were detected in the control experiment without a catalyst (Table S1). The yield of HMF decreases slightly with the further increase in the catalyst, indicating that the increase in the catalyst could promote the further conversion of HMF to EMF at the expense of slight decreases in total yields from 92.7% to 81.4%. With the increase in fructose loading from 1 wt% to 9 wt% (Figure 7c), the product distributions evolved from EMF as leading products to the mixed products of HMF and EMF. In previous studies [28], relatively long reaction times (10–24 h) are typically used for EMF production (Table S2) and the accumulation of HMF has been commonly observed, indicating that rate of EMF generation is remarkably lower than HMF formation. Thus, the evolution of product distributions at high fructose loading is due to the lower rate of HMF etherification toward EMF relative to fructose dehydration to HMF, leading to the accumulation of unreacted HMF. At a fructose loading as high as 9 wt%, the total HMF/EMF yield was still high (75.1%). Such a high yield from a high concentration substrate has rarely been achieved in a green solvent like ethanol. These results demonstrate that the 1.5VP/0.64SPSS/0.37DVB is powerful for the conversion of high concentration substrates to furan compounds.

The catalytic performance of these catalysts for fructose conversion to HMF in tetrahydrofuran (THF) was also compared. It is notable that 1.5VP/0.64SPSS/0.37DVB exhibit a HMF yield remarkably higher than that obtained with Amberlyst-15 (Figure 7d). As is in good agreement with previous reports, Amberlyst-15 exhibited a low HMF yield despite its high acid amounts due to serve side-reactions. In contrast, 1.5VP/0.64SPSS/0.37DVB afforded a HMF yield around 80% due to the inhibited degradation process. Although the improvement of the HMF yield is limited, such an improvement established on a high datum line is highly desired in view of the severely adverse effect of humins on the industrial production of HMF. The conversion of inulin also gave a high EMF yield of 73.7% (Table S1), but the conversion of sucrose and glucose delivered an inferior performance.

Furthermore, we conduct benchmarking analysis via comparing the catalytic performance of 1.5VP/0.64SPSS/0.37DVB with the current state of the art catalytic systems designed for the conversion of fructose in ethanol. Since the conversion of fructose to EMF universally suffers from the trade-off between the low reaction rate under mild temperatures and the rapid product degradation under high temperatures, the optimal product yields were usually reported in previous studies as a predominant parameter to evaluate the performance. For an unambiguous comparison, we collected data of the optimal product yields over different catalysts obtained at their corresponding optimized reaction conditions (Supporting Information, Table S2) [31,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52]. As shown in Figure 8, the use of acidic ionic liquid 1-butyl-3-methylimidazolium hydrogen sulfate (abbreviated to [Bmim][HSO_4_] [34] or [C_4_mim][HSO_4_] [35] in different references) as both the homogeneous catalyst and reaction medium gave the highest EMF yield around 79%. In view of the high cost and the difficulty of ionic liquid recycling, the use of a heterogeneous catalyst is highly desired. Among the collected data on the previously reported heterogeneous catalysts, most of the catalysts gave maximal EMF yields no more than 69% in the medium of pure ethanol with total HMF/EMF yields no more than 80%, except for MOP-SO_3_H [36], SNC-chitin-2 [40] and 25wt%TPA/OAC [42] that deliver maximal EMF yields of 72.2%, 74.6% and 71%, respectively. It is very challenging to further improve the optimal yield, especially when using ethanol as the reaction medium with a heterogeneous catalyst. Therefore, although the reaction conditions are different, the improvement of the optimal yield and selectivity could represent the enhanced catalytic performance. According to this analysis, the 1.5VP/0.64SPSS/0.37DVB catalyst reported here exceeded previous homogeneous and heterogenous catalytic systems for both the EMF yield and total HMF/EMF yield [31,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52]. As for the conversion of fructose to HMF, the 1.5VP/0.64SPSS/0.37DVB catalyst also revealed remarkable superiority in low-boiling point solvent THF. In fact, although high HMF yields approaching 100% have been extensively claimed in previous studies, they depended highly on the use of high-boiling point solvents, including dimethyl sulfoxide (DMSO) [53,54,55], ionic liquids [56], deep eutectic solvent [20] and other complicated mixed solvents (Supporting Information, Table S3) [61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78].

To validate the stability of the catalyst, the spent catalyst was collected, washed, dried and then reused in the recycling experiment. The HMF and EMF yields were well retained in the second run’s recycling experiment. Although the EMF yield decreased slightly from the third to the fifth run, the total HMF/EMF yield could be maintained (Figure 9a). During the fifth run, the EMF yield could still reach up to 80.0% when the reaction time was extended from 40 to 70 min. The FTIR spectra of the spent catalyst were similar to the original material (Figure 9b). The elementary composition of the spent catalyst was also close to the original material (Table 2). In addition, the NH_3_ temperature programmed desorption (TPD) analysis (Figure S2) also showed that the acid amount of the recovered catalyst was close to the original catalyst. These results suggest that the catalyst could be reused in recycling experiments. The gradual decrease in catalytic performance is probably a result of carbon deposition due to the unavoidable generation of humins.

We also perform additional experiments to reveal the reason for the improved performance of 1.5VP/0.64SPSS/0.37DVB over previous catalysts for the conversion of fructose to EMF and HMF. It should be noted that the conversions of fructose to EMF and HMF are fundamental dehydration reactions predominantly driven by strong Brønsted acid. As discussed above, both the fructose conversion rate and the EMF formation rate over 1.5VP/0.64SPSS/0.37DVB are slightly lower than that over 3.2SPSS/2DVB, as is consistent with the lower SO_3_H contents in the former materials, while the trend of conversion over these two materials is comparable. In other words, 1.5VP/0.64SPSS/0.37DVB had slightly lower activity than 3.2SPSS/2DVB but ultimately gave obviously higher maximal yields. These results indicated that the improved maximal yield is not a result of the differences in catalytic activity. In view of the trade-off between the target reaction and side-reactions, we induced that the improvement in the product yield is probably a result of the reduced degradation of products. We tested the stability of EMF and HMF to verify the proposed hypothesis. As shown in Figure 10a, the retention rate of EMF over 1.5VP/0.64SPSS/0.37DVB was more than 90% after 90 min reaction in ethanol at 120 °C, as is remarkably higher than that (73.1%) over 3.2SPSS/2DVB, proving the inhibition of product degradation by 1.5VP/0.64SPSS/0.37DVB. The catalysts may also influence the degradation of intermediates under reaction conditions, but it’s difficult to observe the stability of HMF in ethanol since the etherification of HMF with ethanol is more pronounced than the degradation. Alternatively, the degradation of HMF over different catalysts was simulated using THF as solvents. The results showed that the retention rate of HMF over 1.5VP/0.64SPSS/0.37DVB (Figure 10b) is 2.6 times larger than that over 3.2SPSS/2DVB after 90 min reaction at 120 °C. These results demonstrate that 1.5VP/0.64SPSS/0.37DVB could stabilize HMF and EMF to a great extent relative to the traditional Brønsted acid catalysts due to the incorporation of solvent moieties, as the major reason for the improvement of product yield. The stabilization effect of high-boiling point solvents, such as NMP, DMSO and ionic liquids on HMF have been observed for a long time [23,58,79]. In the present study, the solvent moieties have been incorporated to the sulfonated polymer, in order to overcome the challenge of recycling high-boiling point solvents. Based on the experimental results and previous studies [23,79,80], we deduced that the stabilization effect of the target product is due to the strong interaction between the product and intermediate, as could decrease their susceptibility to nucleophilic attack and thus minimize the undesirable hydration and condensation reactions. In addition to NMP, we have also attempted to synthesize material via the co-polymerization of 1-vinylimidazole, DVB and SPSS, or to directly implant polyvinylpyrrolidone into the porous structure of 3.2SPSS/2DVB. However, the obtained materials all gave low HMF and EMF yields, probably due to the difficulty of precisely introducing solvent moieties while retaining a certain structure.

In summary, we demonstrate that the sulfonated polymers with solvent moieties (1.5VP/0.64SPSS/0.37DVB) enable the highly efficient conversion of fructose to furan compounds using low-boiling-point ethanol as a green reaction medium, as has been seldom achieved in previous studies. It is notable that the high HMF yields reported in previous studies depend heavily on the employment of high-boiling-point solvents. Compared with the previously reported homogenous and heterogenous catalysts, the EMF yield and the total HMF/EMF yield have been substantially improved along with the reduction in HMF degradation and the replacement of the solvent. Both the improvement of the optimal yield and the use of a green solvent represent obvious improvements in the production efficiency. The characterizations indicated that 1.5VP/0.64SPSS/0.37DVB has a slightly lower amount of acid sites than the common sulfonated materials, and low specific surface area and pore volume, which are formerly considered as an adverse factor for catalytic reaction. In fact, rigorous experiments showed that the adsorptive interaction between HMF and polymers depends strongly on the their surface polarity, instead of the specific surface area and pore volume [81]. Accordingly, we inferred that the superior performance is attributed to the inhibited product degradation due to the incorporated solvent moieties. Sun et al. also demonstrated that the introduction of specific moieties to regulate the interaction between the reactant and solid catalyst could remarkably improve the catalytic performance [58,79]. Therefore, we believe that this catalyst design strategy presented here will facilitate the development of more active and selective catalysts to boost the efficiency and greenness of biomass conversions. Nevertheless, the catalyst can not achieve the effective conversion of glucose in ethanol due to the absence of effective active sites to convert glucose to fructose. More endeavors are required to overcome the obstacle of glucose conversion and to further improve the reusability of thecatalyst.

3. Experimental Section

3.1. Reagent and Materials

5-hydroxymethylfurfural (HMF, 99%), 5-ethoxymethylfurfural (EMF, 97%), fructose (99%), Amberlyst-15,and Sodium p-styrenesulfonate (SPSS, 98%) were obtained from Shanghai Macklin Biochemical Technology Co., Ltd. (Shanghai, China). Divinylbenzene (DVB, 80%) and azobisisobutyronitrile (AIBN, 99%) were purchased from Sigma-Aldrich (Shanghai, China). Ethanol (anhydrous, 99.9%) and N, N-Dimethylformamide (DMF, 99%) were obtained from LABGO Chemical Reagent Co., LTD. (Tianjin, China). N-Vinyl-2-pyrrolidone (VP, 99%) was purchased from Heowns Chemical Reagent Co., LTD (Tianjin, China).

3.2. Synthesis of Sulfonated Polymers

The sulfonated polymers were synthesized from monomers and crosslinkers via the radical induced co-polymerization approach. 3.2SPSS/2DVB were prepared by the co-polymerization of DVB and SPSS using azobisisobutyronitrile (AIBN) as initiator according to previous report [36,57]. Typically, 3.2 g of SPSS was dissolved in a solution consisting of 4.5 mL deionized water and 19.6 mL tetrahydrofuran under stirring. It should be noted that 3.2 g of SPSS is insoluble in pure THF. When the mixture of 4.5 mL deionized water and 19.6 mL THF was used as solvent, SPSS could be completely dissolved in the mixed solvent to form homogeneous solution after 10 min of stirring (Figure S1). When SPSS was completely dissolved, 2 g of DVB and 55 mg of AIBN were added sequentially. After stirring for 20 min, the solution was transferred to the autoclave and heated to 100 °C for 24 h. After the reaction, the generated polymer was washed five times with methanol, concentrated sulfuric acid, and methanol in sequence. After drying (80 °C, 24 h) and milling, the obtained solid was named as 3.2SPSS/2DVB.

Similar to the above process, 1.5VP/0.64SPSS/0.37DVB was synthesized by co-polymerization of SPSS, VP and DVB using AIBN as initiator in the medium of N, N-Dimethylformamide (DMF). A measurement of 0.64 g SPSS was added to 20 mL of DMF, and then 0.37g of DVB and 55 mg of AIBN were added sequentially. Upon stirring for 20 min, the solution was transferred to the autoclave and heated to 100 °C for 24 h. The obtained polymer was washed, dried, milled and named as 1.5VP/0.64SPSS/0.37DVB. It should be emphasized that the ratios of SPSS and DVB in 3.2SPSS/2DVB and 1.5VP/0.64SPSS/0.37DVB are the same, as is also consistent with the optimized ratio in previous reports [36]. As for 1.5VP/0.64SPSS/0.37DVB, the initial loading of VP is superfluous to introduce solvent moieties as much as possible, but only limited VP could be incorporated into the resultant material. These specific formulations are necessary to guarantee that they have similar ratios of DVB with SPSS as well as comparable -SO_3_H content to verify the facilitative effect of incorporated solvent moieties on the catalytic conversion of fructose to HMF and EMF.

3.3. Catalyst Characterization

TEM images were performed on a FEI-Talos F200S transmission electron microscope operating at 200 kV. SEM was carried on a Hitachi SU8600 (Hitachi High-Tech Corporation, Tokyo, Japan) electron microscope with an acceleration voltage of 5 kV. Element contents, including sulfur, nitrogen and carbon was determined by Elementar Vario Micro Cube (Elementar Analysensysteme GmbH, Hanau, Germany) and calculated by mass difference. FTIR spectra were measured using Thermo Fisher Nicolet iS50 (Thermo Fisher Scientific, Waltham, MA, USA) with a resolution of 4 cm^−1^ from 4000 to 400 cm^−1^. Thermogravimetric analyses (TG) were recorded by Hitachi STA 200 (Hitachi High-Tech Corporation, Tokyo, Japan) in flowing air (25 mL/min) with heating rate of 10 °C/min from 25 to 800 °C. Nitrogen adsorption-desorption isotherms were collected on Micromeritics Tristar II (3020) (Micromeritics Instrument Corporation, Norcross, GA, USA), the samples were outgassed for 10 h at 150 °C before the test, and the pore-size distribution was calculated using the Barrett-Joyner-Halenda (BJH) model. XPS spectra were tested on Thermo Fischer ESCALAB 250Xi (Thermo Fisher Scientific, Waltham, MA, USA) equipped with Al Kα X-ray source. The NH_3_ temperature-programmed desorption (NH_3_-TPD) data were collected with temperature no more than 300 °C to avoid the obvious decomposition of material.

3.4. Catalytic Conversion of Fructose

Typically, fructose (45 mg, 0.25 mmol), solvent (1 mL, ethanol or THF), and catalyst were added into a 15 mL glass reactor and then sealed. The reaction was performed in an oil bath with stirring. Subsequently, the reactor was heated to a certain temperature and reacted for a desirable time. After the reaction, the reactor was quenched in ice water to terminate the reaction. Finally, the resulting mixture was filtered, diluted with the water and then subjected to the high-performance liquid chromatography (HPLC) analysis. The concentrations of products, including HMF and EMF were analyzed by HPLC (Thermo Fisher Scientific, U3000, Thermo Fisher Scientific, Waltham, MA, USA) equipped with a C18 column and a UV detector. The concentration of fructose was analyzed by HPLC equipped with a HyperREZ XP Carbohydrate H^+^ column (300 × 7.7 mm, 8 μm) and a refractive index detector. The thermal catalytic performance of these materials was also tested via carrying reaction in oil bath as the control.

Substrate conversion, product yield and product selectivity were calculated according to the following Equations (1)–(3).

$$[eqn]$$ $$[eqn]$$ $$[eqn]$$

To test the stability of the catalyst, the spent catalyst was collected, washed and dried after the reaction and then reused in the recycling experiment. From the first to the fourth run, the recovered catalyst was reused at constant reaction conditions (10 mg of fructose, 20 mg of catalyst, 1 mL of ethanol, 130 °C, 40 min). For the fifth run, 40 and 70 min’s reactions were conducted parallelly. In addition, the FTIR spectra, elementary analysis and NH_3_-TPD curve of the recovered catalyst were recorded.

4. Conclusions

In this study, we develop a facile approach to synthesize sulfonated polymers with solvent moieties via the facile co-polymerization of VP, DVB and SPSS, to achieve the highly efficient conversion of fructose to furan compounds in ethanol. A series of characterizations proved the successful co-polymerization of the three precursors toward sulfonated polymers with pyrrolidinone moieties (1.5VP/0.64SPSS/0.37DVB), which have slightly lower acid contents and poor pore structures relative to 3.2DVB/2.0SPSS. The 1.5VP/0.64SPSS/0.37DVB catalyst exhibited a maximal EMF yield of 81.9% with a total HMF/EMF yield of 92.7%, exceeding the previously reported homogenous and heterogenous catalysts. Moreover, the 1.5VP/0.64SPSS/0.37DVB catalyst also gave a high HMF yield in the medium of THF, as is different from the catalytic systems depend on high boiling point solvents. The control experiment suggested that the superior performance is attributed to the inhibited product degradation. We anticipate that this work will facilitate the development of more active and selective catalysts for industrial production of biomass-derived products.