Furfuryl acetate (99%) and 2-(ethoxymethyl)furan (98%) were purchased from Rhawn Chemical Technology Co., Ltd. (Shanghai, China). Paraformaldehyde (98%), Na₂HPO₄·H₂O (98%) and NaH₂PO₄ (98%) were purchased from Maclin Biochemical Co., Ltd. (Shanghai, China). Pt on carbon (Pt 5 wt%), n-hexane (98%), n-hexadecane (99.5%, GC), chloroform-d (99.8%), formic acid (99%), trifluoromethanesulfonic acid (98%) and MgSO₄ (AR) were ordered from Aladdin^® Biochemical Technology Co., Ltd. (Shanghai, China). Phosphotungstic acid anhydrate (99%) was purchased from Titan Scientific Co., Ltd. (Shanghai, China). Anhydrous acetic acid (99.5%), formaldehyde (37 wt%), chloroform, concentrated sulfuric acid (98 wt%) and concentrated hydrochloric acid (36 wt%) were all obtained from XiLong Scientific Co., Ltd. (Shantou, China). All reagents were used as received without further purification.

Initially, 0.4 mmol of 2-(ethoxymethyl)furan (EMF) was mixed with 10 mmol of formaldehyde (37 wt%), 1 ml of 1,4-dioxane and 10 μL of concentrated hydrochloric acid (36 wt%) in a 15-ml vial. The vial was placed in oil bath, magnetically stirred and heated to a desired temperature (30–50 °C) for 1‒8 h. After the reaction, the vial was immediately taken out from the oil bath and cooled by tap water to room temperature. Around 0.25 mg n-hexadecane was added into the reaction liquor as an internal standard and 50 μL of the liquid was sampled and extracted by 1 ml of chloroform. The extract was dehydrated by around 200 mg anhydrous MgSO₄. The resulting solution was analyzed by a gas chromatograph-mass spectrometer (GC-MS, SCION 436GC-SQ, Techcomp group, Shanghai, China) to identify products and the identified products were further quantitatively analyzed by a gas chromatograph (GC, SCION 436C, Techcomp group, Shanghai, China).

A mixture of furfuryl acetate (FA, 0.4 mmol), paraformaldehyde (50 mg) and anhydrous acetic acid (2 ml) was loaded into a 15-ml vial. The vial was magnetically stirred and heated at 100–120°C in an oil bath for 1‒7 h. After the reaction, the vial was immediately taken out from the oil bath and cooled by tap water to room temperature. Around 0.25 mg n-hexadecane was added into the reaction liquor as an internal standard. 50 μL of the liquid was sampled and analyzed by a GC-MS (SCION 436GC-SQ, Techcomp group, Shanghai, China) to identify products and the identified products were further quantitatively analyzed by a GC (SCION 436C, Techcomp group, Shanghai, China).

The solvent in the final product mixture was carefully removed with a rotary evaporator under vacuum. The monomeric furanic products in the residual mixture were extracted with 10 ml hexane three times and the residual was considered as condensed furanics and used for oil production via hydrodeoxygenation.

The condensed furanics separated in last section was used as raw materials for production of hydrocarbon fuels (also termed as oil) through the following hydrodeoxygenation reactions.

About 300 mg of the residual product was re-dissolved with methanol (25 ml) in a 50-ml stainless steel reactor (WZD-50, Wuzhou Dingchuang Technology Co., Ltd., Beijing, China) followed by the addition of Pt/C catalyst (200 mg). The reactor was flushed three times with hydrogen gas, and then pressurized with hydrogen gas to 3 MPa. The mixture was mechanically stirred, and then heated to 250°C and kept at this temperature for 4 h. After the reaction, the reactor was cooled to room temperature and depressurized carefully. The catalyst was separated by filtration, and the reaction liquor was collected. Methanol in the reaction liquor was removed by a vacuum rotary evaporator. The residual organics was further mixed with Pt/C catalyst (200 mg), phosphotungstic acid (Li et al., 2021) (200 mg) and n-hexane (25 ml) in the reactor (WZD-50, Wuzhou Dingchuang Technology Co., Ltd., Beijing, China). The reactor was also flushed and pressurized with hydrogen gas to 3 MPa. The mixture was mechanically stirred at 300 rpm, and then heated to 250°C and reacted at the temperature for 4 h. The reactor was cooled to room temperature and depressurized carefully after the reaction. After the catalyst was separated via filtration, n-hexane in the resultant filtrate was removed by a vacuum rotary evaporator. The resultant viscous oil was used for further characterization.

To evaluate the efficiency of hydrodeoxygenation reactions and the quality of hydrocarbon fuels, the contents (wt%) of C and H elements in the prepared oil samples were analyzed by an element analyzer (Vario EL Cube, Elementar, Germany). Prior to the elemental analysis, solvent-free viscous oils obtained in the Section of “Hydrodeoxygenation of condensed furanics to hydrocarbon fuels” were further dried in a vacuum drying oven at 80°C for 48 h and then ground to powder in an agate mortar. The contents of O element in the samples was calculated by assuming that the total content of C, H and O elements for each sample was 100%.

The results show that the strength of the acid in the reaction system has a great influence on the reaction. When the acidity used in the reaction medium was too high, the substrate was easily carbonized due to the acid-catalyzed ring opening of furan rings and severe condensation of the ring-opening products such as aldehydes or ketones. As such, strong acidity-induced carbonization reactions (e.g., entries 7,8,10, and 11 in Supplementary Table S1; entries 3,4,6,8,9, and 10 in Supplementary Table S2) resulted in dark colored reaction liquors and no hydroxymethylated product was detected in these reactions (Supplementary Table S1, Supplementary Table S2). Low-concentration hydrochloric or and anhydrous acetic acid exhibited better balance between acid-catalyzed hydroxymethylation and carbonization reactions, resulting in decent product yields. Therefore, these two catalytic systems were further studied for hydroxymethylation of furfuryl alcohol derivatives.

The results above demonstrated that low-concentration HCl aqueous solution was more beneficial than other reaction systems to facilitate the hydroxylmethylation reaction of EMF with formaldehyde. Since the acidity would be an important factor affecting the hydroxylmethylation reaction efficiency, the effect of HCl aqueous solution loading was thereby further investigated. 5-(ethoxymethyl)furan-2-methanol (EMFM), a hydroxylmethylated product of EMF, was identified by MS (Supplementary Figure S1) and ¹H-NMR (Supplementary Figure S2) and quantified by GC (Supplementary Figure S3), when the reaction between EMF (0.4 mmol) and formaldehyde (1 ml of 37 wt% aqueous solution, 10 mmol) was catalyzed by 10 µL of 36 wt% HCl aqueous solution (0.1 mmol H⁺) in 1 ml of 1,4-dioxane at 50°C for 7 h. However, without the change of other reaction conditions, increasing the loading of 36 wt% HCl aqueous solution to 20 µL (0.2 mmol H⁺) resulted in a dark reaction solution and reduced hydroxylmethylated product. This could be mainly caused by the increased concentration of H⁺ in water and 1, 4-dioxane mixture, which could accelerate the ring opening reactions of furanics (EMF and/or EMFM) and the condensation (e.g., aldol condensation) of corresponding ring-opening products (Nakagawa and Tomishige, 2012). Meanwhile, we did not observe any hydroxylmethylated product with a decreased acid loading of 1 µL (0.01 mmol H⁺) and a reaction time of 4 h at 60°C, further indicating that acid-catalyzed ring opening and condensation reactions was kinetically faster than the hydroxylmethylation reaction of EMF With an appropriate loading (10 μL, 0.1 mmol H⁺) of HCl aqueous solution, the effects of reaction temperature and time on the hydroxylmethylation efficiency of EMF were further investigated. Overall, increasing the reaction temperature and time facilitated the EMF conversion and the formation of hydroxylmethylated product (EMFM). When the reaction temperature was 30°C, no desired product was detected in the early stage of the reaction (<4 h) (Figure 1B); when the reaction temperature was increased to 40°C, the EMFM yield gradually increased with the increase of the reaction time (Figure 1C). At an elevated reaction temperature of 50°C, the maximum yield (28.3 mol%) of EMFM was achieved at 5 h (Figure 1D) but the EMFM yield slightly decreased at an extended reaction time (6‒8 h) likely due to the condensation of furanics. With the same loading (10 μL, 0.1 mmol H⁺) of HCl aqueous solution, severe condensation or carbonization was observed when a higher temperature of 60°C was used. High-molecular-weight products (300–1,000 Da) (Entry 1 in Supplementary Table S1) observed in the GPC results (Supplementary Figure S9) confirmed the occurrence of condensation during the reaction. In the reaction system of HCl and formaldehyde aqueous solution, these results indicated that a high reaction temperature could significantly increase the rate of side reactions (furan ring opening and condensation), resulting in a serious decrease in the yield of the hydroxylmethylated product.

Under the similar conditions, above results promoted us to conduct the hydroxymethylation reaction of FA (entry 1 in Supplementary Table S1) with formaldehyde and HCl aqueous solutions in 1,4-dioxane at 50 °C for 1 and 5 h. However, no hydroxymethylated products were detected after replacing the EMF with FA (entry 11 in Supplementary Table S2). In addition to the furan ring opening (Liang et al., 2017) and aldol condensation of ring opening products, we infer that in the presence of water, acid could also catalyze the hydrolysis of the ester linkage of FA to form furfuryl alcohol, which could easily condense. Such side reactions could occur for the newly formed hydroxylmethylated products of FA. The hydroxyl groups could be protonated to form a carbonium ion, which could attack the electron-rich furan ring to form condensed products (He et al., 2021).

To avoid the hydrolysis of the ester linkages of FA and condensation reactions caused by strong acid in the aqueous solution, we therefore explored the hydroxymethylation reaction of FA using a weak anhydrous acid (i.e., anhydrous acetic acid) and paraformaldehyde (Entry 1 in Supplementary Table S2) instead of formaldehyde and HCl aqueous solutions. In an anhydrous reaction medium, a hydroxymethylated product of FA (2,5-bis(hydroxymethyl)furan monoacetate (BHMFM)) was identified by MS (Supplementary Figure S4) and ¹H-NMR (Supplementary Figure S5) under an elevated temperature from 80 to 120°C. With acetic acid in the reaction system, BHMFM was further esterified by acetic acid to produce another derivative, 2,5-bis(hydroxymethyl)furan diacetate (BHMFD), which was also confirmed by MS (Supplementary Figure S6) and ¹H-NMR (Supplementary Figure S7). The results indicated that anhydrous acidic environment was favorable to the hydroxymethylation of FA while the aqueous reaction system (e.g., formaldehyde and HCl aqueous solution) was indeed not favorable to the hydroxylmethylation of FA under acidic conditions. At a mild reaction temperature of 80°C, no BHMFM was detected within 3 h. After that, BHMFM yield gradually increased with the extension of the reaction time (Figure 2B). When the reaction temperature was elevated to 100 °C, BHMFM was quickly formed within 2 h. Further increasing the reaction time resulted in decreased BHMFM yields (Figure 2C). This was mainly caused by the esterification of BHMFM to form BHMFD (Figure 2A). The acetylation of BHMFM to form BHMFD was more evident at 120°C than those at 100°C (Figure 2D). At a reaction time of 3 h at 120°C, BHMFM was almost completely esterified to form BHMFD while the total yield of BHMFM and BHMFD was not improved at elevated reaction temperatures (>120 °C).

In-situ esterification could be considered as a protection strategy to prevent the condensation of BHMFM in acidic reaction system. However, the yield of BHMFM or BHMFD was not beyond 30 mol% under the investigated conditions. Similar to the hydroxymethylation reaction of EMF, this would be mainly caused by different reaction rates of target (hydroxymethylation and esterification) and side reactions (dehydration, furan ring opening and condensation). First, it is difficult to improve the yields of BHMFM and BHMFD if the ring-opening rate of furan rings is faster than the hydroxymethylation rate of FA. Second, even the reaction rate of hydroxymethylation is faster than that of furan ring opening, newly formed hydroxymethyl group could be immediately protonated by hydrogen proton and dehydrated to produce carbonium ions in an acidic medium. Carbonium ions are active components that would initiate methylene-bridging condensation products, making the esterification of BHMFM difficult and also resulting in low yields of BHMFM and BHMFD.

From the perspective of kinetics, it can be seen that no matter how to optimize the reaction conditions, part of the furanics would condense to form higher-molecular-weight products (Supplementary Figure S9). Such an issue is also associated with many other studies involving HMF, furfural and their derivatives (Randolph et al., 2018; Chang et al., 2020; Modugno and Titirici, 2021). In addition to the synthesis of hydroxylmethylated products, we tentatively hydrogenated the condensed furanic products into hydrocarbon fuels in order to improve carbon utilization. Oligomerization of furanics followed by hydrodeoxygenation (HDO) reactions to produce hydrocabron fuels and lubricants have been intensively studied (Li et al., 2021). Herein, we investigated the possibility of hydrogenating the condensed furanics into hydrocarbon fuels.

The residue after the removal of furanic monomers via hexane extraction was considered as condensed mixture and used to produce hydrocarbon fuels via hydrodeoxygenation. According to a previously reported procedure (Huber et al., 2005), we used two-step HDO reactions to remove oxygen-containing functional groups from the condensed furanics. The first step was to partially hydrogenate the oxygenated furanic products to increase its solubility in the hydrophobic solvents that was used in the second step; the second step was to near completely remove oxygen from the oxygenated furanic products to obtain hydrocarbon fuels. After these HDO reactions, the color of EMF- and FA-derived oily products changed from dark to colorless and light yellow (Figure 3), respectively. Moreover, the oily products produced by HDO reactions could be well dissolved in n-hexane to form homogeneous and transparent solutions (Figure 3), indicating that these condensed furanics were successfully hydrodeoxygenated and the resultant oily products can be possibly used as drop-in hydrocarbon fuels or additives in gasoline or diesel.

The element contents of the condensed furanics before and after HDO reactions were further analyzed to confirm the occurrence of hydrodeoxygenation. It can be seen from Figure 3 that, the oxygen content of EMF-derived oily products was significantly decreased from 38.9 to 5.9% after HDO reactions, at which the contents of C and H in HDO products were increased from initial 55.5 and 5.6%–83.7 and 10.4% (Supplementary Table S3), respectively. Similar results were also obtained for that of FA-derived oily products (Supplementary Table S3), where O content decreased from 42.0 to 9.7% and C and H contents increased from 51.8 to 6.2%–81.0 and 9.3%, respectively. These results indicate that oxygen was efficiently removed from EMF- and FA-derived condensates to form hydrophobic hydrocarbon fuels with good solubility in n-hexane. As it was known, HMF-involved reactions generally lead to many condensed products, this preliminary study opened up a new way to produce hydrocarbon fuels from condensed furanics products which has been considered as waste, improving the carbon utilization efficiency.