Partition coefficient of furfural
The partition coefficient of FUR was investigated by conducting hydrothermal reactions wherein a solution of 5 wt% FUR in aqueous phase was heated with SBP for 30 min at 190 °C at seven ratios of aqueous phase to SBP: 1:5, 1:3, 1:2, 1:1, 2:1, 3:1, and 5:1 (by volume). Fig. A1 (in the Supplementary Information) exhibits the FUR partitioning (P) achieved, where P was calculated using equation 10 [28].
$$ P=\frac{{\left[\mathrm{FUR}\right]}_{\mathrm{org}}}{{\left[\mathrm{FUR}\right]}_{\mathrm{aq}}} $$
(10)
A FUR partition coefficient 36 was obtained with an aqueous-to-SBP phase ratio of 1:5 and 1:3 (v/v). This value decreased to 28, 22, 19, 9, and 4 as the aqueous-to-SBP fraction ratio increased to 1:2, 1:1, 2:1, 3:1, and 5:1 (v/v). Similarly, partition coefficient decreases when the aqueous-to-organic phase ratio increases has been also observed when using 2-MTHF, CPME, and isophorone [30]. It is assumed that at high aqueous-to-organic phase ratios, the organic solvent is not able to extract FUR from the aqueous phase; thus, FUR remains in the aqueous phase where it could undergo degradation reactions.
Effect of aqueous-to-organic phase ratio
According to the National Center for Biotechnology Information [43], SBP is insoluble in water. However, mutual solubilities of water and SBP have been measured in a recently published paper at various temperatures from 30 to 210 °C [18]. It is observed that the mutual solubility of water and SBP used in the present paper is minimal under the given experimental conditions. Lin et al. [18] studied the mass fraction mutual solubilities for water and SBP from 303.15 to 483.15 K and a pressure of 2.5 × 106 Pa. They reported that at 443.15, 463.15, and 483.15 K, the solubility of water in SBP is 0.03, 0.05, and 0.07, respectively; and the solubility of SBP in water at the same temperature values is 0.02, 0.04, and 0.06.
The effect of aqueous-to-SBP phase ratio on xylose conversion and FUR formation was examined. Therefore, five ratios of aqueous-to-SBP phase (1:5, 1:2, 1:1, 2:1, 5:1 by volume) were proposed employing an aqueous solution containing 186 mol m−3 and SBP in biphasic systems at a temperature of 190 °C for 0.5 h. The FUR yields obtained are presented in Fig. A2 (in the Supplementary Information) and are determined utilizing equation 2. FUR yield builds up as the aqueous-to-organic ratio increases from 1:5 up to 1:1 (by volume, Fig. A2). At ratios of aqueous-to-organic volumes from 2:1 to 5:1, we propose that a larger FUR yield is halted by the formation of increased degradation products. The highest FUR yield (17%) is obtained at 190 °C in 0.5 h when using SBP in an aqueous-to-SBP phase ratio of 1:1 (by volume).
Selectivity to FUR and xylose conversion can be observed in Fig. A3 (in the Supplementary Information). The xylose conversion varied from 30 to 43%. The selectivity to FUR increases as the aqueous-to-SBP phase ratio raises from 1:5 to 1:2 (by volume), from there on it declines when increasing the aqueous-to-SBP phase ratio to 5:1. A recent paper using isophorone, 2-MTHF, and CPME corroborated similar results [30]. This could occur as a result of the saturation of the SBP to extract FUR; hence, FUR remains in the aqueous phase and degradation reactions might take place.
Effect of reaction time and temperature in the biphasic system
The effect of reaction time and temperature on the formation of FUR was investigated by performing reactions from 0.5 to 3 h at 170, 190, and 210 °C in a biphasic system including SBP and a xylose solution of 186 mol m−3. SBP was employed as organic solvent in an aqueous-to-organic phase ratio of 1:1 (by volume). Figure 1 presents the effect of reaction time when employing SPB on FUR yield, xylose conversion, and selectivity to FUR. As it can be seen in Fig. 1b, FUR yield and xylose conversion are significantly influenced by the reaction temperature, which is in accordance with previous reports on this field [44].
As observed in Fig. 1a, after the first 1 h of the hydrothermal reaction, the FUR yield exhibited a threefold increase by raising the reaction temperature from 170 to 190 °C. The highest FUR yield (59%) was reached at 190 °C in 3 h. The peak of the selectivity to FUR (Fig. 1c) was 60%, 59%, and 47% at temperatures of 170, 190, and 210 °C, respectively. Reaction temperature has also an effect on selectivity to FUR under these laboratory conditions as it has been observed in similar biphasic systems using cyclopentyl methyl ether (CPME), isophorone, and 2-methyltetrahydrofuran (MTHF) [30]. It can be observed that selectivity to FUR at the lowest temperature studied (170 °C) displays the highest value in the biphasic system.
An interesting effect, that can be seen in Fig. 1a, is that at times greater than 1.5 h, the FUR yield reached at 190 °C exceeds the FUR yield achieved at 210 °C. At high reaction temperatures (210 °C), it is assumed that SBP no longer extracts FUR as efficiently. Hence, FUR is prone to remain in the aqueous phase rather than in SBP; therefore, degradation reactions of FUR take place faster. A similar effect can be seen when employing CPME, isophorone, and MTHF [30].
Furfural degradation in the biphasic system
To broaden understanding of the behavior of FUR under the conditions of microwave-assisted reaction in the presence of SBP, it is indispensable to understand its degradation rate. The FUR degradation experiments were determined employing SBP at the reaction temperatures of 170, 190, and 210 °C in an auto-catalyzed system. The experimental data exhibiting the residual fractions of FUR encountered in the aqueous and organic phases at different reaction times are displayed in Fig. A4 (in the Supplementary Information). The figure illustrates the effect of the reaction temperature and time when employing 1:1 aqueous-to-SBP phase ratio on the degradation rate of FUR. The results demonstrate a clear dependency of FUR degradation on the temperature and time, as it can be noticed that when rising the reaction temperature and time, the FUR degradation advances. The highest degree of degradation, 28%, was detected at 210 °C after 3 h. In a similar manner, published papers have presented likewise data in monophasic systems [36, 45, 46]. Similarly, data published recently when employing isophorone as organic solvent in biphasic systems described a similar effect on FUR formation [30]. Contrastingly, when employing 2-methyltetrahydrofuran (2-MTHF) and cyclopentyl methyl ether (CPME) as organic solvents for biphasic reactions, FUR yields reached 71% and 78%, respectively. Besides, a lower FUR degradation degree (12%) was reported under similar conditions when using CPME. This could be due to reactions happening between FUR and unsaturated hydrocarbons via condensation, e.g., SBP and isophorone. Markevich et al. [47] reported the reactivity of double bonds in compounds with functional groups; especially, they noted that FUR could be as reactive as acceptor of dimethylcyanomethyl radicals.
Auto-catalyzed dehydration of birch hydrolysate
Hydrothermal reactions of birch hydrolysate were assessed. In addition to xylose and arabinose, birch hydrolysate contains unhydrolyzed xylose and arabinose polymers (arabinoxylan). The presence of both mono-xylose and oligo-xylan has been noted in earlier studies [48, 49]. Figure 2 shows the FUR yield, pentose (xylose and arabinose) conversion, and selectivity to FUR under various reaction times (between 10 and 180 min) at temperatures of 170, 190, and 210 °C. As it can be seen in Fig. 2b, FUR yield and pentose conversion are significantly influenced by the reaction time and temperature, which is in accordance with previous reports on this field [12, 16, 17, 43, 50, 51].
As observed in Fig. 2a, after the first 0.5 h of the hydrothermal reaction the FUR yield was increased by a factor of four by raising the reaction temperature from 170 to 190 °C. The largest FUR yield (46%) was obtained at 190 °C in 1 h. At temperatures of 170, 190, and 210 °C, the highest selectivity to FUR (Fig. 2c) was 65%, 51%, and 43%, respectively. Reaction temperature has also an effect on selectivity to FUR under these conditions. It can be observed that selectivity to FUR at the lowest temperature studied (170 °C) displays the highest value in the present system.
In a work published recently, when using a xylose solution (28 g l−1) under auto-catalyzed conditions FUR yield increased significantly when increasing the reaction temperature from 190 to 210 °C. Contrastingly, when birch hydrolysate liquor is used and dehydrated under auto-catalyzed conditions at 210 °C FUR yield does not surpass the FUR yield obtained at 190 °C. Under high reaction temperatures (210 °C), it is possible that the reaction takes place faster. Hence, FUR is prone to decompose in a shorter reaction time at high temperatures via condensation (reactions between FUR and intermediates, i.e., pentose and hexose isomers) or resinification (reactions between FUR molecules).
Table 2 Process energy balance and utility requirements When using birch hydrolysate liquor (3 ml) at 190 °C in 60 min, 5.3 mg of humins were formed in the auto-catalyzed system (Fig. A5 in the Supplementary Information). These insoluble polymers can be valorized as recently published literature demonstrates [52, 53]. High-value applications of humins include CO2 sequestration, development of catalysts, and soil improvement. The formed humins were analyzed using N2-physisorption, and the sample displays a low surface area of 4 m2 g−1 (Table A2 in the Supplementary Information).
Furfural formation from birch hydrolysate in the biphasic system
The formation of FUR from birch hydrolysate was investigate under optimized conditions for the dehydration of pentoses as it was determined in the previous sections (190 °C, 1:1 aqueous-to-SBP phase ratio (by volume), under microwave irradiation). The initial composition of the birch hydrolysate is given in Table A1 (in the Supplementary Information).
The highest FUR yield (54%) was reached at 190 °C in 3 h using an aqueous-to-SBP phase ratio of 1:1 (by volume) under complete pentose conversion (Fig. 3a and b). Pentose conversion (Fig. 3b) increased from 80 to 94% when increasing reaction time from 0.5 to 1 h at 190 °C.
Under similar conditions, 3 h at 190 °C with an aqueous-to-CPME phase ratio of 1:1 (v/v), a recent published article reported a FUR yield of 68% using a birch hydrolysate liquor containing similar xylo-oligosaccharide concentration [30]. This high yield was reached due to the absence of degradation reactions between FUR and CPME.
Process simulation
Process simulation model was developed in Aspen Plus® v8.8 (Aspen Technology, Inc., USA) by using the universal quasi-chemical (UNIQUAC) thermodynamic method and is shown in Fig. 4.
The pre-hydrolysate stream (S1) from the bio-refinery after lignin separation consisting mostly of xylose (64% xylo-oligosaccharides and 36% monomeric xylose) and water is introduced at 25 °C into agitated reactors (3 reactors in parallel) operating at temperature of 190 °C and 12.1 bar of pressure with a residence time of 3 h. The stoichiometric reactor model (RSTOIC) available in Aspen Plus is used to simulate the auto-catalyzed reaction of xylose to yield FUR. The outlet stream from the reactor is in vapor phase and is condensed using 2 condensers in parallel. The resulting liquid stream at 98 °C is then sent to a decanter for separation of the organic and aqueous phases using 2-sec-butylphenol (SBP). A separator block is utilized to model the phase separation in the decanter with a block split fraction of 0.972 for FUR in the organic phase. The aqueous phase pre-dominantly consisting of water (99 wt%) is separated and can be reused in the bio-refinery for pre-hydrolysis.
The organic phase consisting of FUR and SBP is introduced into a distillation column operating with 20 ideal stages at atmospheric pressure and a reflux ratio of 1. The RADFRAC rigorous distillation column in Aspen Plus is used to model the distillation process. A 97.8 wt% FUR is recovered in the top fraction at 161 °C and the bottom fraction consisting of pure SBP is recycled back for reuse in the decanter for phase separation. The process flow diagram and mass balance are shown in Fig. 4.
The energy balance and utility requirements are shown in Table 2. The total heating duty indicates the total energy supplied for the heating of process streams and the total cooling duty indicates the total energy removed by cooling. The process includes the use of utilities such as high-pressure steam for the reboiler, cooling water for heat exchangers, and electricity for pumps. It is assumed that the cooling water is recirculated to a cooling tower for reuse in the process.
Process economics
The total investment cost for a plant with a FUR production capacity of 5 kt/year is found to be 14.12 M€ (Table 3). The contribution of individual process equipment to the total direct cost (TDC) is shown in Fig. 5. The reactor units are the most expensive accounting for 46% of the TDC followed by heat exchangers and distillation column which contribute to 24% and 15% of the TDC respectively. The annual operating cost consists of fixed costs, variable costs, and general expenses, and is calculated as 7.23 M€. The revenue for the process comes mainly from selling FUR and additional revenue from high-pressure steam condensate that is sold as district heat.
Table 3 Estimation of fixed capital and total investment Bbosa and Brown [54] recently completed a techno-economic analysis of a corn stover-ethanol bio-refinery concept where they set a market price of FUR of 933 €/t. Currently, the prices available in Alibaba are in a range from 910 to 1630 €/t [55].Footnote 2 Dalvand et al. [56] determined the market potential of FUR and its derivatives in a recent study. They identified a set of six FUR derivatives to determine the best combination of FUR derivatives and what proportion of FUR should be converted into each derivative. The authors reported a FUR market price of 1359 €/t. A recent techno-economic evaluation of bioethanol and FUR coproduction is where they reported a price of 2.37 €/gallon and 1540 €/t, respectively [57]. These two chemical compounds were produced from corn stover via biochemical and thermochemical routes. Olcay et al. [58] used aqueous phase processing to produce furfural as one of the products from biomass and reported a FUR MSP of 1.53 €/kg. MSP, reported in this paper, is quite close to price reported in literature and the existing market price of petrochemical based furfural. The minimum selling price of FUR is calculated at NPV equal to zero with a discount rate of 10% and is found to be 1.62 €/kg with the payback period being 8.9 years. However, when the FUR selling price is increased to 1.93 €/kg, the payback period is 5 years with an NPV of 9.5 M€ at the end of the project lifetime as shown in Fig. 6. The internal rate of return is determined by adjusting the discount rate until the NPV at the end of year 20 is equal to zero and is calculated as 20.7%. The equations used for the calculation of NPV and payback period are given in the Supplementary Information.
The effect of production capacity on the MSP of FUR, total investment, and annual operating cost is shown in Table 4. When the production capacity is doubled, it is observed that there is an increase in the total investment and operating costs and at the same time, the MSP of FUR drops to 1.57 €/kg. Tables A3 and A4 (in the Supplementary Information) show the raw materials, utility prices, and cost factors, respectively.
Table 4 Effect of minimum selling price (MSP), total investment (CAPEX), and operating cost (OPEX) for two production capacities (PC) Sensitivity analysis
A sensitivity analysis is carried out to evaluate the effect of various economic factors on the minimum selling price of FUR as shown in Fig. 7. It was observed that annual operating cost, discount rate, and total investment had the biggest impact on the MSP. For instance, when the operating costs and the fixed capital investment increased by 20%, there was an increase of 11.1% and 4.2% in MSP respectively. Similarly, when the discount rate was increased by 50%, the MSP increased by 7.8%. Variation in the taxation rate and project lifetime had comparatively smaller influence on the MSP (Fig. 7).