Tingting
Cai
ab,
Chao
Liu
a,
Jie
Liang
a,
Yan
Ma
a,
Jun
Hu
c,
Jianchun
Jiang
ab and
Kui
Wang
*ab
aInstitute of Chemical Industry of Forest Products, Chinese Academy of Forestry; Key Lab. of Biomass Energy and Material, Jiangsu Province; National Engineering Lab. for Biomass Chemical Utilization, Nanjing 210042, China. E-mail: wangkui@caf.ac.cn
bCo-Innovation Center of Efficient Processing and Utilization of Forest Resources, Nanjing Forestry University, Nanjing 210037, China
cSchool of Chemical Engineering, Northwest University, Xi'an 710069, China
First published on 5th December 2022
Herein, based on the studies on intermolecular weak interactions, we propose a novel perspective on the design of a biphasic system from a polar aprotic solvent (PAS) and saline water, which plays a vital role in the efficient production of furans from a lignocellulosic biomass. Interestingly, molecular dynamics simulation reveals the formation mechanism of the biphasic system from the original miscible PAS and water. The stronger interaction of Na+ and Cl− with water is proven to be responsible for the dissociation of water from water-miscible γ-valerolactone (GVL) to form the biphasic system. Similarly, the outstanding effect on the extraction and stabilization of furfural (FAL) could also be attributed to the better interaction of PAS with FAL. As a result, a remarkable FAL yield of 80.5 mol% could be obtained in the biphasic system of GVL/NaCl aqueous solution (aq.) with 0.3 mmol AlCl3 at 150 °C for 1 h. Moreover, the synergistic catalytic effect between NaCl and AlCl3 was confirmed by blank experiments and in situ23Na NMR spectroscopy. Besides, this process enables the GVL/NaCl aq. biphasic system and AlCl3 catalyst to exhibit excellent recovery with no significant FAL loss after three times of recycling. This work provides a feasible strategy for efficient biorefinery in biphasic systems.
In 1922, Quaker Oats first produced FAL in the industry using oat husks as the raw material and concentrated sulfuric acid as the solvent in a homogeneous hydrothermal system.13 First, oat husks are treated with dilute sulfuric acid to yield FAL, and then FAL is separated by high pressure steam, followed by neutralization and distillation.16 Likewise, Sánchez17et al. hydrolysed corncob in a homogeneous system with 37.06% of FAL yield obtained using sulfuric acid and hydrochloric acid as catalysts at 180 °C. Despite its large-scale application, the acid digestion process still suffers from low FAL yield and high energy consumption, particularly due to serious equipment corrosion and environmental pollution by the application of sulfuric acid and hydrochloric acid.18 In order to solve the problem, solid acid catalysts were introduced in the research of FAL production in homogeneous systems (e.g., SAPO-34/5A, HZSM-5, Sn-MMT, H-SAPO-34, and amorphous Nb2O5),19–26 which were conducive to environmentally benign processes with a higher FAL yield. Moreover, cheap and available metal salts (e.g., FeCl3, Al2(SO4)3, and AlCl3) also exhibited excellent catalytic activity and selectivity in the production of FAL. Zhang27et al. adopted FeCl3 as a catalyst for the conversion of untreated corncob to FAL in a GVL solvent, with 79.6% FAL yield obtained at 185 °C for 100 min. Besides, our group reported similar research for FAL production by metal salt catalysis. Liu28et al. compared the FAL production from xylose in GVL aqueous solutions using different sulfate and chloride salts as catalysts, with the optimized 50.16% FAL yield by Al2(SO4)3 catalysis. However, the difficult separation of FAL from the homogeneous system is still unsolved, leading to a complex purification process with high energy consumption.29 Notably, around 2005, Professor James A. Dumesic30 initially proposed a biphasic solvent system of methyl isobutyl ketone (MIBK) and water for selective production of HMF from fructose. The spontaneous migration of HMF from water to an organic phase was believed to be responsible for its high efficiency and selectivity. Then, in 2007, they conducted research on the production of dimethylfuran from fructose and found that the addition of NaCl could significantly improve the extraction efficiency of the organic phase.31 Similar work for the production of FAL from hemicellulose was also conducted by Dumesic group32 in 2012, based on the design of a biphasic system by addition of 2-sec-butylphenol (SBP), 4-n-hexylphe-nol (NHP) and 4-propyl guaiacol (PG) into an aqueous solution, respectively, which further confirmed the improvement of product separation efficiency by the biphasic system from the incompatible organic solvent and water. Moreover, Hu33,34et al. proposed a novel tetrahydrofuran (THF)/H2O biphasic system for FAL production, from which 51–66% FAL yield can be obtained from various lignocellulosic biomass under microwave heating at 140 °C for 60 min. Despite its excellent performance, THF still suffers from low flash point (−17 °C), high flammability and volatility.
Owing to its excellent thermo-chemical stability, GVL has attracted more attention than THF for the production of FAL.35–38 More than 70.2–80.4% FAL yield has been reported when xylose was used as the substrate in GVL–H2O.35–37 However, the system exhibited unsatisfactory adaptability of raw materials with lower FAL yields from xylan and lignocellulosic biomass. Luo38et al. conducted an interesting biorefinery work on the separation of hemicellulose from corn stover in the GVL–H2O system with 5%–30% NaCl as a catalyst, which proposed a viable strategy for the solution of substrate adaptability by enhancing the interactions of NaCl with the –OH group in hemicellulose and G units in lignin. However, the facilitated mechanism of the biphasic system on FAL production needs to be further discussed. In order to reveal the promoted effect of the biphasic system, we initially constructed a novel biphasic system with GVL and molten salt hydrates (LiCl·3H2O and LiBr·3H2O) in our previous research work.18 As a molten salt hydrate was used as both the solvent and catalyst, 77.2% of FAL was obtained from xylan reacted at 140 °C for 2 h. Considering the industrial-scale application, NaCl, as a cheap and abundant salt, was adopted to establish the biphasic system instead of molten salt hydrates. Interestingly, GVL is originally miscible with water, which is difficult to form a biphasic system; despite its excellent performance in the biorefinery process,39 stratification began after 10 wt%–25 wt% NaCl was added. Although the solvent effect on FAL in the biphasic system has been well discussed,40 the formation mechanism of the biphasic system has been barely studied yet, particularly for the weak interaction among the solvent, water and solute.
Herein, based on the discussion of intermolecular weak interactions, we propose a novel perspective on the design of a biphasic system from a polar aprotic solvent (PAS) and a sodium chloride solution. A control group experiment was conducted to evaluate the efficiency of the biphasic system from different PAS in the xylan conversion. Notably, the radial distribution function (RDF)40 was introduced to demonstrate the weak interaction effect between PAS and NaCl aq. and verified for the formation of the biphasic system. Besides, the interaction forces between the solute and solvent molecules were discussed to reveal the pathway of FAL extraction from the aqueous phase to the organic solvent. Moreover, in situ27Al NMR, in situ23Na NMR and ESI-MS/MS tests were carried out to elucidate the synergistic catalytic mechanism of NaCl–AlCl3 in the isomerization of xylose in the biphasic system. Furthermore, the recovery of the biphasic system was also involved and applied in the conversion of real biomass substrates (i.e., corncob and moso bamboo) with mass balance, which provides a new perspective for the biorefinery industrial process with efficiency and specificity.
The retention time and peak area of FAL in the chromatographic column were determined by an external standard method, and the FAL yield was calculated using the following equations:
Levulinic acid (LA); 5-Hydroxymethylfurfural (HMF).
The interactions of liquid–liquid molecular systems were calculated based on the COMPASSII forcefield and Flory–Huggins model. The energy bin width was 0.02 kcal mol−1, with the selected 1 × 107 pair configurations to find the 100 lowest energy frames, and the reference temperature was set at 423 K. The cluster sample is 1 × 105 and the iterations per cluster is 20.
In the traditional Flory–Huggins model, each component occupies a lattice site. For a lattice with a coordination number Z, the mixing energy (Emix) and interaction parameter (x) can be calculated using eqn (1) and (2):
Emix = 0.5Z(Ebs + Esb − Ebb − Ess) | (1) |
x = Emix/RT | (2) |
The ESI-MS spectra were recorded in the positive ion mode at 1.60 kV detector voltage, 1.56 kV ionization voltage, and 100 kPa gas drying pressure. The source temperature and solvent removal temperature were set as 200 °C, and N2 was used as a carrier gas at a flow rate of 1.5 L min−1.18
In situ 27Al NMR spectrum and in situ23Na NMR spectrum were recorded and tested at 156.38 MHz using an Advance III HD 500 MHz NMR spectrometer (Bruker Biospin). The detection conditions were set as follows: 10.50 μs pulse width, 0.1311 s scanning frequency, and 128 times scanning frequency. The data were collected every 10 min and lasted for 2 h at 60 °C.29
Entry | Organic solvent | Density [g mL−1] | FAL yield [mol%] | FAL yield [wt%] |
---|---|---|---|---|
Reaction conditions: 0.5 g xylan, 0.0725 g AlCl3·6H2O, 0.0979 mol organic solvents, 20.0 g 20 wt% NaCl aq., 150 °C, 1 h. References density values in previous literature.44 | ||||
1 | GVL | 1.04 | 72.60 | 52.80 |
2 | MIBK | 0.80 | 58.20 | 42.33 |
3 | THF | 0.89 | 40.55 | 29.49 |
4 | EA | 0.90 | 60.41 | 43.93 |
5 | PC | 1.20 | 7.79 | 5.67 |
Entry | Catalyst | FAL yield/mol% | FAL yield/wt% |
---|---|---|---|
Reaction conditions: 0.5 g xylan, 20.0 g 20 wt% NaCl aq., 10.0 g GVL, catalyst (based on 0.3 mmol Mn+), 150 °C, 1 h. | |||
1 | None | 16.14 | 11.74 |
2 | ZnCl2 | 16.14 | 11.74 |
3 | CaCl2 | 16.70 | 12.15 |
4 | FeCl3 | 54.10 | 39.34 |
5 | CuCl2 | 57.53 | 41.84 |
6 | AlCl3 | 72.60 | 52.80 |
7 | H2SO4 | 48.98 | 35.62 |
8 | HCl | 48.09 | 34.97 |
Fig. 1 Catalytic conversion of xylan into FAL at different temperatures and time points (reaction conditions: 0.5 g xylan, 0.072 g AlCl3·6H2O, 10.0 g GVL, 20.0 g 20 wt% NaCl aq.). |
Fig. 2(a) exhibits that the FAL yield varied with different catalyst loadings. Normally, 72.6 mol% FAL yield can be obtained at a lower AlCl3 loading of 0.072 g under the above-mentioned reaction conditions, accompanied by a spot of sugar intermediates. Notably, more than 80.50 mol% of FAL could be obtained when increasing the AlCl3 loading to 0.080 g. However, the FAL yield slightly decreased with excessive loading of AlCl3. The unnecessary high density of active sites will accelerate the condensation of FAL, which is believed to be responsible for the decrease in FAL yield.7,48
Notably, 0.080 g loading of AlCl3·6H2O was selected for the following study on the influence of mass ratio of GVL and NaCl aq. on FAL production. As shown in Fig. 2(b), a moderate FAL yield of 64.89 mol% could be observed with 25/5 mass ratio of GVL/NaCl aq. The continuous increase in the proportion of NaCl aq. will yield a slight lift in the FAL yield. Notably, as high as 80.50 mol% of FAL yield could be observed when the mass ratio of GVL/NaCl aq. was set as 10/20, which could be attributed to the positive effects on the isomerization of xylose to xylulose by the Lewis acid center yield from NaCl and AlCl3 at a certain concentration. The mechanism of different solvent ratios leading to changes in the activation free energy of xylose isomerization into xylulose has been reported in our previous work.18 Continuously increasing the portion of NaCl aq. to 5/25, the FAL yield visibly decreased to 75.70 mol%, which originated from the reduction of GVL content, followed by the inhibition of extraction of FAL from the aqueous phase. In brief, GVL/NaCl aq. with a mass ratio of 10/20 exhibited the most efficient synergistic effect on the production and stabilization of FAL.
Notably, molecular dynamics simulation of Na+ in the solvent was introduced to evaluate the weak interaction of solvent molecules, which provides a reliable view for detailed explanation on the formation of the biphasic system. As shown in Fig. 3(a), the local arrangement of solvent molecules around Na+ was revealed by calculating the central of mass radial distribution function (RDF) of ions and molecules. Interestingly, Fig. 3(a) exhibits the different layering process accompanied by the absence (1) and existence (2) of NaCl aq. (20 wt%) in a GVL/H2O solution. Moreover, the first solvation peak for water was 2.25 Å, followed by GVL of 2.33 Å and Cl− of 2.65 Å, and the peak height sorted as Cl− (20.10r) > H2O (10.42r) > GVL (2.15r), namely, the “Na–Cl” and “Na–H2O” curves are much higher than that of “Na–GVL” curves, which indicated a stronger interaction between Na+ and H2O when compared with GVL.
Fig. 3(b) shows more intuitively that the number of H2O around Na+ is far greater than that of GVL molecules in the system of GVL/NaCl aq. In other words, Na+ and Cl− can extract H2O from the miscible H2O–GVL solution by the so-called wrapped effect, yielding the biphasic system. However, Fig. 3(c) provides the number of molecules in the first solvation peak around Na+ of the system in different organic solvents, including GVL, MIBK, THF, EA and PC. It is observed that the amount of “Na–H2O” is much higher than that of “Na–organic Solvent” in the first solvation peak, indicating that the attraction of Na+ to H2O molecules is stronger than that of organic solvents, which is consistent with the observed phenomenon that an initial homogeneous solution exhibited obvious stratification boundaries in a few minutes between organic solutions and NaCl aq.
Table 3 illustrates the permittivity of different solvents and the solubility with water and NaCl aq., which indicated that pure water and organic solutions are miscible or slightly soluble initially. However, it will change after NaCl is added into the solution, which is consistent with computational data. Due to the weak solvation hydration of NaCl in organic solvents with small polarity, it is difficult to ionize free ions. Conversely, the interaction between water and NaCl is strong due to solubility, and NaCl completely ionizes the free hydrated ions to form a stable dispersed solution. Therefore, the smallest water molecule can easily enter the internal solvation peak of Na+. In this case, the imbalance of charge will make Cl− arrange orderly in the outer layer of Na+via electrostatic attraction. In Table 3, the permittivity of GVL and water is 33.2 and 81.0, respectively. The great difference in permittivity between GVL and water manifests that the attraction between GVL and water will be weak. Due to the strong mutual attraction between Na+, Cl− and H2O molecules by solubility and electrostatic attraction but the weak interaction between H2O and GVL, GVL will move from inside to outside, forming a biphasic system.
Entry | Solvents and solutes | Permittivity | Solubility with water | Solubility with NaCl aq. |
---|---|---|---|---|
a References permittivity in previous literature.18 | ||||
1 | GVL | 33.2a | miscible | insoluble |
2 | MIBK | 13.4a | Slightly Soluble | insoluble |
3 | THF | 8.3a | Slightly Soluble | insoluble |
4 | EA | 6.1 | Slightly Soluble | insoluble |
5 | PC | 69.0 | miscible | insoluble |
6 | FAL | 41.9 | miscible | — |
7 | H2O | 81.0 | — | — |
In order to understand the selective extraction effect of organic solvents on FAL, the weak interaction between different solvents (GVL, MIBK, THF, EA, PC, H2O) and FAL was also calculated, and is exhibited in Fig. 4. It is well known that the interaction parameter x is a measure of solvent capacity in a solvent system in the lattice theory of Flory–Huggins. As shown in Fig. 4, the x values are in the following order of EA ≈ MIBK < THF ≈ GVL < PC < H2O. Obviously, organic solvents can extract FAL from the aqueous phase due to the relatively stronger binding energy of organic solvents to FAL when compared with that of H2O, which were basically consistent with the following experiment data and characterization results.
Fig. 4 Molecular interactions in the solvent, (a) GVL/FAL, (b) MIBK/FAL, (c) THF/FAL, (d) EA/FAL, (e) PC/FAL, and (f) H2O/FAL. |
Fig. 5 In situ 27Al NMR and in situ23Na NMR spectroscopy of the substrates. Detection conditions: 8.00 g D2O, 0.25 g xylose, 2.0 g NaCl, and 0.04 g AlCl3·6H2O. |
In order to further identify the catalytic active species form of Al3+, the ESI-MS method was adopted. As shown in Fig. 6(a), a xylose peak occurs at m/z = 151. In Fig. 6(b), the two peaks at m/z = 79, 97 belong to [Al(H2O)3 − 2H]+ and [Al(OH)2(H2O)2]+ species, respectively. In Fig. 6(c), the two characteristic peaks at m/z = 77 and 113 could be assigned to [Na(H2O)3]+ and [Na(H2O)5]+ species, respectively. Fig. 6(d) exhibits the ESI-MS spectrum of xylose, AlCl3, and NaCl mixed solution. Notably, the [Na(H2O)5]+ species peak was still observed at m/z = 113, while the [Xylose + H]+ species peak disappeared. Instead, two [Al(OH)2(Xylose)]+ and [Al(OH)2(Xylose) (H2O)]+Al3+ binding peaks were observed at m/z = 211 and 229, which were believed as the catalytic active species form of Al3+ during isomerization of xylose.7
To evaluate the sustainability of the proposed process, the biphasic system with a AlCl3 catalyst was recovered three times. During each run, the upper organic phase containing FAL could be directly separated by a separation funnel, followed by distillation to recover FAL and GVL, which could be reused after facile purification by active carbon adsorption of soluble humins. However, the lower aqueous phase containing NaCl and AlCl3 could be directly reused by addition of a small amount of fresh NaCl solution to make up 20 g. The recovery of the NaCl-AlCl3 system was determined by measuring the yield of FAL in GVL after the reaction. As shown in Fig. 7, the FAL yield in GVL was 66.66% in the original solution. After three times run, 61.34% of FAL yield could still be observed, which indicated that the biphasic system with AlCl3 exhibited favourable stability and catalytic activity during the reaction.
The mass transfer flow for the conversion of corncob was also involved, and is summarized in Fig. 9. Notably, after the reaction, the system automatically formed into the organic upper layer and aqueous lower layer with 34.14 g of solid residue obtained in the bottom, including 30.24 g glucan and 3.90 g lignin without xylan observed, which indicated that xylan in corncob was completely converted with 27.58 wt% of lignin and 74.69 wt% of glucan remaining. Then, 13.65 g of the dissolved humins in the organic phase could be precipitated by dilution and centrifugation, followed by distillation to separate 18.05 g FAL and other derivates. However, 8.78 g FAL remaining in the aqueous phase could also be extracted by EA. The left GVL and aqueous phase with trace of sugars and metal halide salts could be directly retrieved without further purification. Overall, this process in the biphasic system provides a feasible strategy for the conversion of agricultural and forestry wastes into profitable chemicals.
Fig. 9 Mass balance of FAL production from corncob (reaction conditions: 100.00 g corncob 2439.00 g GVL, 4878.00 g 20 wt% NaCl aq., 19.51 g AlCl3·6H2O, 150 °C for 1 h). |
This journal is © The Royal Society of Chemistry 2023 |