Switchable Deep Eutectic Solvents for Lignin Dissolution and Regeneration

Deep eutectic solvents (DESs) are promising for lignin dissolution and extraction. However, they usually possess high polarity and are difficult to recycle. To overcome this drawback, a variety of switchable ionic liquids (SILs) composed of 1,8-diazabicyclo[5.4.0]undec-7-ene (DBU) and alcohols was synthesized and screened. According to the thermodynamic modeling suggestions, the selected DBU–HexOH SIL was coupled with hydrogen-bond donors to form switchable-DES (SDES) systems with moderated viscosity, conductivity, and pH while maintaining switchability. The SDESs produced a well-improved lignin and lignin model compound solubility compared with those of SILs; charging CO2 into SDES (SDESCO2) caused a further increase in solubility. The solubility (25 °C) of syringic acid, ferulic acid, and milled wood lignin in SDESCO2 reached 230.57, 452.17, and 279.12 mg/g, respectively. Such SDES-dissolved lignin can be regenerated using acetone as an anti-solvent. The SDES-regenerated lignin exhibited a well-preserved structure with no noticeable chemical modifications. Furthermore, the SDESCO2 lignin possessed a higher molecular weight (Mw = 10,340 g/mol; Mn = 7672 g/mol), improved uniformity (polydispersity index = 1.35), and a higher guaiacyl lignin unit content compared with the original milled wood lignin. The SDES system proposed in the present work could benefit the fractionation of lignin compounds and facilitate downstream industrial processes.


Introduction
Abundant reserves of lignocellulosic biomass make it an important renewable resource [1].As a main component within lignocellulose, lignin is a three-dimensional amorphous polymer composed of methoxylated phenylpropane units [2].Its aromatic structure causes lignin to be recognized as an important natural source of phenolic chemicals [3]; it is estimated that around 20 billion tons/year of such chemicals are produced by plants [4].However, the chemical and physical interactions between lignin and carbohydrate components within lignocellulosic biomass make the separation of lignin difficult to achieve [5,6].Traditional lignin separation approaches, such as chemical treatment, organic solvent treatment, mechanical treatment, and ionic liquid (IL) treatment methods [7], are well developed but, nevertheless, reported to have several drawbacks.Chemical treatment methods, for example, acid or alkali treatment, can achieve lignin separation through the degradation and dissolution of lignin macromolecules [8]; however, their applications are limited by severe lignin deconstruction and unavoidable chemical modifications.In addition, other shortcomings, such as the corrosion of equipment, wastewater treatment, and chemical recycling, also prohibit further application [9].Organic solvents, such as N-methylmorpholine-N-oxide [10], can be used to dissolve and separate lignin from lignocellulose.However, most of these solvents are toxic and volatile.Because of this, such methods are commonly used for laboratory research and small-scale production [11,12].Lignin can also be separated through a ball mill treatment [13]; however, this approach is extremely energy-consuming.ILs can selectively remove lignin from the lignocellulose [14], effectively overcoming the lignocellulose obstinacy [12].Although it is reported to produce lignin with an increased chemical activity [15], the IL method is limited by its high cost [16], poor reusability [17], and, in some cases, it even exhibits greater toxicity than organic solvents [1].Therefore, it is in the interests of both academic and industrial fields to develop efficient and sustainable methods for lignin dissolution and separation.
A deep eutectic solvent (DES) is a eutectic solvent formulated by hydrogen-bond donors (HBDs) and hydrogen-bond acceptors (HBAs) [18].ILs can sometimes act as typical HBAs in DES systems [19,20].Indeed, the DESs share the advantages of ILs, which are low vapor pressure, easy synthesis, and adjustable physical and chemical properties [21].With the addition of HBDs, a DES can be designed to be low cost [22], with reduced viscosity [23], good biodegradability [24], and low toxicity [25].In recent years, DESs have been applied in many areas, for example, in organic reactions [26], biotransformations [27], sensor development [28], and pharmaceutical formulation [29].Furthermore, DESs have been reported as promising solvents for lignocellulosic biomass fractionation due to their selective lignin extraction nature [30,31].The extracted lignin is usually very pure and uniformly distributed.The selective cleavage of the ether bond in lignin macromolecules during DES treatment has been widely reported [32].For example, choline chloride-lactic acid DES produced a mild acid-based catalytic environment, which triggered the cleavage of ether bonds between phenylpropane units, resulting in lignin depolymerization and dissolution [33].The strong hydrogen-bond network within the DES system also facilitated the selective extraction of lignin from the lignocellulose [34,35].It was reported that acidic DESs with strong HBDs caused the proton-catalyzed cleavage of the ether, ester, and glycosidic linkages present in the lignin-carbohydrate complex, leading to the extraction and depolymerization of lignin and xylan [36,37].The DES treatment also led to the controllable cleavage of C-O-C and C-C bonds within the lignin macromolecules to produce a lignin stream with improved values [38].Zhu et al. used a choline chloride-ethylene glycol DES to extract very pure lignin and well-preserved β-O-4 linkages [39].Xu et al. used an ultrasound-assisted choline chloride-formic acid DES to extract lignin with a small molecular size, narrow distribution, and well-preserved β-O-4 bonds [1].Although these DES treatments have been proven as efficient lignin dissolution and extraction methods, they still present defects of poor reusability, mainly caused by their high polarities [40,41].It is, therefore, worth developing a "switchable" DES system, which could act as a polar solvent during the lignin dissolution and extraction processes and could also transfer into a non-polar state during recycling.
Mixing organic superbases and alcohols, followed by bubbling with CO 2 , generates a so-called switchable ionic liquid (SIL) [42,43], whose chemical and physical properties can be altered by the addition of acid gases, such as CO 2 and SO 2 [43].These SILs can be recycled by reversibly transforming to their molecular precursors by heating or by bubbling N 2 [44].Lam Phan et al. found that the exposure of a 1:1 mixture of two neutral liquids, 1,8-diazabicyclo [5.4.0]undec-7-ene (DBU) and monohydric alcohols, to gaseous CO 2 at 1 atm caused an exothermic conversion of the liquid phase to an ionic-form substance (Figure 1) [42,45].Anugwom et al. tested the possibility of using these SILs for lignocellulosic biomass treatment [46].However, like IL defects, SILs are also expensive and highly viscous.Recently, several switchable DESs (SDESs) with a reversible phase transition nature, triggered by CO 2 , pH, and heat, were also developed [47].Therefore, it would be beneficial to use the concept of DESs, taking SIL as an HBA and coupling it with a selected HBD, to generate a brand-new switchable solvent system, thereby overcoming the defects mentioned previously.A detailed review of the possibility of using switchable solvents for lignin dissolution and extraction was presented in our previous work [48].The proposed solvent system could provide a low-cost, high-efficiency, and sustainable method for lignin dissolution and extraction; this could further facilitate the development of plant fiber pretreatment technology.The reaction scheme for the switchable ionic liquids (SILs) [42].
A variety of SILs composed of DBU and different alcohol compounds was synthesized and screened in this work.Optimized SIL was coupled with an HBD to form stable SDES systems; this was followed by using the systems to dissolve lignin and its model compounds.Lignin solubility was checked using ultraviolet spectroscopy, and characterizations such as Fourier-transform infrared (FTIR) spectrum, heteronuclear single-quantum coherence nuclear magnetic resonance (HSQC NMR), gel permeation chromatography (GPC), and thermogravimetric analysis (TGA) were carried out, to confirm the possibility of using the proposed SDES system for lignin dissolution and extraction processes.
A variety of SILs composed of DBU and different alcohol compounds was synthesized and screened in this work.Optimized SIL was coupled with an HBD to form stable SDES systems; this was followed by using the systems to dissolve lignin and its model compounds.Lignin solubility was checked using ultraviolet spectroscopy, and characterizations such as Fourier-transform infrared (FTIR) spectrum, heteronuclear single-quantum coherence nuclear magnetic resonance (HSQC NMR), gel permeation chromatography (GPC), and thermogravimetric analysis (TGA) were carried out, to confirm the possibility of using the proposed SDES system for lignin dissolution and extraction processes.

Preparation of Switchable Deep Eutectic Solvents
The switchable deep eutectic solvents (SDES) were synthesized according to the method employed by [51], by using SIL as the hydrogen-bond acceptor (HBA) and water (H 2 O) as the hydrogen-bond donor (HBD).The HBA and HBD were mixed at a 1:10 to 10:1 weight ratio and stirred at 25 • C. A transparent and uniform SDES system was obtained, namely, DBU-HexOH/H 2 O. DBU-HexOH/H 2 O demonstrated switchable physical and chemical properties upon the absorption or release of CO 2 .Polar SDES (SDES CO2 ) can be obtained by bubbling CO 2 into SDES at room temperature for 30 min until the viscosity of the system increases significantly, whereas non-polar SDES (SDES ) can be restored by heating the SDES CO2 at 80 • C for 10 h.

Characterizations of Solvent Properties
The pH of the testing sample was detected using a pH meter (PHS-3G, Shanghai INESA Scientific Instrument Co., Ltd., Shanghai, China) at 25 • C. The conductivity of the testing sample was detected using a conductivity meter (DDS-307A, Shanghai INESA Scientific Instrument Co., Ltd., Shanghai, China).The viscosity of the DES was measured according to the method described in [49], by placing 1 g of sample solution on the fixture of a rotational rheometer (ARES-G2, TA Instruments Vorster, New Castle, DE, USA), operating at 30 rpm.The polarity of the sample to be measured is illustrated by their miscibility in a low polarity reagent (decane) [42].

Preparation of Milled Wood Lignin
The milled wood lignin (MWL) was prepared according to the method recorded in [52].A specific amount of 80~40 mesh poplar powder was extracted by a solution with a volume ratio of benzene to ethanol of 2:1, and the dried raw material was placed in a planetary ball mill (Pulverisette 5, FRITSCH, Markt Einersheim, Germany).The milled sample was extracted with 1,4-dioxane and deionized water at a volume ratio of 96:4 (v/v).The extract was centrifuged, and the supernatant was evaporated, concentrated, and freeze-dried to obtain the MWL.

Solubility of the Lignin Model Compound
Lignin solubility was detected according to the method described in [53].Lignin and its model compounds were added to the testing solvent under continuous stirring at room temperature-25 • C-to fully dissolve until saturation.The liquid phase was separated by filtration using a 0.45 µm organic filter, and the absorption of the dissolved lignin was determined using an ultraviolet-visible spectrophotometer (Agilent Cary 8454, Agilent Technologies, Santa Clara, CA, USA).The lignin concentration was then calculated using Lambert-Beer's law.The characteristic absorption peaks for lignin and lignin model compounds were listed as follows: alkali lignin 280 nm, syringaldehyde 307 nm, vanillic acid 256 nm, syringic acid 265 nm, and ferulic acid 314 nm [54].

Dissolution and Regeneration of Lignin
Lignin dissolution and regeneration were carried out according to the method employed in [9,55].MWL was added into the testing solvents (1:100, w/w) under continuous stirring at room temperature.The fully dissolved lignin solution was filtrated with a 0.45 µm organic filter.Acetone was added to the lignin solution (1:10, v/v) as an antisolvent and then kept at 4 • C for 12 h to enable lignin regeneration.The precipitate was recovered using centrifugation (8000 rpm, 10 min) followed by filtration and then oven-dried at 105 • C until a constant weight was achieved.

Characterizations of Lignin
The number average molecular weight (Mn), weight average molecular weight (Mw), and polydispersity index (PDI) were measured by gel permeation chromatography (GPC, e2695, Agilent Technologies Inc., Palo Alto, CA, USA) equipped with an Agilent 1200 series high-performance liquid chromatograph (HPLC) and an ultraviolet detector (UV) [34,56].Acetylated lignin samples (2 mg) were dissolved in THF (1 mL) and filtered through a 0.45 µm filter.The injection volume was 100 µL, and the wavelength of the UV detector was 280 nm.THF was used as the mobile phase under a flow rate of 100 mL/min.A calibration curve was prepared using polystyrene in the range of 1480~1,233,000 g/mol.The thermal oxidative degradation and stability of the lignin samples were subjected to thermogravimetric analysis (TGA SDT650, TA Instruments, Milford, MA, USA) [57].The lignin samples were placed in aluminum crucibles and tested between room temperature and 800 • C at a heating rate of 10 • C/min under nitrogen conditions.
The structural characterization of the lignin samples was carried out using Fouriertransform infrared spectroscopy (FTIR, Bruker ALPHA, Ettlingen, Germany) [3].A total of 1 mg lignin sample was mixed with 100 mg dried KBr, ground, and pressed into tablets.Samples were scanned 32 times over a range of 400~4000 cm -1 at a resolution of 4 cm −1 .Detailed structural characterization was performed using nuclear magnetic resonance (NMR) spectroscopy (HSQC NMR, BRUKER AVANCE III HD 500 M, Karlsruhe, Germany); around 50 mg of lignin was dissolved in 0.5 mL of DMSO-d 6 [58].NMR spectra of lignin samples were obtained using a Bruker Avance III HD500 MHz spectrometer at a room temperature of 25 • C. The 1 H- 13 C heteronuclear single-quantum coherence (HSQC) spectral standard pulse sequence was as follows: spectral width 1 H of 11 ppm with 2048 sampling points; 13 C spectral width of 190 ppm with 256 data points; 64 scans; and 1 s scan delay.Volume integration of the signals in the HSQC NMR spectra was performed in Bruker Top Spin 2.1 software.

Molecular Simulation
The molecular simulation was conducted using the conductor-like screening model for real solvents (COSMO-RS) model (BIOVIA COSMOtherm 2020, Version 20.0.0,Dassault Systèmes, Paris, France), in which quantum chemical calculations were combined with statistical mechanics to explore the thermodynamic behaviors of the chemicals used in this work.
The structure of DBUH and HexCO 3 was drawn by Turbomole (BIOVIA TmoleX 2021, Version 21.0.0,Dassault Systèmes, Paris, France).The geometry optimizations were performed at the density functional theory (DFT) level and utilized the BP function with resolution of identity (RI) approximation; a triple-ξ valence polarized basis set (TZVP) was utilized [59,60].All the other chemicals were obtained from the built-in database.For the COSMO-RS calculations, it was assumed that ILs were set as a mixture of an equimolar composition of cations and anions, and the DESs were set as a mixture of an equimolar composition of HBAs and HBDs [61].

Characterizations of Switchable Ionic Liquids
As shown in Figure 2a, a transparent and homogeneous SIL was obtained by simply mixing DBU and monohydric alcohol at room temperature.According to Figure 1, bubbling CO 2 into SIL caused the formation of its ionic state (SIL CO2 ), leading to a significant increase in viscosity, and sometimes even the formation of solids (Table A1).Among the testing SIL systems, most of the SIL CO2 remained as a transparent and homogeneous liquid.For example, DBU-PrOH CO2 , DBU-BuOH CO2 , DBU-HexOH CO2 , and DBU-OctOH CO2 were still viscous liquids in the presence of CO 2 .However, DBU-MeOH CO2 and DBU-ETA CO2 formed rigid solids.The CO 2 in SIL CO2 was removed by heating at 80 • C, after which a transparent SIL' was obtained.As the presented work focuses on the dissolution and extraction of lignin, the proposed solvent system should maintain the liquid state either with or without the presence of CO 2 .The solvent properties of the SIL system were then investigated; there were notable changes in SIL polarity, conductivity, pH, and viscosity triggered by introducing and removing CO 2 (Figure 2).and extraction of lignin, the proposed solvent system should maintain the liquid state either with or without the presence of CO2.The solvent properties of the SIL system were then investigated; there were notable changes in SIL polarity, conductivity, pH, and viscosity triggered by introducing and removing CO2 (Figure 2).The polarity of the SIL system can be regulated reversibly by introducing and removing CO2.This polarity conversion was verified by mixing SILs with a low-polarity reagent (decane).Taking the DBU-HexOH SIL system as an example, the DBU-HexOH is miscible with decane, whereas its ionic form (DBU-HexOHCO2) is not.As shown in Figure 2b, the conversion in polarity remained after three CO2 charging-discharging cycles.Pumping CO2 through the homogenous SIL-decane solution caused the formation of emulsion, while simply removing CO2 by heating restored miscibility.DBU-PrOH, DBU-BuOH, and DBU-OctOH SIL systems (Table A1) demonstrated a similar polarity conversion performance to that of DUB-HexOH.Other SIL systems, however, encountered difficulties in restoring their non-polar state from SILCO2 during the heating process, possibly due to the strong interaction formed between the alcohol compound and CO2 [62].Our study, therefore, focused on the analysis of DBU-PrOH, DBU-BuOH, DBU-HexOH, and DBU-OctOH SIL systems.
The formation of ionic compounds was confirmed by analyzing the conductivity and viscosity of the SIL systems.Generally, SIL and SIL′ demonstrated significantly lower conductivity than that of SILCO2.Alcohol and DBU (0.012 µS/cm) both possess low The polarity of the SIL system can be regulated reversibly by introducing and removing CO 2 .This polarity conversion was verified by mixing SILs with a low-polarity reagent (decane).Taking the DBU-HexOH SIL system as an example, the DBU-HexOH is miscible with decane, whereas its ionic form (DBU-HexOH CO2 ) is not.As shown in Figure 2b, the conversion in polarity remained after three CO 2 charging-discharging cycles.Pumping CO 2 through the homogenous SIL-decane solution caused the formation of emulsion, while simply removing CO 2 by heating restored miscibility.DBU-PrOH, DBU-BuOH, and DBU-OctOH SIL systems (Table A1) demonstrated a similar polarity conversion performance to that of DUB-HexOH.Other SIL systems, however, encountered difficulties in restoring their non-polar state from SIL CO2 during the heating process, possibly due to the strong interaction formed between the alcohol compound and CO 2 [62].Our study, therefore, focused on the analysis of DBU-PrOH, DBU-BuOH, DBU-HexOH, and DBU-OctOH SIL systems.
The formation of ionic compounds was confirmed by analyzing the conductivity and viscosity of the SIL systems.Generally, SIL and SIL demonstrated significantly lower conductivity than that of SIL CO2 .Alcohol and DBU (0.012 µS/cm) both possess low electroconductivities.As expected, the SILs generated by mixing DBU and alcohols produced low conductivities.The conductivity of SIL, shown in Figure 2c, decreased with the growth of carbon chains, and the conductivity of DBU-PrOH, DBU-BuOH, DBU-HexOH, and DBU-OctOH gradually decreased from 0.148 µS/cm to 0.021 µS/cm.Introducing CO 2 allowed the SIL to convert into its ionic form, producing a dramatic increase in conductivity in the SIL CO2 .Among all the tested SIL systems, DBU-HexOH CO2 presented the highest conductivity of 1.250 µS/cm.Discharging CO 2 allowed the electroconductivity to be restored.However, the conductivity of SIL was generally higher than that of the SIL system; this could be attributed to the incomplete removal of CO 2 during the heating process.In terms of their viscosities, SILs presented slightly higher viscosities than those of DBU (0.0005 Pa•s) and alcohols (~10 −4 Pa•s).As shown in Figure 2d, the viscosities of SILs were within the range of 0.007 Pa•s-0.0126Pa•s.Similar to conductivity, charging with CO 2 caused a surge in SIL viscosity, especially for the DBU-BuOH CO2 (0.9127 Pa•s) and DBU-HexOH CO2 (0.8531 Pa•s) samples.Similarly, viscosity was restored by removing CO 2 from the system.The significant increase in viscosity relates to the formation of an ionic compound: This created more and stronger ionic interactions, enhancing the internal friction in SIL CO2 .As shown in Figure 2, the DBU-BuOH and DBU-HexOH SIL systems demonstrated the most notable switching ability for conductivity and viscosity, whereas DBU-BuOH CO2 (Figure 2c) presented some difficulties in terms of CO 2 release when heated.Although there was only a minor difference in their SIL and SIL' viscosities, the conductivity of DBU-BuOH' (0.616 µS/cm) was much higher than that of DBU-BuOH (0.126 µS/cm), indicating the incomplete removal of CO 2 .Therefore, DBU-HexOH was deemed the preferred SIL system, as its solvent properties can easily be restored from 1.250 µS/cm to 0.240 µS/cm, demonstrating a preferred switching ability.
DBU is a strong organosuperbase with a pH of 15.01; SILs composed of DBU and alcohols, therefore, also present strong basicity.As shown in Figure 2e, no noticeable variations in the pH of DBU-PrOH, DBU-BuOH, DBU-HexOH, or DBU-OctOH, which ranges from 14.96 to 14.75, could be obtained.Charging with CO 2 caused the formation of SIL CO2 and decreased its pH to around 12, while removing CO 2 restored the SIL pH to about 14, which is close to that of SIL.Therefore, the SIL system pH was also regulated by charging and discharging CO 2 .For example, DBU-HexOH had a pH of 14.76; charging CO 2 decreased the pH to 11.92 (DBU-HexOH CO2 ), and removing CO 2 by heating restored the pH to 14.24 in DBU-HexOH'.However, it is proposed that a strong basicity environment may affect the aromatic structure of lignin, destroy the C-O single bond in lignin, and lead to a significant chemical modification of lignin macromolecules [63].It would be wise to introduce a cosolvent to moderate the alkaline operating condition, as this may ease the side reactions.In addition, the cosolvent would help reduce the high viscosity of SIL CO2 , which may prohibit lignin dissolution.

From Switchable Ionic Liquids to Switchable Deep Eutectic Solvents
The use of cosolvent has been reported to enhance the solubility of lignin derivatives [54,64].However, for this specific use, it should be carefully selected to stabilize the SIL system and promote the hydrogen-bond network within.The molecular simulation was, therefore, conducted with the conductor-like screening model for real solvents (COSMO-RS), in which quantum chemical calculations were combined with statistical mechanics to explore the thermodynamic behavior of the DBU and HexOH used in this work [61].As was previously demonstrated, charging CO 2 into SIL caused the formation of ionic SIL CO2 .The molecule structures and their charge distributions of the chemicals in the DBU-HexOH SIL system are shown in Figure 3.As can be seen in Figure 3a, charging CO 2 caused the formation of a strong hydrogen-bond donor (HBD) center (blue-purple area) on the DBUH from DBU.In addition, a strong hydrogen-bond acceptor (HBA) zone (red area) was created on the HexCO 3 from the HexOH.Therefore, in addition to the strong ionic interaction in SIL CO2 , possible hydrogen bonds could also exist in the DBU-HexOH CO2 mixtures.Figure 3b shows the charge distribution on the molecular surface, where the σ values lower than −0.0082 e/Å 2 are the HBD region, σ values between −0.0082 e/Å 2 and 0.0082 e/Å 2 represent the non-polar region, and values higher than 0.00821 e/Å 2 are the HBA region.As shown in Figure 3b, DBU and HexOH acted as mild polar molecules, as most of their peaks were located in the non-polar and weak HBD (−0.0082~−0.015e/Å 2 )/weak HBA (0.0082~0.015 e/Å 2 ) regions.According to the reaction scheme (Figure 1), charging with CO 2 converted them into DBUH and HexCO 3 , where a strong HBD peak at −0.021 e/Å 2 and an HBA peak at 0.020 e/Å 2 are noticed.The sigma potential results in Figure 3c indicate that only a small change in the mixture was induced by CO 2 charging, both DBU-HexOH and its ionic form (DBU-HexOH CO2 ) presenting a strong HBD affinity.Due to charging with CO 2 , the formation of ionic substances in SIL CO2 , therefore, mainly contributes to the internal ionic interactions and only produces a minor effect on the overall hydrogen-bonding ability of the mixture.In addition, based on the COSMO-RS modeling results, introducing a cosolvent such as an HBD is an effective method of enriching the hydrogen-bonding network inside the proposed SIL solvent system.Water normally acts as both an HBD and HBA, presenting characteristic peaks located at −0.016 and 0.017 e/Å 2 , respectively.The addition of water allowed it to perform as an ideal HBD in both the nonionic (SIL) and ionic (SIL CO2 ) states of the DBU-HexOH system.As shown in Figure 3b, water presented peaks both in the strong (−0.016e/Å 2 ) and weak (−0.013 e/Å 2 ) HBD regions, both of which overwhelmed the HBD assignments within the DBU-HexOH and DBU-HexOH CO2 (DBUH and HexCO 3 ) systems.Therefore, the addition of an HBD as a cosolvent enriched the hydrogen-bond network within the DBU-HexOH SIL system.In this case, the SIL/SIL CO2 worked as the HBA and water acted as the HBD within the mixture system.
DBU-HexOHCO2 mixtures.Figure 3b shows the charge distribution on the molecular surface, where the σ values lower than −0.0082 e/Å 2 are the HBD region, σ values between −0.0082 e/Å 2 and 0.0082 e/Å 2 represent the non-polar region, and values higher than 0.00821 e/Å 2 are the HBA region.As shown in Figure 3b, DBU and HexOH acted as mild polar molecules, as most of their peaks were located in the non-polar and weak HBD (−0.0082~−0.015e/Å 2 )/weak HBA (0.0082~0.015 e/Å 2 ) regions.According to the reaction scheme (Figure 1), charging with CO2 converted them into DBUH and HexCO3, where a strong HBD peak at −0.021 e/Å 2 and an HBA peak at 0.020 e/Å 2 are noticed.The sigma potential results in Figure 3c indicate that only a small change in the mixture was induced by CO2 charging, both DBU-HexOH and its ionic form (DBU-HexOHCO2) presenting a strong HBD affinity.Due to charging with CO2, the formation of ionic substances in SILCO2, therefore, mainly contributes to the internal ionic interactions and only produces a minor effect on the overall hydrogen-bonding ability of the mixture.In addition, based on the COSMO-RS modeling results, introducing a cosolvent such as an HBD is an effective method of enriching the hydrogen-bonding network inside the proposed SIL solvent system.Water normally acts as both an HBD and HBA, presenting characteristic peaks located at −0.016 and 0.017 e/Å 2 , respectively.The addition of water allowed it to perform as an ideal HBD in both the non-ionic (SIL) and ionic (SILCO2) states of the DBU-HexOH system.As shown in Figure 3b, water presented peaks both in the strong (−0.016e/Å 2 ) and weak (−0.013 e/Å 2 ) HBD regions, both of which overwhelmed the HBD assignments within the DBU-HexOH and DBU-HexOHCO2 (DBUH and HexCO3) systems.Therefore, the addition of an HBD as a cosolvent enriched the hydrogen-bond network within the DBU-HexOH SIL system.In this case, the SIL/SILCO2 worked as the HBA and water acted as the HBD within the mixture system.Based on the COSMO-RS thermodynamic modeling results, a switchable deep eutectic solvent (SDES) system was proposed by adding HBDs as cosolvents into the SILs to improve their hydrogen-bonding networks.The addition of the HBD should be able to reduce the viscosity and cost of the solvent, expand the liquid range, and enhance the solubility [36,37].In this work, SIL and various HBDs were mixed (Figure A1); water was selected as an ideal HBD for the formation of a stable homogeneous and transparent DBU-HexOH/H 2 O system with a 1:5 mass ratio, either with or without the presence of CO 2 (Figure A2).It can be seen from Table A1 that charging CO 2 into SDES caused the elevation of conductivity from 4.848 µS/cm (DBU-HexOH/H 2 O) to 23.030 µS/cm (DBU-HexOH/H 2 O CO2 ), and removing CO 2 allowed the electroconductivity to be restored.However, the conductivity of DBU-HexOH/H 2 O (10.066 µS/cm) was higher than that of DBU-HexOH/H 2 O; this could be attributed to the incomplete removal of CO 2 due to the presence of water.Charging with CO 2 also increased the viscosity of the SDES system, from 0.0186 Pa•s (DBU-HexOH/H 2 O) to 0.0825 Pa•s (DBU-HexOH/H 2 O CO2 ).Moreover, releasing CO 2 by heating allowed the viscosity of the DBU-HexOH/H 2 O to be restored to 0.0466 Pa•s.It should be noted that the addition of the HBD significantly reduced the viscosity in the ionic state to about 1/10 of that for the SIL system (0.8531 Pa•s for SIL CO2 vs. 0.0825 for SDES CO2 ).In addition, the pH of the DBU-HexOH/H 2 O was 14.26: Charging with CO 2 decreased the pH to 9.50 (DBU-HexOH/H 2 O CO2 ), which was much lower than that of SIL CO2 (DBU-HexOH CO2 , pH 11.92).Therefore, the proposed SDES, composed of a mixture of DBU-HexOH and water, could moderate the viscosity and pH of the SIL system, while allowing it to maintain its switchable nature, making the SDES more suitable for the lignin dissolution process.

Lignin Dissolution in Switchable Solvents
The solubility of lignin model compounds in the SIL and SDES systems is shown in Table 1.The DBU-HexOH SIL system demonstrated limited lignin solubility, dissolving 12.12 mg/g syringic acid, 2.01 mg/g vanillic acid, 2.24 mg/g syringaldehyde, 22.87 mg/g ferulic acid, and 10.88 mg/g alkaline lignin at room temperature.However, its ionic form (DBU-HexOH CO2 ) did not demonstrate any lignin solubility.This phenomenon is very similar to that of DBU, which is a major component in SIL.DBU dissolved 22.79 mg/g syringic acid, 4.54 mg/g vanillic acid, 1.17 mg/g syringaldehyde, 2.26 mg/g ferulic acid, and 1.91 mg/g alkaline lignin at room temperature but did not demonstrate lignin solubility after being charged with CO 2 .As the other component in SIL, monohydric alcohol, however, demonstrated a different lignin solubility.Both HexOH and HexOH CO2 presented a very small but similar solubility to vanillic acid, syringaldehyde, and ferulic acid.However, charging with CO 2 increased their solubility for syringic acid, and alkaline lignin increased from 34.9 mg/g and 9.72 mg/g to 40.24 and 16.30, respectively.Although water presented limited lignin solubility and no notable change after charging with CO 2 , the proposed DBU-HexOH/H 2 O SDES system, mixed with SIL and water, demonstrated a notable increase in lignin solubility.The solubility of syringic acid, vanillic acid, syringaldehyde, and ferulic acid in DBU-HexOH/H 2 O achieved 207.58 mg/g, 21.95 mg/g, 7.98 mg/g, and 58.12 mg/g, respectively.These could be further increased by charging CO 2 into the system.The solubility of syringic acid, vanillic acid, syringaldehyde, and ferulic acid in the DBU-HexOH/H 2 O CO2 system was 230.57mg/g, 78.43 mg/g, 11.64 mg/g, and 452.17 mg/g, respectively.Therefore, the lignin solubility in the SDES system could also be regulated by the addition of CO 2 , and this would vary depending on the type of lignin model compounds.Particularly, for vanillic acid and ferulic acid, after charging CO 2 into SDES, their solubility increased by 357% and 778%, respectively.However, the solubility of alkaline lignin in SDES decreased from 5.67 mg/g to 4.44 mg/g after forming its ionic state.Therefore, the proposed SDES mixed with DBU-HexOH and H 2 O demonstrated improved lignin solubility, and this was also regulated by charging and discharging CO 2 .The proposed SDES system was further tested for the dissolution of milled wood lignin (MWL).The DBU-HexOH SIL dissolved 9.05 mg/g MWL, while the DBU-HexOH/ H 2 O SDES dissolved up to 213.35 mg/g MWL.Charging CO 2 into the SDES further increased MWL solubility to 279.12 mg/g.The SDES-dissolved lignin was easily regenerated using acetone as an anti-solvent, with a yield of 84.31% (SDES CO2 -MWL).The variations in weight average molecular weight (Mw), number average molecular weight (Mn), and the polydispersity index (PDI) before and after SDES treatment were tested using gel permeation chromatography (GPC); the results are given in Table 2.In comparison with the MWL lignin, both regenerated lignin samples demonstrated a higher molecular weight.The native lignin sample (MWL) was 7701 g/mol Mw and 2959 g/mol Mn, with a PDI of 2.60.The SDES-treated MWL sample had a similar PDI to that of its raw material (MWL), but the regenerated lignin demonstrated an elevated molecular weight of 9823 g/mol Mw and 3397 g/mol Mn.SDES CO2 treatment resulted in a lignin stream with a much higher molecular weight (Mw = 10,340 g/mol, Mn = 7672 g/mol) and with an improved uniformity (PDI = 1.35).Compared with the traditional DESlignin (Table 2), the lignin stream produced in this work had a well-preserved long-chain structure and a much higher molecular weight.Therefore, the GPC results indicate that the SDES treatment is a promising method for lignin dissolution and extraction, as it not only regulated the lignin solubility with CO 2 but also produced a lignin stream with a high molecular weight and improved uniformity, both of which may benefit the downstream process.The thermal oxidative degradation and stability of the MWL before and after SDES treatments were investigated using the thermogravimetric analysis (TGA) [65].As shown in Figure 4a, all the testing samples exhibited a similar thermal performance, divided into three stages.The first stage was the initial degradation stage (80~200 • C), most likely caused by the removal of moisture and volatile components [66].The second stage was the main degradation stage of lignin (200~400 • C), where lignin degradation occurred.Carboxylation breakdown of aliphatic hydroxyl groups and ether bonds present in the structure of lignin occurred, while degradation can be attributed to the side chain dehydrogenation reaction [67].Both SDES-MWL and SDES CO2 -MWL samples demonstrated a more significant mass loss, which is possibly due to the partial breakage of connections between lignin macromolecules after regeneration.The third stage was the carbonization stage (400~600 • C), where methoxy groups and C-C bonds of lignin were disrupted, with the release of volatiles and production of bio-oils [68].The decomposition of the lignin sample in Figure 4a occurred slowly at this stage, resulting in char residue [69].This could be attributed to the production of highly branched and extremely condensed aromatic structures.The MWL sample produced a high char residue rate of 66.7%; this could be attributed to the fact that the sample had a highly complex and condensed lignin structure.MWL is easily converted to char residue owing to its structural resemblance [68].Once the MWL had reached a temperature of 600 • C, the quality of the coke was stable and no further reaction occurred.However, both SDES-and SDES CO2 -treated MWL produced lower char residue rates: 28.3% and 1.3%, respectively.This indicates the lower thermal stability of SDES-MWL and SDES CO2 -MWL compared with that of the MWL sample.Carboxylation breakdown of aliphatic hydroxyl groups and ether bonds present in the structure of lignin occurred, while degradation can be attributed to the side chain dehydrogenation reaction [67].Both SDES-MWL and SDESCO2-MWL samples demonstrated a more significant mass loss, which is possibly due to the partial breakage of connections between lignin macromolecules after regeneration.The third stage was the carbonization stage (400~600 °C), where methoxy groups and C-C bonds of lignin were disrupted, with the release of volatiles and production of bio-oils [68].The decomposition of the lignin sample in Figure 4a occurred slowly at this stage, resulting in char residue [69].This could be attributed to the production of highly branched and extremely condensed aromatic structures.The MWL sample produced a high char residue rate of 66.7%; this could be attributed to the fact that the sample had a highly complex and condensed lignin structure.MWL is easily converted to char residue owing to its structural resemblance [68].Once the MWL had reached a temperature of 600 °C, the quality of the coke was stable and no further reaction occurred.However, both SDES-and SDESCO2-treated MWL produced lower char residue rates: 28.3% and 1.3%, respectively.This indicates the lower thermal stability of SDES-MWL and SDESCO2-MWL compared with that of the MWL sample.Structural characterization of lignin before and after SDES treatment was performed using Fourier-transform infrared (FTIR) spectroscopy; the assignments of the characterizing peaks are shown in Table A2.As shown in Figure 4b, the stretching vibration for hydroxyl is observed at 3422 cm −1 [70].The stretching vibrations for the lignin aromatic ring skeleton at 1600 cm −1 and 1507 cm −1 can be seen in all testing samples; this indicates that no significant change in the aromatic ring structure of MWL occurred during the SDES treatments [34,71].The peak at 2936 cm −1 was attributed to the C-H vibration in CH3-/CH2 = groups.The characteristic peaks at the wave number of 1461 cm −1 were assigned to the C-H asymmetric vibration of CH2 = groups and C-H transformation in the aromatic rings [3].All the characteristic peaks mentioned above were found in all the testing samples.Stretching vibrations for C=O bonds are typically found between 1740 and 1700 cm -1 ; this is where signals attributable to C=O bonds in unconjugated ketones, carbonyls, and ester groups are normally observed.Lower absorption energies (around 1700 cm -1 ) were reported for conjugated aldehydes and carboxyl acids [72].MWL produced a characteristic peak at 1710 cm -1 , while SDES-MWL and SDESCO2-MWL demonstrated lower intensities within this range, indicating lower carbonyl and carboxyl group content.The peak at 1327 cm −1 was attributed to the stretching vibration of the C-O in the syringyl (S) lignin unit, whereas the peaks at 1270 cm −1 represent the stretching vibration of the C-O in the guaiacyl (G) lignin unit [73].The peak at 1124 cm −1 demonstrates the presence of syringyl moieties in lignin because it represents the C-H deformation of the lignin S unit.Therefore, a Structural characterization of lignin before and after SDES treatment was performed using Fourier-transform infrared (FTIR) spectroscopy; the assignments of the characterizing peaks are shown in Table A2.As shown in Figure 4b, the stretching vibration for hydroxyl is observed at 3422 cm −1 [70].The stretching vibrations for the lignin aromatic ring skeleton at 1600 cm −1 and 1507 cm −1 can be seen in all testing samples; this indicates that no significant change in the aromatic ring structure of MWL occurred during the SDES treatments [34,71].The peak at 2936 cm −1 was attributed to the C-H vibration in CH 3 -/CH 2 = groups.The characteristic peaks at the wave number of 1461 cm −1 were assigned to the C-H asymmetric vibration of CH 2 = groups and C-H transformation in the aromatic rings [3].All the characteristic peaks mentioned above were found in all the testing samples.Stretching vibrations for C=O bonds are typically found between 1740 and 1700 cm -1 ; this is where signals attributable to C=O bonds in unconjugated ketones, carbonyls, and ester groups are normally observed.Lower absorption energies (around 1700 cm -1 ) were reported for conjugated aldehydes and carboxyl acids [72].MWL produced a characteristic peak at 1710 cm -1 , while SDES-MWL and SDES CO2 -MWL demonstrated lower intensities within this range, indicating lower carbonyl and carboxyl group content.The peak at 1327 cm −1 was attributed to the stretching vibration of the C-O in the syringyl (S) lignin unit, whereas the peaks at 1270 cm −1 represent the stretching vibration of the C-O in the guaiacyl (G) lignin unit [73].The peak at 1124 cm −1 demonstrates the presence of syringyl moieties in lignin because it represents the C-H deformation of the lignin S unit.Therefore, a feature of GS-type lignin was demonstrated in the FTIR spectra.Additionally, the signal at 1036 cm −1 , corresponding to the C-O deformation of primary -Ohs, was detected in all lignin fragments [74].The FTIR spectra did not show any noticeable chemical modifications occurring in MWLs during the SDES treatments.
In the aromatic region, the C 2,6 -H 2,6 correlations from S-type units and the sum of C 2 -H 2 , C 5 -H 5 , and C 6 -H 6 correlations from G-type units were used to estimate the S/G ratio [74].A strong signal at δ C /δ H 103.8/6.71belonged to the S unit at C 2,6 -H 2,6 , whereas the G units showing the cross-peak signals for C 2 -H 2 (δ C /δ H 110.9/6.98),C 5 -H 5 (δ C /δ H 114.9/6.77), and C 6 -H 6 (δ C /δ H 119.0/6.80)can be found in all the lignin samples.It can, therefore, be confirmed that the MWL used in this work was predominantly comprised of S and G units.The S/G ratios of MWL, SDES-MWL, and SDES CO2 -MWL were 0.88, 0.73, and 0.76, respectively, indicating that the SDES and SDES CO2 treatments caused slightly more Stype lignin loss than G-type lignin loss.Other signals observed in the aromatic regions could be assigned to p-hydroxycinnamyl alcohol end groups (I), cinnamaldehyde end groups (J), and p-hydroxybenzoate substructures (PB).The C-H correlated signals for I and J were found at δ C /δ H 128.2/6.44 and 126.1/6.76 ppm, respectively.The C 2,6 -H 2,6 correlations for PB were observed as a strong signal at δ C /δ H 131.2/7.67.These characterization results clearly demonstrate, again, that the lignin macromolecule framework remained intact following both the SDES and SDES CO2 treatments.In the aromatic region, the C2,6-H2,6 correlations from S-type units and the sum of C2-H2, C5-H5, and C6-H6 correlations from G-type units were used to estimate the S/G ratio

Polymers 2023 ,
15, x FOR PEER REVIEW 3 of 21 method for lignin dissolution and extraction; this could further facilitate the development of plant fiber pretreatment technology.

Figure 2 .
Figure 2. Switchable solvent properties of SILs triggered by CO2: (a) effect of CO2 on the state of DBU-HexOH SIL system; (b) effect of CO2 on the miscibility of DBU-HexOH in decane; (c) effect of CO2 on the electric conductivity of DBU-based SILs; (d) effect of CO2 on viscosity of DBU-based SILs; and (e) effect of CO2 on pH of DBU-based SILs.

Figure 2 .
Figure 2. Switchable solvent properties of SILs triggered by CO 2 : (a) effect of CO 2 on the state of DBU-HexOH SIL system; (b) effect of CO 2 on the miscibility of DBU-HexOH in decane; (c) effect of CO 2 on the electric conductivity of DBU-based SILs; (d) effect of CO 2 on viscosity of DBU-based SILs; and (e) effect of CO 2 on pH of DBU-based SILs.

Figure 3 .
Figure 3. Molecular modeling results from COSMO-RS: (a) surface charge distribution for DBU and HexOH before and after charging CO2; (b) sigma profiles of water and the chemicals in the SIL system; and (c) sigma potentials for the DBU-HexOH and DBU-HexOHCO2 mixtures.

Figure 3 .
Figure 3. Molecular modeling results from COSMO-RS: (a) surface charge distribution for DBU and HexOH before and after charging CO 2 ; (b) sigma profiles of water and the chemicals in the SIL system; and (c) sigma potentials for the DBU-HexOH and DBU-HexOH CO2 mixtures.

Table 1 .
Solubility of switchable solvent system to lignin model compounds (25 • C, mg/g).Data with extremely low solubility and which could not be detected.

Table 2 .
Molecular weights and polydispersity indices of lignin.

Table A1 .
Cont.Data that cannot be determined as the sample is solid; * Data test with the presence of water.

Table A2 .
Assignment of Fourier-transform infrared spectrum analysis of lignin samples.