Tunable and functional deep eutectic solvents for lignocellulose valorization

Stabilization of reactive intermediates is an enabling concept in biomass fractionation and depolymerization. Deep eutectic solvents (DES) are intriguing green reaction media for biomass processing; however undesired lignin condensation is a typical drawback for most acid-based DES fractionation processes. Here we describe ternary DES systems composed of choline chloride and oxalic acid, additionally incorporating ethylene glycol (or other diols) that provide the desired ‘stabilization’ function for efficient lignocellulose fractionation, preserving the quality of all lignocellulose constituents. The obtained ethylene-glycol protected lignin displays high β-O-4 content (up to 53 per 100 aromatic units) and can be readily depolymerized to distinct monophenolic products. The cellulose residues, free from condensed lignin particles, deliver up to 95.9 ± 2.12% glucose yield upon enzymatic digestion. The DES can be recovered with high yield and purity and re-used with good efficiency. Notably, we have shown that the reactivity of the β-O-4 linkage in model compounds can be steered towards either cleavage or stabilization, depending on DES composition, demonstrating the advantage of the modular DES composition.


Materials and Methods
Materials: All the chemicals were used as received from TCI or Sigma Aldrich without any purification. Choline chloride (ChCl), oxalic acid dihydrate (OA) and ethylene glycol (EG) used for deep eutectic solvents (DESs) preparation and n-octadecane and 3, 5-dimethylphenol were used as an internal standard were purchased from Sigma-Aldrich. Lignin model compounds were prepared as previously reported (Supplementary Section 4.1). Birch The lignins ChCl/OAL and DPL10 were further characterized by Agilent PY-3030D/7890B-5977A pyrolysis GC-MS (Agilent Technologies Co. Ltd, USA). The pyrolysis was performed at 500℃ for 30s, the oven temperature was programmed from 40℃ to 220℃ at 5℃/min, Helium (99.999%, 1mL/min) was used as carrier gas, all the compounds were identified by comparing the mass spectra with those of the Wiley and NIST libraries.
NMR spectroscopy: 1 H and 13 C NMR spectra were recorded on a Varian Mercury Plus 400, Agilent MR 400 (400 and 100.59 MHz, respectively) and Bruker Advanced NEO 600 (600 and 150.92 MHz, respectively) using CDCl3 or acetone-d6 as a solvent. 1 H and 13  Molecular simulations were used to further investigate the hydrogen bond interaction in the DES systems. Gromacs 2016.1 software package was used for the simulations with a temperature of 373K using a time step of 2fs. The V-rescale thermostat 1 and Parrinello-Rahman barostat 2 were used for controlling the temperature and pressure. Hydrogen containing atoms were constrained using LINCS algorithm. 3 Periodic boundary conditions were used, the Particle Mesh Ewald (PME) algorithm was used for long-range electrostatic interactions, the switch function for the van der Waals and electrostatic interactions was 1.0 nm. Gromos54A7 field was used for choline chloride (ChCl), oxalic acid (OA) and ethylene glycol (EG) from ATB library and SPC water model was employed. 4 Typically, two binary DES systems (100 ChCl and 200 EG or 100 ChCl and 20 OA molecules in the box, respectively) and a ternary DES system (100 ChCl, 200 EG and 20 OA molecules in the box, representing DP-DES10) were created by Packmol package and subsequently simulated, respectively. After energy minimization, the simulations were run for 500 ns under the NPT system 373 K and ambient pressure, the last 100ns simulation was extracted for analysis. All the data were analyzed and visualized with the help of Gromacs' tools and Visual Molecular Dynamics (VMD) package.
The density functional theory (DFT) calculations were used to investigate the interaction between HBA and HBD in DES. All DFT calculations were performed at the Gaussian 16 software package. The geometries of DESs were fully optimized using B3LYP/6-311+G** basis set. 5

Composition analysis of the cellulose residue and DP-DES or CS-DES fractionated lignin
The composition of the solid samples was determined according to the NREL procedure. 6 The monosaccharide was determined by HPLC (1260 Agilent Technologies, USA) with a refractive index detector (RID). An aminex column HPX-87H (Bio-Rad, USA) was used to analyze the monosaccharides at 50 °C with 5 mM sulfuric acid at a flow rate of 0.6 mL/min.
In these condition, xylose, mannose and galactose was eluted at the same retention time which were integrated at a single peak. The composition of birch is: 39.4% cellulose, 27.9% hemicellulose, 19.3% acid insoluble lignin (AIL) and 1.1% acid soluble lignin (ASL).
Elemental analysis (C, H and N) of protected lignin was performed using an Elementar VarioMICRO Cube, 3-4 mg of protected lignin was used to determine the C, H and N content.

Milled wood lignin isolation procedure
The ball wood lignin was extracted by following from published steps. 8 20 g wood powders were suspended in 400 mL dioxane/water (v/v, 96:4) and stirred for 24 h under dark, and the residue was collected by centrifugation and extracted by fresh solvent for another 24 h. The liquid was combined and condensed to approximately 30 mL, and further precipitated in 3x the volume of 96% ethanol, the ethanol phase was condensed to around 30 mL (this step was repeated) and precipitated in 10 volume times acid water (pH=2), the isolated lignin was further washed with acidic water and freeze-dried.

Enzymatic hydrolysis of the DES fractionated cellulose residues
Enzymatic hydrolysis of the cellulose residues was performed as follows: 7.5 mL of 4% (w/v) solid samples (0.3 g) in a buffer (sodium acetate, 50 mmol/L, pH 5.5, 2.88 mg tetracycline chloride) was added into a 50 mL tube with cap, and the tubes were kept at 50 °C in a VWR incubating orbital shaker (Model 3500 L) at 250 rpm for 72 h. Enzymatic hydrolysis was conducted with CTec2 (Novozymes, Denmark, 0.1 mL/g substrates) and 0.3 mL of the hydrolysate was sampled periodically to determine the released sugar amount. After the inactivation of the enzyme (5 min 100 °C), the glucose and xylose were determined by HPLC (Agilent 1260 series, Agilent Technologies, USA) with an autoinjector and RID detector. An HPX-87H (BIO-RAD, USA) with a 4 mM H2SO4 eluent at 0.6 ml/min was used. Here, the ternary DES (guanidine hydrochloride/EG/p-TSA and also ChCl/EG/p-TSA) was used for lignocellulose fractionation showing xylan and lignin removal from switchgrass and cellulose retention. The 2D NMR of the lignins showed significant change in lignin interunit linkages. In this paper, the ternary DES (ChCl-Glycerol-AlCl3·6H2O) was used for removing lignin from garlic skin and green onion root. No lignin characterization or tunability of the system was shown.
The aqueous portion was extracted with Et2O (2 x 100 mL). The combined organic extracts were washed with brine, dried over MgSO4, filtered, and concentrated in vacuo. Purification solid. 13 C NMR spectrum of 1a.

Compound4-(3-hydroxy-1-(2-hydroxyethoxy)-2-(2-methoxyphenoxy)propyl)-2-methoxyphenol (4da)
The mixture of products was obtained after extraction workup with dichloromethane and water. The crude mixture containing 4da was purified by column chromatography (methanol: dichloromethane, 2:98) to give a mixture of 4daas an oil in 82% yield.4da were identified separately by NMR.  Supplementary Figure 19 shows the 1 H NMR of the DES mixtures. The signals for all the aliphatic protons of the components can be identified in both the 1 H and 13 C NMR spectra. In all cases, for the protons belonging to the hydroxyl groups of ChCl and EG (4 -ChCl, 7 -EG) and carboxylic group of OA (5), respectively, only one signal can be seen in the NMR spectra recorded on the neat DES. This coalescence of the protons belonging to a hydroxyl group (including water) can also be seen in the literature in the case when water is present in the system. In the present case, DES was prepared starting from dry choline chloride and anhydrous oxalic acid and ethylene glycol, but water is generated by the ester formation side reaction between the oxalic acid and choline chloride or ethylene glycol. The ester formation was previously shown in the literature. 33,34 In the NOESY spectrum (Supplementary Figure 20) of ChCl:OA:EG (1:1:2, neat) several correlations between the peak corresponding to the hydroxyl groups and the rest of the molecules, the -CH2-groups of ChCl and EG and CH3 of ChCl, can be observed (green oval).
Due to the fact that all -OH and COOH overlap, these can also be intramolecular correlations and intermolecular correlations. Of particular interest is however a correlation between the peaks corresponding to the -CH2-groups of EG (3.55 ppm) and the ones corresponding to the CH3 groups of choline chloride (3.20 ppm) (marked with blue circle). This is strongly suggesting a steric proximity of these two components, supporting the involvement of EG in the ternary DES.
According to the literature, the viscosities of the studied DES systems are as presented in the  (1) where D is diffusivity, k is the Boltzmann constant, T is the temperature, η is the viscosity, and r is the radius of the diffusing solute molecule.
From this, r will be Equation 2 The values obtained for the radius of the diffusing molecule suggest that the ternary mixture

IR measurements of DES
The prepared DES samples were also analyzed by FTIR-ATR spectroscopy, the data is presented in Supplementary Figure 23  Generally, ρ and ▽ 2 ρ were used to describe the properties of hydrogen bond. 5 As the results in Supplementary Table 10 showed, all the ρ and ▽ 2 ρ are within the range of electron density

Molecular dynamic simulation of DESs
The simulation of different DES systems 4 showed that in ChCl/EG DES, the average number of hydrogen bonds (HBs) between chloride and ethylene glycol (Cl-EG) is 258 ( Supplementary Figures 26 and 27). When oxalic acid (OA) was added for preparing DP-DES10, it can be observed that the HBs of Cl-EG decreased to 241 and the HBs of Cl-OA are 17, this is because the hydrogen bonding interaction of Cl-OA is stronger than Cl-EG as suggested by DFT calculation.

Supplementary Note 5: DES for lignocellulose fractionation studies
We have now performed a series of control experiments using microcrystalline cellulose  Figure 51). These experiments strongly suggest that the nanosized particles are condensed lignin. The ability of DP-DES10 to 'protect' the lignin from condensation is likely responsible for this behavior, compared to ChCl/OA DES which has higher acid content and no stabilization function.

Thermal degradability
The thermal stability of two lignin samples (DPL10 and ChCl/OAL) was determined by TG-FTIR, respectively ( Supplementary Figures 45 and 46) The yield of solid residues for DPL10 and ChCl/OAL after 800℃ degradation was 24.5% and 40.8% respectively, consistent with the degree of condensation. The results of TG-FTIR at 220℃ also showed differences between the two samples. Both samples displayed typical peaks of water with the characteristic bands at 3500-3964 cm -1 and 1300-1800 cm -1 as well as characteristic bands at 2271-2391 cm -1 and 586-726 cm -1 indicative of CO2 formation. 40 These signals were more pronounced for DPL10 and less intense for ChCl/OAL, as expected from the respective -O-4 contents and level of condensation.

Pyrolysis GC-MS
In addition, these lignins ChCl/OAL and DPL10 are further characterized by pyrolysis Table 14 and Figure 47). The stabilized DPL10 lignin clearly demonstrated higher yield of pyrolytic monomers compared to ChCl/OAL, consistent with TG-FTIR analysis.  Figure 51d), was fibrillated and without formation of nanoparticles.

Analysis of the behavior of xylan and MCC in DES
Microcrystalline cellulose (MCC) alone or xylan were separately treated in different DES at 100 ℃ for 24 h (Supplementary Figures 52a and 52b). Gratifyingly, MCC was very stable in DP-DES10 (99% retention) and displayed high stability in DP-DES20 (containing more OA) as well (93.5% retention). This is in good agreement with the lignocellulose fractionation data see Supplementary Tables 16 and 18.
Next, 0.35g xylan was stirred in DP-DES10 at 100℃ for 24h as hemicellulose model (Supplementary Figure 49c). A clear liquid was obtained after the indicated time, without any coloring, indicating full dissolution of xylan. In order to determine whether xylan has further reacted, the mixture was diluted with 10 mL water, filtered through a 0.22 μm microfilter and further analyzed by HPLC in terms of xylose content. Partial hydrolysis of xylan was confirmed by detecting 7.2±0.42% yield of xylose.

In-depth composition analysis of the obtained CR
The hemicellulose retention and composition residues after different DES treatments were analyzed with HPAEC-PAD. The obtained residues were hydrolyzed via acid hydrolysis (samples were diluted 20 times before injection). Therefore, about 15 mg of each pellet was mixed with 0.45 mL 72% (w/w) sulfuric acid and incubated for 1h at 30°C. After diluting the acid to 1 M, the incubation was continued for 3h at 100°C to fully hydrolyze the carbohydrates into monomers. Arabinose, galactose, glucuronic acid, mannose, rhamnose, and xylose were quantified by integrating the peak area of corresponding standards. Total hemicellulose content was calculated as a sum of all neutral sugars and glucuronic acid.
Elution of monosaccharides (0.25 mL min -1 ) was performed with a multi-step-gradient using The obtained hemicellulose composition of Birchwood was in a similar range of previously published data 41,42 . It contains xylose (25% w/w) in the polymeric form of xylan.
Around 2% w/w of galactose was detectable and traces of arabinose and rhamnose. The applied DES treatments lead to a loss in total hemicellulose content. However, the ternary DES treatment retain more hemicellulose (~38% w/w) if compared to the binary DES treatment (23% w/w). The total hemicellulose content decreased from 28% w/w in raw birchwood down to 12% w/w after ChCl/OA DES treatment. It was shown that xylan was debranched completely and partially degraded as can be seen in the loss in xylan content and complete loss of galactose, rhamnose and arabinose. indicating an approximately 30% loss of EG during the fractionation process.

Quantification of losses of Oxalic acid (OA) during recycling
In order to quantify potential losses of oxalic acid (OA) during recycling and DES treatment, the pH values of the fresh DP-DES10 and the recycled RDP-DES10 were determined according to the literature. 55 The pH value of the fresh DP-DES10 is 1.06, which changed to 2.57 after one reaction/recycling step (Entry 2). The pH value could be easily recovered by adding 107 mg OA (Entry 3) to RDP-DES10. When this DES was used for a new fractionation run and recycled again, the pH value appeared to be similar to that obtained previously 2.51 (Entry 4), showing good reproducibility of the pH change after processing.
A control recycled DES just after mixing DP-DES10 and lignocellulose without heating showed a less significant pH change to 1.59 (Entry 5), this indicates that the loss of OA may occur both in the fractionation and recycling procedure.
As comparison, the other DES compositions with more acid content, such as the ChCl/OA as well as the CS-DES designed for acidolysis or DP-DES20, had pH values of 0.62, 0.74, 0.9 respectively, as expected.

Investigation of purity of the recycled DES and its effectiveness for further fractionation
In order to determine the purity of the DES after lignocellulose fractionation and recycling, we have next performed semi-quantitative 1 H NMR analysis of the used and recycled DES.
First, we could assign all signals in the 1 H NMR spectrum of the fresh DP-DES10, which beside the individual components, also showed small peaks of esters from the DES components, as already described in literature [6,7] . Semi-quantitative analysis was used to determine the "estimated purity" of DES according to the 1 H NMR of DES (this does not take into account changes in inorganics content). All signals were assigned, integrated and set the total integration to 100. The "estimated purity" of DES was calculated as follows: proton of all the unknown impurities/total protons in recycled DES×100%. It was seen, that the estimated NMR purity of freshly prepared DP-DES10 is 100%. The purity of the DESs was shown in Supplementary Table 12

Supplementary Note 7: Characterization of CRs by XRD and BET
The crystallinities of the untreated wood and fractionated cellulose residues (ChCl/OACR and DPCR10) were characterized by X-ray diffraction (XRD). All the samples showed typical cellulose Ⅰ structure as shown in Supplementary Figure 56, the crystallinity of untreated wood was 57% according to Segal's method, while the ChCl/OACR and DPCR10 showed increased crystallinities of 68% and 70%, respectively. The increase in crystallinity is typical for DES fractionated samples, and is attributed to the removal of lignin and some of the hemicellulose (amorphous components) during fractionation.
The BET surface area of the cellulose residues was tested by Micromeritics ASAP2020 gas adsorption analyzer. The results are summarized in Supplementary Table 21