Transformations of the Lignin Molecular Structure During the Acidic and Alkaline DES Treatments

The DES wheat straw lignins, LCL extracted using acidic Lac–ChCl DES and KGL extracted using alkaline Gly–K₂CO₃ DES, exhibited distinct physicochemical properties. LCL had 38% higher purity but 21% lower yield than KGL (Table S3). However, their monosaccharide contents were only slightly different (Table S3). The thermogravimetric results indicated that KGL degraded at a lower rate than LCL (Table S4). Moreover, LCL exhibited a relatively narrow molar-mass distribution (Figure S2).

Two-dimensional heteronuclear single-quantum correlation nuclear magnetic resonance (2D HSQC NMR) analysis was used to explore the structural properties of the aromatic units as well as the interunit linkages of the DES-isolated lignins. The Lac–ChCl and Gly–K₂CO₃ DES treatments substantially altered the molecular structure of lignin, as depicted in Figure 1a–d and detailed in the corresponding cross-signal assignments in Table S5. The aliphatic region in the 2D HSQC NMR spectra of LCL (Figure 1a) displayed a characteristic signal at δ_C/δ_H 68–72/4.5–5.5, which corresponded to Hibbert's ketones (HK). These ketones, which are typical products of lignin acidolysis, were derived from the arylglycerol–β–aryl ether (β–O–4) substructures.^[41-43] Thus, acidic DES triggered simultaneous lignin degradation and dissolution, making it a promising solvent with great potential for industrial-scale lignin fractionation.^[16] In contrast, conventional acid hydrolysis of lignin is challenging to scale up due to limited lignin solubility and high process costs.^{[44, 45]}

Compared with LCL, KGL displayed a structural characteristic, exhibiting more β–O–4 linkages, including erythro and threo side-chain structures (Figure 1b), as corroborated by its lower phenolic hydroxyl group contents (1.57 mmol g⁻¹ in KGL vs. 1.80 mmol g⁻¹ in LCL, Figure S3). These findings indicate that the mechanism of lignin degradation in alkaline Gly–K₂CO₃ DES is similar to that of the delignification process of traditional alkaline pulping.^{[46, 47]} However, the alkaline-degradation mechanisms were more complex than the acidolysis ones, and the KGL structure was more similar to that of intact, natural lignin than to that of lignin extracted by alkaline pulping.^[48-50] These changes yielded lignin fractions with unique characteristics, impacting their potential subsequent utilization. LCL, with high purity and phenolic hydroxyls (Table S3 and Figure S3), is ideal for bio-based adhesives, phenolic resins, or antioxidants.^{[2, 51, 52]} KGL, with high thermal stability and structural strength, suits applications like composite reinforcement, thermal insulation, or carbon material precursors.^[53-56]

Dissolution Mechanisms of Lignin in the Acidic and Alkaline DESs

The molecular-level dissolution mechanisms of lignin in the various DES environments were further elucidated by computational simulations using the guaiacylglycerol–β–guaiacyl ether (GGE) model compound. Particularly, we demonstrated the interrelationship between the structural transformation and dissolution behavior of LCL and KGL. The β–O–4 structure in lignin includes erythro and threo stereostructures, each bearing a pair of enantiomers (Figure 2a). The ratio of these stereostructures varies with the biomass origin.^[57] The erythro isomer of GGE was used to simulate the molecular NCIs between lignin and the DES molecules, as it is more abundant in grass lignin, e.g., wheat straw lignin, which was utilized here as our raw material.^[58]

The RDG scatter graph versus electron density multiplied by the sign of the second eigenvalue (sign λ₂)ρ was plotted to identify the properties and strengths of the intra/intermolecular NCIs in the whole GGE–solvent molecular systems (Figure 2b). The RDG scatter points presented the corresponding strong attractive interactions (blue region), van der Waals interactions (vdW, green region), and strong repulsive interactions (steric effects, red region). Moreover, the vdW attractions and steric repulsions dominated in the GGE–ChCl–Lac, GGE–Gly–K₂CO₃, GGE–THF, and GGE–DMSO systems, as shown by the large, color-coded areas. Particularly, the largest scattering was concentrated in the green region, with multiple spikes in the −0.02<sign(λ₂)ρ<0.01 range, which is characteristic of strong vdW interactions, as shown in the scatter plots of all the GGE–solvent systems.^{[59, 60]} A spike appeared exclusively in the GGE–ChCl–Lac system around sign(λ₂)ρ of −0.04 a.u., indicating strong H-bonding interactions.^[61] The isosurface maps corresponding to the NCIs of the most stable GGE configurations under distinct solvent effects are shown in Figure 2c. The disk-shaped color domains indicate NCIs within GGE in the mixed molecular systems. The large green–red domain between the two aromatic rings of the GGE molecule displayed the intramolecular π–π stacking interactions (a strong vdW interaction) in the GGE–THF system, which folded the aromatic rings, generating the offset-stacked GGE configuration. In other GGE–solvent systems, the π–π stacking effect between aromatic rings was relatively weak, and strong repulsive interactions (red domains) existed despite the presence of weak H-bonding interactions (blue–red domains) between the oxygen and hydrogen atoms.

The effects of different solvent environments on the structural conformations of GGE were further illustrated by plotting the NCI isosurfaces of the GGE–solvent systems (Figure 3a–d). The GGE molecule was well solvated in all the investigated solvents via strong intermolecular attractive interactions, as characterized by the large green isosurface domains. The strong H-bonding interactions existed between GGE and the acidic ChCl–Lac or alkaline Gly–K₂CO₃ DES molecules, as revealed by several small, green disk-like domains, and these interactions reduced the intramolecular stacking. These intra/intermolecular NCIs contributed to the most stable lateral-shifted conformation of GGE presented in the anhydrous ChCl–Lac and Gly–K₂CO₃ DES systems, accounting for ∼91% and 100% of the Boltzmann distributions, respectively (Figure 3a,b). The GGE–THF system exhibited very strong π–π interactions between two GGE rings, as indicated by the large green–red isosurface disks (Figure 3c), keeping the aromatic rings stacked in an offset mode.^[62] Two offset-stacked conformations of GGE with the lowest energy were observed in the THF environment, accounting for 67% and 33% of the Boltzmann distribution. Correspondingly, DMSO induced the stretching of two aromatic rings within GGE. The stretched conformation of GGE with the second lowest energy accounted for more than 24% of the Boltzmann distribution. Additionally, two other stable lateral-shifted conformations were formed (∼37% and 18% of the Boltzmann distribution), mainly exhibiting intramolecular aromatic C─H···π interactions. Thus, GGE dissolution in THF was almost triggered by vdW interactions, and the intramolecular π–π stacking interactions were simultaneously well preserved. In contrast, the DMSO molecule most efficiently disturbed the GGE stacking (Figure 3d). Figure 3e–h further reveals the NCIs of the GGE clusters in four solvent systems (GGE–ChCl–Lac, GGE–Gly–K₂CO₃, GGE–THF, and GGE–DMSO systems). In the GGE–DES systems, the lateral shifting of the GGE molecules prevented their tight assembly, and the solvent molecules filled the spaces within GGE clusters, partially disrupting their π-stacking interactions. In contrast, in the GGE–THF system, GGE clusters exhibited more compact, offset-stacked configurations, primarily driven by intramolecular π–π interactions, suggesting that THF favored π-stacking compared to DESs. Further, in the DMSO molecular system, intramolecular stacking of GGEs was disrupted, leading to a stretched configuration of GGE clusters.

Formation Mechanisms of Ultrasmall LNPs in Different Solvents

The LNPs were prepared from a direct self-assembling process in aqueous acidic (ChCl–Lac/H₂O) and alkaline (Gyl–K₂CO₃/H₂O) DES systems via dialysis; THF and DMSO were employed as reference solvents. The particle size distributions of the LNPs (Figure 4a–f), as determined by the statistical analyses of the field-emission scanning electron microscopy (FESEM) images (Figure 4g–l), indicated the notable variations in the particle sizes of LNPs: 100–180 nm (∼90%) for ChCl–Lac-dialyzed LNPs (LCLNPs), 500–2000 nm (∼82%) for THF-dialyzed LNPs derived from LCL (T_LCLNPs), 40–90 nm (∼82%) for Gly–K₂CO₃-dialyzed LNPs (KGLNPs), and 200–1000 nm (∼84%) for THF-dialyzed LNPs derived from KGL (T_KGLNPs), respectively. Thus, the average diameters of the particles obtained from acidic and alkaline DES-mediated dialysis (142 nm for LCLNPs and 66 nm for KGLNPs) were approximately one-ninth and one-seventh, respectively, of the diameters of particles obtained from THF-mediated dialysis (1,330 nm for T_LCLNPs and 492 nm for T_KGLNPs). Additionally, the ultrasmall sphere-like nanoparticles formed in DES-dialyzed systems (Figure 4g and j) exhibited a more homogeneous size distribution than that of the spherical LNPs formed in THF-dialyzed systems (Figure 4h and k). When DMSO was exchanged with water molecules during dialysis, it did not trigger lignin aggregation into uniform LNPs (Figure 4i and l). The stability of the LNPs was monitored based on the zeta potential (ζ–potential) of particle dispersions as a function of time (Figure 4m). All the LNPs displayed a high ζ–potential (less than −40 mV) and remained highly stable for 120 days. The ζ–potentials of the DES-obtained LNPs were more negative than those of the THF-dialyzed samples. Additionally, the alkaline DES-dialyzed LNPs exhibited highly negative ζ–potentials, indicating enhanced dispersion stability.

The results indicated that the intra/intermolecular π–π interactions were the key driving forces behind the formation of compact LNPs. Specifically, offset-stacked and lateral-shifted conformations stabilized lignin aggregation, promoting nanoparticle formation. In acidic and alkaline DESs, lignin was strongly solvated through H-bonding and vdW interactions. However, despite this solvation, the lateral-shifting interactions between lignin molecules remained sufficient to drive the formation of nanoparticles. In view of this, the strong weakening of the π–π interactions of lignin induced by the solvent molecules (DMSO) prevented the lignin self-assembling as well as nanoparticle formation during dialysis. These findings highlight the crucial role of solvent properties in determining the formation and characteristics of LNPs (Figures 4n–o). Moreover, DESs used in the dialysis process can be recovered and reused at least five times for the fabrication of LNPs without significant loss of production. In contrast, traditional organic solvents such as THF and DMSO are non-recyclable (Table S6). The yields of DES-LNPs ranged from 96% to 98%, while DES recovery yields were maintained between 90% and 94% over five fabrication cycles (Figure S4). The superior performance of DESs in terms of high dissolution efficiency, excellent solvent recyclability, and overall fabrication sustainability compared to conventional solvents is highlighted in Table S7.

To elucidate the underlying mechanism of the experimental results, we performed MD simulations to investigate the self-assembly behavior of lignin molecules and the formation dynamics of LNPs in the solvent-exchange process. The simulations employed in this study focused on the determination of the structural conformations of a high molar-mass grass-like lignin model (4,182 g mol⁻¹, Figure S5) in different aqueous solvent systems. At a high H₂O content (90% by volume), the degree and morphology of the lignin molecules assembled into clusters varied distinctly in different solvent systems (Figures 5a–d and S6). The dynamic trajectories of the lignin molecules over 200 ns are presented in the supplementary animated images. Figure 5e–h show the average number of H-bonds between lignin and other components in the lignin–solvent/H₂O mixed systems. In the presence of H₂O, the number of H-bonds between lignin and H₂O molecules fluctuated between 76 and 116 throughout the simulation, indicating the rapid formation of H-bonding interactions between the lignin and H₂O molecules. The number of H-bonds between lignin and H₂O decreased in the following sequence: DMSO>ChCl–Lac DES≥Gly–K₂CO₃ DES>THF. This suggests that the H₂O molecules effectively disrupted the existing H-bonds between DESs and lignin, leading to higher lignin dispersion compared to conventional solvent exchange with THF, although still weaker than in DMSO. Further, the number of H-bonds between the lignin and solvent molecules decreased in the following sequence: Gly–K₂CO₃ DES>ChCl–Lac DES>THF>DMSO, indicating that H-bonding remained in DES/H₂O systems after solvent exchange. At equilibrium (after 50 ns), the number of H-bonds between lignin and H₂O molecules was much higher than that between the lignin and solvent molecules. This highlights the criticality of H-bonding interactions between H₂O and lignin in lignin stacking. This finding is also supported by the number of H-bonds determined in the GGE–solvent/H₂O simulation systems (Figures S7a–d).

The RMSDs were calculated from 200 ns simulation trajectories to measure the degrees of conformational changes in the lignin clusters at different time steps in relation to the initial conformation (0 ns) (Figure 5j). The fluctuations in the deviation also reflected the stability of the lignin cluster throughout the simulation. The obtained RMSDs of self-assembled lignins increased rapidly with fewer fluctuations, stabilizing at a conformational deviation (13.1Å) after 17 ns in the THF/H₂O system. In contrast, the RMSD values in the ChCl–Lac/H₂O, Gly–K₂CO₃/H₂O, and DMSO/H₂O systems increased gradually with fluctuations, stabilizing after 135 ns, with RMSDs of ∼9.1, 12.7, and 12.9Å, respectively. This indicated that the THF/H₂O system was the fastest to reach equilibrium among the other solvent/H₂O systems. However, appreciable fluctuations were observed (from 11.5 to 17.6Å) in the 85–120 ns range, indicating significant conformational changes in the lignin clusters in the aqueous THF environment. Large changes were also observed in the lignin configurations in the aqueous Gly–K₂CO₃ (∼15Å) and DMSO (∼15.5Å) systems. For the ChCl–Lac/H₂O system, the RMSD values remained lower than 10Å throughout the 200 ns trajectory, indicating that the configurational structures of the lignin cluster exhibited overall stability. In contrast, THF, DMSO, and the Gly–K₂CO₃ DES significantly influenced the structural stacking of the lignin molecules in the presence of water (>90 volume%), even though these three systems initially equilibrated before 50ns.

The mean smallest distances between the aromatic units of lignin (Figure 5k–n) were calculated to reflect the specific π–π interactions within the lignin cluster, further characterizing the aggregation tendency of lignin molecules in various solvent/H₂O environments. In ChCl–Lac/H₂O, Gly–K₂CO₃/H₂O, and THF/H₂O environments, most intramolecular aromatic-unit pairs in lignins exhibited distances of less than 0.562 nm, while intermolecular aromatic-unit pairs were spaced at over 1.312 nm (Figure 5k–m). In contrast, in the DMSO/H₂O system (Figure 5n), intramolecular aromatic-unit distances of lignin exceeding 1.312 nm had the highest distribution density. Simultaneously, the intermolecular-unit distances of lignin below 0.375 nm were the most frequent (Figure 5n). This indicated the weakening of the intramolecular π–interactions in the presence of DMSO and H₂O molecules. Consequently, the lignin molecules aggregated into large clusters, limiting the formation of separate particles. In the ChCl–Lac/H₂O (Figure 5k) and Gly–K₂CO₃/H₂O systems (Figure 5l), the average minimum interunit distances within the lignin molecules were predominantly in the 0–0.188 nm, indicating the presence of stronger intramolecular π–interactions. However, the density of the lignin intermolecular-unit pairs with distances of less than 0.188 nm in the ChCl–Lac/H₂O system significantly exceeded that in the Gly–K₂CO₃/H₂O (Figure 5l) system, indicating stronger intermolecular π-interactions and leading to the interconnection of some particles during solvent exchange as shown in Figure 4g. Contrarily, the lignin molecules in the THF/H₂O environment displayed moderate intra/intermolecular π–stacking interactions with intramolecular-unit distances greater than 0.188 nm and intermolecular-unit distances greater than 0.375 nm (Figure 5m), inducing the formation of larger spherical LNPs as depicted in Figure 4h and k. Thus, the lignin molecules formed relatively small, tightly packed LNPs in the acidic and alkaline DES-mediated dialysis driven by high intramolecular π–π-interaction degrees as well as appropriate intermolecular π–π interactions, which confirmed the stable lateral-shifted conformations of the lignin molecules as shown in Figure 5j. The radical-distribution-function plots of the noncovalent aromatic-carbon pairs within the GGE clusters in the different aqueous-solvent systems revealed similar π–interaction changes (Figures S7e–h).

The aromatic ring–ring distances/stacked angles are calculated as a crucial criterion for illustrating the π–stacking interactions (Figures 6 and S8).^[63-67] In the deployed ChCl–Lac/H₂O and Gly–K₂CO₃/H₂O systems, the π–stacking interactions were observed as indicated by the distances and stacked angles between the aromatic rings, which were distributed in relatively narrow regions of 5–7Å and 15°–60° (Figures 6a–j). As a result, these intra/intermolecular π–stacking interactions promoted the formation of a single dominant laterally shifted conformation in DES/H₂O systems. Specifically, throughout the 200 ns simulation, a high-density region was clustered around 6.2Å and 31° in ChCl–Lac/H₂O system (Figures 6a–e), while in the Gly–K₂CO₃/H₂O system, the dominant conformation was observed around 6Å and 42°(Figures 6f–j). However, in the THF/H₂O and DMSO/H₂O systems, more spread-out density distributions were observed within a wide range of 4–7Å and 30°–90°, suggesting that the π-interactions were relatively flexible and multiple stacked configurations emerged (Figures 6k–t). In the 200 ns simulation of the THF/H₂O system (Figures 6k–o), the major stacked mode of the aromatic rings with a distance centered on 6Å and a wide range of stacked angles (30°–65°), indicating the diverse conformational distributions with lateral shifting and stretching. Additionally, the aromatic rings of lignin were also densely distributed around 4.3Å and 30° from 100 ns onwards (Figures 6n and o), indicating that the lignin cluster maintained its offset-stacked configurations with strong π-stacking interactions in the equilibrium THF/H₂O system.^[68] In the DMSO/H₂O system, the aromatic interunit stacked angle was concentrated at 30°, and the stacked distances were relatively broad, with two primary distribution centers at 4.5 and 5.8Å, exhibiting offset-stacked and lateral-shifted configurations (Figures 6p–t). Remarkably, within the 100 ns simulation (Figures 6q–s), the stacked aromatic rings were mainly distributed around 6.4Å and 30°, showing stretched conformations, consistent with the results of the GGE model compound simulation (Figure 3d). These results indicate that strong π–stacking interactions in offset-stacked conformations promoted lignin-molecule aggregation into large clusters during THF/H₂O and DMSO/H₂O exchange (Figure 6n, o, s, and t). In contrast, the absence of offset-stacked configurations of lignin molecules led to the formation of smaller LNPs during DES/H₂O exchange, as lateral-shifted conformations dominated (Figure 6d, e, i, and j). Due to the prevalence of stretched configurations, lignin molecules preferentially assembled into plate-like aggregates during DMSO/H₂O exchange (Figures 6q–s).

The molecular-density distributions revealed that in acidic (Figure 7a and e) and alkaline (Figure 7b and f) DES/H₂O, as well as DMSO/H₂O (Figure 7d and h) simulated systems, the trajectories of the solvent molecules were not significantly affected by the lignin molecules, and all the solvent molecules exhibited high miscibility with water. Conversely, the molecular-motion trajectory of THF highly corresponded to that of lignin in the THF/H₂O environment (Figure 7c and g). THF molecules mainly accumulate around the lignin cluster, particularly at the trajectories marked by red arrows, at which points lignin–THF complexes form.^[69] The results indicated that acidic ChCl–Lac, alkaline Gly–K₂CO₃, and DMSO were easily removed along with the H₂O molecules during solvent exchange, whereas the THF molecules were relatively difficult to remove due to their stronger interaction with lignin.

Further analysis of the SASA plots over 200 ns of MD simulation revealed the dynamic behavior and aggregation tendencies of lignins in different aqueous solvent systems (Figures 7i–l). Overall, the total SASA values (blue line) varied across the four mixed solvent systems. THF/H₂O (Figure 7k) and DMSO/H₂O (Figure 7l) exhibited higher total SASA values (∼61.89 and ∼58.04 nm², respectively) than DES/H₂O systems (∼53.12 nm² in ChCl–Lac/H₂O (Figure 7i) and ∼55.70 nm² in Gly–K₂CO₃/H₂O (Figure 7j)). This suggests that DES/H₂O exhibited weaker solvation effects, promoting the formation of smaller LNP.

In all systems, the hydrophilic SASA values (red line) were around 20.0 nm² (Figure 7i–l), suggesting weak hydrophilic interactions between lignin and solvent molecules in the presence of water (90% volume). In contrast, the hydrophobic SASA values (yellow line) observed in the THF/H₂O (∼40.21 nm²) and DMSO/H₂O (∼36.8 nm²) systems were larger than those in the Gly–K₂CO₃/H₂O (∼33.3 nm²) and ChCl–Lac/H₂O (∼34.9 nm²) systems, indicating stronger hydrophobic interactions in aqueous THF and DMSO environments, which facilitated larger cluster formation.

Dynamic fluctuations in total SASA values further supported these observed trends. The ChCl–Lac/H₂O system (Figure 7i) showed the slightest fluctuations (between 46.6 and 62.4 nm²), indicating stable solvation conformations of lignin over 200 ns simulation. In contrast, the total SASA values in the THF/H₂O system (Figure 7k) showed the most significant fluctuations (between 50.8 and 78.5 nm²), indicating frequent conformational changes. Gly–K₂CO₃/H₂O (46.4–70.4 nm²) and DMSO/H₂O (47.3–69.9 nm²) showed intermediate variations (Figure 7j and l).

These results suggest that the addition of water rapidly weakens the interactions between DES molecules and both the hydrophilic and hydrophobic regions of lignin. Consequently, this weakening of interactions inhibits the extensive aggregation of lignin molecules, resulting in the formation of smaller LNPs during solvent exchange. In contrast, stronger hydrophobic interactions in THF/H₂O and DMSO/H₂O promote larger clusters.