Activity-Coefficient-Guided Heuristics for Deep Eutectic Solvent Fractionation of Kraft Lignin Authors: Gloria Agyapong, Regan Ceaser, C. Luke Williams, M.Clayton Wheeler, Sampath Gunukula Published: 2026 ABSTRACT The valorization of technical lignin requires fractionation strategies that deliver chemically and thermally distinct lignin streams with predictable structure–property relationships while minimizing hazardous solvent use. Here, we propose a preliminary, exploratory thermodynamic design framework for lignin fractionation based on deep eutectic solvents (DESs), using activity coefficients as quantitative descriptors of solvent–solute interactions. Six chloride-based DESs were synthesized and characterized through freezing-point depression, FTIR spectroscopy, and Schröder–Le Chatelier analysis. All eutectic compositions exhibited negative deviations from ideality (γ < 1), consistent with strong hydrogen-bond interactions. Importantly, lignin solubility and solvent stability were highest among the systems studied within an apparent, exploratory activity-coefficient window (0.2 < γ < 1), supporting rational DES pre-screening that should be validated across a broader range of DES chemistries without extensive empirical screening. Guided by these heuristics, two optimized, sulfur-free DESs (ChCl/levulinic acid and tetrabutylammonium chloride/ethylene glycol) were applied in sequential fractionation of industrial softwood kraft lignin, yielding four distinct fractions. Structural and elemental analyses suggested reduced apparent molecular weight, lower methoxyl content, and increased hydrogen content, while thermal analysis showed substantially decreased glass transition temperatures (<100 °C). These results suggest that activity-coefficient-guided DES selection offers a promising strategy for upgrading heterogeneous kraft lignin into thermally compatible fractions, representing an early step toward predictive solvent engineering for lignin valorization. Introduction The transition to a circular bioeconomy demands the effective valorization of all biomass fractions, particularly lignin, the largest renewable source of aromatic carbon. Industrial pulping processes, particularly the kraft process, generate 100 million tons of technical lignin annually. Despite this abundance, kraft lignin remains underutilized due to its high structural heterogeneity, broad molecular-weight distribution, and variable functional-group chemistry, which limit its direct incorporation into value-added materials such as thermoplastics, composites, and carbon-based products. Overcoming these limitations necessitates fractionation strategies capable of producing more compositionally uniform lignin fractions with clearer structure–property relationships, while adhering to principles of green chemistry and engineering. Solvent-based fractionation has emerged as an effective route to reduce lignin heterogeneity by selectively dissolving and precipitating lignin populations according to molecular weight and functionality. The most common approaches are sequential dissolution/extraction, controlled antisolvent precipitation, and solvent dissolution coupled with membrane separation. While the first two strategies successfully separate lignin into defined molecular weight fractions, they are often limited by empirical solvent selection. This reliance on trial-and-error lacks established principles linking solvent properties, lignin structure, and fractionation outcomes, thereby impeding reproducibility, scalability, and industrial adoption. For example, antisolvent precipitation demands large solvent volumes and arbitrary solvent choices, resulting in high energy and separation costs. Similarly, sequential organic solvent fractionation is often hindered by operational complexity and the use of hazardous solvents. Furthermore, studies using solvents like γ-valerolactone coupled with ultrafiltration have revealed that solvent polarity strongly influences lignin morphology and molecular behavior. , Unexpected molecular weight distributions observed in these systems, corroborated by molecular dynamics simulations, suggest that solvent-induced conformational changes complicate separation. These limitations motivate predictive fractionation platforms guided by solvent–lignin interaction principles. In response, deep eutectic solvents (DESs) have gained attention as tunable, low-volatility, and potentially recyclable alternatives for biomass and lignin processing. Formed through strong hydrogen-bond interactions between a hydrogen-bond acceptor (HBA) and a hydrogen-bond donor (HBD), DESs offer adjustable polarity and solvation environments well-suited for lignin dissolution and fractionation. While numerous studies have demonstrated the efficacy of DESs for lignin extraction, depolymerization, and fractionation, , DES selection remains largely empirical, and transferable design heuristics linking solvent thermodynamics to lignin solubility and fractionation performance are still lacking. − Wang et al. characterized lignin dissolution thermodynamics in DESs using enthalpy, entropy, and Gibbs free energy of dissolution derived from Flory–Huggins polymer solution theory, providing system-specific energetic insight into individual DES–lignin pairs. However, their approach does not provide an experimentally simple pre-screening criterion before solubility testing. Zhong et al. applied explainable machine learning to predict DES lignin removal efficiency from herbaceous biomass, offering a statistically powerful correlative tool, but one that requires a large training dataset and yields a predictive output without direct physical interpretability at the level of intermolecular interaction strength. The present work differs in both purpose and method: it uses a single, low-cost, experimentally accessible measurement (freezing-point depression) to estimate a physically interpretable descriptor of DES non-ideality, evaluated prospectively, prior to fractionation, to guide solvent selection and sequencing. This positions the activity-coefficient approach as a complementary, mechanistically grounded screening toolnarrower in current validated scope than either of the above studies but distinguished by its low experimental cost and direct physical interpretability, both of which are advantageous for rapid, low-resource DES pre-screening prior to deployment of more resource-intensive thermodynamic or machine-learning analyses. , A central challenge lies in quantifying the intermolecular interactions governing DES formation and lignin solvation. Activity coefficients, which capture deviations from ideal solution behavior, provide a thermodynamically rigorous descriptor of hydrogen-bond-driven non-ideality in eutectic systems. The Schröder–Le Chatelier relationship has been widely applied to evaluate melting-point depression and interaction strength in DESs and eutectic mixtures, particularly in systems exhibiting strong associative interactions. , However, despite their relevance, activity coefficients have rarely been used as predictive metrics for lignin solubility or solvent stability in DES-based fractionation processes, representing a critical knowledge gap. In this study, we propose an exploratory thermodynamic framework for DES-guided lignin fractionation, using activity coefficients derived from freezing-point depression to define solvent–solute interaction regimes relevant to lignin solubility and stability. Six chloride-based DESs were synthesized and systematically characterized to extract preliminary heuristics linking eutectic non-ideality to lignin dissolution behavior. Guided by these preliminary design heuristics, two sulfur-free DESs, choline chloride/levulinic acid and tetrabutylammonium chloride/ethylene glycol, were applied in sequential fractionation of industrial softwood kraft lignin. The resulting fractions were evaluated through structural, elemental, and thermal analyses to establish direct correlations between solvent thermodynamics, lignin structure, and material properties. By advancing DES-based lignin fractionation from empirical solvent screening toward activity-coefficient-guided solvent engineering, this work provides an initial set of transferable design heuristics that, while preliminary, are intended to support scalable, atom-efficient lignin upgrading and strengthen the role of DESs in renewable aromatic biorefineries. Materials and Methods Materials Indulin AT kraft lignin was purchased from Ingevity and oven-dried at 100 °C for 8 h prior to use. Choline chloride (ChCl) (98%) and tetrabutylammonium chloride (TBACl) (95%) were sourced from Sigma-Aldrich and dried at 100 °C before use. All other chemical reagents, including pyrocatechol (Py), vanillin (Van), p-coumaric acid (pCA), hexanol (Hex), ethylene glycol (EG), levulinic acid (Lev) and acetic acid (Ac) were obtained at ≥98% purity from Sigma-Aldrich, Acros Organics, or Fisher Chemicals and used as received. Preparation of DESs Six binary DES systems were synthesized using a standard heating method (Table ). Individual components were weighed into capped vials using a high-precision analytical balance with a readability of ±10–5 g. The mixtures were heated with constant stirring at atmospheric pressure at temperatures just below the boiling point of the components for 1 h until a clear, homogeneous liquid formed. The resulting DESs (Table ) were cooled and stored at room temperature for subsequent characterization and fractionation experiments. Detailed hydrogen bond donor (HBD) and hydrogen bond acceptor (HBA) combinations, along with their molecular structures, are provided in Table S1 and Figure S1. 1Hydrogen-Bond Acceptor (HBA) and Hydrogen-Bond Donor (HBD) Combinations Used to Synthesize DESsComponents HBA (A)HBD (B)DES AbbreviationsCholine ChlorideEthylene GlycolChCl/EGCholine ChlorideLevulinic acidChCl/LevCholine ChlorideAcetic acidChCl/AcCholine ChloridePyrocatecholChCl/PyTetrabutylammonium chlorideEthylene GlycolTBACl/EGTetrabutylammonium chlorideHexanolTBACl/Hex Hydrogen-Bond Acceptor (HBA) and Hydrogen-Bond Donor (HBD) Combinations Used to Synthesize DESs Characterization of Deep Eutectic Solvents Determination of Freezing Point Depression The freezing points of all binary mixtures at different molar compositions were measured using an ethanol bath and dry ice. The vial containing the sample was partially submerged in the ethanol bath, with a K-type thermocouple thermometer (S220-T8, Huato) inserted for temperature measurement. Dry ice was gradually added to the ethanol bath to lower the temperature until the first crystal or solid formed in the sample. The temperature at which the first visible crystal appeared was recorded as the freezing point. The binary mixtures with the lowest freezing points were identified as the eutectic compositions for further evaluation. Fourier Transform Infrared Spectroscopy (FTIR) FTIR was employed to confirm the presence of hydrogen bonding interactions between the two components in the binary mixtures at the deep eutectic point. The analysis was conducted using a Nicolet i20 FTIR-ATR with a SMART iTX diamond ATR (Thermo Scientific), utilizing 128 scans per sample over a range of 4000 to 600 cm–1. FTIR analysis was performed on both the pure components and the corresponding binary mixtures by placing a small amount of each sample on the diamond plate of the ATR. Determination of Activity Coefficient Activity coefficients of the DES components were estimated using the Schröder–Le Chatelier (S–LC) equation at the eutectic composition (eq ). Eutectic temperatures were experimentally determined from freezing-point depression measurements, while melting temperatures and enthalpies of fusion for the pure components were obtained from literature or differential scanning calorimetry (DSC). The S–LC equation relates melting point depression to deviations from ideal solution behavior and has been widely employed to quantify non-ideality in eutectic and DES systems. , In this study, the calculated activity coefficients were used exclusively as a comparative thermodynamic metric to assess the strength of intermolecular interactions between hydrogen bond donors and acceptors. While the assumptions underlying the S–LC equation such as ideal solid phases and negligible solid–solid interactions are not strictly valid for strongly associated DES systems, the approach remains appropriate for evaluating relative non-ideality at the eutectic point. The resulting activity coefficients were used to establish predictive heuristics linking DES non-ideality to lignin solubility and solvent stability, rather than to describe absolute phase equilibria. Because ΔCp corrections were not included, the calculated values are treated as apparent comparative activity coefficients. Importantly, no extrapolation beyond the eutectic composition or temperature range was performed. 1 γi=exp[ΔmhiR(1Tmi−1T)]xi where, γi represents the activity coefficient of component i in the mixture, Δmhi and Tmi, are enthalpy of fusion and temperature of pure component i respectively, and T is the Freezing temperature of the mixture. Because the S–LC treatment excludes ΔCp, the absolute magnitude of the reported activity coefficients should not be interpreted as a rigorous phase-equilibrium quantity. We did not perform new calorimetric measurements of component ΔCp values in this study; however, the qualitative ranking of DES non-ideality used to motivate solvent selection (TBACl/EG as the more strongly non-ideal system; ChCl/Lev as moderately non-ideal) is independently corroborated by FTIR spectroscopy, measured lignin solubility, and the resulting fractionation and thermal behavior described below, all of which are mutually consistent with this ranking without depending on the absolute numerical precision of the apparent activity coefficients. A ΔCp sensitivity analysis using literature-based estimates is identified as a valuable direction for future work. Stability Test To assess the stability of the DESs during lignin fractionation, we conducted experiments using selected DES samples with lignin-derived compounds (vanillin, p-coumaric acid, and pyrocatechol) as well as kraft lignin. The objective was to examine how the addition of a third component, acting as a hydrogen bond donor, affects the DES structure. Lignin-derived compounds were added at 10–50 wt % and stirred for 1 h before FTIR analysis. The functional group changes of the resulting mixtures were analyzed using a Nicolet i20 FTIR-ATR instrument. The stability of the DESs was also tested with acetone, a potential antisolvent, to study its interaction with the DES. Lignin Solubility in DES The solubility of lignin in DES was measured by saturating the DES with lignin at 30 °C and atmospheric pressure. The saturation point of the different DES was determined by adding kraft lignin and stirring continuously at 30 °C until the DES reached saturation. Initially, lignin equivalent to 1 wt % of the DES mass was added and stirred until no visible solids remained. Additional lignin was then added in 1 wt % increments and stirred until visible solids persisted. , The last lignin loading at which no visible solids remained was recorded as the operational solubility limit. The values reported here represent operational saturation limits based on visual persistence of solids, consistent with established protocols in the DES–lignin solubility literature, rather than thermodynamic equilibrium solubilities. , We did not independently confirm saturation by turbidimetry, dynamic light scattering, or microscopy, and we cannot exclude the possibility that colloidal lignin dispersions persisted beyond the visually defined saturation point in some systems. Solubility values throughout this manuscript should accordingly be interpreted as practical, process-relevant saturation limits rather than equilibrium quantities. Sequential Lignin Fractionation Sequential lignin fractionation was performed using either a single DES or multiple DESs (Figure ). Kraft lignin (1 g) was contacted with DES (10 g) at 30 °C for 15 min, followed by vacuum filtration to separate soluble and insoluble fractions. Repeated extraction steps yielded multiple fractions (F1, F2 and F3). In multi-DES fractionation, insoluble residues from the first DES (TBACl/EG) were reprocessed using a second DES (ChCl/Lev). Soluble lignin fractions were precipitated using deionized water or acetone, centrifuged, filtered, and air-dried prior to analysis. Schematic representation of kraft lignin fractionation by deep eutectic solvent and with water or acetone as antisolvent. Characterization of Lignin Fractions Thermal Analysis The glass transition temperature (T g) of the lignin fractions was measured by differential scanning calorimetry (DSC). DSC was performed using a TA Instruments DSC 2500 under nitrogen flow (20 mL min–1). Samples (5–15 mg) were heated from 25 to 200 °C at 10 °C min–1. TGA was performed using a TA Instruments Q500. Approximately 15 mg of sample was loaded onto a platinum pan and heated from room temperature to 900 °C at 10 °C min–1. 1H-NMR of Lignin Fractions 1H-NMR spectra were recorded using a Bruker AVANCE 400 spectrometer with 128 scans. Approximately 50 mg of sample was dissolved in 0.6 mL of DMSO-d 6 in a 5 mm NMR tube. The sample was allowed to dissolve in the solvent before running the analysis. Elemental Analysis of Lignin Fractions Elemental compositions were measured using a Flash 2000 CHNS/O elemental analyzer. For each fraction, C, H, and N contents were measured in triplicate, and oxygen was calculated by difference. Replicate measurements were limited to elemental analysis in this study; thermal, spectroscopic, and solubility measurements were performed as single determinations per sample/fraction, consistent with the exploratory, preliminary scope of this work. Expanded replicate statistics across all characterization techniques are identified as a priority for future validation studies. Results and Discussion Formation and Screening of Deep Eutectic Solvents Deep eutectic solvents were synthesized using a standard heating-and-mixing protocol to evaluate the propensity of selected HBA and HBD combinations to form stable eutectic liquids (Table ). While several combinations produced homogeneous liquids at elevated temperature, only a subset remained stable at ambient conditions (Table S1). These results indicate that DES formation is highly sensitive to intermolecular compatibility, including hydrogen-bonding capacity, component molar ratio, and mutual miscibility. For example, choline chloride (ChCl) formed eutectic liquids with vanillin at multiple molar ratios; however, these systems rapidly solidified upon cooling to ambient temperature, indicating that hydrogen-bond interactions sufficient to induce melting-point depression at elevated temperatures were not maintained at room temperature (Table S1). Similar temperature-dependent behavior has been reported for weakly associated DES systems and highlights the need to distinguish mixtures that form homogeneous liquids only at elevated temperature from DESs that remain stable liquids under ambient operating conditions. In contrast, combinations such as ChCl/pCA did not form DES under the same conditions, consistent with prior observations that steric rigidity and limited miscibility can suppress eutectic behavior despite the presence of hydrogen-bonding functionalities. , Likewise, the absence of DES formation in ChCl/Hex mixtures reflects insufficient polarity matching and poor ion–dipole stabilization. These results suggests that DES formation depends on cooperative hydrogen bonding, polarity compatibility, and compositional balance, rather than HBD acidity alone. , Importantly, the observed variability reinforces the limited availability of experimentally grounded DES-selection heuristics and motivates the analysis developed below. Spectroscopic Evidence of Hydrogen-Bond Network Formation Fourier transform infrared spectroscopy was employed to confirm the formation of DESs and probe intermolecular interactions at the eutectic composition (Figure ). Compared to the spectra of the pure HBD components, all stable DESs exhibited pronounced broadening of the O–H stretching band in the 3500–3000 cm–1 region. This broadening reflects the formation of extensive hydrogen-bond networks and reduced O–H bond strength due to cooperative association, a defining characteristic of DES systems. FTIR spectra of (a) pure HBA/HBD components and (b) corresponding DESs. In addition, the characteristic C–N stretching vibration of choline chloride at 955 cm–1 was preserved in the DES spectra, indicating physical association rather than chemical transformation. , Carbonyl bands near 1702 cm–1 were retained in ChCl/Lev and ChCl/Ac. In TBACl/Hex, the carbonyl band was absent, suggesting strong acid–salt or ion-pair interactions, potentially involving partial carboxylate character. This spectral change indicates that strong DES component interactions can alter the local bonding environment of the carboxylic acid HBD. This is supported by the activity coefficient studies discussed subsequently. Collectively, the FTIR results corroborate the freezing-point depression data and support the interpretation that strong, yet reversible, hydrogen-bond interactions underpin the observed eutectic behavior, consistent with prior DES–lignin studies emphasizing hydrogen-bond-driven solvation mechanisms. Freezing-Point Depression and Identification of Eutectic Compositions The solid–liquid equilibrium behavior of each binary mixture was quantified through systematic freezing-point measurements across the full composition range. All stable DES systems exhibited substantial melting-point depression relative to their pure components, with well-defined eutectic compositions characterized by minimum freezing temperatures (Table ). This behavior indicates that intermolecular interactions in the liquid phase are energetically favored relative to self-association in the pure components, a prerequisite for DES formation. 2Eutectic Compositions (EC), Enthalpies of Fusion (ΔHm), Melting Points (MP), Measured Freezing Points, and Calculated Activity-Coefficient Trends for the Studied DES Systems ΔHm (kJ/mol)MP of component (°C) Activity coefficientDESEC of HBD:HBAABABMP of mixture (°C)γA γB ChCl/EG1:44.311.2324–12.9–64<1<1ChCl/Lev1:2 9.2232433–52<1<1ChCl/Ac1:2 11.7232416.6–51<1<1ChCl/Py1:1 22.87324100–103–8.2<1<1TBACl/EGxA ≤ 0.214.6911.270–12.9<−72<1<1TBACl/HexxA ≤ 0.2 15.4870–47<−61<1<1 Eutectic Compositions (EC), Enthalpies of Fusion (ΔHm), Melting Points (MP), Measured Freezing Points, and Calculated Activity-Coefficient Trends for the Studied DES Systems Phase diagrams constructed from these measurements reveal pronounced negative deviations from ideal mixing behavior near the eutectic point, particularly in systems containing polyols or organic acids as HBDs (Figures S2–S4). Such deviations are characteristic of strongly associating systems and provide a thermodynamic basis for subsequent activity-coefficient analysis. Importantly, the magnitude of freezing-point depression varied significantly across systems, suggesting differences in hydrogen-bond strength and network cooperativity that cannot be inferred from composition alone. These findings establish freezing-point depression as an important experimental input for quantifying non-ideality and motivate the use of activity coefficients as a comparative thermodynamic descriptor for evaluating DES suitability in lignin fractionation. Heuristic I: Eutectic Point Activity Coefficient as a Descriptor of DES Non-Ideality The defining thermodynamic feature of deep eutectic solvents is their pronounced deviation from ideal mixing behavior, manifested experimentally as significant melting-point depression at the eutectic composition. To quantify this non-ideality, activity coefficients of individual DES components were estimated at the eutectic point using the Schröder–Le Chatelier relationship. Across all stable DESs examined, both HBAs and HBDs exhibited activity coefficients less than unity (γ < 1) at the eutectic composition, indicating favorable intermolecular interactions relative to the pure liquid phases (Table ). 3Apparent Activity Coefficients and Lignin Solubility of DES Mixtures Activity coefficient DESγA γB Lignin solubility at 31 °CChCl/EG0.9920.3430.504 g/gChCl/Ac0.6990.2900.563 g/gChCl/Lev0.6960.3710.344 g/gChCl/Py0.6760.0910.002 g/gTBACl/EG0.2520.2700.396 g/gTBACl/Hex0.1880.4660.005 g/g Apparent Activity Coefficients and Lignin Solubility of DES Mixtures Solid–liquid equilibrium diagrams and corresponding activity-coefficient profiles consistently showed negative deviations from ideality localized near the eutectic composition, confirming that eutectic formation arises from cooperative hydrogen bonding and ion–dipole interactions. In ChCl-containing DES, the HBD exhibited significantly lower activity coefficient compared to the HBA (0.091–0.373 vs. 0.676–0.992), while the opposite was found to occur in the TBACl-containing DES (0.270–0.466 vs. 0.188–0.252). This trend could be due to the cations formed from TBACl being larger and more hydrophobic than that of ChCl affecting the charge density and the interaction of the chloride ions with the HBD. In addition, earlier research has shown that quaternary ammonium salts with longer alkyl chains can cause stronger HBD-acceptor associations in specific binary mixtures. Importantly, negative deviation from ideality at the eutectic point is a necessary thermodynamic condition for DES formation, in agreement with prior analyses of eutectic and DES systems. This observation constitutes an initial, preliminary Heuristic I: DES formation is consistently associated with negative deviation from ideality at the eutectic point. While this criterion does not by itself predict lignin solubility, it establishes a necessary thermodynamic foundation for solvent stability and associative behavior. This is consistent with recent thermodynamic analyses of DES systems and supports the use of S–LC–derived activity coefficients as a comparative metric for interaction strength rather than an absolute descriptor of phase equilibria. , Heuristic II: Correlating Activity Coefficient to Lignin Solubility While negative deviation from ideality is required for DES formation, the magnitude of this deviation was found to play a critical role in determining lignin solubility. Comparison of activity coefficients at the eutectic point with experimentally measured lignin solubility revealed a nonmonotonic relationship. DESs exhibiting moderate negative deviations (0.2 < γ < 1) for both components displayed the highest lignin solubility (0.34–0.56 g g–1), whereas systems in which either the HBA or HBD exhibited very strong non-ideality (γ ≤ 0.2) showed minimal lignin dissolution (Table ). The strongest HBD interaction (γ = 0.091) was observed in ChCl/Py whereas the strongest HBA interaction (γ = 0.188) was observed with TBACl/Hex. However, the presence of these extremely strong interactions may have limited effective lignin–solvent interactions thereby resulting in the observed minimal solubility. This behavior suggests that excessively strong hydrogen-bond interactions between DES components can reduce lignin solubility by effectively forming a strongly self-associated solvent network, thereby limiting the availability of hydrogen-bonding sites for lignin–solvent interactions. Such systems are also typically associated with higher viscosity, further restricting mass transport and solvation efficiency. In contrast, DESs within an intermediate non-ideality window provide sufficient interaction strength to disrupt lignin–lignin associations while maintaining solvent flexibility. This interpretation is consistent with reports showing that at activity coefficients below or equal to 0.2, the predictive solubility of lignin in DES is negatively impacted. ,, We note that the present study does not experimentally decouple thermodynamic non-ideality from viscosity- or mass-transfer-related effects, both of which could plausibly contribute to the low solubility observed at γ ≤ 0.2. Direct rheological characterization of the DES systems, and solubility measurements at controlled, matched viscosity, would be required to isolate these contributions and are identified as important directions for future mechanistic work. Accordingly, Heuristic II emerging from this study is that higher operational lignin solubility was observed within a preliminary, exploratory activity-coefficient range, rather than at maximal non-ideality. Given that this trend is derived from only six DES systems, two of which were carried forward into fractionation, the (0.2 < γ < 1) window should be regarded as an exploratory range motivating further investigation rather than a validated general design rule. Because this trend is based on a limited solvent set, broader validation across additional HBA, HBD, and anion chemistries is needed before the window can be considered generalizable. Heuristic III: Structural Stability of DESs in the Presence of Lignin and Model Compounds Solvent stability under high solute loadings is a prerequisite for robust lignin fractionation processes. , In this context, DESs must maintain their hydrogen-bond network and eutectic integrity in the presence of lignin and lignin-derived aromatics to ensure process robustness. To evaluate this requirement, selected DESs were systematically challenged with increasing concentrations of lignin model compounds and kraft lignin, and their structural responses were assessed by FTIR spectroscopy. At low to moderate loadings (≤20 wt %), the addition of pyrocatechol (Py) to ChCl/EG resulted in an increase of the broad O–H stretching band (3500–3000 cm–1) (Figure a), consistent with altered hydrogen-bonding interactions. This behavior suggests possible incorporation of Py into a ternary hydrogen-bonded network in which Py acts as an auxiliary hydrogen-bond donor, integrating into the DES matrix rather than disrupting it. Increasing the Py content to 30 wt % initially attenuated the O–H stretching intensity before partially recovering at higher loadings, suggesting competitive hydrogen-bonding between ethylene glycol and Py for chloride coordination. This transition reflects a shift from cooperative solvation to donor competition at elevated solute concentrations. FTIR spectra of (a) DES with different weight percentages of pyrocatechol (Py) (b) Addition of 20 wt % of Py, vanillin (VAN) and p-Coumaric acid (pCA) to DES (c) Addition of 20, 30, and 50 wt % of Kraft lignin to DES. Similar stabilization trends were observed upon the addition of vanillin (VAN) and p-coumaric acid (pCA) at 20 wt % (Figure b). In addition to strengthening the O–H stretching region, these compounds introduced characteristic carbonyl absorptions (1700 cm–1) and aromatic ring vibrations (1605 and 1508 cm–1), as well as C–O stretching associated with guaiacyl units (1286 cm–1). The preservation of the DES spectral envelope alongside these new features confirms that the DES framework remains intact while accommodating secondary hydrogen-bond donors, consistent with prior reports on polyol-based DES adaptability. When challenged with kraft lignin, the DES showed limited spectral changes upon kraft lignin addition. At 20 wt % lignin loading, no significant perturbation of the DES hydrogen-bond network was observed (Figure c), aside from the appearance of aromatic CC stretching bands near 1600 cm–1, attributable to partial lignin solubilization. Even at 50 wt % lignin, the FTIR spectra showed minimal deviation from the parent DES, suggesting saturation-limited interactions rather than solvent degradation or phase instability. Collectively, these observations support Heuristic III, which posits that DESs exhibiting moderate non-ideality, identified through activity-coefficient analysis, maintain structural integrity under chemically and industrially relevant lignin loadings. These FTIR results show that the selected DESs retain their main hydrogen-bonding features under lignin loading, supporting their use as stable fractionation media under the conditions studied. This stability is a prerequisite for solvent recovery, although recycling experiments are required to confirm reuse performance. Sequential Fractionation by Single and Multiple DES Building directly on the thermodynamic heuristics established above, ChCl/Lev and TBACl/EG were selected because their activity coefficients fall within the apparent non-ideality range (0.2 < γ < 1.0) associated with higher lignin solubility in the studied systems. DESs operating in this γ-range balance hydrogen-bond donation, chloride coordination, and solvent self-association, enabling controlled solubilization rather than indiscriminate dissolution. TBACl/EG was employed as the primary fractionation solvent owing to its lower apparent activity coefficient (γA = 0.252 and γB = 0.270) relative to ChCl/Lev, coupled with a high lignin solubility (0.396 g g–1). These low activity coefficients reflect a highly cooperative hydrogen-bond network dominated by EG–Cl––TBA+ interactions, which stabilizes the solvent structure even under high lignin loadings (Figure c). ChCl/Lev exhibits moderate non-ideality (γA = 0.696 and γB = 0.371) and a lower lignin solubility (0.344 g g–1), consistent with a less deeply bound solvent network that favors selective extraction of lignin fractions requiring intermediate hydrogen-bond strength. This thermodynamic complementarity supports their sequential deployment. Thermal Analysis of Lignin Fractions DSC was employed to identify the glass transition temperature (T g) and specific heat capacity (Cp), properties that influence the suitability of lignin as a precursor for carbon fibers or thermoplastics. The parent kraft lignin exhibited two distinct T g values at 71 °C and 131 °C. Following fractionation with either ChCl/Lev or TBACl/EG, all soluble fractions (F1–F3) displayed T g values below 100 °C (Figure ). ChCl/Lev-derived fractions showed T g values between 65 and 70 °C, while TBACl/EG fractions ranged from 61 to 73 °C. This systematic T g depression correlates directly with the activity coefficients of the solvents: DESs with lower γ-values preferentially solubilize lower number-average molecular weight (M n) lignin populations, reducing chain entanglement and increasing free volume in accordance with the Flory–Fox relationship. , The absence of a discernible T g in the undissolved residues from both DES platforms indicates enrichment in highly condensed, rigid aromatic structures that resist both solvation and thermal softening, consistent with their exclusion from DESs with γ ≤ 0.2. DSC heating curves of lignin fractions obtained from (a) TBACl/EG with water, (b) TBACl/EG with acetone, (c) ChCl/Lev with water, and (d) sequential TBACl/EG–ChCl/Lev fractionation with water as the antisolvent for determination of T g and Cp. Thermogravimetric analysis (TGA) and derivative thermogravimetry (DTG) further substantiate this activity-coefficient-associated structural sorting (Figure ). Kraft lignin displayed a broad, multistep degradation profile, whereas DES-fractionated lignins exhibited sharper DTG max peaks consistent with improved apparent thermal uniformity. , In the ChCl/Lev system, soluble fractions decomposed primarily between 200 and 400 °C, while the undissolved residue yielded a significantly higher char fraction (45% at 800 °C), suggesting selective retention of condensed aromatic domains (Figure a and b). By contrast, TBACl/EG fractions showed an intense degradation event near 250 °C (Figure c and d), which may arise from lower-molecular-weight lignin fragments and/or partial volatilization of retained alkylammonium species. These features are consistent with the strong solvent–solute interactions implied by γ ≤ 0.2 and corroborated by NMR and elemental signatures. TGA and DTG curves of various lignin fractions from single sequential fractionation using (a, b) ChCl/Lev DES and (c, d) TBACl/EG DES. Antisolvent choice further modulated these thermodynamic outcomes. DI water precipitation produced sharp, well-defined T g transitions, whereas acetone precipitation broadened T g ranges and increased Cp by up to 150% for select fractions, indicating enhanced thermal resilience. This effect suggests that acetone selectively recovers lignin populations with broader configurational freedom, complementing the activity-coefficient-controlled solubilization step. Sequential fractionation using TBACl/EG followed by ChCl/Lev yielded lignin fractions with more clearly differentiated thermal behavior, demonstrating the cumulative impact of staged activity-coefficient control. DSC analysis showed that soluble fractions from the two-step process retained T g values well below 100 °C (73 °C for TBACl/EG-derived fractions and 58 °C for ChCl/Lev-derived fractions), enhancing compatibility with thermoplastic matrices. Notably, the ChCl/Lev-soluble fraction exhibited a substantially higher Cp at T g (2.2 J g–1 °C–1) than the TBACl/EG-soluble fraction (0.8 J g–1 °C–1), reflecting refined partitioning of lignin populations according to hydrogen-bond density and chain flexibility. These results support the use of activity-coefficient heuristics to guide solvent sequencing and thermal-property differentiation during sequential DES fractionation. By pairing a deeply non-ideal DES (TBACl/EG) for initial disruption with a moderately non-ideal DES (ChCl/Lev) for secondary refinement, the process advances beyond empirical solvent selection toward a thermodynamically guided, design-led lignin fractionation strategy suitable for scalable biorefinery integration. Structural Evolution of Lignin Fractions The molecular-scale structural evolution of lignin fractions obtained via single-DES and sequential multi-DES fractionation was elucidated using 1H-NMR spectroscopy (Figure ), providing molecular-level support for the thermal and degradation trends observed. Together, these analyses demonstrate how DES thermodynamic non-ideality guides lignin–DES interactions, molecular sorting, and downstream thermal behavior. 1H-NMR spectra of (a) ChCl/Lev lignin fractions (b) TBACl/EG lignin fractions from single sequential fractionation of kraft lignin and (c) TBACl/EG and ChCl/Lev soluble lignin fractions from multiple sequential fractionations of kraft lignin. The parent kraft lignin exhibits spectral features typical of softwood lignin, including guaiacyl (G) aromatic protons at 6.7–7.5 ppm, p-hydroxyphenyl (H) units at 7.4–8.0 ppm, prominent methoxyl resonances at 3.3–4.0 ppm, and broad aliphatic side-chain signals between 1.2 and 1.8 ppm. , This combination of overlapping aromatic and aliphatic signals reflects the high molecular weight, broad polydispersity, and chemical heterogeneity consistent with the wide T g distribution and multistep thermal degradation profile observed for kraft lignin. Across all DES-derived fractions, the preservation of aromatic units, methoxyl groups, and oxygenated aliphatic protons (3.3–4.5 ppm) suggests that the main lignin structural motifs are retained after DES processing. Instead, selective redistribution of structural motifs occurs, consistent with the narrowing of T g transitions and sharpening of DTG max peaks following fractionation. Systematic changes in aromatic-to-aliphatic signal intensity are consistent with structural sorting inferred from thermal analysis. In the ChCl/Lev system (Figure a), soluble fractions F1 and F2 retained strong aromatic proton signals in the 6.5–8.0 ppm region, consistent with their higher carbon content from elemental analysis and their intermediate T g values (65–70 °C). As fractionation progressed toward F3 and the undissolved residue, increased aliphatic proton intensity (1.2–1.6 ppm) and reduced aromatic density were observed, indicating enrichment in hydrogen-rich, condensed lignin domains. This structural shift is consistent with the absence of a detectable T g for the undissolved fraction and its elevated char yield, suggesting that ChCl/Lev appears to preferentially solubilize oxygenated, moderately aromatic lignin while excluding rigid, thermally stable structures. In contrast, TBACl/EG induced markedly different structural outcomes (Figure b), consistent with its deeper thermodynamic sink (γA ≈ γB ≤ 0.2). Soluble fractions displayed enhanced aliphatic resonances at 0.8–2.0 ppm alongside attenuated aromatic signals, indicative of preferential extraction of flexible, alkyl-rich lignin domains. Sharp triplet-like peaks at ∼0.9, 1.3, and 1.5 ppm arise from the butyl chains of the tetrabutylammonium cation, indicating possible retention of tetrabutylammonium species in the recovered fraction. This aliphatic enrichment correlates with the lower T g values and intense DTG peak centered near 250 °C observed for TBACl/EG fractions, consistent with lower apparent molecular weight or increased chain mobility. The strong DES self-association implied by low activity coefficients may limit the disruption of π–π stacking interactions, thereby disfavoring solubilization of highly aromatic lignin domains and contributing to the persistence of condensed residues with no discernible T g. Sequential fractionation using TBACl/EG followed by ChCl/Lev resulted in a stepwise structural differentiation (Figure c) that directly parallels the staged thermal transitions reported. The initial TBACl/EG step removed aliphatic-rich, low-rigidity fractions associated with lower T g and earlier thermal degradation. Subsequent ChCl/Lev treatment selectively extracted fractions enriched in aromatic and methoxyl functionalities, including detectable aromatic methyl signals, yielding lignin with higher Cp values and more predictable T g behavior. The final undissolved residue exhibited broad, low-intensity aromatic signals characteristic of highly condensed lignin, consistent with its high thermal stability and char yield. The 1H-NMR results support the activity-coefficient heuristic proposed here and show that sequential DES fractionation can direct lignin structural differentiation under the conditions studied. DESs exhibiting moderate negative deviations from ideality enable balanced lignin–DES interactions that yield fractions with narrow T g distributions and sharp DTG profiles, whereas excessively low activity coefficients favor DES self-association and selective extraction of flexible, aliphatic lignin domains. The sequential DES fractionation enables thermodynamically guided control over lignin structure and thermal performance, advancing lignin processing from empirical solvent screening toward rational, application-specific design. The structural and thermal trends described above are interpreted as inferred molecular-weight redistribution, based on Tg depression, DTG sharpening, NMR signal redistribution, and elemental H/C shifts, rather than direct molecular-weight measurement. We did not perform GPC/SEC analysis in this study, and we therefore cannot rule out contributions from viscosity, mass-transfer, or colloidal effects to the observed fractionation behavior, in addition to the activity-coefficient-driven solvation mechanism proposed here. Direct molecular-weight characterization of these fractions via GPC/SEC is planned as a follow-up study. Elemental Composition and Implications for Material Applications Elemental analysis reveals systematic shifts in carbon and hydrogen content across all DES-fractionated lignin samples relative to the parent kraft lignin, providing quantitative support for the molecular and thermal trends identified. In general, fractionation led to a decrease in carbon content accompanied by an increase in hydrogen content, consistent with the preferential recovery of hydrogen-rich lignin domains, potentially associated with lower apparent molecular weight or increased aliphatic character. These compositional shifts are consistent with the reduced glass transition temperatures, enhanced thermal flexibility and increased aliphatic proton intensity detected, suggesting that DES-mediated fractionation selectively disrupts condensed aromatic domains while preserving lignin’s core functionality. The efficiency and selectivity of this elemental redistribution are guided by the extent of solvent non-ideality at the eutectic composition. In the ChCl/Lev system, a progressive decrease in carbon content from 62.64% in F1 to 54.26% in the undissolved residue, coupled with a corresponding increase in hydrogen content from 5.61% to 6.82% (Table ), reflects selective solubilization of oxygenated and moderately aliphatic lignin domains. These domains are stabilized in DESs that maintain a balanced hydrogen-bonding strength. Specifically, the activity coefficient for this system (γA = 0.696) suggests HBA/HBD interactions that are sufficiently robust to disrupt the complex intermolecular associations of the lignin polymer, yet not so extreme as to immobilize the internal solvent structure. This balance is further evidenced by nitrogen levels in the soluble fractions (F1–F3), which remain close to the parent lignin (0.39%–0.81%), indicating minimal incorporation of DES components into the lignin matrix. A significant nitrogen spike (2.17%) was only observed in the undissolved residue, suggesting limited DES incorporation into the soluble fractions. 4Elemental Analysis of Kraft Lignin and ChCl/Lev DES Lignin FractionsSample%C%H%NO%H/CO/CKraft lignin63.465.510.5130.521.030.36F162.645.610.3931.581.070.38F262.145.750.5431.571.100.38F359.495.840.8133.861.170.43Undissolved54.266.822.1736.751.500.51 Elemental Analysis of Kraft Lignin and ChCl/Lev DES Lignin Fractions In contrast, the TBACl/EG system highlights the consequences of deeper thermodynamic non-ideality on lignin composition (Table ). With both components exhibiting low activity coefficients (γA = 0.252 and γB = 0.270), this DES generates fractions with substantially higher hydrogen contents, reaching 8.52% in the undissolved residue. Elevated H/C ratios, coupled with pronounced aliphatic resonances in 1H–NMR spectra, reflect the extraction of flexible, alkyl-rich lignin domains as well as partial association with the tetrabutylammonium cation. These elemental features are consistent with the lower T g values and intense DTG peaks near 250 °C observed for TBACl/EG fractions, indicative of reduced molecular weight and increased thermal volatility. However, the concurrent nitrogen enrichment across all fractions (up to 1.72%) suggests greater solvent retention, consistent with the stronger DES self-association inferred from activity-coefficient analysis. 5Elemental Analysis of Kraft Lignin and TBACl/EG DES Lignin FractionsSample%C%H%NO%H/CO/CKraft lignin63.465.510.5130.521.030.36F164.896.871.0527.191.260.31F256.416.620.9636.011.280.38F362.957.331.2628.461.390.34Undissolved59.758.521.7230.011.700.38 Elemental Analysis of Kraft Lignin and TBACl/EG DES Lignin Fractions These results suggests that the elemental profile of fractionated lignin can be systematically directed by selecting DES components according to their activity coefficients. While the ChCl/Lev platform is better suited for producing high-purity, low-nitrogen fractions suitable for advanced materials and bio-oil production, the TBACl/EG platform offers a pathway for isolating highly aliphatic domains that require lower thermal processing temperatures. , The presence of quaternary ammonium groups may be relevant to producing bio-based resins or antimicrobial coatings without requiring secondary nitrogenation steps. These findings highlight the utility of activity-coefficient-based heuristics in designing more rational solvent-selection strategies for lignin fractionation that move beyond empirical trial-and-error. Furthermore, to ensure robustness of these heuristics, future studies should expand the dataset to clarify the activity-coefficient window and its relationship to lignin properties. Conclusions This work proposes a preliminary thermodynamically motivated strategy for selecting DESs for kraft lignin fractionation, directly aligned with the objective of transforming heterogeneous technical lignin into compositionally distinct fractions with application-relevant thermal properties. By using activity coefficients derived from freezing-point depression as a quantitative descriptor of DES non-ideality, we explore a direct link between hydrogen-bond interaction strength, lignin solubility, solvent stability, and fractionation performance. Systematic characterization using freezing-point analysis, FTIR spectroscopy, and Schröder–Le Chatelier modeling shows that negative deviations from ideality (γ < 1) at the eutectic point are both a defining feature of DES formation and a comparative metric for lignin–DES interactions. Based on a dataset of six DES systems, the results suggest an apparent, exploratory window of moderate non-ideality in which lignin dissolution and solvent robustness appear to be maximized; this window should be treated as a hypothesis for further testing rather than an established design rule. Guided by this principle, ChCl/Lev and TBACl/EG were rationally selected and applied in single and sequential fractionation schemes, enabling controlled deconstruction of kraft lignin into chemically distinct fractions. Elemental analysis and 1H–NMR spectroscopy reveal systematic structural evolution toward features consistent with lower apparent molecular-weight, hydrogen-rich domains with reduced aromatic condensation, while DSC and TGA support enhanced thermal flexibility, including glass transition temperatures below 100 °C and increased specific heat capacities relative to the parent lignin. The complementary behavior of TBACl/EG and ChCl/Lev suggests that sequential DES platforms can be selected based on activity-coefficient heuristics. Several limitations should be considered in interpreting these results: the dataset underlying the proposed activity-coefficient window comprises only six DES systems, of which two were carried forward to fractionation. The reported activity coefficients are apparent, comparative values that do not account for ΔCp. Molecular-weight-related claims are inferred from thermal, spectroscopic, and elemental trends rather than direct GPC/SEC measurement. The reported solubility values are operationally defined saturation limits rather than suggested equilibrium quantities and viscosity/mass-transfer effects have not been experimentally decoupled from thermodynamic solvation effects. In summary, this study supports a more preliminary, exploratory approach to DES selection for lignin fractionation and provides a basis for future validation across broader DES families and lignin sources. By linking DES non-ideality with lignin structure and thermal performance, this approach provides a plausible exploratory pathway for producing tailored lignin feedstocks suitable for thermoplastics, carbon fiber precursors, and other advanced materials central to chemical engineering. Supplementary Material