Abstract

The valorization of pyrolysis bio-oil into fuels and renewable chemicals relies on separation steps that recover its most valuable constituents, and liquid-liquid extraction with biobased solvents offers a sustainable route to isolate lignin-derived methoxyphenols. This study combines in silico solvent screening, experimental liquid-liquid equilibrium measurements, and thermodynamic modeling to recover guaiacol and 2,6-dimethoxyphenol from a seven-component synthetic bio-oil. Quantum-chemical predictions with COSMO-RS narrowed more than 1700 candidates to 156 solvents immiscible with water at 313.15 K, which a five-step sustainability-and-availability cascade and a multicriteria decision-analysis ranking reduced to a shortlist of 40. Dimethyl carbonate, cyclopentyl methyl ether, and 2-methyltetrahydrofuran were retained for experimental validation, with 2-butanol as a thermodynamic negative reference. Dimethyl carbonate matched or exceeded the conventional extractants n-hexane and toluene in predicted capacity while keeping a favorable safety, health, and environmental profile. Ternary phase diagrams confirmed wide miscibility gaps in both synthetic and real bio-oil, the latter showing pyrolytic-lignin precipitation that constrains the operating window. Across 20 ternary systems, the PC-SAFT equation of state was the most accurate (RMSD of 0.0310), ahead of the purely predictive COSMO-RS (0.0478) and the Modified UNIFAC 2.0 model (0.0749). In multicomponent extractions, guaiacol reached a distribution coefficient as high as 50.73 while d-fructose stayed in the water-rich phase, and the fitted PC-SAFT parameters transferred to the seven-component mixture (RMSD of 0.0267), supporting a combined screening-and-modeling strategy for sustainable bio-oil fractionation.