The selection of effective demulsifiers for separating water-in-oil (W/O) emulsions remains largely guided by the empirical Hydrophilic-Lipophilic Balance (HLB) system, which is sensitive to temperature, salinity, and crude composition. Here we introduce a physics-based Polarity Index (PI), derived from molecular dipole moments predicted by a Gaussian Process Regression (GPR) model trained on Density Functional Theory (DFT) calculations for 50 anchor compounds. PI values for a broader set of 124 surfactant/demulsifier-relevant molecules are obtained by GPR prediction. The PI scale (0–100) is anchored by n-hexadecane (PI = 0, μ = 0.00 D) and the dataset maximum BMIMCl (PI = 100, μ = 1.95 D), with water at PI = 95 as an intermediate reference. Provisional performance thresholds derived from bottle tests on 20 representative compounds (Pearson r = −0.91) demonstrate the PI's utility for rational demulsifier selection and blend formulation. A supplementary validation study across three additional emulsion systems — rapeseed oil/beeswax–brine/borax (pH 9.2, 4.3% beeswax), Iraqi Basra waste crude (high asphaltene, waste treatment context), and North Sea oil-based mud — reveals a consistent pattern: multi-sulfonated fatty acids and structurally homologous C18 sodium soap blends achieve 45–95% water recovery across systems where PI predicts 80% predicted). A proposed two-parameter refinement — PI combined with a Surface Activity Score (SAS) encoding interfacial anchor groups and structural chain-match — is introduced to address these system-specific deviations.
Mohammad Jerrow (2026) studied this question.