Oil desulfurization using deep eutectic solvents as sustainable and economical extractants via liquid-liquid extraction: Experimental and PC-SAFT predictions

摘要

The reduction of the sulfur content in crude oil is of utmost importance in order to meet the stringent environmental regulations. Thiophene and its derivatives are considered key substances to be separated from the crude oil. In previous works, six deep eutectic solvents (DESs) based on tetraethylammonium chloride, tetrahexylammonium bromide and methyltriphenylphosphonium bromide as hydrogen bond acceptors (HBAs) and polyols (ethylene glycol and glycerol) as hydrogen bond donors (HBDs) were successfully applied for the extraction of thiophene from {n-alkane + thiophene} mixtures via liquid liquid extraction. One of the objectives of this work was to study the effect of the aliphatic hydrocarbon type length (e.g. n-hexane vs n-octane) on the extraction performance of the same DESs. Extraction performance was evaluated by the selectivity and the thiophene distribution coefficient. Based on new experimental data, higher selectivities and lower thiophene distribution coefficients were obtained when thiophene was extracted from n-octane instead of n-hexane. Another objective was to predict the phase behavior of the ternary systems {n-alkane + thiophene + DES) using Perturbed-Chain Statistical Associating Fluid Theory (PC-SAFT). The PC-SAFT "pseudo-pure component" approach was applied, in which a DES was considered as a pseudo-pure compound (not a mixture). The pure-component parameters of the DESs were obtained by fitting to liquid density data, which were measured at temperatures between 298.2 K and 323.2 K. Binary interaction parameters were fitted to experimental binary LLE data for the systems {n-alkane + DES) and {thiophene + DES) at 298.2 K and atmospheric pressure, while the LLE data of the ternary systems {n-alkane + thiophene + DES) were fully predicted. It was found that the distribution coefficients and selectivity of the ternary systems containing DESs could be qualitatively well predicted using this model. (C) 2018 Elsevier B.V. All rights reserved.