چكيده به لاتين
Due to the many problems caused by the presence of nitrogen compounds in fuels, denitrogenation of fuel is important. Considering the advantages of the extractive denitrogenation method using ionic liquids (ILs), this method has been investigated in the present study. In this regard, the distribution of pyridine between the ionic liquid-rich phase and hydrocarbon-rich phase has been predicted using the quantitative structure-property relationship (QSPR) approach. After a comprehensive survey in the literature, a large dataset including 51 ternary systems (i.e., IL (1), pyridine (2), and hydrocarbon (3)) was collected. The present dataset covers 19 cations, 14 anions, and 10 hydrocarbons. Therefore, the structural effect of each cation, anion, or hydrocarbon has been taken into account on the distribution of pyridine between the IL-rich phase and hydrocarbon-rich phase. Multiple linear regression (MLR) and multi-layer perceptron (MLP) have been employed to develop the predictive/descriptive linear and non-linear models, respectively. Results showed that the final linear model (R2 = 0.9165, %AARD = 23.2722) and non-linear model (R2 = 0.9892, %AARD = 9.1936) have an acceptable capability for Y2 prediction. By interpreting the selected descriptors of the QSPR model, the effect of changes in the cation, anion, and hydrocarbon structures on the extraction of pyridine was also studied. It was found that the size, branches, and cyclic or acyclic structure of hydrocarbons were important parameters affecting the pyridine distribution. Also, the size, branches, and position of the electronegative atoms (especially oxygen atoms) in the structure of anions as well as the alkyl side chain length and C-N bonds in the structure of cations could also affect the pyridine distribution.
