Development of Generalized Correlation for Electrical Conductivity Prediction of Pure Ionic Liquid

• Main Authors: Arguelles, Carlos Gabriel A., Dionio, Carlos John B., Enciso, John Matthew C., Go, Ar Jettherson T., Jardiolin, Paul Andrei M.
• Format: text
• Published: Animo Repository 2021
• Subjects: pure ionic liquids; electrical conductivity; generalized correlation; data trimming; data mining
• Online Access: https://animorepository.dlsu.edu.ph/conf_shsrescon/2021/paper_men/2; https://animorepository.dlsu.edu.ph/context/conf_shsrescon/article/1629/viewcontent/ME_Development_of_Generalized_Correlation_for_Electrical_Conductivity.pdf
• Institution: De La Salle University

Ionic liquids are salts in liquid form that are composed of short-lived ion pairs. They are the new trend of solvent because of their very low vapor pressure, good chemical and thermal stability, and melting temperatures lower than 100°C. Pure ionic liquids contain ions that can conduct electricity or serve as electrolytes. But experimentation using ionic liquids would be expensive. This study aims to develop a generalized correlation for the electrical conductivity prediction of pure ionic liquids. The researchers gathered data of pure ionic liquids that involved the electrical conductivity property from the ThermoIL Database. The collected data were then trimmed based on a developed scheme and classifications. After trimming the data, the researchers evaluated the data using MATLAB software. The residual value was calculated, and a parity plot was constructed to test the models’ accuracy. The researchers gathered 2,425 data points from 310 references and were trimmed to 220 data points from 21 references. The parity plot and graph of the residuals plotted against pressure showed that the experimental and calculated values were close. Results showed that the electrical conductivity of pure ionic liquids could be predicted using a model patterned to Pitzer correlation with reduced temperature and reduced pressure as variables. Data with two or more references and low uncertainty made a good result on the models to create a generalized correlation via curve fitting.