Data Analysis for Electrolyte Systems: A Method Illustrated on Alkali Halides in Water

Abstract

In the recent years, the need to improve the thermodynamic models, particularly making them more precise and predictable for complex systems, has increased [Industrial Requirements for Thermodynamics and Transport PropertiesHendriksE. Hendriks, E. Ind. Eng. Chem. Res.2010491113111141 Industrial Requirements for Thermodynamic and Transport PropertiesKontogeorgisG. Kontogeorgis, G. Ind. Eng. Chem. Res.20216049875013 The complexity of electrolyte systems is due to the strong interactions between ions, the hydration forces that take place during salt dissociation, and the physical forces at high concentrations of solute. Consequently, the existing thermodynamic electrolyte models show some limitations and are far from being completely optimized for industrial simulations. Thermodynamics based on electrolyte mixtures and mixed solvents requires further development and research [PolingB. E. Poling, B. E. The Properties of Gases and LiquidsMcGraw-Hill Education2001The EleTher Joint Industrial Project (JIP) aims at promoting research in this field. A practical workflow is under development where the first stage consists in analyzing the data. The present work is part of this larger scope and presents a strategy to analyze the internal and external consistency of experimental data for electrolyte mixtures. The focus is on alkali halide salts in water. Internal consistency allows testing the quality of data by confronting them to each other and using the thermodynamic relationships between these data. For that purpose, the data are confronted with a model that was initially optimized and the deviations are analyzed. For this purpose, the electrolyte non-random two-liquid (eNRTL) thermodynamic model was selected. Some deviations are attributed to the model inadequacy (high temperature or high molalities). Others are to be attributed to data inconsistencies. On the other hand, the objective of external consistency is to evaluate the consistency between different systems of the same family. The Bromley model is used for this purpose, as it contains a single adjustable parameter. Investigating the trend of this parameter with type of compound or with temperature may provide valuable information regarding the underlying physics, in addition to identifying outliers.

Predicting the properties of salt water using neural network potentials and continuum solvent theory

“…[8] This is a crude solution as there are large gaps and uncertainties in existing experimental databases. [9,10] For example, the activity coefficients of even relatively simple electrolyte solutions such as rubidium hydroxide and lithium bicarbonate have never been measured to the best of our knowledge.…”

Towards predictive design of electrolyte solutions by accelerating ab initio simulation with neural networks

“…Unfortunately, these oen contain unreliable data or are missing key pieces of information. 11,12 For instance, the activity coefficients of many fundamental electrolytes such as lithium bicarbonate and rubidium hydroxide in water have never been reported in the experimental literature to the best of our knowledge. 13 In recent years, researchers have focused on developing predictive approaches to the design of electrolyte solutions for battery applications using large datasets 14,15 such as the electrolyte genome project.…”

Towards predictive design of electrolyte solutions by accelerating ab initio simulation with neural networks.

“…Unfortunately, these often contain unreliable data or are missing key pieces of information. 11,12 For instance, the activity coefficients of many fundamental electrolytes such as lithium bicarbonate and rubidium hydroxide in water have never been reported in the experimental literature to the best of our knowledge. 13 In recent years, researchers have focused on developing predictive approaches to the design of electrolyte solutions for battery applications using large datasets 14,15 such as the the electrolyte genome project.…”