Characterization of Mixtures Part 1: Prediction of Infinite‐Dilution Activity Coefficients Using Neural Network‐Based QSPR Models

Abstract

The major problem in building QSAR/QSPR models for mixtures lies in their characterization. It has been shown that it is possible to construct QSPR models for the density of binary liquid mixtures using simple mole fraction weighted physicochemical descriptors. Such parameters are unsatisfactory; however, from the point of view of interpretation of the resultant models. In this paper, an alternative mechanism-based approach to the characterization of mixtures has been investigated. It has been shown that while it is not possible to build significant linear models using these descriptors, it has been possible to construct satisfactory artificial neural network models. The performance of these models and the importance of the individual descriptors are discussed.

Characterization of Mixtures. Part 2: QSPR Models for Prediction of Excess Molar Volume and Liquid Density Using Neural Networks

“…As a result, modifications to the descriptors in order to encode likely intermolecular interactions were investigated. This resulted in a successful QSPR model for infinite dilution activity coefficients in binary liquid mixtures4 and now, in this work, successful QSPR models for excess volumetric properties of liquid mixtures. Could these modified descriptors be applied more generally to the modelling of other properties of mixtures?…”

“…The seven descriptors reported here and in the previous paper4 encode the most important non‐covalent intermolecular interactions and may be sufficient to model a wide variety of properties of binary and possibly even more complex mixtures. Furthermore, it may be possible to formulate a similar set of descriptors to account for the more complex intermolecular interactions which occur between ligands and receptors.…”

“…In order to address this problem we devised a set of mixture descriptors which were designed to specifically account for interactions between the components of a mixture. These have been successfully applied to the prediction of infinite‐dilution activity coefficients using neural networks (part 1 of this series4). In this paper we report the application of this approach to modelling V E .…”

Molecular Design

“…An alternative technique has been proposed in which the descriptors are based on mechanistic theories concerning the property to be modeled [29]. The models can be used for prediction and this is fine if that is all that is required but the descriptors themselves relate to two or more molecules and the modeling process, using an ensemble of neural networks, is at best opaque.…”

Modeling of non-additive mixture properties using the Online CHEmical database and Modeling environment (OCHEM)

“…However, in recent years several studies have attempted to develop QSPR models to predict non-additive properties (density [1], infinite dilution activity coefficient [2], bubble temperature [3], azeotropic behavior [4], or excess molar volume [5]) of mixtures. There is also very considerable interest in modeling the toxicity of chemical mixtures.…”