Empirical models for service life (break-through time) are valuable tools in the selection and design of activated carbon filters for military as well as industrial applications. Such predictive models should ideally be based on easily physical properties of the adsorbent-adsorbate system. In this paper, we present quantitative-structure-affinity models (QSAfR) for the break-through time of halogenated hydrocarbons, monofunctional amines, ketones and aldehydes in dry and humid carbon. The modeling strategy is based on statistical experimental design in the selection of the training-set followed by multivariate data analysis of a multitude of property descriptors to establish adsorbate properties that are influential during physisorption. The final QSAfR models were simplified to include only a few easily obtainable property descriptors.
展开▼