International Journal of Engineering and Technical Research (IJETR) ISSN: 2321-0869 (O) 2454-4698 (P), Volume-7, Issue-8, August 2017 6 www.erpublication.org Abstract— Nanofiltration is currently applied in many industrial processes. The separation efficiency of nanofiltration systems is related to complex phenomena occurring at membrane surface and within the nanopores. The nature of these phenomena is still a subject of debate and there is a real need to better reproduce these phenomena through simple and accurate predictive models. In this paper, interfacial and dielectric properties of two commercial nanofiltration membranes are investigated with the modeling of the permeation of ions typically found in seawater. The membrane charge density was estimated using zeta potential measurements and the dielectric exclusion was represented by the Born model. The predictions of rejection and permeate flux were in good agreement with experimental results when the dielectric effect was considered, indicating that the calculation of membrane charge with zeta potential data is appropriate. Based on simulation results, dielectric constants inside nanopores were calculated and results show that the ion solvation model is appropriate for these membranes. Index Terms— Dielectric exclusion, Ionic rejection, Nanofiltration, Zeta potential. I. INTRODUCTION Nanofiltration (NF) is a pressure-driven membrane separation process with characteristics between those of reverse osmosis and ultrafiltration and is currently applied in many industrial processes such as desalination [1]. The separation efficiency of nanofiltration systems is related to a complex mechanism including steric, dielectric and electrostatic partitioning effects between membrane and solutions [2], [3]. During the last two decades, the prediction of membrane performance has been a relevant area of research [1]. There is an increasing need for developing of model-based tools to design new membrane systems or to optimize existing membrane installations. These models should predict fluxes and rejections as a function of transmembrane pressure for a given membrane system. Also, they should be able to determine the membrane properties necessary to attain a Marcela Costa Ferreira, School of Chemistry, Federal University of Rio de Janeiro, Rio de Janeiro, Brazil. João Victor Nicolini, Chemical Engineering Program / COPPE, Federal University of Rio de Janeiro, Rio de Janeiro, Brazil. Heloísa L. S. Fernandes, Chemical Engineering Department, School of Chemistry, Federal University of Rio de Janeiro, Rio de Janeiro, Brazil. Fabiana Valéria da Fonseca, Inorganic Processes Department, School of Chemistry, Federal University of Rio de Janeiro, Rio de Janeiro, Brazil. desired retention and be a realistic predictive tool with a limited number of experiments [3], [4]. The most widely and successfully adopted NF predictive models are based on the extended Nernst – Planck (ENP) equation to describe the mass transfer across the membrane [3], [5], [6]. This model considers the three important mechanisms of ionic transport in membranes: (a) diffusion, (b) electromigration as a result of concentration and electrical potential gradients and (c) convection caused by the pressure difference across the membrane [7]. One of the most studied models is the Donnan-Steric Pore Model (DSPM) [8], [9]. This model describes the transport of ions in terms of an effective membrane thickness (Δx), a membrane charge density (X d ) and an effective pore radius (r p ) [5], [6], [8], [9]. It also takes into account the effects of hindrance to diffusion and convection within the pore and the equilibrium partitioning due to a combination of Donnan and sieving mechanisms at membrane / solution interfaces [6]. Although this model has been reported to successfully describe simple systems such as those constituted by organic molecules, it has not been very successful for multivalent cations. To improve the prediction capability, some modifications for DSPM model have been suggested by Bowen and Welfoot [9] such as the incorporation of dielectric constant variations between bulk and pore solutions, which has shown better prediction of divalent ions rejection. Bandini and Vezzani [10] proposed a more general model, called Donnan-Steric Pore Model & Dielectric Exclusion (DSPM&DE), which is basically an extension of the DSPM model, in which the primary effect of the DE is considered as the most relevant in determining ion partitioning, together with steric hindrance and Donnan equilibrium. The membrane charge density is obtained by fitting rejection data in the DSPM and DSPM&DE models, being an empirical function related to the feed electrolyte concentration in terms of a Freundlich isotherm [10], [11]. This model is independent of the electrolyte type and does not consider any pH effect. It has been demonstrated to be appropriate in the case of single salts and multicomponent mixtures for some membranes, but it failed in some other cases [2]. Hence, it has been suggested that the membrane charge density is related to zeta-potential data by measuring the streaming potential of nanofiltration membranes, considering the influence of the ionic strength and pH. Other possibility is to develop physico-chemical models to describe the mechanism of charge formation, considering dissociations of functional groups [2], [12]. In this study, a model based on the DSPM models equations is used to predict the rejection of various ions Modeling of ionic transport through nanofiltration membranes considering zeta potential and dielectric exclusion phenomena Marcela Costa Ferreira, João Victor Nicolini, Heloísa L. S. Fernandes, Fabiana Valéria da Fonseca