Effect of Feed Water pH on the Partitioning of Alkali Metal Salts from Aqueous Phase into the Polyamide Active Layers of Reverse Osmosis Membranes

Abstract

The partitioning of solutes into the polyamide active layers of reverse osmosis (RO) membranes is a key membrane property determining solute permeation. Quantification of partition coefficients and their dependence on feedwater pH would contribute to the development of predictive transport models of contaminant transport through RO membranes; however, neither solute partitioning nor the effect of feed solution pH on partitioning has been thoroughly characterized in the literature. Accordingly, we characterized the partitioning of all chloride salts of alkali metals (CsCl, RbCl, KCl, NaCl, and LiCl) from the aqueous phase into the polyamide active layers of five polyamide RO membranes, including one prepared in-house and four commercial membranes. We evaluated the effect of pH on the partitioning of alkali metal salts and whether the effect of pH on salt partitioning and rejection is consistent with Donnan theory predictions. Results showed that for all membranes, the partition coefficients of all salts were less than one and did not differ substantially among RO membranes. Results also indicated that for all membranes tested, Donnan theory provided an appropriate theoretical framework to estimate the effect of pH on salt partitioning (evaluated for all chloride salts of alkali metals) and salt rejection (evaluated for NaCl). Thus, we conclude that changes in salt rejection resulting from feed solution pH are primarily driven by changes in salt partitioning with comparatively small changes in salt diffusion coefficients.

Enthalpic and Entropic Selectivity of Water and Small Ions in Polyamide Membranes

“…More specifically, the rejection trend in the denser NF90 membrane correlates to the order of the hydrated radius of the cations (Li + > Na + > K + > Cs + ; Table ), suggesting that dehydration of the cations at the narrower constrictions of the NF90 membrane is more hindered compared to the NF270 membrane and size exclusion dominates the rejection trend. These rejection trends, which are partially supported by recently published results on cation partitioning into the NF90 membrane and simulated carbon nanopores, are conserved when concentration polarization is taken into consideration (Figure S7).…”

Significance of Co-ion Partitioning in Salt Transport through Polyamide Reverse Osmosis Membranes

“…Following eqs 13a−17, K tot can be determined from the QCM measurements. As shown in Figure 2B, K tot decreases as the feed salt concentration increases (the scatter plot is shown in the Supporting Information, Figure S3), in agreement with previous studies measuring the partitioning coefficients of salt by electrochemical impedance, 13,15 Rutherford backscattering spectroscopy, 16 and QCM-D. 10,11 The decreasing trend of K tot is explicable in terms of the Donnan equilibrium. Increasing the salt concentration decreases the magnitude (i.e., absolute value) of the Donnan potential, resulting in a reduced K tot (eq.…”

“…An uncoated sensor was also tested with the same solutions as a control to account for the changes in viscosity and density in the solutions. , The mass of the partitioned salt ( m salt ) was determined from the frequency change recorded by the QCM after subtracting the frequency change detected by the uncoated sensors. This mass change ( m salt ) was used to calculate the concentration of the salt partitioned into the membrane ( c s,p ) as described elsewhere , normalΔ m = δ ( c s , v f v M W v + c s , p false( 1 − f normalv false) M W p ) where δ is the membrane thickness, c s,v is the salt concentration in the PA membrane voids, f v is the void fraction ( f v = 0.3), and MW v and MW p are the molecular weights of the salt in the voids and in the membrane, respectively. Given that the voids in the rough, nodule-like PA active layer surface of RO membranes are tens of nanometers in size, the salt concentration inside the voids ( c s,v ) is the same as that in the bulk solution (i.e., no salt exclusion between bulk solution and voids).…”

“…It is further assumed that MW v is the molecular weight of the unhydrated NaCl, while MW p is the hydrated molecular weight. The assumption might be valid considering that the PA film is not rigid and can accommodate the hydrated water molecules without displacing pre-existing water from the film, based on the work conducted by Coronell et al 10,11 Additionally, MW p is calculated from…”

“…We also note that K co is conceptually equivalent to the mobile salt partitioning coefficients in previous studies. 10,11 ■ MATERIALS AND METHODS…”