Tuning the permselectivity of polymeric desalination membranes via control of polymer crystallite size

Membrane desalination is a leading technology for treating saline waters to augment fresh water supply. The need for high-performance desalination membranes, particularly with high water/salt selectivity, has stimulated research into the fundamental structure-property-performance relationship of state-of-the-art membranes. In this study, we utilize a facile method for tuning properties of a polymeric desalination membrane to shed light on water and salt transport mechanisms of such membranes. A desalination membrane made of cellulose triacetate is treated in a plasticizer solution, followed by water rinsing. The modified membranes exhibit reduced salt flux without compromising water flux, indicating enhanced water/salt selectivity. An inspection of material characteristics using a model film system reveals a plasticizing-extracting process in changing the polymeric structure, which leads to the reduction of crystallite size in the polymer matrix, consequently affecting the transport properties of the membranes. Our findings highlight the potential of the plasticizing-extracting process in fabricating membranes with desired desalination performance.

where Δm is the measured weight change of the draw solution, ρ is the density of water, A is the membrane area, and t is the time.
A conductivity meter probe was immersed in the feed solution side to determine the increase of NaCl concentration in the feed solution at three-minute intervals. Reverse salt flux, Js, was calculated by: where CF is the NaCl concentration in the feed solution, Initial F V is the initial volume of the feed solution, A is the membrane area, and t is the time.

Calculation of membrane transport properties
The transport properties of the membranes were determined based on the desalination performance data in FO. In particular, salt permeability, B, was first calculated using 1 : where CD is the concentration of the NaCl draw solution; Pe δ is the Peclet number of the boundary layer in the draw solution side with external concentration polarization (ECP): where k is the feed side mass transfer coefficient, which can be estimated from a correlation for the rectangular cell geometry and given operating conditions. Pe S is the Peclet number of the support layer in the feed side with internal concentration polarization (ICP): where S is the structural parameter of the CTA membrane, which was determined to be ~595 μm according to a previous publication using the same membrane 2 . Then, water permeability, A, was determined using 3,4 : where n is the number of dissolved species created by the draw solute, Rg is the ideal gas constant, and T is the absolute temperature. The calculated transport properties of membranes are listed in Supplementary Table 1. S4

Methodology for deconvoluting X-ray diffraction data into amorphous and crystalline components
Both amorphous and crystalline regions in the cellulose triacetate films contribute to the X-ray diffraction patterns. The diffraction intensity, I, as a function of scattering vector, q, which was obtained from the measurement and can be written as: where IA(q) and IC(q) are the amorphous and crystalline contributions, respectively.
In order to analyze the Bragg peak profiles, one must separate the amorphous contribution from the total intensity vs. q. The broad hump centered at = 1.46 Å −1 in the 1-D integrated curve, corresponding to the amorphous halo shown in the 2-D diffraction pattern (insets of Figure 2d), can be simply fitted using a Gaussian function: where I0 is the offset, qc is the center, w is the width, and A is the area.
As shown in Figure 2d, only the left section of the amorphous hump from 1.0 to 1.

Characterization of membrane transport properties in RO
Membrane transport properties, i.e., water and salt permeabilities, were determined in a lab-scale crossflow RO unit 5 . The effective membrane area available for permeation was 20.02 cm 2 (7.7 cm × 2.6 cm), the crossflow velocity was fixed at 21.4 cm s −1 , and the temperature was 25 ± 0.2 C.
The loaded membrane was first compacted with DI water at an applied pressure of 31 bar (450 psi) until the permeate flux reached steady state. The applied pressure was then decreased to 27.6 bar (400 psi) to determine the pure water flux. The water permeability coefficient, A, was then calculated from: where Jw is the measured pure water flux, and ΔP is the applied pressure. Salt rejection was determined at 27.6 bar (400 psi) using a calibrated conductivity meter (Oakton Instruments, USA).
The observed NaCl rejection, R, was calculated from: where Cp is the permeate NaCl concentration and Cf is the feed NaCl concentration. The salt permeability coefficient, B, was determined from 6,7 : where NaCl w J is the measured water flux with NaCl feed solution, k is the mass transfer coefficient in the crossflow cell, which can be obtained from a correlation with rectangular channel geometry in laminar flow: where D is the NaCl diffusion coefficient, Re is the Reynolds number, Sc is the Schmidt number, dh is the hydraulic diameter, and L is the length of the channel.

Analysis of hydrogen bond interaction between p-nitrophenol with cellulose triacetate
The interaction of p-nitrophenol (PNP) with cellulose triacetate (CTA) material is important in determining the effect of plasticization. Specifically, CTA polymer is abundant in oxygencontaining groups, such as carbonyl (C=O) and hydroxyl groups (C−OH), which could serve as hydrogen-bond acceptors. In the molecular structure of PNP, the nitro group in the para position of the aromatic ring tends to withdraw electrons, thereby reducing the electron density of the phenolic hydroxyl group to serve as a hydrogen-bond donor. As such, the hydrogen atom of the phenolic hydroxyl of PNP will favorably form a hydrogen bond with oxygen-containing groups on the CTA chain, thus facilitating the penetration of PNP into the CTA polymer matrix to induce the plasticizing effect. To verify the presence of hydrogen bonding, we characterized the fabricated CTA films using Fourier transform infrared (FTIR) and the data are presented in Supplementary wavenumbers for all the oxygen-containing groups. For example, the peak of C=O shifts to 1731 cm −1 . These red shifts are likely ascribed to formation of hydrogen bonds between the hydroxyl group of PNP with the oxygen-containing groups of CTA, thereby inhibiting the stretch of these groups to lower frequencies 10,11 . After rinsing with water, the characteristic peaks of PNP disappeared on the deswelled film (blue curve), consistent with the release of PNP observed in other measurements (i.e., POM, DSC, and WAXD). Notably, the peaks of oxygen-containing groups also recovered to their corresponding positions and shape in the pristine samples. Therefore, these results provide further support for the plasticizing-extracting mechanism. Additionally, forming hydrogen bonding requires the hydrogen donor group in its protonated form rather than deprotonated form, which could explain the ineffectiveness of the deprotonated PNP in modifying CTA membranes ( Supplementary Figures 2 and 3).

Analysis of the effect of membrane transport properties on desalination performance
(1) Analysis of water flux In osmotically-driven membrane processes, water flux, Jw, across the membrane is determined by: where A is the water permeability, Δπm is the effective osmotic pressure across the active layer of the membrane.
Due to the detrimental effects of external concentration polarization (ECP) in the draw solution, internal concentration polarization (ICP) within the porous support, and reverse salt flux, Js, across the membrane, the Δπm is lower than the osmotic pressure difference between the bulk draw and feed solutions and could be determined by 12-14 : , , where We rephrase Supplementary Equation 15 by moving A to the denominator: (2) Analysis of reverse salt flux In osmotically-driven membrane processes, reverse salt flux, Js, across the membrane is where B is the salt permeability, ΔCm is the effective concentration gradient across the active layer of the membrane, which could be determined by: 12, 13, 14 We rephrase Supplementary Equation 19 by moving B to the denominator: In the denominator, the decrease of B value through PNP treatment led to a larger value in term 1/B, thereby resulting in a larger value in the denominator that gave a decreased Js.

S9
In summary, the water flux is affected by both water and salt permselectivities, whereas the reverse salt flux is mainly determined by the salt permeability. Therefore, we observed unchanged water flux with decreased reverse salt flux for the membranes treated with low concentrations of PNP solutions.     .