Polyamide-based membranes with structural homogeneity for ultrafast molecular sieving

Structural homogeneity of polyamide barrer layer can maximize the water permeability without compromising the solute rejection of thin-film composite membranes by minimizing the mass fluctuation, greatly surpassing the permeability-selectivity tradeoff.

Polyamide-based thin-film composite membranes formed by conventional interfacial polymerization generally suffer from the depth heterogeneity of the selective layer. However, 3D models of the nanoscale PA density maps of various reverse osmosis (RO) membranes constructed by Culp et al. revealed that the density fluctuations were detrimental to water transport1,2. They suggested that controlling over the internal nanoscale inhomogeneity could maximize the water permeance by minimizing mass fluctuations without sacrificing solute rejections. Therefore, improving the nanoscale homogeneity of the PA layer is presumably to be of great significance in surpassing the permeability-selectivity tradeoff. A simple, low-cost and effective, readily available method to tune the nanoscale homogeneity of TFC membranes for the co-improvement of permeability and selectivity is preferred.

In this study, we demonstrate a facile and versatile approach to tune the nanoscale homogeneity of polyamide-based thin-film composite membranes via inorganic salt-mediated interfacial polymerization process. We added common inorganic salts into the amine monomer solution during the IP process to confine and regulate the diffusion behavior of amine monomers toward the water-oil interface.

Molecular dynamics simulations is employed here to investigate the IP behavior for both the control and salt-participating systems. It can be observed in simulation boxes that Na+ and Cl ions tend to accumulate gradually and surround MPD molecules near the interface to retard the diffusion of amine monomers toward the water-oil interface (Figure 1a). As the result, self-diffusion coefficient of MPD in the system with NaCl (D = 0.23 × 10-9 m2 s-1) is nearly one-third of its counterpart in the system without NaCl (D = 0.6 × 10-9 m2 s-1) as confirmed by Figure 1b, indicating the lower diffusivity of MPD molecules in the presence of NaCl. Such a difference in the self-diffusion coefficient results in fewer MPD molecules in the water/hexane interface in the NaCl-containing system, as confirmed by the number density (ρN) distribution of MPD (Figure 1c). According to the distribution of MPD in Figure 1c, the distinct peak located at ~70 Å indicates the accumulation of amine monomers in the reaction zone. It can be found that the reaction zone width of the system with NaCl (z = 63-78 Å) is thinner than that of the system without NaCl (z = 58-78 Å). This shrunken reaction zone caused by the decreased diffusivity of MPD ultimately favors the thinner PA layer formed in the presence of NaCl. It can be observed in Figure 1d that the number variation of MPD molecules (∆N) from the bulk aqueous solution to the water-hexane interface steadily increases in the presence of NaCl, indicating the enhanced uniformity of the MPD diffusive flux, which therefore contributes to the formation of structurally homogeneous PA layers3 via the salt-participating IP process.

Figure 1. MD Simulation results. (a) The initial (0 ns) and final (10 ns) snapshots of two systems with (right) and without (left) NaCl, respectively. (b) The MSD profiles of MPD molecules in two systems with and without NaCl, respectively. (c) The number density (ρN) distribution of MPD molecules (green and red lines), Na+ (blue line) and Cl (yellow line) ions in the two systems with and without NaCl, respectively. (d) The increased number variation of MPD molecules (∆N = N N0) compared with the initial value (N0).

The structural homogeneity can be testified by the Positron Annihilation Lifetime Spectroscopy results (Figures 2a-2b). It can be seen that salt-modified PA layers possess narrowly distributed free volume pore sizes, confirming our speculation of uniform dense layer. Meanwhile, the in-depth O/N ratio profiles shown in Figure 2c reveal that the O/N ratios of the modified membrane vary slightly against the detection depth in contrast to the fluctuating O/N ratio profiles of the pristine membrane, confirming the nanoscale homogeneity of salt-modified membranes again.

Figure 2. Microstructural properties of the pristine and modified membranes. Free volume pore size distribution of the pristine and modified membranes formed by (a) MPD/TMC and (b) PIP/TMC. (c) In-depth O/N ratio profiles of membranes Pristine@MPD and NaCl@MPD.

Inspection of cross-sectional SEM and TEM images reveals that the transparent thicknesses of the modified MPD-TMC and PIP-TMC PA layers show the thickness reduction of 62-67% and 44-59% respectively, as compared to their counterparts, correlating well with the simulation results. Moreover, the intrinsic thicknesses marked by red lines (PA wall) of modified membranes are even slightly thinner than that of the pristine membrane for both MPD/TMC and PIP/TMC monomer systems. In addition, surface roughness decrements of 57-62% and 35-62% are achieved for modified MPD-TMC and PIP-TMC PA layers respectively compared to their counterparts.

Figure 3. Morphology and topology of the pristine and modified membranes. (a) Surface (scale bar: 1 μm and 100 nm) and (b) cross-sectional SEM images (scale bar: 100 nm), (c) cross-sectional TEM images (scale bar: 200 nm) and (d) topologic AFM images (scale bar: 1 μm) of the pristine and modified membrane.

The separation performances of these membranes fabricated via the conventional and salt-participating IP processes were then evaluated by FO, RO and NF processes. It can be seen in Figures 4a, 4b that the water fluxes of the modified membranes in the FO process increase by ~56-86% (Figure 5a) compared to that of the pristine membrane. In addition, the reverse salt fluxes of modified membranes are lower than that of the pristine membrane, overcoming the permeability-selectivity tradeoff relationship. Consistent results were also obtained for RO tests. Figure 4b shows that the water permeances of modified membranes increase by ~92-165% while higher NaCl rejections are well maintained (~97-99% versus ~95%), compared to those of the pristine membrane. Meanwhile, the results in Figure 4c reveals that the water permeances of NaCl- and NaHCO3-modified membranes increase by 119% and 63% respectively in NF process, , coupled with higher solute (NaCl and Na2SO4) rejections, compared to that of the pristine membrane. Especially for NaCl rejections of modified membranes, a significant improvement of ~160-169% is achieved.

Figure 4. Separation and antifouling properties of the pristine and modified membranes. (a) Water flux (Jv) and reverse salt flux (Js) of the pristine and modified membranes formed by MPD and TMC applied in FO process, under PRO mode using DI water and 2 M NaCl solution as the feed and draw solutions. (b) Water permeance (A) and NaCl rejections (Rs-NaCl) of the pristine and modified membranes formed by MPD and TMC using 1000 ppm NaCl solution as the feed applied in RO separation. (c) Water permeance (A) and solute rejections (Rs) of the pristine and modified membranes formed by PIP and TMC using 1000 ppm NaCl or Na2SO4 solution as the feed applied in NF process. (d) Dynamic fouling test results of the pristine and modified membranes formed by MPD and TMC in FO process. (e) Water permeance (A) and NaCl rejections (Rs-NaCl) of various salt-modified membranes formed by PIP and TMC using 1000 ppm NaCl or Na2SO4 solution as the feed applied in NF process. (f) S values of various salt-modified membranes formed by PIP and TMC. The error bars represent the standard deviation and were calculated on the basis of at least three data points measured from different samples.

Apart from the greatly surpassed permeability-selectivity tradeoff, the hydrophilic and smooth surface also renders the salt-modified membrane improved antifouling capacity.  Figure 4d shows that the water flux of the pristine membrane drops 44% and only 14% (recovery ratio ~70%) is recovered after physical cleaning. Instead, the flux drops of modified membranes are only 23-27%, while the recovery ratio can reach ~91-93%.

We also found that the separation performance of modified membranes is affected by the type of inorganic salt. The salt with the larger ionic radius is added, the modified membrane shows the higher water permeance (Figures 4e), possibly due to the removal of the larger salt that leaves the larger free volume pores in the PA layer, as confirmed by the larger S value (Figure 4f).

Overall, we demonstrated a facile and versatile approach to tune the nanoscale homogeneity of PA-based TFC membranes via a salt-regulated IP process for various monomer systems and salt types to improve separation performance. The behavior of inorganic salts accumulated near the water-oil interface confines and regulates the diffusion of amine monomers into the reaction zone, resulting in the uniform diffusive flux of amine monomers, thus the spatially homogeneous polymerization and the formation of smooth and thin PA layers with structural homogeneity. The resulting membranes therefore show greatly improved water permeance and/or solute rejection for all three monomer systems in FO/RO/NF separations, as well as the decreased fouling propensity. Moreover, the water permeance and solute rejection of the resulting membranes can be reasonably tuned by selecting suitable inorganic salts. Therefore, this work sheds insights into the fabrication of TFC membranes with structurally homogeneous PA selective layers for molecular sieving.

For more details, please refer the paper with the link: https://doi.org/10.1038/s41467-022-28183-1

References

1          Culp, T. E. et al. Nanoscale control of internal inhomogeneity enhances water transport in desalination membranes. Science 371, 72-75 (2021).

2         Geise, G. M. Why polyamide reverse-osmosis membranes work so well. Science 371, 31-32 (2021).

3          Liang, Y. et al. Polyamide nanofiltration membrane with highly uniform sub-nanometre pores for sub-1 Å precision separation. Nature communications 11, 1-9 (2020).