Overcoming the permeability-selectivity trade-off of desalination membranes via controlled solvent activation

Min Gyu Shin, Jin Young Seo, Hosik Park, You In Park, Jung Hyun Lee

Research output: Contribution to journalArticlepeer-review

43 Citations (Scopus)


Here, we present a facile solvent activation method for significantly enhancing the desalination performance of reverse osmosis (RO) membranes. Polyamide (PA)-thin film composite (TFC) RO membranes were activated with a dimethyl sulfoxide (DMSO)/water mixture, whose solvency power was carefully controlled by adjusting the DMSO volume fraction. A DMSO/water mixture with a DMSO volume fraction of 0.3 effectively activated the PA selective layer while marginally deforming the polysulfone support of the lab-made PA-TFC membrane, thus considerably enhancing its water permeance by ~43% while maintaining its NaCl rejection (~99.4%). All the commercial membranes activated with the optimized DMSO/water activation protocol also exhibited dramatically enhanced water permeance (26–155%) with unchanged or even higher NaCl rejection, surpassing the conventional permeability–selectivity trade-off. A careful characterization of the structures and properties of the model PA film under various solvent environments revealed the thermodynamics and kinetics associated with the activation-induced structural deformation of the PA network, which governs its structural density, consequently affecting the separation properties of the membrane. Our strategy provides a commercially viable means for the fabrication of high-performance membranes together with shedding light on the underlying the structure-property relationship of polymeric membranes.

Original languageEnglish
Article number118870
JournalJournal of Membrane Science
Publication statusPublished - 2021 Feb 15

Bibliographical note

Funding Information:
This research was supported by the National Research Foundation of Korea (NRF) grant funded by the Korean government ( 2019R1A2C1002333 , 2019M3E6A1064103 and 2018R1A4A1022194 ) and the Technology Innovation Program ( 20010914 ) funded by the Ministry of Trade, Industry & Energy (MOTIE, Korea).

Funding Information:
Fig. 4 presents the ATR-FTIR spectra of the PA-TFC membranes activated with DMSO/water of different ?DMSO. The pristine PA-TFC membrane exhibited characteristic PA peaks at 1,668 cm?1 (C=O stretching), 1,610 cm?1 (hydrogen?bonded C=O stretching) and 1,542 cm?1 (N?H bending) [24,25], which were absent for the PSF support. Regardless of ?DMSO, all the DMSO/water-activated membranes showed the same ATR-FTIR spectra as those of the pristine control, implying that the chemical structure of the bulk PA network is not altered by solvent activation [32,33]. The relative PA peak intensity, defined by the intensity ratio of the PA characteristic peak at 1,668 cm?1 to the PSF characteristic peak at 1,488 cm?1, can be used as a rough estimation of the relative amount of the PA layer in the membrane. A close examination of the ATR-FTIR spectra revealed that the membrane activated with ?DMSO = 0.9 exhibited the slightly attenuated relative PA peak intensity, while other membranes showed the similar values (Supporting information S8). This result indicates that severe PA swelling at ?DMSO = 0.9 can lead to the leaching of a noticeable amount of PA fragments from the PA layer (i.e., partial etching). Notably, no characteristic DMSO peak at 1,044 cm?1 (S=O stretching vibration) [56] was observed for all the pristine and activated membranes (Supporting information S9). The absence of DMSO in the membranes was further evidenced by the thermal gravimetric analysis (TGA) results that all the membranes exhibited identical TGA curves without showing the weight loss in the temperature range from 180 to 200 ?C corresponding to DMSO with the boiling point of 189 ?C (Supporting information S9). Hence, we speculated that DMSO bound to the membrane during activation was completely removed by the thorough water washing step employed in this study due to its high affinity with water.Interestingly, the trend of A/B was quite different from that of A. For example, the membrane activated with ?DMSO = 0.3 displayed the highest compaction-induced enhancement of A/B, which recovered to the level of the pristine membrane (?DMSO = 0). A further increase in ?DMSO (>0.3) led to a less enhancement of A/B, resulting in a stabilized A/B that was lower than that of the pristine control. The PA structure moderately deformed by activation with ?DMSO = 0.3 would be effectively compacted and retightened by pressurization in a reversible manner due to its increased chain flexibility, which represents a defect-healing mechanism (Fig. 6b) [25]. This beneficial mechanism is believed to be responsible for the remarkable recovery of the initially impaired A/B to the level of the pristine control via compaction for the membrane activated with ?DMSO = 0.3. As a result, the fine balance between the generation and healing of the defects in the PA layer activated with ?DMSO = 0.3 can account for its higher stabilized A with a similar stabilized A/B compared to those of the pristine counterpart [25]. To further confirm our claim, A and A/B of the pristine and DMSO/water-activated (?DMSO = 0.3) membranes were monitored with increasing the operating pressure (Supporting information S14). As the operating pressure was elevated, A decreased while A/B increased for both membranes mainly due to the enhanced membrane compaction at a higher pressure (Supporting information S14), as also reported by other researchers [58]. Importantly, A/B of the activated membrane was lower than that of the pristine membrane at low pressures, but recovered to the pristine value at higher pressures (>10 bar). This result further supports our claim that the loosened network of the activated PA layer is more effectively compacted and densified by the high applied pressure. However, stronger solvency power (?DMSO > 0.3) can cause the severe and viscous deformation of the PA layer [25], leading to a significant reduction in initial A/B together with a drastic increase in A. This viscous irreversible deformation of the PA network cannot be completely restored and retightened even under highly pressurized conditions [24], resulting in a lower stabilized A/B than that of the pristine control. In addition, the significant structural deformation of the PSF support activated with ?DMSO > 0.3 (Fig. 2) could generate large PA?support interfacial stress and thus interfacial defects, which would presumably further reduce A/B while enhancing A [34].This research was supported by the National Research Foundation of Korea (NRF) grant funded by the Korean government (2019R1A2C1002333, 2019M3E6A1064103 and 2018R1A4A1022194) and the Technology Innovation Program (20010914) funded by the Ministry of Trade, Industry & Energy (MOTIE, Korea).

Publisher Copyright:
© 2020 Elsevier B.V.


  • Dimethyl sulfoxide
  • Polyamide
  • Reverse osmosis
  • Solvent activation
  • Thin film composite membrane

ASJC Scopus subject areas

  • Biochemistry
  • General Materials Science
  • Physical and Theoretical Chemistry
  • Filtration and Separation


Dive into the research topics of 'Overcoming the permeability-selectivity trade-off of desalination membranes via controlled solvent activation'. Together they form a unique fingerprint.

Cite this