Fig. 5
2D sheets of PAP[5] channels could be assembled into composite membranes and demonstrated enhanced performance compared with current commercial membranes. a SEM images showed ~ 100% coverage of PAP[5] 2D sheets on PC and PES membranes after three and four cycles of layer-by-layer deposition, respectively. Scale bar, 1 μm. b After three and four cycles of deposition on PC and PES membranes, the permeabilities were 3.0 ± 1.2 LMH bar−1 and 64.8 ± 11.3 LMH bar−1, respectively, (mean ± s.d., n = 3). The permeability of the modified PES membranes is approximately one order of magnitude higher that of commercial nanofiltration membrane N30F (3.4 ± 0.3 LMH bar−1) and NDX (4.6 ± 0.1 LMH bar−1) with a similar molecular weight cutoff (MWCO). c The molecular weight cutoff (MWCO) was ~ 450 Da, 370 Da, and 360 Da for the modified PES membrane, commercial N30F and NDX membranes, respectively, as determined from filtration of dyes of various molecular weights. The molecular weight cutoff probability distributions derived from fitting to a sigmoidal model showed the standard deviation (σ) of modified PES (132 Da) membranes was similar to that of N30F (146 Da) and NDX (125 Da). The pore size distribution is tighter for the PES membrane with a σ/MWCO ratio of 0.29 compared with 0.35–0.39 for the commercial membranes. The measured rejection values were corrected using a concentration polarization model and all the cutoff data were fit to a sigmoidal function, as described in Supplementary Methods. d A comparison of PAP[5] channel-based membrane to other commercial NF membranes shows that within the cutoff range (400~500 Da), the channel-based membrane has one order of magnitude higher permeability and in general is several times higher that of commercial nanofiltration membranes across the complete range of available data on such membranes from literature50,51,52,53,54,55,56,57,58,59,60,61,62,63,64. e Photographs of feed (F) and permeate (P) containing different dye molecules (decreasing MW from left to right in image) for modified PC membranes. Both positively and negatively charged dyes had similar rejections as shown in Supplementary Fig. 27. Data shown are the average of triplicate measurements with standard deviation
