Molecular Filtration by Ultrathin and Highly Porous ... - ACS Publications

Sep 28, 2016 - Copyright © 2016 American Chemical Society. *E-mail: [email protected]. Homepage: http://mypage.zju.edu.cn/binsu. Cite this:Anal. Chem...
0 downloads 0 Views 1MB Size
Subscriber access provided by CORNELL UNIVERSITY LIBRARY

Article

Molecular Filtration by Ultrathin and Highly Porous Silica Nanochannel Membranes - Permeability and Selectivity Qian Yang, Xingyu Lin, and Bin Su Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.6b02968 • Publication Date (Web): 28 Sep 2016 Downloaded from http://pubs.acs.org on September 28, 2016

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Analytical Chemistry is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 8

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Molecular Filtration by Ultrathin and Highly Porous Silica Nanochannel Membranes - Permeability and Selectivity Qian Yang, Xingyu Lin, Bin Su* Institute of Analytical Chemistry, Department of Chemistry, Zhejiang University, Hangzhou 310058, China ABSTRACT: An ideal molecular filtration membranes should be highly permeable and selective, thus desiring the membranes to be ultrathin, highly porous and consists of small and uniform pores or channels. In this work we report the molecular filtration by free-standing ultrathin silica nanochannel membranes (SNMs) using a U-shaped cell and spectrophotometric detection, focusing on the quantitative evaluation of permeability and selectivity of SNMs. Thanks to the ultrasmall channel size, namely ca. 2  3 nm, and the negatively charged channel surface arising from the deprotonation of silanol groups, the SNM displayed excellent size and charge selectivity for molecular filtration. The selectivity coefficient for separation of small methyl viologen from large cytochrome c is as high as 273, because of the uniform pore/channel size. The charge-based filtration can be modulated by the salt concentration and solution pH, which control the overlap of radial electrical double layer and surface charge sign/density, respectively. Owing to the high relative pore density, namely 16.7%, and the straight and vertical channel orientation, the SNM is highly permeable, displaying a molecule flux much higher than commercially available dialysis membrane and others reported previously. In addition, we demonstrated that, by biasing a small voltage across the SNM, both the flux and separation selectivity could be significantly enhanced.

INTRODUCTION Molecular filtration is a fundamental procedure in many chemical, biological, pharmaceutical and petrochemical processes.1 Conventional membranes consisting of a random network of pores with a wide distribution of pore sizes, as well as a large membrane thickness, do not present sufficient selectivity and permeability for filtration at the molecular level.2 Exploration of new technologies that fabricate ultrathin nanoporous membranes capable of both precise and fast molecular filtration is thus of great significance.3-4 An ideal molecular separation membrane desires adequate stability, high permeability and precise selectivity.5 The stability describes the ability of the membrane to withstand a certain flux without breaking, as well as pressure, thermal and chemical durability. The permeability describes the ability of the membrane to achieve a high throughput, which expects the membrane to be highly porous, ultrathin and less channel tortuosity. The selectivity refers to the precise capability of the membrane to separate the target molecules from the mixture, which desires sufficiently small and uniform pore/channel size and effectively selective surface chemistry. In nature, biological protein channels in vivo regulate the flow of certain chemical species, such as ions, water and nucleic acids, in and out of cells with an ultrafast flux and precise selectivity.6-8 Inspired by this near perfect function, many well-ordered nanochannel membranes have been artificially fabricated for molecular filtration,3,9-12 as well as chemical/biomolecular discrimination and purification,13-18 drug delivery,19-22 desalination and sensing.23-25 Nanolithographic membranes are usually characterized by uniform pore size and highly ordered pore distribution,18,26-27 which however require expensive facilities to fabricate and thus have only limited applications. Track-etched polymer membranes (i.e. polycar-

bonate, polyimide and polystyrene) are commercially available, whereas low pore density usually yields a low throughput for molecular filtration.28-31 Another commercially available membrane is porous anodic alumina, which is however usually thick, mechanically fragile and consists of channels in between ~30 nm and 400 nm,32-33 thus reducing channel size by post-modification is necessary to achieve molecular filtration.34-37 Nanoporous block copolymer membranes have been widely prepared but the chemical and thermal stability is insufficient for operation in harsh conditions.38-39 Other types of nanoporous membranes, such as vertically aligned carbon nanotubes and nanopores on single layer graphene, have also been fabricated,40-41 which often require complicated fabrication conditions/operations and are thus difficult to scale up. Although significant progress has been achieved, fabrication of ideal molecular filtration membranes in a simple and economic way still remains challenging. Very recently, we have reported a facile approach to fabricate ultrathin (10  200 nm in thickness) silica nanochannel membrane (SNM) consisting of straight channels with a uniform size (2  3 nm in diameter, and can be further reduced by surface modification) and a high density (4  1012 cm2, corresponding to relative pore density of 16.7%).42-43 These structural characteristics make it ideal for molecular filtration based on size and charge. In this work, we performed a detailed quantitative investigation on the molecular filtration by this SNM, focusing on the assessment of the permeability and selectivity. The permeability was examined by singlecomponent permeation experiments and compared with other nanoporous membranes. The separation selectivity was investigated by two-component permeation experiments according to the molecular size and charge. In addition, we demonstrated that both flux and selectivity of molecular separation could be remarkably improved by the external electric field.

ACS Paragon Plus Environment

Analytical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

EXPERIMENTAL SECTION Chemicals and Reagents. All chemicals were used as received without further purification. Ultrapure water (18.2 Mcm) was used in all experiments. Methyl viologen (MV2+) dichloride, 1,5-naphthalene disulfonate (NDS2) disodium salt, fluorescein (FL2) disodium salt were obtained from Aladdin. Cytochrome c (Cyt c) and poly(methyl methacrylate) (PMMA, Mw = 996000) were purchased from Sigma-Aldrich. Standard regenerated cellulose dialysis membrane (molecular cut off 6 – 8 kD) was purchased from Spectrum Laboratories, Inc. Tracketched polyethylene terephthalate (PET) membrane (2 m of pore diameter) was purchased from Haoxia Co., Ltd. (Shanghai). Indium tin oxide (ITO) coated glass (thickness: 100 nm, resistance of < 17 /square) was obtained from Zhuhai Kaivo. Porous silicon nitride (p-SiN) chip was purchased from Risun Instrument (Shanghai). Preparation and Characterization of SNM/p-SiN. The SNM was primarily grown on the indium tin oxide (ITO) glass by the StÖber-solution growth approach.44 Then it was exfoliated and transferred to the p-SiN by a PMMA-assisted transfer method as have reported recently.42 The details can be found in the supporting information. As illustrated in Figure 1a, the p-SiN chip consists of a 150-nm-thick SiN layer evaporated on 1 cm  1 cm silicon wafer, in the center of which a 700 m  700 m SiN window was prepared by etching the beneath silicon layer. On this window, there is an array of perforated pores of 4 m in diameter arranged in a square lattice (24  24) with a pore-to-pore distance of 30 m. After supporting the SNM, a structure in the form of nanochannels-on-micropores, designated as SNM/p-SiN, was obtained.

Page 2 of 8

a HT7700 transmission electron microscope (Hitachi, Japan) at 100 kV, respectively. Molecular Permeation Experiment. A home-made Ushaped cell was used for molecular permeation experiments, as shown in Figure 1b. The SNM was fixed in between two Teflon half-cells (designated as feed and permeate cells, respectively), each of which has two quartz glass windows with an optical length of 1 cm for light irradiation and transmission. In between the SNM and Teflon cell, a silicon rubber gasket was used for sealing. Two Teflon cells were positioned in a Teflon holder with a screw, which could be rotated to press two cells tightly contacted. Probe molecules used in this study include anionic FL2 and NDS2, cationic MV2+ and Cyt c. Their structures are shown in Figure S1. For the single-component permeation experiment, the feed cell was filled with 1 mM KCl or 1 M KCl (pH 6.7) and 5 mM probe molecules (unless otherwise specified), while the permeate cell contained only the same electrolyte solution without probes. In the case of twocomponent permeation experiment, a mixture with either of probe molecules at 2.5 mM was added into the feed cell with electrolyte (unless otherwise mentioned). The electric field controlled molecular permeation was performed by biasing a dc voltage supplied by Keithley 6487 picoammeters (KeithleyTektronix, USA) in between the feed and permeate cells using two silver/silver chloride (Ag/AgCl) electrodes. The molecular permeation process was monitored by an optic fiber ultraviolet-visible (UV-visible) spectrophotometer (QEPro, Ocean Optics). The UV-visible absorption spectrum of permeated molecules was continuously recorded. In the course of measurements, both feed and permeate solutions were stirred by small magnetically rotated stirrers (550 rmp, IKARCT basic), in order to avoid concentration-polarization at the membrane surface. Calibration plots of each probes were used to determine their concentrations in the permeate cell. The slope of permeation curve, namely the concentration in permeate solution versus time, was considered as the molecular flux across the SNM. All the average fluxes and standard deviations were calculated from three series of data.

RESULTS AND DISCUSSION

Figure 1. (a) Illustration of the SNM/p-SiN chip with perforated channels for molecule separation. The left is the photograph of a real SNM/p-SiN chip. Note that the size in the form of nanochannels-on-micropores does not represent the real dimension since the silica nanochannel is significantly smaller than the SiN micropore. (b) The home-made cell for molecule permeation using UV-visible spectroscopic detection.

The microscopic structure and morphology of SNM were characterized by scanning electron microscopy (SEM) and transmission electron microscopy (TEM). They were performed on a SU-8010 field-emission scanning electron microscope (Hitachi, Japan) at an accelerating voltage of 3.0 kV and

Characterization of SNM/p-SiN. The SNM primarily grown on the ITO electrode surface was characterized by SEM and TEM (see images in Figure S2, the supporting information). The SNM has a thickness of ca. 94 nm and consists of vertical nanochannels with a uniform size (2  3 nm in diameter) and a high density (up to 4.0  1012 cm2, corresponding to relative pore density of 16.7%). The SNM/p-SiN with perforated channels was prepared by the PMMA-assisted transfer approach.42 As shown by the optical images in Figure 2a and 2b, the p-SiN window, appeared as yellow, consists of an array of perforated micropores in a square lattice pattern (24  24) with a pore-to-pore distance of 30 m (Note that more micropores can be seen on the window edges, which however are not perforated). Before and after supporting SNM, the p-SiN window did not change its color significantly but turned a bit blurred, because the SNM is ultrathin and highly transparent. From the SEM images (see Figures 2c and 2d), we can see that all micropores are covered with SNM without any cracks. Given the SNM is ultrathin, the beneath micropores on the p-SiN window are still visible. This structure

ACS Paragon Plus Environment

Page 3 of 8

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

can be identified more clearly with a single micropore before and after supporting SNM, as shown in Figures 2e and 2f.

Figure 2. Metallographs (a, b) and SEM images (c, d) of the pSiN window before (a, c) and after (b, d) supporting SNM. (e, f) SEM images of a single SiN pore before (e) and after (f) supporting SNM.

Figure 3. (a) TEM image of 3 mm-in-diameter p-SiN chip for insitu TEM measurement. The scale bare is 20 m. (b, c) TEM images of a SiN micropore before (b) and after (c) supporting SNM. (d) Magnified TEM image showing the edge of SNM and SiN pore.

The surface morphology and pore structure of SNM/p-SiN were also characterized by TEM. The specimen was prepared by transferring the SNM to a 3-mm-diameter p-SiN chip, which was then positioned directly on the microscope sample holder for observation. Figure 3a shows the TEM image of a bare p-SiN, on which an array of 4 m pores can be found (Note that this p-SiN chip was only used for TEM, which has a small pore-to-pore distance and is different from those used in the molecular separation). Figures 3b and 3c compare the TEM images of a single p-SiN pore before and after supporting SNM, showing obviously that the ultrathin SNM indeed covers the micropore without cracks. The micropore bright-

ness turned a bit dark due to the electron transmission was partially blocked by the SNM. From the high-magnification TEM image (see Figure 3d), the free-standing SNM possesses a uniform pore size (2  3 nm in diameter) and a high pore density. In comparison with the TEM image of SNM mechanically scraped from the ITO electrode (see Figure S2b), the PMMA-assisted transfer did not vary the surface morphology or pore structure of SNM. Size-Based Molecular Separation. Considering the size of silica nanochannels is uniform, we can expect a precise separation resolution in terms of molecular size. The permeation of two cationic compounds, Cyt c and MV2+, was comparatively studied. The dimension of Cyt c (pI = 9.8)45 is 2.6 nm  3.3 nm  3.2 nm, corresponding to an equivalent diameter of 4.1 nm,13 while the size of MV2+ is only 0.53 nm (see Figure S1).9

Figure 4. (a) The UV-visible spectra of permeate solution recorded for MV2+ permeation at different times: 31, 60, 128 and 184 min (from bottom to top). The inset is the permeation kinetic plot showing the variation of concentration of MV2+ in the permeate solution with time. (b) The UV-visible spectra of 4.04 M Cyt c (the black curve) and permeate solution after 24 h of permeation experiment (the red curve). (c) The variation of concentrations of MV2+ and Cyt c in the permeate solution versus time for a permeation experiment with both 2.5 mM MV2+ and 20.2 M Cyt c in the feed solution. In all cases, the feed and permeate solutions contained 1 M KCl.

The results for single-component permeation are shown in Figures 4a and 4b. In both cases, 1 M KCl was added to both feed and permeation solutions in order to screen the electrostatic effect of membrane surface charges (the charge effect will be studied in the next section). As seen from Figure 4a, the UV-visible spectra of permeated solution display two signature bands of MV2+ at 209 nm and 257 nm, which continuously increase with time. The absorbance at 257 nm was used to calculate the concentration of MV2+ in the permeate solution and to construct the kinetic permeation plot (as displayed in the inset of Figure 4a). Apparently, the concentration of MV2+ linearly increased with time. The slope of this line denotes the molecular flow rate across the SNM, from which the molecular flux (J, mol s1 cm2) can be calculated by multiplying the permeate volume and dividing the effective area of SNM. Given the flow rate is dependent on the initial concentration of MV2+ in the feed solution, it can be concluded that MV2+ permeate the SNM via the Fick diffusion instead of single file diffusion (see supporting information section S3 and Figure S3 for details).22,46 This is reasonable since the diameter of MV2+ is much smaller than the size of channels. In this case, J can thus be expressed as,47 J D

c x

ACS Paragon Plus Environment

(1)

Analytical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

where D is the diffusion coefficient of permeated molecule (cm2 s1), c the initial concentration difference between the feed and permeate solutions (mol cm3), x the membrane thickness (cm), respectively. The molecular flux of MV2+ thus calculated was 4.66  0.25 mol cm2 min1, from which the diffusivity of MV2+ was calculated to be 1.46  107 cm2 s1. Although the SNM is highly porous and the channel size is larger than MV2+, the permeation of MV2+ will be still hindered physically. In this case, its diffusivity can be estimated in terms of the Renkin equation, 2 (2) Dh  D 1    1  2.104  2.09 3  0.95 5  where  is defined as the ratio of molecular diameter versus pore diameter. The value derived from eq. 2 is 2.47  106 cm2 s1, to which the experimental one is close. Moreover, both values are indeed smaller than the bulk one, namely 7.74  106 cm2/s,48 proving the transport of MV2+ was hindered by the nanochannel membrane. In contrast with MV2+, no permeation of Cyt c was observed, which is reasonable as the size of Cyt c is larger than that of silica nanochannels. As seen in Figure 4b, the signature band of Cyt c at 400 nm was not observed in the permeation solution even after 24 h. The result also proves that the SNM is able to separate molecules based on size. Thus, twocomponent permeation experiment was performed with both 2.5 mM MV2+ and 20.2 M Cyt c present in the feed solution. As displayed in Figure 4c, the concentration of MV2+ in the permeation solution continuously increased with time, whereas Cyt c was completely undetectable. Additionally, no permeation lag time was displayed for MV2+ in comparison with the hybrid mesoporous silica membrane.49 Note that the flux of MV2+ here (2.30  0.50 mol cm2 min1) was one half of that in the single-component permeation experiment simply because its initial concentration was half, in agreement with the Fick’s law, namely eq. 1. This value also suggests that the coexisted Cyt c did not influence the permeation of MV2+ demonstrating the antibiofouling ability for filtration of SNM. To quantify the size separation ability of SNM, the separation selectivity of MV2+ versus Cyt c, α(MV2+/Cyt c), is defined as the ratio of permeation fluxes,

α  MV 2+ Cyt c  =

J MV 2+ J Cyt c

(3)

Given the flux of Cyt c could not be determined, a minimal selectivity coefficient (αmin) was used to assess the size separation ability of SNM.4 Considering the Cyt c concentration was below the detection limit of the UV-visible spectrophotometer (1.1  107 M), αmin(MV2+/Cyt c) was calculated to be 273, indicating an excellent size separation ability of SNM. This large value can be attributed to the ultrasmall and uniform pore size, as well as the high pore density and ultrathin thickness, of SNM. Electrostatic Effect on the Molecular Permeation. The silica nanochannel surface is negatively charged due to the deprotonation of silanol groups with an isoelectric point of 3  4 in KCl solution (pH = 6.7).42 The surface charges can induce the formation of a radial electric double layer (EDL), whose thickness, also called the Debye length (D), can be in principle varied from ca. 3.9 to 0.3 nm by increasing the concentration of KCl from 1 mM to 1 M.47 Given the nanochannel size is only 2  3 nm, the radial EDL thus overlapped significantly at the low electrolyte concentration, leading to the permselec-

Page 4 of 8

tive transport. The transport of negatively charged molecules (i.e. so-called co-ions) through nanochannels is electrostatically repelled, whereas that of positively charged ones (i.e. socalled counter-ions) favored. This scenario was confirmed by the permeation results of NDS2, MV2+ and FL2. As shown in Figure 5a and Figure S4, the flux of NDS2 (calculated with the absorbance at 286 nm) at a low electrolyte concentration is significantly smaller than that at a high one, due to a stronger electrostatic repulsion between NDS2 and channel wall in the former case. MV2+ has the similar size with NDS2 but opposite charge, so that its permeation was faster than NDS2. Moreover, the flux (calculated with the absorbance at 256 nm) did not vary that much with the electrolyte concentration (increased from 4.66 to 7.08 mol cm2 min1 with decreasing the electrolyte concentration, as shown in Table 1; see also Figures 5a and S5), due to a stronger electrostatic attraction between MV2+ and channel wall. In the case of FL2 that is twice larger than NDS2 (1.1 nm versus 0.55 nm, although bulk diffusivity is similar, 5.77  106 cm2/s for NDS2, 5.1  106 cm2/s for FL2)9,48,50 its permeation at the low electrolyte concentration was remarkably suppressed and that at the high one was also slower than NDS2 (see blue points in Figure 5a and the spectra in Figure S6).

Figure 5. (a) Kinetic plots for single-component permeation of NDS2 (black), MV2+ (red) and FL2 (blue) at different electrolyte concentrations (1 mM: open points; 1 M: solid points). (b) Kinetic plots for two-component permeation of MV2+ (black) and FL2 (red) at a low electrolyte concentration (1 mM KCl). (c) Kinetic plots for single-component permeation of NDS2 (red) and MV2+ (black) at pH 3 (1 mM HCl in both feed and permeate solutions).

Table 1. Ion permeation flux across the SNM and selectivity coefficient obtained from the single-component experiments. Flux (mol cm2 min1) Ion α

1 M KCl

1 mM KCl

1 mM HCl

NDS2

3.80 ±0.57

0.26 ±0.04

5.51

13.4 ±0.7

MV2+

4.66 ±0.25

7.08 ±0.32

4.32

1.5 ±0.1

FL2

3.08 ±0.25

0.027 ±0.003



114.1 ±11.1

“—”: not determined To access the charge effect quantitatively, we also define here the selectivity coefficients, respectively, for anions (α1M/1mM) and cations (α1mM/1M),

ACS Paragon Plus Environment

Page 5 of 8

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry 1 M/1 mM 

J Anion 1 M KCl  J Anion 1 mM KCl 

(4)

1 mM/1 M 

J Cation 1 mM KCl  J Cation 1 M KCl 

(5)

According these two equations, the selectivity coefficient corresponds to the ratio of high flux versus low one, which is expected to be larger than unity. The selectivity coefficients obtained for NDS2, MV2+ and FL2 are 13.4, 1.5 and 114.1, respectively, as summarized in Table 1. A significantly small α for MV2+ indicates that the separation capacity of SNM toward cations by varying the electrolyte concentration is weak, in agreement with that reported previously by Adrien Plecis et al.51 By contrast, the membrane can effectively separate anions, with the selectivity coefficient one or two orders of magnitude larger than that of cations. Moreover, the  value of FL2 was ca. 8-fold larger than that of NDS2, because the electrostatic repulsion is more effective for big anions. Based on the molecular flux obtained in the singlecomponent permeation (see Table 1), the separation selectivity of MV2+ versus FL2 at a low electrolyte concentration according to the molecular charge is excellent, with an expected selectivity coefficient ( α  MV 2+ FL2    J MV2+ J FL2 ) as high

as 262. To prove it experimentally, two-component permeation experiments were performed by adding both 2.5 mM MV2+ and 2.5 mM FL2 into the feed solution (1 mM KCl was also dissolved in both feed and permeate solutions to keep the total ionic strength the same as that in the single-component permeation). According to the Fick diffusion (namely eq. 1), the respective molecular fluxes are expected to be one half of that in the single-component permeation experiments. However, as shown by the kinetic plots in Figure 5b (see also the UV-spectra in Figure S7), the flux of FL2 was increased by 16 times, whereas that of MV2+ reduced by twice, eventually yielding a selectivity coefficient of 8.4, which is significantly smaller than that expected, namely 262. This phenomenon was also confirmed by adding another oppositely charged probe in the course of single-component permeation experiment. As shown in Figure S8, the flux of FL2 was immediately increased by 14-fold after adding MV2+, whereas that of MV2+ was instantaneously decreased twice upon addition of FL2. The reason behind this phenomenon remains unexplored, which are probably associated with charge and steric effects in the ultrasmall channels. Its rationalization requires further and deep studies. Apart from the electrolyte concentration, the charge polarity of channel surface can also regulate the permselectivity of silica nanochannels. The pI of silica is ca. 3 ~ 4 as mentioned above, the channel surface will be positively charged at a solution pH below 3. As shown in Figure 5c, at pH 3 the flux of NDS2 became higher than that of MV2+. In comparison with the fluxes at nearly neutral pH and the similar electrolyte concentration, namely 1 mM KCl (see data in Table 1 and Figure S9 for details), the flux NDS2 was increased from 0.26 to 5.51 mol cm2 min1, while that of MV2+ was slightly decreased from 7.08 to 4.32 mol cm2 min1. This variation can be ascribed to the reversal of electrostatic interaction of ions with channel surface from repulsion to attraction for NDS2 and vice versa for MV2+. Similar variation of permeation selectivi-

ty induced by solution pH has also been observed previously with cysteine modified gold/polycarbonate membrane.9

Figure 6. (a) Schematic illustration of molecular separation driven by a dc voltage. The electrolyte was 1 mM KCl in both cells. (b) Kinetic permeation plots of MV2+ (solid points) and FL2 (open points) when applying a voltage (feed versus permeate) of +0.5 V (red), +0.8 V (blue) and +1.0 V (black). (c) Kinetic permeation plots of FL2 (solid points) and MV2+ (open points) when applying a voltage (feed versus permeate) of 0.5 V (red), 0.8 V (blue) and 1.0 V (black). Table 2. Permeation flux and selectivity coefficient for twocomponent permeation of MV2+ and FL2 driven by externally applied voltages Flux (mol cm2 min1) (MV2+/FL2)

Vapp / V MV2+

FL2

0

1.82  0.23

0.22  0.03

8.4  0.3

0.5

9.28

0.04

224.0

0.8

10.55





1.0

16.71





 0.5

0.013

1.40

107.7a

 0.8



1.24

a

 1.0



1.41

a

“—”: not determined; a (FL2/MV2+). On the other hand, a remarkably improved separation selectivity of SNM for mixture was achieved by biasing an appropriate voltage between the feed and permeate solution (as illustrated in Figure 6a). Figure 6b shows that the flux of MV2+ increased with the magnitude of positive voltage (feed versus permeate), whereas that of FL2 was significantly suppressed. When the voltage was higher than +0.8 V, FL2 in the permeate solution was hard to detect and the spectra was only dominated by the characteristic band of MV2+ at 257 nm (see Figure S10). At +1.0 V, the flux of MV2+ was enhanced by ca. 9 times in comparison with the pure diffusion flux (see values in Table 2). The enhancement can be ascribed to the external electric field force exerted by the applied positive voltage, which facilitated the migration of MV2+ from feed to permeate but hindered that of FL2. Note that the electroosmotic flow also contributed to the ion transport, which indeed has been observed but will be reported elsewhere soon later. Furthermore, as shown in Figure 6c, the inversion of the voltage,

ACS Paragon Plus Environment

Analytical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

namely applying a negative voltage (feed versus permeate), suppressed completely the transport of MV2+. It means that the electrostatic field force exerted by the channel wall (driving force) was neutralized and surpassed by the external electric field force (resisting force). However, the flux of FL2 was not significantly varied with the external voltage (see Figures 6c and S11), although the external electric field force favored the transport of FL2. It suggests that the electrostatic resisting force from charged nanochannels was pretty strong for FL2. Flux Comparison with Commercial and Other Membranes. The SNM is characterized by a high pore density, straight channel structure and ultrathin thickness, so a high molecular flux is anticipated. We experimentally compared the molecular flux across the SNM with the commercial dialysis membrane (6 kD – 8 kD, similar pore diameter with SNM). The electrolyte solution used was 1 M KCl, in order to screen the effect of EDL. Figure S12a shows the UV-visible spectra for the permeation of FL2 across the dialysis membrane. Figure S12b compared the permeation fluxes across the SNM and dialysis membrane normalized by the effective membrane area. Apparently, the flux across the SNM was higher than that across the dialysis membrane by ca. 830-fold (3.0810 versus 0.0037 mol cm2 min1, as shown in Table 3). Additionally, the molecular permeation across the dialysis membrane displayed an obvious lag time at the beginning while no any lag time was observed for SNM (see the inset of Figure S12b). The reason can be attributed to the ultrathin thickness of SNM (only about 94 nm). Note that the thickness of dialysis membrane is in the micrometer scale. Table 3. Comparison of permeation flux through SNM with the commercial available dialysis membrane and other membranes reported in literatures Flux (mol cm2 min1) Membrane FL2

NDS2

MV2+

3.08 ±0.25

3.80 ±0.57

4.66 ±0.25

0.0037 ±0.0004





Au/PC



0.0033

0.025

PET



~0.00092

~0.046

CNT





0.00023

SNM Dialysis membrane

“—”: no reported. PC: i.d. = 1.9 nm, 6 m;9 PET: i.d. ~ 18 nm, 12 m;52 CNT: i.d. = 6.7 nm, 5 – 10 m.53

In Table 3, we also compare the fluxes of NDS2 and MV2+ across the SNM with other types of membranes found in literatures. Apparently, the molecular flux through the SNM is much higher than others. For instance, the flux of MV2+ through the SNM is about 180 – 280 fold higher than that of Au plated polycarbonate membrane.

Page 6 of 8

tively evaluated the permselectivity and molecular permeability of SNMs. The molecular size-based sieving effect of SNMs was firstly demonstrated in single-component experiments, which showed that the restriction to transport through pores increased with increasing the molecular size. An excellent separation selectivity (αmin(MV2+/Cyt c) = 273) was obtained for their mixture. Since the SNMs was negatively charged, it allowed the selective transport of positive charges and hindered that of negative charges, which was especially obvious at a low electrolyte concentration (namely 1 mM KCl), in comparison with that at a high one, namely 1 M KCl. The selectivity was reversed in 1 mM HCl by changing the surface charge from negative to positive. Moreover, the selectivity coefficient can be significantly increased when external electric field was imposed and the flux of cations was able to be enhanced 9-times when applied +1.0 V (feed versus permeate). In comparison with commercially available membranes and those reported previously, the molecular flux across the SNM was found to about 2 orders of magnitude higher. This enhancement can be attributed to the ultrathin thickness and high pore/channel density. All these results suggest that the SNM is an excellent filter for highly selective and fast separation of molecules. We also believe that the SNM is very useful in chemical analysis, for instance gated sampling for microfluidics, and in fundamental studies to understand the mass transport in the nanoscale.

ASSOCIATED CONTENT Supporting Information The Supporting Information is available free of charge on the ACS Publications website: the synthesis and characterization of SNM; confirmation of permeation of MV2+ and FL2 through SNM by Fick diffusion; more UV spectra of single-component and two-component permeation; more kinetic plots of MV2+ and NDS2 at different pH and ionic strength; UV spectra results of ion transport through SNM under electrochemical control; and molecule permeation across commercial membrane. (PDF)

AUTHOR INFORMATION Corresponding Author *E-mail: [email protected]; Homepage: http://mypage.zju.edu.cn/binsu.

Notes The authors declare on competing financial interest.

ACKNOWLEDGMENT This work was supported by the Nature Science Foundation of China (21335001 and 21575126) and the Zhejiang Provincial Natural Science Foundation (LR14B050001).

REFERENCES (1) (2) (3) (4)

CONCLUSIONS

(5)

In summary, the SNM consisting of a high density of straight and uniform nanochannels enables the selective and fast molecular separation in terms of size and charge. We quantita-

(6) (7)

Zhang, H.; Tian, Y.; Jiang, L. Nano Today 2016, 11, 61-81. Warkiani, M. E.; Bhagat, A. A. S.; Khoo, B. L.; Han, J.; Lim, C. T.; Gong, H. Q.; Fane, A. G. ACS Nano 2013, 7, 1882-1904. Striemer, C. C.; Gaborski, T. R.; McGrath, J. L.; Fauchet, P. M. Nature 2007, 445, 749-753. Jirage, K. B.; Hulteen, J. C.; Martin, C. R. Science 1997, 278, 655658. Vlassiouk, I.; Apel, P. Y.; Dmitriev, S. N.; Healy, K.; Siwy, Z. S. Proc. Natl. Acad. Sci. USA 2009, 106, 21039 –21044. Baker, L. A.; Bird, S. P. Nature Nanotech. 2008, 3, 73-74. MacKinnon, R. Angew. Chem. Int. Ed. 2004, 43, 4265-4277.

ACS Paragon Plus Environment

Page 7 of 8 (8)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) (21) (22) (23) (24) (25) (26) (27) (28) (29) (30) (31)

Analytical Chemistry Fang, H.; Wan, R.; Gong, X.; Lu, H.; Li, S. J. Phys. D: Appl. Phys. 2008, 41, 103002-1030017. Lee, S. B.; Martin, C. R. Anal. Chem. 2001, 73, 768-775. Nishizawa, M.; Menon, V. P.; Martin, C. R. Science 1995, 268, 700-702. Hou, Z. Z.; Abbott, N. L.; Stroeve, P. Langmuir 2000, 16, 24012404. Savariar, E. N.; Krishnamoorthy, K.; Thayumanavan, S. Nature Nanotech. 2008, 3, 112-117. Yamauchi, Y.; Kimura, T. Chem. Commun. 2013, 49, 11424-11426. Yu, S.; Lee, S. B.; Kang, M.; Martin, C. R. Nano Lett. 2001, 1, 495498. Lee, S. B.; Mitchell, D. T.; Trofin, L.; Nevanen, T. K.; Soederlund, H.; Martin, C. R. Science 2002, 296, 2198-2200. Kohli, P.; Harrell, C. C.; Cao, Z.; Gasparac, R.; Tan, W.; Martin, C. R. Science 2004, 305, 984-986. Gaborski, T. R.; Snyder, J. L.; Striemer, C. C.; Fang, D. Z.; Hoffman, M.; Fauchet, P. M.; McGrath, J. L. ACS Nano 2010, 4, 6973-6981. Ileri, N.; Faller, R.; Palazoglu, A.; Letant, S. E.; Tringe, J. W.; Stroeve, P. Phys. Chem. Chem. Phys. 2013, 15, 965-971. Jeon, G.; Yang, S. Y.; Kim, J. K. J. Mater. Chem. 2012, 22, 14814. Chilkoti, A.; Dreher, M. R.; Meyer, D. E.; Raucher, D. Adv. Drug. Deliv. Rev. 2002, 54, 613-630. Adiga, S. P.; Jin, C. M.; Curtiss, L. A.; Monteiro-Riviere, N. A.; Narayan, R. J. 2009, 1, 568-581. Yang, S. Y.; Yang, J.-A.; Kim, E.-S.; Jeon, G.; Oh, E. J.; Choi, K. Y.; Hahn, S. K.; Kim, J. K. ACS Nano 2010, 4, 3817-3822. Surwade, S. P.; Smirnov, S. N.; Vlassiouk, I. V.; Unocic, R. R.; Veith, G. M.; Dai, S.; Mahurin, S. M. Nature Nanotech. 2015, 10, 459-464. Howorka, S.; Siwy, Z. Chem. Soc. Rev. 2009, 38, 2360-2384. O'Hern, S. C.; Jang, D.; Bose, S.; Idrobo, J. C.; Song, Y.; Laoui, T.; Kong, J.; Karnik, R. Nano Lett. 2015, 15, 3254-3260. Tong, H. D.; Jansen, H. V.; Gadgil, V. J.; Bostan, C. G.; Berenschot, E.; Rijn, C. J. M. v.; Elwenspoek, M. Nano Lett. 2004, 4, 283-287. Kwak, E.-S.; Henzie, J.; Chang, S.-H.; Gray, S. K.; Schatz, G. C.; Odom, T. W. Nano Lett. 2005, 5, 1963-1967. Price, P. B.; Walker, R. M. J. Appl. Phys. 1962, 33, 3407-3412. Martin, C. R. Science 1994, 266, 1961-1966. Apel, P. Radiat. Meas. 2001, 34, 559-566. Jiang, Y.; Gao, J.; Guo, W.; Jiang, L. Chem. Commun. 2014, 50, 14149-14152.

(32) (33) (34) (35) (36) (37) (38) (39) (40) (41) (42) (43) (44) (45) (46) (47) (48) (49) (50) (51) (52) (53)

Masuda, H.; Fukuda, K. Science 1995, 268, 1466-1468. Li, A. P.; Muller, F.; Birner, A.; Nielsch, K.; Gosele, U. J. Appl. Phys. 1998, 84, 6023-6026. El-Safty, S.; Shenashen, M. A. Anal. Chim. Acta 2011, 694, 151161. El-Safty, S.; Shahat, A.; Awual, M. R.; Mekawy, M. J. Mater. Chem. 2011, 21, 5593. Lee, S.; Zhang, Y.; White, H. S.; Harrell, C. C.; Martin, C. R. Anal. Chem. 2004, 76, 6108-6115. Yamada, K.; Gasparac, R.; Martin, C. R. J. Electrochem. Soc. 2004, 151, E14-E19. Yu, H.; Qiu, X.; Moreno, N.; Ma, Z.; Calo, V. M.; Nunes, S. P.; Peinemann, K. V. Angew. Chem. Int. Ed. 2015, 127, 1-6. Jackson, E. A.; Hillmyer, M. A. ACS Nano 2010, 4, 3548-3553. Holt, J. K.; Park, H. G.; Wang, Y.; Stadermann, M.; Artyukhin, A. B.; Grigoropoulos, C. P.; Noy, A.; Bakajin, O. Science 2006, 312, 1034-1037. Hinds, B. J.; Chopra, N.; Rantell, T.; Andrews, R.; Gavalas, V.; Bachas, L. G. Science 2004, 303, 62-65. Lin, X.; Yang, Q.; Ding, L.; Su, B. ACS Nano 2015, 9, 11266– 11277. Yan, F.; Lin, X.; Su, B. Analyst 2016, 141, 3482-3495. Teng, Z.; Zheng, G.; Dou, Y.; Li, W.; Mou, C. Y.; Zhang, X.; Asiri, A. M.; Zhao, D. Angew. Chem. Int. Ed. 2012, 51, 2173-2177. Vinu, A.; Streb, C.; Murugesan, V.; Hartmann, M. J. Phys. Chem. B 2003, 107, 8297-8299. Martin, F.; Walczak, R.; Boiarski, A.; Cohen, M.; West, T.; Cosentino, C.; Shapiro, J.; Ferrari, M. J. Controlled Release 2005, 102, 123-133. Bard, A. J.; Faulkner, L. R. Electrochemical Methods: Fundamentals and Applications; Wiley, 1980. Majumder, M.; Chopra, N.; Hinds, B. J. ACS Nano 2011, 5, 3867– 3877 Chen, Y.; Wu, L.; Zhu, J.; Shen, Y.; Gan, S.; Chen, A. J. Porous Mater. 2010, 18, 251-258. Corti, H. R.; Frank, G. A.; Marconi, M. C. J. Solution Chem. 2008, 37, 1593-1608. Plecis, A.; Schoch, R. B.; Renaud, P. Nano Lett. 2005, 5, 11471155. Nguyen, Q. H.; Ali, M.; Nasir, S.; Ensinger, W. Nanotech. 2015, 26, 485502. Majumder, M.; Chopra, N.; Hinds, B. J. J. Am. Chem. Soc. 2005, 127, 9062-9070.

ACS Paragon Plus Environment

Analytical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

TOC graphic

ACS Paragon Plus Environment

Page 8 of 8