Nanoporous silica colloidal membranes suspended in glass

Nanoporous silica colloidal membranes suspended in glass

Andrew K. Bohaty, Alexis E. Abelow & Ilya Zharov

Journal of Porous Materials ISSN 1380-2224 Volume 18 Number 3 J Porous Mater (2011) 18:297-304 DOI 10.1007/s10934-010-9379-z

1 23

Your article is protected by copyright and all rights are held exclusively by Springer Science+Business Media, LLC. This e-offprint is for personal use only and shall not be selfarchived in electronic repositories. If you wish to self-archive your work, please use the accepted author’s version for posting to your own website or your institution’s repository. You may further deposit the accepted author’s version on a funder’s repository at a funder’s request, provided it is not made publicly available until 12 months after publication.

1 23

Author's personal copy J Porous Mater (2011) 18:297–304 DOI 10.1007/s10934-010-9379-z

Nanoporous silica colloidal membranes suspended in glass Andrew K. Bohaty • Alexis E. Abelow Ilya Zharov

•

Published online: 21 April 2010 Ó Springer Science+Business Media, LLC 2010

Abstract We prepared silica colloidal membranes suspended in glass openings and containing no major mechanical defects. The surface of these colloidal membranes was modified with amine groups. The diffusion rate of Fe(bpy)2? 3 through the suspended amine-modified colloidal membranes was attenuated by adding acid to the solution. The amine-modified colloidal membranes displayed an average selectivity (the ratio of diffusion rates in the absense and presence of the acid) of 2.6 for Fe(bpy)2? 3 . This selectivity is believed to result from the electrostatic repulsion between the protonated amine-modified membrane surface and positively charged Fe(bpy)2? and was 3 confirmed by observing no change in (1) the diffusion rate of Fe(bpy)2? through an unmodified suspended colloidal 3 membrane, and (2) the diffusion rate of a neutral molecule through the amine-modified colloidal membrane with and without the acid present in solution. Keywords Colloidal crystal  Membrane  Nanopore  Molecular transport

1 Introduction Nanofluidics is a relatively new but highly promising field dealing with molecular transport phenomena at the nanometer size confinement and with devices utilizing these phenomena [1]. The promise of nanofluidic devices stems from the advantages provided by the miniaturized components and systems for separations [2] and sensing [3]. A. K. Bohaty  A. E. Abelow  I. Zharov (&) Department of Chemistry, University of Utah, 315 South 1400 East, Salt Lake City, UT 84112, USA e-mail: [email protected]

The reduction in physical dimensions leads to the reduction of sample size and material consumption, and shortens analysis time due to enhanced mass transport. The unique character of nanofluidics results from the fact that the electric double layer (EDL), characterized by the inverse Debye length (K), the cross-sectional dimension of a nanochannel and the size of molecules become comparable in these systems [4]. Changing the EDL or channel dimension can thus significantly affect the molecular transport. Reduction of channel size down to the nanoscale also causes a considerable increase in a surface-to-volume ratio. As a result, surface processes are more pronounced and the molecular transport can be effectively controlled by the surface properties of the nanochannel. Nanofluidic architectures provide a great addition to microfluidic devices for applications in chemical analysis [1]. The pre-concentration [5, 6] and separation [7, 8] capabilities provided by the nanofluidic elements within the microfluidic architectures expand the ability to perform complex analytical operations on mass-limited samples while retaining sample integrity. Single nanofluidic channels or nanopore arrays sandwiched between microfluidic channels can also serve as nanovalves for selective sample injection and collection after electrophoretic separation [5]. Finally, nanochannels offer rapid mixing based on convective flow from nanochannels to the microchannel [9]. To create nanofluidic architectures, a variety of nanoporous materials have been used including track-etched polycarbonate [10–15], anodized alumina [16–18], silicon nitride[19] and carbon nanotubes [20, 21]. A nanoporous material that has been the focus of our work is silica colloidal crystals [22–28]. Silica colloidal crystals form when nanometer-sized silica spheres self-assemble into a closepacked face-centered cubic (fcc) lattice. The three-dimensional interconnected pores found in the colloidal crystal

123

Author's personal copy 298

form highly ordered arrays [29]. The pore size is dependent on the silica sphere size, namely it is 15% the size of the silica sphere used [30]. An additional characteristic of silica colloidal crystals that makes them an attractive nanoporous material is that the diffusive flux of molecules normal to the (111) plane is only ca. 10 times smaller than that in the free solution, independent of the size of spheres used to form the colloidal crystal [31]. Using colloidal crystals creates a new paradigm within the field of nanoporous materials because they (1) form through selfassembly, (2) allow straightforward control of nanopore size in a broad range, (3) possess exceptionally high molecular throughput, and (4) offer a large variety of surface chemistries. Interest in the incorporation of colloidal crystals in analytical devices is reflected in the fact that several methods of patterning monodisperse nanospheres on top of various substrates have been recently reported. They include solvent assisted micromolding in capillaries [32], electrostatic assembly [33], and growth of colloidal crystals through entropy-driven crystallization of micro-spheres on a patterned poly(methylmethacrylate) substrate [34]. Microfluidic and spin-coating assembly were used to prepare planarized colloidal crystals patterned within oriented silicon substrates [35, 36]. Colloidal crystals inside a microchannel network were prepared using a combination of electrocapillary forces and evaporation induced selfassembly [37]. High-quality colloidal crystals were fabricated in hydrophilic trenches of silicon wafers [38]. Colloidal crystals have been formed inside microfluidic channels using unmodified silica spheres [39, 40]. Silica colloidal crystals suspended in copper wire for the use in photonics have been also reported [41]. Our interest is in the incorporation of surface-modified silica colloidal as nanofluidic components inside microfluidic channels. As the initial step in this direction, we have demonstrated that openings in silicon substrates could be used to support a silica colloidal membrane [42]. These suspended colloidal crystals ranged in thickness from 40 to 300 lm and both SEM images and diffusion measurements showed that the crystals contained no major cracks or defects. We have also shown that thin silica colloidal films and thick free-standing colloidal membranes [43] can be surface modified with both small molecules [22–25] and polymer brushes [26, 27] to control the transport of molecules through the nanopores. The colloidal films modified with amines demonstrated good permselectivity for cationic species in both aqueous and non-aqueous solutions [22, 23]. The goal of the present work was to establish the possibility of incorporating silica colloidal crystals into glass channels, to evaluate the possibility of the surface modification and the stability of the resulting nanofluidic

123

J Porous Mater (2011) 18:297–304

component. Thus, we describe the preparation of 150-lmthick silica colloidal crystals suspended in round openings in glass, their surface-modification with amines, and the permselective behaviour of the resulting nanofluidic devices as a function of pH.

2 Experimental section 2.1 Materials Butylamine, (99%, Aldrich), dansyl chloride, (99%, Aldrich), 3-aminopropyltriethoxysilane, APTES (98%, Aldrich), 2,2’dipyridyl (99%, Aldrich), ammonium iron(II) sulfate hexahydrate (99%, Aldrich), ammonium hexafluorophosphate (99.5%, Acros), ammonium hydroxide, NH4OH (30%, Mallinckrodt), trifluoroacetic acid, TFA (99%, Aldrich), tetrabutylammonium hexafluorophosphate, TBAP (98%, Aldrich), and TEOS (99.9%, Alfa Aesar) were all used as received. Iron trisbipyridine hexafluorophosphate was synthesized according to the previously reported procedures [44, 45]. All water used was 18 MX cm water and was obtained from a Barnstead ‘‘E-pure’’ water purification system. All ethanol used was 200 proof. Acetonitrile (HPLC grade, Mallinckrodt) and triethylamine, TEA (100% J. T. Baker) were freshly distilled from calcium hydride. Column chromatography was carried ˚ , 230–400 mesh under slight out on silica gel (Silicycle) 60 A pressure. Silica TLC was performed on aluminium foil plates coated with 0.1 mm Merck silica gel 60 F254. TLC plates were visualized using UV or potassium permanganate. NMR spectra were recorded on Varian VXL-300 MHz in CDCl3 at 300 MHz (1H NMR) or 75 MHz (13C NMR). 1H NMR were referenced to residual CHCl3 (d 7.27 ppm) and 13C NMR were referenced to CHCl3 (d 77.23 ppm). A Branson 1510 sonicator was used for all sonication. All scanning electron microscopy (SEM) images were taken using a Hitachi S3000N instrument and all samples were coated with gold. UV/Vis measurements were performed using an Ocean Optics USB2000 or USB4000 instrument. 2.2 Glass substrates Glass microscope slide covers (2.2 9 2.2 cm2) 0.15-mmthick with a single round opening with a diameter ranging from 250 to 450 lm were purchased from Technical Glass, Inc.

2.3 Preparation of 239 nm silica spheres Silica spheres were prepared according to the previously reported procedures [46, 47]. All glassware was cleaned

Author's personal copy J Porous Mater (2011) 18:297–304

with 18 MX cm water and dried prior to use. Silica spheres were prepared by dissolving TEOS (24 g, 0.10 mol, 0.20 M) in absolute ethanol (250 mL total volume) and dissolving NH4OH (13 mL, 0.40 M final concentration of ammonia) and water (140 g, 8.0 mol, 17 M) in absolute ethanol (250 mL total volume). These two solutions were simultaneously poured into a 1 L Erlenmeyer flask and vigorously stirred at room temperature. After 30 min of stirring the solution became cloudy indicating the silica sphere formation and the reaction was allowed to proceed for 24 h. The silica spheres were isolated by centrifugation in conical tubes at 1,1639g for 15 min. After all the spheres were collected as pellets at the bottom of the tubes, the supernatant was decanted. The silica spheres were purified by suspending the spheres in 10 mL of absolute ethanol and sonicating for 30 min. The colloidal suspension was then centrifuged for 15 min at 1,1639g. The supernatant was decanted and the purification steps were repeated 5 more times. After the final centrifugation, the supernatant was decanted and the silica spheres were dried in a stream of nitrogen for 2 h. SEM images were taken and 100 different individual silica spheres were measured to determine the average size to be 239 ± 17 nm. 2.4 Suspended colloidal membranes A single glass substrate was placed vertically into a 10 mL beaker (22 mm inner diameter, 33 mm height, Kimble) containing a 10 mL dispersion of 239 nm silica spheres in absolute ethanol (3.0 wt%) [42]. The 10 mL beakers containing the substrate were placed in a vibration free environment under a 190 9 100 crystallization dish raised 6 cm off the bench top using four evenly distributed supports. The solvent was allowed to evaporate completely at room temperature, which would usually proceed overnight. Colloidal membranes were suspended inside the opening in the glass substrate with *80% success rate. 2.5 Surface modification of suspended colloidal membranes A colloidal membrane comprised of 239 nm silica spheres suspended in a glass substrate was placed into a 30 mL wide-mouth jar containing a mixture of 15 mL of dry acetonitrile and 0.20 mL of APTES (0.85 mmol, 0.06 M). The reaction was set up under the nitrogen atmosphere. After the substrate was added, the jar was covered with parafilm and the colloidal membrane was allowed to stand in the APTES solution for 24 h. The amine-modified suspended colloidal membranes were then rinsed in 10 mL of acetonitrile for 2 h two times.

299

2.6 Synthesis of N-butyl-5(dimethylamino)naphthalene-1-sulfonamide (1) N-Butyl-5-(dimethylamino)naphthalene-1-sulfonamide (1) was synthesized according to the previously reported procedures (Fig. 1) [48, 49]. Dansyl chloride (0.25 g, 0.93 mmol) was dissolved in 25 mL of dry acetonitrile in a 50 mL round bottom flask equipped with nitrogen. Butylamine (1.4 mL, 1.3 mmol) was added to the yellow solution and the reaction mixture was stirred under nitrogen for 1.5 h. When butylamine was added the solution changed from yellow to green. After 1.5 h the solvent was removed in vacuo and the residue was purified with flash chromatography using 80:20 hexane/ethyl acetate. Pure 1 was isolated in 91% yield. 1H NMR (300 MHz, CDCl3): d 0.73 (t, 3 H, J = 7.0 Hz), 1.98 (m, 2 H), 2.35 (m, 2 H), 3.85 (m, 8 H), 5.05 (t, 1 H, J = 5.1 Hz), 7.18 (d, 1 H, J = 7.3 Hz), 7.54 (m, 2 H), 8.26 (d, 1 H, J = 7.1 Hz), 8.33 (d, 1 H, J = 8.3 Hz), 8.54 (d, 1 H, J = 8.3 Hz). 13C NMR (75 MHz, CDCl3): d 13.5, 19.7, 31.7, 43.1, 45.5, 115.3, 118.9, 123.3, 128.5, 129.6, 129.8, 123.0, 130.4, 135.0, 152.0. 2.7 Diffusion measurements Diffusion through the suspended colloidal membranes in glass was measured by placing a 0.15-mm-thick microscope slide cover supporting a single suspended colloidal membrane made with 239 nm silica spheres between two connected 1 cm quartz cuvettes. One of the cuvettes contained 4 mL of the feed solution, either 9.55 mM solution of Fe(bpy)2? 3 in acetonitrile or 8.53 mM solution of 1 in acetonitrile, and the other cuvette contained 4 mL of pure acetonitrile. The microscope slide cover was placed between two Kalrez o-rings to prevent leaking and a clamp was used to hold the cuvettes, o-rings and the glass substrate in place. Each cuvette contained a stir bar and was covered with ParafilmÒ to prevent evaporation. The cuvette containing the pure solvent was then placed in the cuvette holder between two fiberoptic cables and the solution was blanked. Both solutions were stirred and the analyte diffusion rate was monitored by recording the absorbance in the receiving cuvette at 520 nm for Fe(bpy)2? 3 and 255 or 283 nm for 1 for at least 18 h. When selectivity

Fig. 1 Synthesis of 1

123

Author's personal copy 300

J Porous Mater (2011) 18:297–304

measurements were conducted the solutions in both cuvettes contained either 50 mM acetic acid or 50 mM TFA. For diffusion measurements of 1 performed with solutions containing 50 mM TFA the absorption at 255 nm was monitored, while for the measurements conducted without the TFA present the wavelength at 283 nm was monitored. The salt concentration was varied from 10 to 50 mM using TBAP. The absorbance of the analyte molecules was measured using Ocean Optics USB2000 or USB4000 spectrometer. Data points were recorded every 300 s with an initial delay of 300 s. All measurements were repeated in triplicate for each suspended colloidal membrane.

3 Results and discussion 3.1 Suspended colloidal membranes in glass In order to model the incorporation of suspended silica colloidal membranes in glass microfluidic devices we decided to use 150-lm-thick glass as a substrate. Figure 2 shows an SEM image of an unfilled opening in a glass microscope slide cover that was used as a substrate to support a silica colloidal membrane. This substrate was placed vertically into a dispersion of 239 nm silica spheres in absolute ethanol (3.0 wt%) and the solvent was allowed to evaporate completely at room temperature overnight. Colloidal membranes were suspended inside the opening in the glass substrate with *80% success rate. Figure 3 shows a light microscope image of a silica colloidal membrane suspended in openings in glass microscope slide covers. The image shows that the

Fig. 3 Light microscope image of a representative colloidal membrane suspended in glass with a 455 9 364 lm opening (size bar is 100 lm)

colloidal membrane contains no major cracks through the middle of the membrane. There are small cracks visible at the periphery of the colloidal membrane, which could result from the edge of the colloidal membrane not connecting tight to the roughly drilled opening in the microscope slide. To examine the packing of the silica spheres in the glass openings, SEM images of the filled openings in glass were obtained. Figure 4a show a typical image where the silica spheres are packed in an fcc orientation as was seen for the colloidal membranes in the silicon wafer. The SEM images showed point defects in the colloidal membranes, but no major cracks or other mechanical defects that would extend through the membrane. The SEM images of both sides of the slides showed the membranes to be flush with the glass or slightly elevated above the glass substrate surface. 3.2 Diffusion measurements through unmodified colloidal membranes suspended in glass

Fig. 2 SEM image of a representative glass microscope slide with a 455 9 364 lm opening (size bar is 250 lm)

123

Diffusion measurements provide a way of calculating the thickness of the suspended colloidal membranes and testing the membranes for cracks and pinholes. Diffusion through colloidal crystals has been the subject of numerous theoretical treatments [50] and of a recent investigation by electrochemical methods [31]. To determine if there are any defects in the colloidal membranes, the observed diffusion rate, RD of Fe(bpy)2? 3 was compared to the calculated diffusion rate for a silica colloidal membrane that is

Author's personal copy J Porous Mater (2011) 18:297–304

301

Fig. 4 SEM images of suspended colloidal membranes in glass microscope slide covers comprised of 239 nm silica spheres. Left: unmodified (size bar is 2 lm). Right: modified with APTES (size bar is 2.5 lm)

RD ¼ Jcolloid  S

ð1Þ

where Jcolloid is the molecular flux across the colloidal membrane and S is its area [51]. The molecular flux can be expressed as Jcolloid ¼

DC e   Dsol L s

ð2Þ

where DC is the concentration of the dye, L is the thickness of the membrane (0.015 cm), Dsol is the diffusion coefficient in solution (1.0 9 10-5 cm2 s-1 for Fe(bpy)2? in 3 acetonitrile), e, the void fraction, is 0.26, and s, the tortuosity, is ca. 3.0 [31]. The diffusion coefficient of Fe(bpy)2? was calculated 3 using the limiting current obtained from a known size microelectrode and using the equation ilim ¼ 4nFaDsol C

ð3Þ

where n is the number of electrons transferred per molecule, F is Faraday’s constant, and a is the electrode radius and C is the concentration of the diffusing species [52]. The diffusion rate of Fe(bpy)2? 3 through a silica colloidal membrane with an area of 1.30 9 10-3 cm2 (the opening shown in Fig. 2) was 1.14 ± 0.03 9 10-12 mol s-1 cm-2. The diffusion rate for the same size membrane can be estimated as 0.72 9 10-12 mol s-1 cm-2. The observed rate is in a reasonable agreement with the estimated rate, suggesting that the colloidal membrane did not contain major defects, as also indicated by SEM results. A factor that may have led to the higher observed diffusion rate is the presence of small cracks around the periphery of the colloidal membranes. The flux through the suspended colloidal membranes is 1–2 orders of magnitude higher than those observed for polycarbonate membranes [10–13].

colloidal membrane, as confirmed by SEM (Fig. 4b). Our previous experiments demonstrated that allowing the reaction to proceed for longer periods of time (i.e., 48 h) produces a polymeric film that cover the entire membrane surface and blocks the nanopores [22]. The membranes modified for shorter periods of time did not show as good a selectivity as the membranes modified for 24 h. When the suspended silica colloidal membranes were modified with amines the average diffusion rate of Fe(bpy)2? decreased from 1.27 ± 0.03 9 10-12 mol s-1 3 to 1.01 ± 0.02 9 10-12 mol s-1 (Fig. 5). This clearly demonstrates that the nanopores were covered with a polymeric film reducing the pore size and the void fraction, and thus decreasing the flux through the membrane [26]. However, this film is not very thick based on the relatively small (*20%) decrease in the molecular flux of Fe(bpy)2? 3 . Next, the effect an acid had on the diffusion rate of Fe(bpy)2? was studied. When 50 mM acetic acid was 3 present in solution the diffusion rate of Fe(bpy)2? 3 decreased by *30%, to 7.07 ± 0.07 9 10-13 mol s-1. This drop in diffusion rate is likely due to the electrostatic

6E-08

mol of Fe(bpy)32+

150 lm thick, i.e., the thickness of the substrate. The diffusion rate through a colloidal crystal is calculated as

4E-08

2E-08

0

3.3 Amine-modified suspended colloidal membranes

0.E+00

3.E+04

5.E+04

time, s

The surface of the silica colloidal membranes suspended in glass was modified by treatment with APTES for 24 h. After the surface modification with the silane there was no evidence of a polymeric film covering the surface of the

Fig. 5 Representative plots for diffusion of 9.55 mM Fe(bpy)2? 3 through suspended colloidal membrane in glass, unmodified (circles), modified with APTES for 24 h with no acid present (triangles), with 50 mM acetic acid (squares) and 50 mM TFA present (crosses)

123

Author's personal copy 302

J Porous Mater (2011) 18:297–304

repulsion of the positively charged Fe(bpy)2? from the 3 protonated amine surface. When a stronger acid, 50 mM TFA, was used the diffusion rate decreased by *60% to 4.21 ± 0.88 9 10-13 mol s-1. These diffusion rates were obtained from measurements for three different suspended colloidal membranes. Similar decreases in diffusion rate were observed for four additional suspended membranes modified with APTES when in the presence of TFA. Our observations described above are similar to those reported earlier by our group for amine-modified colloidal films [22, 23]. The larger drop in current in the presence of TFA is believed to result from a larger positive surface charge on the surface of amine-modified colloidal membrane. We have shown that the pKa of the protonated amines surfacebound to a colloidal film in aqueous solution is 5.7 [22], which is lower than the solution pKa of alkyl ammonium ions (*10) [53]. This drop in pKa is due to the electrostatic repulsion between the neighbouring alkyl ammonium groups, resulting in their easier deprotonation [53]. Since the pKa of the surface-bound amines is only ca. one pKa unit greater than the pKa of acetic acid (4.76) [54], the amine surface is not expected to be fully protonated, leading to a small decrease in the diffusion rate. When TFA is used (pKa 0.6) [22] the amine-modified surface is expected to be fully protonated leading to a greater decrease in the diffusion rate. Although this quantitative reasoning is used to explain the change in diffusion rate in aqueous media, our results suggest that amine-modified membranes qualitatively behave in a similar fashion in acetonitrile. The diffusion rates could be restored to the original value by soaking the colloidal membranes in 0.1 M TEA for 3 h. Figure 6 shows the reversibility of the diffusion rate of Fe(bpy)2? before and after the protonation with 3 TFA. When membranes were not soaked with TEA before

RD of Fe(bpy)32+, mol·s-1

1.E-12 1.E-12 8.E-13 6.E-13 4.E-13 2.E-13 No TFA

50 mM TFA

No TFA

50 mM TFA

No TFA

50 mM TFA

No TFA

Fig. 6 Diffusion rate, RD, of Fe(bpy)2? with and without 50 mM 3 TFA present through an amine modified suspended colloidal membrane in glass

123

performing the diffusion measurement in the absence of the acid, the diffusion rates were *30% lower than the initial rates observed, indicating that the colloidal surface was still partially positively charged. These results provide additional evidence that the change in the diffusion rate results from the electrostatic repulsion of the positively charged permeant by the protonated surface. The acid-controlled selectivity of the transport of Fe(bpy)2? through the amine-modified silica colloidal 3 membrane can be defined as a ratio of the diffusion rates observed for the membrane in the absence and in the presence of TFA. The selectivity of 2.6 was calculated for Fe(bpy)2? 3 . This value is smaller than the selectivity values found for cations diffusing through polycarbonate membranes that have shown a selectivity of 6.9 for Ru(bpy)2? 3 . However, the latter measurements were performed through membranes containing pores with 3.0 nm inner diameter, which is about 13 times smaller than the pores present in our colloidal membranes (38 nm for 239 nm silica spheres) [55]. These smaller pores allow the surface charge to interact stronger with the molecules diffusing through the pores, leading to a greater selectivity. An additional factor that may have led to the smaller selectivity values is the presence of small cracks around the periphery of the colloidal membranes (see Fig. 3). The diffusion rate of Fe(bpy)2? in the presence of 3 50 mM TFA as a function of supporting electrolyte (TBAP) concentration was studied as well, but the colloidal membranes were not stable at salt concentrations higher than 50 mM and fell out of the glass support. The entire concentration range (0–500 mM) could not be studied, but measurements at lower electrolyte concentrations showed that as the salt concentration increased the diffusion rate increased. The increase in diffusion rate in the presence of the supporting electrolyte results from screening the charge on the surface of the colloidal nanopores allowing more Fe(bpy)2? 3 through the charged membrane [22, 23]. To further demonstrate that the selectivity observed for the amine-modified silica colloidal membranes indeed results from the electrostatic repulsion of Fe(bpy)2? 3 by the positively charged colloidal nanopore surface in the presence of the acid, the diffusion rates of a neutral molecule 1 in acetonitrile with and without the acid present was studied, as well as the diffusion rate of Fe(bpy)2? 3 through the unmodified suspended colloidal membrane in glass with and without TFA present. The diffusion of rate of 8.55 mM, obtained for 1 in the absence of TFA (Fig. 7), did not change when 50 mM TFA was added. This indicated that molecule 1 was not being electrostatically repelled from the charged surface. When the diffusion rate of Fe(bpy)2? through an unmodified 3 colloidal membrane suspended in glass was measured no change in the rate was observed when 50 mM TFA was

Author's personal copy J Porous Mater (2011) 18:297–304

303

9E-08

mol of 1

6E-08

3E-08

0 0.E+00

2.E+04

4.E+04

6.E+04

time, s Fig. 7 Representative plots of diffusion rates of 1 through the aminemodified colloidal membrane suspended in glass without (crosses) and with (triangles) TFA present

added. This result was expected since there are no amines present on the surface of the colloidal membrane to become charged and electrostatically repel the charged molecule from entering the nanopores. Also, these membranes showed no change in diffusion rate of Fe(bpy)2? after 3 exposure to TEA, indicating that triethylamine was not affecting the measurements. These observations were obtained for three different silica colloidal membranes suspended in glass.

4 Conclusions We have described the preparation of silica colloidal membranes suspended in glass openings as a model of silica colloidal crystal incorporation as nanofluidic components in microfluidic devices. The diffusion rate of Fe(bpy)2? through the colloidal crystal membranes sus3 pended in glass comprised was in good agreement with the calculated value suggesting that the membranes contain no major mechanical defects, although small defects have been observed at the periphery of the membranes. The surface of these colloidal membranes was modified with amine groups. The aminated membranes are reasonably robust but break at salt concentrations above 50 mM. The diffusion of Fe(bpy)2? through amine-modified colloidal 3 membranes could be controlled by adding acid to the solution. The amine-modified colloidal membranes displayed an average selectivity of 2.6 for Fe(bpy)2? 3 after the addition of an acid. This selectivity is believed to result from the electrostatic repulsion between the protonated amine-modified colloidal membrane surface and positively charged Fe(bpy)2? 3 . This mode of selectivity was confirmed

by observing no change in the diffusion rate of Fe(bpy)2? 3 through an unmodified suspended colloidal crystal membrane in glass with and without acid present. Also, the diffusion rate of a neutral molecule through the aminemodified colloidal membrane was unaltered in the presence of TFA. The suspended colloidal membranes are promising as a nanoporous system to control the diffusion of ions in microfluidic devices. However, work needs to be done to improve the tightness of the contact between the colloidal crystal and the glass surface, and to increase the robustness of these membranes if they are to be used in systems where pressure and salt concentration are of importance. To address the former concern, glass substrates with smoother and better defined openings will have to be used. To increase the robustness of the membranes we will grow polymers on the colloidal surface or will modify the membrane surface with a cross-linking agent that would chemically link the silica spheres to one another. These issues are currently under investigation in our laboratory. Acknowledgments This work was supported by the National Science Foundation CAREER Award (CHE-0642615).

References 1. A. Piruska, M. Gong, J.V. Sweedler, P.W. Bohn, Chem. Soc. Rev. 39, 1060 (2010) 2. M.E. Davis, Science 417, 813 (2002) 3. H. Bayley, C.R. Martin, Chem. Rev. 100, 2575 (2000) 4. P.T. Taney, J.-R. Butuille, T.J. Pinnavai, Chem. Adv. Mater., ed. by L.V. Interrante, M.J. Hampden-Smith (Wiley-VCH, New York, 1998), p. 329 5. T.C. Kuo, L.A. Sloan, J.V. Sweedler, P.W. Bohn, Langmuir 17, 6298 (2001) 6. A.N. Chatterjee, D.M. Cannon Jr, E.N. Gatimu, J.V. Sweedler, N.R. Aluru, P.W. Bohn, J. Nanopart. Res. 7, 507 (2005) 7. Y. Zhang, A.T. Timperman, Analyst 128, 537 (2003) 8. T.-C. Kuo, D.M. Cannon, M.A. Shannon, P.W. Bohn, J.V. Sweedler, Sens. Actuators A 102, 223 (2003) 9. T.-C. Kuo, H.-K. Kim, D.M. Cannon Jr, M.A. Shannon, J.V. Sweedler, P.W. Bohn, Angew. Chem. Int. Ed. 43, 1862 (2004) 10. C.R. Martin, M. Nishizawa, K. Jirage, M. Kang, J. Phys. Chem. B 105, 1925 (2001) 11. M. Nishizawa, V.P. Menon, C.R. Martin, Science 268, 700 (1995) 12. C.R. Martin, M. Nishizawa, K. Jirage, M. Kange, S.B. Lee, Adv. Mater. 13, 1351 (2001) 13. K.B. Jirage, J.C. Hulteen, C.R. Martin, Anal. Chem. 71, 4913 (1999) 14. K.-Y. Chun, P. Stroeve, Langmuir 17, 5271 (2001) 15. Z. Hou, N.L. Abbott, P. Stroeve, Langmuir 16, 2401 (2000) 16. S. Kipke, G. Schmid, Adv. Funct. Mater. 14, 1184 (2004) 17. I. Vlassiouk, A. Krasnoslobodtsev, S. Smirnov, M. Germann, Langmuir 20, 9913 (2004) 18. S.W. Lee, H. Shang, R.T. Haasch, V. Petrova, G.U. Lee, Nanotechnology 16, 1335 (2005) 19. M. Wanunu, A. Meller, Nano Lett. 7, 1580 (2007)

123

Author's personal copy 304 20. B.J. Hinds, N. Chopra, T. Rantell, R. Andrews, V. Gavalas, L.G. Bachas, Science 303, 62 (2004) 21. P. Nednoor, N. Chopra, V. Gavalas, L.G. Bachas, B. Hinds, Chem. Mater. 17, 3595 (2005) 22. M.R. Newton, A.K. Bohaty, H.S. White, I. Zharov, J. Am. Chem. Soc. 127, 7268 (2005) 23. M.R. Newton, A.K. Bohaty, Y. Zhang, H.S. White, I. Zharov, Langmuir 22, 4429 (2006) 24. J. Cichelli, I. Zharov, J. Am. Chem. Soc. 128, 8130 (2006) 25. J. Cichelli, I. Zharov, J. Mater. Chem. 17, 1870 (2007) 26. O. Schepelina, I. Zharov, Langmuir 22, 10523 (2006) 27. O. Schepelina, I. Zharov, Langmuir 23, 12704 (2007) 28. S. Zheng, E. Ross, M.A. Legg, M.J. Wirth, J. Am. Chem. Soc. 128, 9016 (2006) 29. S. Wong, V. Kitaev, G.A. Ozin, J. Am. Chem. Soc. 125, 15589 (2003) 30. H. Zhang, M.J. Wirth, Anal. Chem. 77, 1237 (2005) 31. M.R. Newton, K.A. Morey, Y. Zhang, R.J. Snow, M. Diwekar, J. Shi, H.S. White, Nano. Lett. 4, 551 (2004) 32. P. Yang, T. Deng, D. Zhao, P. Feng, D. Pine, B.F. Chmelka, G.M. Whitesides, G.D. Stucky, Science 282, 2244 (1999) 33. J. Aizenberg, P.V. Braun, P. Wiltzius, Phys. Rev. Lett. 84, 299 (2000) 34. K.-H. Lin, J.C. Crocker, V. Prasad, A. Schofield, D.A. Weitz, T.C. Lubensky, A.G. Yodh, Phys. Rev. Lett. 85, 1770 (2000) 35. G.A. Ozin, S.M. Yang, Adv. Funct. Mater. 11, 95 (2001) 36. S.M. Yang, G.A. Ozin, Chem. Commun. 24, 2507 (2000) 37. J.Y. Shiu, C.-W. Kuo, P. Chen, J. Am. Chem. Soc. 126, 8096 (2004)

123

J Porous Mater (2011) 18:297–304 38. J. Ye, R. Zentel, S. Arpiainen, J. Ahopelto, F. Jonsson, S.G. Romanov, C.M. Sotomayor Torres, Langmuir 22, 7378 (2006) 39. Y. Zeng, D.J. Harrison, Anal. Chem. 79, 2289 (2007) 40. J. Park, D. Lee, W. Kim, S. Horiike, T. Nishimoto, S.H. Lee, C.H. Ahn, Anal. Chem. 79, 3214 (2007) 41. Z.-Z. Gu, D. Wang, H. Mo¨hwald, Soft Matter 3, 68 (2007) 42. A.K. Bohaty, I. Zharov, Langmuir 22, 5533 (2006) 43. A.K. Bohaty, I. Zharov, Langmuir 25, 3096 (2009) 44. D. Sarkar, P.V. Subbarao, G. Begum, K. Ramakrishna, J. Colloid Interface Sci. 288, 591 (2005) 45. T.G. Braga, A.C. Wahl, J. Phys. Chem. 89, 5822 (1985) 46. W. Sto¨ber, A. Fink, E. Bohn, J. Colloid Interface Sci. 26, 62 (1968) 47. W. Wang, B. Gu, L. Liang, W. Hamilton, J. Phys. Chem. B 107, 3400 (2003) 48. M.C. Gennaro, E. Mentasti, C. Sarzanini, V. Porta, Chromatographia 25, 117 (1988) 49. W.A. Summers, J.Y. Lee, J.G. Burr, J. Org. Chem. 40, 1559 (1975) 50. G.D. Mahan, J.O. Sofo, Phys. Rev. B 62, 2780 (2000) 51. E.L. Cussler, Diffusion: mass transfer in fluid systems, 2nd edn. (Cambridge University Press, New York, 1997) 52. A.J. Bard, L.R. Faulkner, Electrochemical methods: fundamentals, applications, 2nd edn. (Wiley, New York, 2001) 53. J.-J. Shyue, M.R. De Guire, T. Nakanishi, Y. Masuda, K. Koumoto, C.N. Sukenik, Langmuir 20, 8693 (2004) 54. A. Miller, Writing reaction mechanisms in organic chemistry (Academic Press, Inc., San Diego, 1992) 55. S.B. Lee, C.R. Martin, Chem. Mater. 13, 3236 (2001)