Liquid Microarrays - ACS Publications

Mark Platt,† Robert A. W. Dryfe,*,† and Edward P. L. Roberts‡. Department of Chemistry, University of Manchester Institute of Science & Technolo...
1 downloads 0 Views 173KB Size
Langmuir 2003, 19, 8019-8025

8019

Voltammetry with Liquid/Liquid Microarrays: Characterization of Membrane Materials Mark Platt,† Robert A. W. Dryfe,*,† and Edward P. L. Roberts‡ Department of Chemistry, University of Manchester Institute of Science & Technology, P.O. Box 88, Manchester, United Kingdom M60 1QD, and Department of Chemical Engineering, University of Manchester Institute of Science & Technology, P.O. Box 88, Manchester, United Kingdom M60 1QD. Received April 29, 2003. In Final Form: July 1, 2003 A simple nondestructive method is presented to characterize the physical properties of ultrafiltration membranes. The technique utilizes voltammetry at the interface between two immiscible electrolyte solutions (ITIES) and has been applied to commercially available γ-alumina membranes. Upon the application of a potential difference across the ITIES, the voltammetric response resulting from an ion transfer is a direct measure of the membrane porosity. The technique has been applied to the measurement of the fractional porous area (porosity) and the membrane thickness. The reported technique offers advantages over existing methods such as being easy to use, reproducible, inexpensive, and nondestructive. Furthermore, unlike many optical techniques, this method is not limited by a minimum nominal pore size and has thus been used to make measurements when the reported nominal pore diameter is as small as 20 nm.

Introduction The characterization of ultrafiltration membranes in general, and of γ-alumina membranes in particular, has been attempted by many researchers using various techniques, such as gas penetration,1 bubble point analysis,2 Hg porosimetry,1 scanning electron microscopy (SEM),3 SEM combined with computerized imaging analysis (CIA),4 Rutherford backscattering spectrometry (RBS),5 hydraulic resistance,6 neutron scattering,7 and atomic force microscopy.8 Electrochemical approaches to membrane characterization in general,9 and to γ-alumina membranes in particular,10 have been reported. Membranes of γ-alumina, known commercially as Anopore membranes, are excellent filtration devices due to their high porosities of ca. 109 pores per cm2, small pore sizes, resistance to a wide range of organic solvents, and stability at high temperature.11 A further advantage is the ability to change the surface functionality of the alumina by reacting terminal hydroxyl groups with compounds such as silanes or carboxylic acids.12 Alumina membranes have also been used as templates to control the electrodeposition * Corresponding author. E-mail: [email protected]. Fax: +44 161 200 4559. Tel: +44 161 200 4522. † Department of Chemistry. ‡ Department of Chemical Engineering. (1) Palacio, L.; Pra´danos, P.; Calvo, J. I.; Herna´ndez, A. Thin Solid Films 1999, 348, 22. (2) Jakobs, E.; Koros, W. J. J. Membr. Sci. 1997, 124, 149. (3) Dalvie, S. K.; Baltus, R. E. J. Membr. Sci. 1992, 71, 247. (4) Herna´ndez, A.; Calvo, J. I.; Pra´danos, P.; Palacio, L.; Rodrı´guez, M. L.; de Saja, J. A. J. Membr. Sci. 1997, 137, 89. (5) Pesiri, D. R.; Snow, R. C.; Elliott, N.; Maggiore, C.; Dye, R. C. J. Membr. Sci. 2000, 176, 209. (6) Labbez, C.; Fievet, P.; Szymczyk, A.; Simon, C.; Vidonne, A.; Foissy, A.; Pagetti, J. Desalination 2001, 141, 291. (7) Steriotis, T.; Mitropoulos, A.; Kanellopoulos, N.; Keiderling, U.; Wiedenmann, A. Physica B 1997, 234, 1016. (8) Bowen, W. R.; Hilal, N.; Lovitt, R. W.; Williams, P. M. J. Membr. Sci. 1996, 110, 233. (9) Brett, C. M. A.; Alves, V. A.; Fungaro, D. A. Electroanalysis 2001, 13, 212. (10) Miller, C. J.; Majda, M. J. Am. Chem. Soc. 1985, 107, 1419. (11) Bluhm, E. A.; Bauer, E.; Chamberlin, R. M.; Abney, K. D.; Young, J. S.; Jarvinen, G. D. Langmuir 1999, 15, 8668. (12) Slavov, S. V.; Chuang, K. T.; Sanger, A. R. Langmuir 1995, 11, 3607.

Figure 1. Cross-sectional SEM image of part of an Anopore membrane, with a nominal pore diameter of 100 nm. The active layer (denoted A) is defined by the small pore and is between 0.5 and 1 µm in length. The support layer (denoted S) contributes to the rest of the membrane and is approximately 59 µm in depth.

of metallic particles.13 However, certain factors exist which complicate the analysis of such commercial membranes: one recurring factor is the asymmetric membrane structure, since both an “active” layer and a “support” layer are present, as shown in Figure 1. The active layer defines the membrane’s properties, as this contains the smallest pores; hence the characterization of this layer is an important step in judging the utility of a given membrane. Problems associated with some of the aforementioned techniques include their inability to distinguish between these two layers. The relatively wide variation seen in the results reported in the literature for the fractional porous area (porosity) is taken as evidence of this. Commercially, three different Anopore membranes are available with nominal pore diameters of 200 nm (denoted A02), 100 nm (A01), and 20 nm (A002). Data obtained by some of the previously listed techniques as applied to the Anopore membranes used in this study are collated in Table 1. (13) Hornyak, G. L.; Patrissi, C. J.; Martin, C. R. J. Phys. Chem. B 1997, 101, 1548.

10.1021/la034726v CCC: $25.00 © 2003 American Chemical Society Published on Web 08/05/2003

8020

Langmuir, Vol. 19, No. 19, 2003

Platt et al.

Table 1. Results Obtained for the Porosity of Anopore Membranes Using Various Techniquesa nominal Hg porosimetry gas penetration SEM with CIA permeabilities

A002

A01

A02

0.25-0.30 0.20 0.293

0.40 0.31 0.531 0.290* 0.352

0.50 0.31 0.558 0.359* 0.293

a The porosity is defined as the fraction of area corresponding to the pores, compared to the total area of the membrane surface. The asterisk denotes a value specific to the “active” face of the membrane. The values quoted are taken from ref 1.

Electrochemical methods have been used previously to determine membrane properties, but such studies have generally focused on the measurement of surface potentials,14 including streaming potentials developed at nonzero pressure gradients,15 or resistance measurements.6 Voltammetric techniques have been used previously, but such studies have generally relied on the electrode/ electrolyte interface and have thus required the membrane to be affixed to the electrode surface, the current response being in turn related to the porosity.16 A specific advantage of the use of voltammetry is that the current response is intrinsically related to the flux of the analyte; hence information on membrane transport is readily deduced. However, a disadvantage of voltammetry is the requirement that a perfect seal is formed between the electrode and membrane, particularly where rough electrode and/ or membrane surfaces are involved, since the lack of an intact seal will lead to considerable error in the results: poor adhesion between alumina membranes and evaporated gold contacts has been noted in earlier voltammetric studies of transport within alumina membranes.17 Data on the transmembrane fluxes of ions can also be obtained using radioisotope measurements,11 although this clearly restricts the range of analytes that can be investigated. Some of the most recent work on membrane characterization has suggested that the active layer pores may possess some tortuosity, rather than being linear.5 Such imaging techniques are highly surface sensitive and have provided useful information on the surface porosities and surface features of the Anopore membranes. However, these techniques rely on relatively expensive equipment, are destructive (for SEM, membranes must generally be coated with conducting substrates, which renders them useless for subsequent filtration purposes), have limits to resolution, and can be time consuming. Despite these restrictions, work carried out to date using SEM and CIA has provided some of the most accurate surface porosity measurements of both the A02 and A01 membranes.4 The use of a liquid/liquid interface to help provide surface information has been described previously.2 This type of experiment has yet to be given a common term and covers methods known as bubble point or biliquid permporometry. Such experiments utilize the capillary action of the perpendicular pores to help stabilize one of the solutions (denoted solution A), which is chosen so that it has a high affinity for the membrane material, thus filling the pores up to their mouths. The determination of membrane properties occurs by placing a second liquid (solution B), which is immiscible with the first and does (14) (a) Bowen, W. R.; Hughes, D. T. J. Colloid Interface Sci. 1991, 143, 252. (b) Kosmulski, M. Langmuir 2002, 18, 785. (15) Herna´ndez, A.; Martı´nez, F.; Martı´n, A.; Pra´danos, P. J. Colloid Interface Sci. 1995, 173, 284. (16) (a) Miller, C. J.; Widrig, C. A.; Charych, D. H.; Majda, M. J. Phys. Chem. 1988, 92, 1928. (b) Ikeda, O.; Ohtani, M.; Yamaguchi, T.; Komura, A. Electrochim. Acta 1997, 43, 833. (17) Miller, C. J.; Majda, M. J. Electroanal. Chem. 1986, 207, 49.

Figure 2. Schematic of the “bubble point” experimental setup.

Figure 3. Cell schematics for the typical liquid/liquid cell. Experiments may be performed in the absence of the membrane, configuration a; with the positions of the two phases inverted in the presence of a membrane, configuration b; or as in (a), but in the presence of a membrane, which results in a thin film being formed on the lower half of membrane, configuration c. An expanded schematic of the interfacial position resulting from configuration c is shown in (d).

not readily wet the membrane surface, on top of the membrane (see Figure 2). The pressure required to force solution B into the pores is used to determine the pore radii. The purpose of the present work is to report an electrochemical variant of the biliquid method, where ions are transferred across the liquid/liquid interface (ITIES, interface between two immiscible electrolyte solutions) and the ensuing voltammetric response is used to characterize the membrane properties. Theory The application of a potential difference at the liquid/ liquid interface is a relatively new electrochemical technique,18 which allows the study of processes such as ion transfer across an aqueous/organic interface. The method generally utilizes a four-electrode potentiostat consisting of two pairs of counter and reference electrodes. A typical cell schematic is shown in Figure 3. The potentiostatically controlled applied potential difference is defined by eq 1:

∆Φ ) Φaq - Φorg

(1)

The essence of the method presented herein is that an ion is transferred from the aqueous phase to the organic phase, via the membrane pores, and is thus used to probe the (18) See for example: (a) Girault, H. H. J.; Schiffrin, D. J. In Electroanalytical Chemistry; Bard, A. J., Ed.; Marcel Dekker: New York, 1989; Vol. 15, p 1. (b) Liu, B.; Mirkin, M. V. Anal. Chem. 2001, 73, 670A.

Voltammetry with Liquid/Liquid Microarrays

Langmuir, Vol. 19, No. 19, 2003 8021

I ) N(4zFDCbulkr)

Figure 4. Schematic of the diffusion layer growing as time progresses, as it changes from linear to radial and back to linear in form.

membrane geometry. The peak current, Ip, observed at the ITIES, for a one-dimensional diffusive ion transfer, is given by the Randles-Sevcˇik equation:19

Ip ) 0.4463

x( )

z3F3 ACbulkD1/2ν1/2 RT

(2)

where Ip is the peak current, z is the charge number of the transferring ion, A is the area of the conducting interface (the ITIES in this case), D is the diffusion coefficient of the analyte species, Cbulk is the concentration of the analyte, and ν is the voltage scan rate (the other symbols have their usual meanings). Modification of the ITIES with porous materials leads to various limiting cases arising for the currents observed. Under the experimental conditions employed with the liquid/liquid voltammetric techniques described here, the ion transferring from the aqueous phase has to travel through the pore, leading to the possible diffusion fields shown in Figure 4. Equation 2 holds if the diffusion fields remain one dimensional, due to overlap between all the individual diffusion fields (right-hand side of Figure 4) or due to linear diffusion through the pores (the case second from the left in Figure 4). However, the Ip values measured through eq 2 will vary depending on which of the linear diffusion subclasses holds. If the overlap case occurs, eq 2 remains valid (with A now representing the total membrane area), whereas if the diffusion fields to the individual pores do not traverse the entire membrane, then eq 2 should be replaced by eq 3:

Ip ) 0.4463

x( )

z3F3 θACbulkD1/2ν1/2 RT

(3)

where the factor θ is defined as the membrane porosity and represents the fraction of the membrane surface that is pore. Equation 4 becomes valid if the individual diffusion fields traverse the membrane but do not overlap and thus radial diffusion dominates (second from the right in Figure 4), resulting in a steady-state current response:20 (19) Bard, A. J.; Faulkner, L. R. Electrochemical Methods: Fundamentals and Applications, 2nd ed.; Wiley: New York, 2001; Chapter 6. (20) Hulteen, J. C.; Menon, V. P.; Martin, C. R. J. Chem. Soc., Faraday Trans. 1996, 92, 4029.

(4)

where r is the mean pore diameter and N is the number of pores; this can be obtained from the product of the quoted pore density and the area of the membrane exposed (A). Variations on eq 4 arise if the length of the pore is significant with respect to r.21 On modification of the ITIES with the γ-alumina membranes, the response recorded will be given by one of eqs 2, 3, or 4, or by behavior between these limiting cases depending on the dimensions of the diffusion field, relative to the mean separation and lengths of adjacent pores. The modification of the ITIES with porous materials and the effects of such micron scale arrays upon the resultant diffusion fields have been reported previously for polyester membranes21,22 and zeolite membranes.23 The previous work on this subject merely correlated the observed voltammetric properties with the (assumed) membrane properties. In the present work, the voltammetry is used to determine the porosity of the membranes employed. The new method for the determination of membrane properties is based upon principles similar to those of the bubble point experiments. In the electrochemical case reported here, solution A is an aqueous electrolyte solution stabilized at the pore mouth. The second solution, B, is an electrolyte solution of 1,2dichloroethane (DCE). Capillary action on the alumina surface forces the less dense solution A into the membrane and counteracts the gravitationally unstable configuration, supporting the denser of the two solutions, B, on top of the cell to give the configuration shown in Figure 3b. When the orientations of the liquids are inverted, under the influence of gravity, solution A is no longer retained at the pore mouth, and a thin film is formed on the underside of the membrane, as depicted in Figure 3d, from the cell configuration in Figure 3c. Recent work from this laboratory on electrodeposition at the ITIES has supported the last statement and has shown that the position of the interface, with respect to the membrane, affects the size and shape of the metal deposits formed at the aluminamodified ITIES.24 Experimental Section Voltammetric experiments were performed using a PCcontrolled EG&G model 273 potentiostat (Princeton Applied Research Corp., Princeton, NJ) operating in four-electrode mode. The reference electrodes were silver/silver chloride wires, produced in-house by oxidation of a 1 mm diameter silver wire in a lithium chloride solution. The counter electrodes were constructed in-house by spot welding an approximately 1 cm2 area of platinum gauze (Advent, Eynsham, U.K.) to a length of platinum wire (Advent). Anopore membranes of 13 mm diameter with nominal pore diameters of 20, 100, or 200 nm (Whatman, Kent, U.K.) were sealed to a 0.96 cm inner diameter glass tube with a silicone sealant (RS Components Ltd). The cell was set up as shown in Figure 3. Once assembled, the total area of the membrane-modified interface exposed (A) was 0.72 cm2. The bottom half of the cell was then filled with the aqueous solution, taking care to ensure that no air bubbles were trapped under the membrane. The organic solution was then placed in the top half of the cell and covered with a small volume of deionized water to prevent the solvent from evaporating. The height of the upper column of liquid was not found to influence the observed (21) Kralj, B.; Dryfe, R. A. W. Phys. Chem. Chem. Phys. 2001, 3, 5274. (22) Wilke, S.; Osborne, M. D.; Girault, H. H. J. Electroanal. Chem. 1997, 436, 53. (23) Dryfe, R. A. W.; Holmes, S. M. J. Electroanal. Chem. 2000, 483, 144. (24) Platt, M.; Dryfe, R. A. W.; Roberts, E. P. L. Chem. Commun. 2002, 2324.

8022

Langmuir, Vol. 19, No. 19, 2003

voltammetry over the height range employed (1-2 cm). When the opposite solution configuration was adopted (aqueous phase uppermost), the aqueous phase traversed the membrane, as noted above. The membrane was treated with trichloromethylsilane (supplied by Aldrich) in an attempt to stabilize the latter configuration, by reducing the hydrophilicity of the alumina surface; however, the effect of this treatment was not sufficient to prevent the formation of the thin film. The position of the interface can also be inferred from its effect on the voltammetric response, as discussed later in this text. The membrane structure is asymmetric, as noted earlier: the membrane was normally oriented with the active layer adjacent to the organic phase (i.e., the active layer was uppermost, with reference to Figure 3b), although one series of experiments was performed with the support layer uppermost. The aqueous phase electrolyte system was lithium chloride (5 × 10-2 mol dm-3, Lancaster Synthesis, Morecambe, U.K.) and lithium sulfate (5 × 10-2 mol dm-3, Aldrich, Gillingham, U.K.) dissolved in water obtained from a Milli-Q purification system. The liquid/liquid interface was formed with water and DCE (HPLC grade, obtained from Lancaster Synthesis). The organic electrolyte, bis(triphenylphosphoranylidene) ammonium tetraphenylborate (BTPPA TPB), was prepared by adapting a reported metathesis procedure25 from equimolar amounts of bis(triphenylphosphoranylidene) ammonium chloride (BTPPACl) (Lancaster, 97%) and lithium tetraphenyl borate (Li TPB) (Lancaster, 98%). The concentration of BTPPA TPB in all cells was 2 × 10-2 mol dm-3. The reference electrode for the organic phase was formed by placing the silver/silver chloride wire in an aqueous solution of BTPPACl and lithium chloride (1 × 10-3 and 1 × 10-2 mol dm-3, respectively). When the probe ion, tetraethylammonium (TEA+), was present initially in the aqueous phase, it was added as the chloride salt (purchased from Aldrich) to give concentrations of either 1 × 10-3 or 2 × 10-3 mol dm-3. The ion was sometimes initially dissolved in the organic phase: in this case, it was present as the tetraphenylborate salt, formed by metathesis of the chloride salt with Li TPB, using the procedure referred to above for BTPPA. The aqueous phase pH was measured with a pH electrode (Hanna Instruments, Leighton Buzzard, U.K.) and adjusted by addition of sodium hydroxide (semiconductor grade, supplied by Aldrich). Cyclic voltammograms were recorded at scan rates ranging from 0.02 to 0.17 V s-1. Between each scan, the cell was left for 10 min to allow the system to return to equilibrium. Numerical simulations of the cyclic voltammetric response were performed on the UMIST UNIX server, using an implicit finite-difference technique to simulate the voltammograms.26 SEM imaging was performed by sticking the membranes to a 12 mm diameter imaging stub, coating with a layer of carbon, and subsequently imaging using a Philips XL30 Field Emission Gun scanning electron microscope.

Results and Discussion Figure 5 shows simulated voltammograms for a series of scan rates, from which the peak current, Ip, can be measured and plotted against the square root of scan rate to give a Randles-Sevcˇik plot. Because of the proportionality between membrane area, or porosity, and Ip, it is clear that comparison of the observed and predicted voltammetric responses can be used to assess the membrane porosity for a given area, as long as the linear diffusion model implicit in eqs 2 and 3 can be assumed to hold. The veracity of these equations can be tested through Randles-Sevcˇik plots of the Ip dependence on scan rate. A linear relationship between porosity and the ion transfer current will be observed if the individual diffusion fields remain within the pores, as depicted on the left of Figure 4, and therefore the scan rates employed need to be sufficiently fast to prevent the diffusion layer from extending beyond the membrane. (25) Shao, Y.; Stewart, A. A.; Girault, H. H. J. Chem. Soc., Faraday Trans. 1991, 87, 2593. (26) Alden, J. A.; Compton, R. G.; Dryfe, R. A. W. J. Electroanal. Chem. 1995, 397, 11.

Platt et al.

Figure 5. Simulated voltammetric responses for linear diffusion, with Cbulk set to 1 × 10-3 mol dm-3, D equal to 8.75 × 10-6 cm2 s-1, and A equal to 0.195 cm2. The voltage scan rates displayed here are (a) 0.10 V s-1, (b) 0.06 V s-1, and (c) 0.02 V s-1.

Figure 6. Voltammetry observed when the ITIES was modified with an A02 membrane: Cbulk was set to 1 × 10-3 mol dm-3, and A was 0.724 cm2. The voltage scan rates shown here are (a) 0.13 V s-1, (b) 0.08 V s-1, and (c) 0.04 V s-1.

Figure 6 shows the voltammograms obtained experimentally when an A02 membrane is placed at the ITIES according to the procedure described. It is noted from the voltammograms that the currents are significantly smaller than those expected if the area available for ion transfer is set equal to the area of membrane exposed to the ITIES (i.e., A corresponds to the area exposed, and θ is assumed to be 1.0). For example, with ν equal to 0.1 V s-1, a peak current of ca. 65 µA was observed for the conditions of Figure 5 except that A was equal to 0.72 cm2, compared to a value of 180 µA expected on the basis of eq 2. The reduction in current is due to the membrane’s porous nature: the total area of the pores can therefore be calculated from comparison of the gradients of the calculated and observed Randles-Sevcˇik plots, Figure 7. The optimal fit between the experimental and simulated peak currents is found when the interfacial area is set to 28% of A, the actual area available for ion transfer; hence θ in this case is deduced to be 0.28. Although linear Randles-Sevcˇik plots were always observed for sufficiently high scan rate values, a nonzero

Voltammetry with Liquid/Liquid Microarrays

Figure 7. Randles-Sevcˇik plot using peak currents (open circles) obtained in Figure 6. The straight line is calculated from eq 3, allowing θ to vary. The optimal fit between the observed and calculated data, for the measured value of A, was found for a θ value of 0.28.

Figure 8. Simulated (a) and observed (b) cyclic voltammetry for TEA+ transfer using an A01 membrane, with Cbulk of 1 × 10-3 mol dm-3, A of 0.724 cm2, ν of 0.1 V s-1, and θ equal to 0.27.

current axis intercept was generally noted, as typified by the data of Figure 7. This effect is conceivably caused by current from the transfer of ions adsorbed on the internal surface of the membrane, although experiments performed explicitly to test this hypothesis have not shown a significant change (vide infra). Notably, a nonzero intercept was observed for a variety of ions (both cations such as TEA+ and tetramethylammonium as well as anions such as nitrate and iodide), suggesting that the effect observed is not dependent on a specific chemical functionality. Numerical simulations of the one-dimensional diffusion equation using the literature value21 of D for TEA+ show that for the scan rates employed experimentally, the diffusion field is smaller than the nominal thickness of the membranes, quoted as 60 µm.27 The growth of the diffusion field and its effect on the current can be seen in the cyclic voltammograms recorded: Figure 8 shows a comparison of simulated and observed voltammograms for a 1 × 10-3 mol dm-3 aqueous solution of TEA+ across an A01 membrane, where the value of θ used in the simulation is 0.27. The simulated voltammetry shows distinct peaks in both the forward and reverse sweeps, whereas the forward sweep in the experimental response, which represents the transfer of TEA+ through the pore and into the organic (27) Technical information supplied by the manufacturer; see http:// whatman.co.uk.

Langmuir, Vol. 19, No. 19, 2003 8023

Figure 9. Cyclic voltammograms recorded when the ITIES was modified with an A02 membrane. Cbulk was 1 × 10-3 mol dm-3, and the total area of membrane exposed was 0.724 cm2. The scan rates shown are (a) 0.12 V s-1, (b) 0.08 V s-1, and (c) 0.04 V s-1.

Figure 10. Randles-Sevcˇik analysis of peak currents obtained for the conditions given in Figure 9 (filled circles). The dashed lines labeled a and b correspond to the predicted response for θ values of 0.29 and 1.0, respectively. Scan rates above 0.06 V s-1 correspond to the diffusion fields being retained within the membrane pores.

phase, shows a less well-defined peak. The enhanced current observed experimentally after Ip is attributed to the diffusion layer beginning to grow beyond the pores, with the current observed thus becoming proportional to the entire membrane area as shown schematically in Figure 4. As a result of the increased current flow, the reverse peak current observed experimentally is much higher than the calculated value. This increase in current due to the diffusion layer growing can be seen to a greater extent if even slower scan rates are employed, therefore allowing the diffusion field to grow, or if the membrane thickness is less than the nominal value of 60 µm. Figure 9 shows the cyclic voltammetric response for an A02 membrane. The response appears similar to that illustrated in Figure 8a; however, a distinct kink is seen in the resultant Randles-Sevcˇik plot (Figure 10). This kink is due to the transition in the current response at the slower scan rates as the response tends from the pore response (θ < 1) to the overlapping response (θ ) 1) and hence is a direct indication that the diffusion layer is growing nonlinearly for the reasons given above. The point at which the transition between the responses predicted by eqs 2 and 3 occurs can be used to determine the corresponding thickness of the diffusion layer at this scan rate, and hence the thickness of the membrane can be found. Deviations

8024

Langmuir, Vol. 19, No. 19, 2003

Platt et al.

Table 2. Mean Porosity Values (θ) and Their Standard Deviations, Determined Using the Active Layers of Each of the Classes of γ-Alumina Membrane Investigated from the Voltammetric Experimentsa

membrane A002 A01 A02

porosity using 1 × 10-3 mol dm-3 TEA+

porosity using 2 × 10-3 mol dm-3 TEA+

mean porosity

0.12 ( 0.03 0.27 ( 0.02 0.30 ( 0.03

0.14 ( 0.02 0.26 ( 0.02 0.31 ( 0.03

0.13 ( 0.02 0.27 ( 0.02 0.30 ( 0.03

a The results quoted for each membrane class are mean values taken from experiments with three different membranes for each concentration of the probe ion (TEA+).

from the θ ) 1 response occur where the diffusion layer thickness, calculated for linear diffusion, becomes comparable with the membrane thickness; hence the RandlesSevcˇik model breaks down (Figure 4). The nominal diffusion layer thickness was determined via numerical simulation of the aqueous phase concentration profile for the scan rate where deviation from the Randles-Sevcˇik behavior was observed. This thickness was taken to be equal to the thickness of the membrane. For the particular membrane used for the data of Figure 8, the gradient at higher scan rates gives a θ value of 0.33 and a mean membrane thickness of 72 µm. For each class of membrane employed, the transfer of TEA+ was repeated six times: three times at a concentration of 1 × 10-3 mol dm-3 and three further times at a concentration of 2 × 10-3 mol dm-3. This process was performed to provide an assessment of the reproducibility of the experimental procedure and is the basis of the uncertainty values quoted in Table 2. The tests performed on each class of membrane gave an idea of variability (e.g., in mean pore diameter) of the membranes and any dependence of the measured θ values on analyte concentration. It should be stressed that all the membranes investigated within a given class came from the same product batch. Table 2 gives the porosities measured at all concentrations for each active face of the membrane. Notably, consistent values of θ are obtained for all three classes of membrane, which do not show significant dependence on the concentration of TEA+. The possibility of significant interaction of TEA+ with the pore walls is not excluded, but at the high ionic strengths employed it is apparent that any such effects have a minimal effect on the transmembrane flux of this analyte (as determined voltammetrically). It is possible that any adsorption of TEA+ occurs over a long time scale; hence a further series of experiments was performed where the aqueous TEA+ solution was left in contact with the alumina membrane for 48 h prior to experimentation. However, no significant change in the calculated θ value was seen. The pH of the aqueous solution was also adjusted, from its “as-made” value of 6.3, to 8.4 (the isoelectric point of Anopore membranes has been reported11 to be around pH 8). Once more, the voltammetric results indicated that the porosity of the membrane, as measured through the transfer of TEA+, was unaffected by this variation, an observation also attributed to the high ionic strengths of the aqueous solutions employed. As mentioned in the Experimental Section, when the aqueous phase was uppermost within the cell (Figure 3c), the aqueous solution tended to traverse the membrane to produce a thin aqueous layer on its underside, as drawn schematically in Figure 3d. The existence of this thin layer was inferred through the voltammetric data, which showed a response intermediate between the “classical” peak separation of ca. 0.06 V and the thin layer response which

Figure 11. Voltammetry obtained under the “thin-layer” conditions illustrated in Figure 3c,d. The voltammograms shown relate to an A01 membrane, with Cbulk equal to 1 × 10-3 mol dm-3. A was 0.724 cm2, and the scan rates shown are (a) 0.01 V s-1, (b) 0.030 V s-1, (c) 0.06 V s-1, and (d) 0.10 V s-1.

tends toward a peak separation of 0 V.28 Typical voltammograms observed experimentally under these conditions are presented in Figure 11. The reduction in peak separation observed is taken as evidence for the formation of a thin layer of aqueous solution on the underside of the membrane, with a depth comparable to that of the diffusion layer established within it. Within 30 min, this layer grows to a size which can be observed with the naked eye. The lowest peak separations observed, at 0.01 V s-1, are of the order of 0.04 V. The porosity of the support layer was investigated by retaining the experimental configuration shown in Figure 3b but inverting the membrane such that the support layer was adjacent to the liquid/liquid interface. The area of contact between the two liquids is thus given by θA, where θ now refers to the support layer porosity, rather than that of the active layer. Comparison of the gradient of the Randles-Sevcˇik plot, as performed for the active layer measurements (see Figures 7 and 10), indicated that θ was 0.35 for an A002 membrane with a 1 × 10-3 mol dm-3 aqueous solution of TEA+ (data not shown). This is in agreement with previous conclusions on the structure of γ-alumina membranes drawn from SEM data,4,29 although the approach presented here is simpler to implement and does not require the deposition of a conducting film on the membrane but provides ready quantification of the porosities of the distinct layers. As was noted in the Introduction, the ability to extract this type of information for membranes with extremely small pores (r of 10 nm) is of considerable utility. Finally, the results presented here support statements made in earlier publications involving experiments such as bubble point analysis2 and previous reports from this laboratory on electrodeposition at the ITIES,24 which have yielded information on the position of the liquid/liquid interface with respect to the surface of the alumina. The interface lies at the pore mouth on the organic side of the membrane, a position which is completely reproducible, implying that the aqueous phase fills the pores. By contrast, when TEA+ was initially present in the organic phase, a current response consistent with transport to the entire membrane area was observed, indicative of θ tending to unity (data not shown). This supports the inference that the liquid/liquid interface is located on the organic side of the membrane, since the diffusion fields (28) Bard, A. J.; Faulkner, L. R. Electrochemical Methods: Fundamentals and Applications, 2nd ed.; Wiley: New York, 2001; p 456. (29) (a) Furneaux, R. C.; Rigby, W. R.; Davidson, A. P. Nature 1989, 337, 147. (b) Crawford, G. P.; Steele, L. M.; Ondris-Crawford, R.; Iannacchione, G. S.; Yeager, C. J.; Doane, J. W.; Finotello, D. J. Chem. Phys. 1992, 96, 7788.

Voltammetry with Liquid/Liquid Microarrays

in the latter case are not affected significantly by the presence of the membrane. Conclusions The use of voltammetric methods to characterize the pore fraction and thickness of membranes is described. A previous report from this laboratory dealing with polymer Nucleopore membranes discussed the possibility of this approach:21 here the concept is put into practice using γ-alumina membranes. The method presented is based on ion transfer at the ITIES and thus has the advantage of being nondestructive (coating with a conductive substrate is not required) and complexities associated with the formation of intact seals are avoided. The membranes may possess relatively complex structure; for example, in the case of the γ-alumina membranes discussed here, the membranes possess asymmetric pores. The method is applicable to membrane materials with very small (10-8 m) pore radii, well below the limits of optical and many microscopic techniques. The only prerequisite for this

Langmuir, Vol. 19, No. 19, 2003 8025

method is that the membranes should, ideally, display preferential wetting by one of the liquid phases employed. The data thus obtained provide information on the membranes themselves and also, conceivably, on the behavior of liquid/liquid interfaces within complex heterogeneous structures, which could be of relevance to oil recovery within porous media. Furthermore, we note that information on the transport of ions within porous structures should be amenable via the techniques developed here. Work currently underway is exploring the ion transfer process at low ionic strengths to explore the influence of adsorption processes on the voltammetric response observed. Acknowledgment. The financial support of the EPSRC is acknowledged. We thank Mr. I. Brough (Manchester Materials Science Centre) for assistance with SEM imaging. LA034726V