Thin Layer Coulometry with Ionophore Based Ion-Selective

Apr 29, 2010 - Paper-Based Thin-Layer Coulometric Sensor for Halide Determination .... GalvaPot, a custom-made combination galvanostat/potentiostat an...
1 downloads 0 Views 570KB Size
Download PDF
Anal. Chem. 2010, 82, 4537–4542

Thin Layer Coulometry with Ionophore Based Ion-Selective Membranes Ewa Grygolowicz-Pawlak and Eric Bakker* Department of Chemistry, Nanochemistry Research Institute, Curtin University of Technology, Perth, WA 6845, Australia We are demonstrating here for the first time a thin layer coulometric detection mode for ionophore based liquid ion-selective membranes. Coulometry promises to achieve the design of robust, calibration free sensors that are especially attractive for applications where recalibration in situ is difficult or undesirable. This readout principle is here achieved with porous polypropylene tubing doped with the membrane material and which contains a chlorinated silver wire in the inner compartment, together with the fluidically delivered sample solution. The membrane material consists of the lipophilic plasticizer dodecyl 2-nitrophenyl ether, the lipophilic electrolyte ETH 500, and the calcium ionophore ETH 5234. Importantly and in contrast to earlier work on voltammetric liquid membrane electrodes, the membrane also contains a cationexchanger salt, KTFPB. This renders the membrane permselective and allows one to observe open circuit potentiometric responses for the device, which is confirmed to follow the expected Nernstian equation. Moreover, as the same cationic species is now potential determining at both interfaces of the membrane, it is possible to use rapidly diffusing and/or thin membrane systems where transport processes at the inner and outer interface of the membrane do not perturb each other or the overall composition of the membrane. The tubing is immersed in an electrolyte solution where the counter and working electrode are placed, and the potentials are applied relative to the measured open circuit potentials. Exhaustive current decays are observed in the range of 10 to 100 µM calcium chloride. The observed charge, calculated as integrated currents, is linearly dependent on concentration and forms the basis for the coulometric readout of ion-selective membrane electrodes.

selectivity and sensitivity in many cases,2,3 it still utilizes traditional transduction principles, including potentiometry and amperometry. These have never been intended for recalibration free and robust sensing approaches. In the clinical laboratory setting, for instance, where sensors based on such principles are employed in vast numbers, a one point recalibration step is typically performed with each sample reading to compensate for any drift in the sensor output signal.4 Clearly, the transduction principles of electrochemical sensors must be reconsidered from fundamental principles to achieve the ambitious goals mentioned above.5 Absolute measurement principles in the analytical sciences are rare, but they do include gravimetry and coulometry. In controlled potential coulometry, the current associated with a specific redox reaction is monitored until it decays to near-zero values.6 The observed current integrated over the time period gives the overall charge, which ideally relates to the number of moles of material converted in the sample using Faraday’s law. The use of thin layer samples, as in thin layer coulometry, improves the speed of analysis relative to the use of voluminous samples.7 The key challenge in employing this readout principle for chemical sensors has been that redox reactions at metal electrodes tend not to be sufficiently selective. Recently, however, a coulometric readout was successfully used in enzymatic glucose bioassays, for example, allowing one to exploit the confined volume for the exhaustive electrolysis of potential interferences before the actual glucose assay.8 The electrochemical detection of ionic species is preferably accomplished with ion-selective electrodes. Polymeric or liquid membrane systems are now particularly well developed, and their sensing selectivities are a function of membrane extraction and complexation processes.9,10 These membranes are traditionally read out by zero current potentiometry, often giving extremely low detection limits.11 Moreover, the use of voltammetric methods on membrane electrodes has increased in recent years, including

Modern directions in chemical sensing demand the realization of self-contained, robust sensing principles that do not require frequent supervision or recalibration in the field. Environmental remote chemical sensing applications with a large network of inexpensive chemical sensors, for instance, would have tremendous promise in monitoring programs. Current technologies are often restricted to passive, nonselective optical systems.1 While state of the art in electrochemical sensing technology may offer adequate

(2) Lee, H. J.; Yoon, I. J.; Yoo, C. L.; Pyun, H.-J.; Cha, G. S.; Nam, H. Anal. Chem. 2000, 72, 4694–4699. (3) Wang, J.; Tian, B. Anal. Chem. 1992, 64, 1706–1709. (4) Jacobs, E.; Ancy, J. J.; Smith, M. Point Care: J. Near-Patient Test. Technol. 2002, 1, 135–144. (5) Bakker, E.; Bhakthavatsalam, V.; Gemene, K. L. Talanta 2008, 75, 629– 635. (6) Bard, A. J. Anal. Chem. 1966, 38, 88R–98R. (7) Oglesby, D. M.; Anderson, L. B.; McDuffie, B.; Reilley, C. N. Anal. Chem. 1965, 37, 1317–1321. (8) Tsujimura, S.; Kojima, S.; Ikeda, T.; Kano, K. Anal. Bioanal. Chem. 2006, 386, 645–651. (9) Buhlmann, P.; Pretsch, E.; Bakker, E. Chem. Rev. 1998, 98, 1593–1687. (10) Bakker, E.; Buhlmann, P.; Pretsch, E. Chem. Rev. 1997, 97, 3083–3132. (11) Bakker, E.; Pretsch, E. Angew. Chem., Int. Ed. 2007, 46, 2–11.

* To whom correspondence should be addressed. (1) Miller, R. L.; Del Castillo, C. E.; McKee, B. A., Eds. Remote Sensing of Coastal Aquatic Environments. Technologies, Techniques and Applications; Springer: Secaucus, NJ, 2005. 10.1021/ac100524z  2010 American Chemical Society Published on Web 04/29/2010

Analytical Chemistry, Vol. 82, No. 11, June 1, 2010

4537

cyclic voltammetry and amperometry,12 stripping voltammetry,13 normal pulse voltammetry,14 and pulsed chronopotentiometry.15,16 When exploring these as sensor readout principles, the membrane was normally chosen not to exhibit ion-exchanger properties in order to achieve a sample-membrane interface that can be effectively polarized by the electrochemical perturbation step. The groups of Sanchez-Pedreno17 and Osakai18 were reportedly the first to use ion transfer voltammetry principles in a flow system to demonstrate the coulometric detection of very lipophilic ions. These two early works did not yet utilize chemical receptors (ionophores), and the coulometric efficiency was not yet optimal. Kihara and co-workers offered the first description of small ion detection with ionophores with this approach, using a porous Teflon tubing that was wholly immersed into an organic solvent that contained a lipophilic electrolyte and a chemical receptor.19 Kihara and co-workers indeed showed that the absolute detection of ionic species is possible with this approach. Despite its elegance, this approach does not yet form the basis for a convincing practical device because of the required large volume of volatile organic solvent and high quantities of specialty chemicals. We believe that a membrane based sensing approach with modern ion-selective electrode materials is required to achieve this goal. We show here that a liquid membrane based coulometric ion detector can be achieved by utilizing membrane components typically used in polymeric ion-selective electrodes and with membranes that exhibit ion-exchanger properties. A common ion that is potential determining at both interfaces greatly simplifies the theoretical understanding of these systems relative to earlier cell designs, where the cation extraction at one interface was typically electrochemically coupled to an anion extraction step at the other side.20 With thin membranes or membranes that have high ion mobilities, as desired in coulometry, this would be problematic because these two ions may eventually meet and result in significant water uptake with associated swelling and opacity increase of the membrane.21 As an important other benefit, the approach introduced here also allows one to monitor the open circuit potential of the cell and to use this information to apply the correct potential for the coulometric readout step. EXPERIMENTAL SECTION Reagents, Materials, and Equipment. Dodecyl 2-nitrophenyl ether (DDNPE), tetradodecylammonium tetrakis(4-chlorophenyl)borate (ETH 500), potassium tetrakis[3,4-bis(trifluoromethyl)phenyl]-borate (KTFPB), calcium ionophore IV (ETH 5234), high molecular weight poly(vinyl chloride) (PVC), tetrahydrofuran (THF), calcium chloride dihydrate, and potassium chloride were purchased in the highest quality available from Fluka. Aqueous solutions were prepared by dissolving the appropriate salts in MilliQ-purified distilled water. (12) (13) (14) (15) (16) (17) (18) (19) (20) (21)

Samec, Z.; Samcova, E.; Girault, H. H. Talanta 2004, 63, 21–32. Kim, Y.; Amemiya, S. Anal. Chem. 2008, 80, 6056–6065. Jadhav, S.; Meir, A. J.; Bakker, E. Electroanalysis 2000, 12, 1251–1257. Shvarev, A.; Bakker, E. Anal. Chem. 2003, 75, 4541–4550. Gemene, K.; Bakker, E. Anal. Chem. 2008, 80, 3743–3750. Sanchez-Pedreno, C.; Ortuno, J. A.; Hernandez, J. Anal. Chim. Acta 2002, 459, 11–17. Sawada, S.; Taguma, M.; Kimoto, T.; Hotta, H.; Osakai, T. Anal. Chem. 2002, 74, 1177–1181. Yoshizumi, A.; Uehara, A.; Kasuno, M.; Kitatsuhi, Y.; Yoshida, Z.; Kihara, S. J. Electroanal. Chem. 2005, 581, 275–283. Jadhav, S.; Bakker, E. Anal. Chem. 2001, 73, 80–90. Jadhav, S.; Bakker, E. Anal. Chem. 1999, 71, 3657–3664.

4538

Analytical Chemistry, Vol. 82, No. 11, June 1, 2010

The polypropylene hollow fiber Accurel PP Q3/2 was supplied by Membrana GmbH (Wuppertal, Germany), and Sterling Silver Round wires of 250 and 500 µm i.d. were supplied by AEMETAL (Sydney, Australia). Unless otherwise specificied, all measurements were performed at room temperature vs a double-junction Ag/AgCl/sat. KCl/1 M LiOAc reference electrode (Mettler-Toledo AG, Schwerzenbach, Switzerland) placed in the outer aqueous solution. The ion-selective hollow fiber cell served as the working electrode, and a high surface area Pt electrode (MT Electrodes, Perth, Australia) served as the counter electrode in coulometric measurements. The potentiometric calibration was performed using a 16-channel EMF monitor (Lawson Laboratories, Inc., Malvern, PA) connected to a personal computer. The coulometric measurements were performed with an Autolab PGSTAT128N (Metrohm Autolab, Utrecht, The Netherlands) attached to a personal computer. Calibrations were performed by feeding the cell with appropriate solutions using an ISMATEC peristaltic pump (Glattbrugg, Switzerland) with a flow rate of 25 µL · min-1. Preparation of the Coulometric System. Silver wires of 15 cm in length and 250 (for the twisted configuation) or 500 µm diameter were cleaned with acetone and water before chlorination. A 9 cm section of the cleaned wire was dipped in a 1 M HCl solution. A potential of 0.5 V against double junction reference electrode was applied between the silver wire and the high surface area platinum counter electrode. The process was carried out for 2 min. The silver/silver chloride wire was rinsed thoroughly with distilled water after chlorination. The polypropylene porous tube of 600 µm inner diameter, 200 µm wall thickness, and 7 cm length was washed inside and out with distilled water and THF. The chlorinated part of the wire was introduced into the polypropylene tube to its entire length. Two polypropylene pipet tips were used as fluidic connections for the flow-through system. The bare, nonchlorinated part of the silver wire was directed through the outlet tip to allow for the electrical connection. After assembly, the porous fiber was doped with 200 µL of the membrane cocktail composed of 266.7 mg of DDNPE, 30 mg (10 wt %) of ETH 500, 2.4 mg (10 mmol · kg-1) of Ca-ionophore IV, 0.9 mg of KTFPB (30 mol % relative to the ionophore), and 1.5 mL of THF. The obtained ion-selective membrane was left to dry overnight, and any excess of the liquid membrane was wiped off with tissue paper. A PVC slurry was used to seal the inlet and outlet of the cell. The resulting inner active surface area of the membrane was about 113 mm2, as estimated from the nominal dimensions. A silicone tube was fixed to the inlet of the cell and connected to the peristaltic pump. The inner side of the cell was washed for a day by continuously flowing 1 mM CaCl2 solution at the flow rate mentioned above. Procedures. A potentiometric calibration was performed using a silver/silver chloride wire immersed in the outer solution of a fixed composition (10-3 M CaCl2 + 10-2 M KCl) connected to the potentiometer as indicator electrode while the silver/silver chloride inside the hollow fiber membrane was connected to the reference port. Sample solutions of a fixed 10-5, 5 × 10-4, 10-4, 5 × 10-3, and 10-3 M CaCl2 concentration and with 10-2 M KCl as background electrolyte were continuously passed through the polypropylene tube, and the associated potential

changes in the cell were recorded. The activities of calcium ion and the liquid junction potential were corrected with the Debye-Hu ¨ ckel approximation and the Henderson equation. All other experiments were performed in a 3-electrode system with the silver/silver chloride wire inside the hollow fiber, which now acted as the working electrode. The same outer solution as above was used (10-3 M CaCl2 + 10-2 M KCl). The peristaltic pump was always off during the following measurements. For the current interruption test, four alternative +1 µA and -1 µA 3 s long pulses were applied with 30 s regeneration pulses at 0 A current following and preceding each pulse. Normal pulse voltammetry was performed within a range of 250 mV starting from the open circuit potential (OCP) and going toward more positive values. Potential pulses (1 s) in 10 mV potential increment were followed by 5 s OCP regeneration pulses (baseline potential). For the coulometric analysis procedure, three pulses were applied. An initial 20 s pulse at the OCP confirmed the stability of the system, followed by a 30 s OCP + n · 25 mV calcium extracting pulse (n ranging from 1 to 10). A third 60 s pulse at OCP was used to regenerate the system before a new portion of sample was delivered to the active electrode area. RESULTS AND DISCUSSION In order to maximize the ion mobility of the membrane, porous polypropylene tubing was doped with a membrane containing plasticizer without the use of traditional poly(vinyl chloride) (PVC). Similar hollow fiber membrane materials were introduced by the group of Buffle as permeation membrane sampling systems.22 Preliminary experiments on similarly formulated 20 µm thick Celgard membranes showed that membranes without any PVC did not exhibit a long operational lifetime with the traditional plasticizer bis(2-ethyl hexyl sebacate) (DOS). This might be due to nonoptimal casting conditions, using the solvent THF in the membrane cocktail slurry. Earlier work on doping 20 µm thin Celgard membranes (based on porous polypropylene as well) with DOS yielded functional membranes for a period of 1 to 2 days,23 which is expected to improve with the much thicker tubular membrane used here (200 µm). To eliminate potential issues with limited lifetime, the plasticizer dodecyl 2-nitrophenyl ether, a more lipophilic analog of o-NPOE, was used. This solvent has been used in ion-selective electrode applications for over 30 years and exhibits an estimated octanol-water partition coefficient of log P ) 7.4, 2 orders of magnitude larger than that of o-NPOE.24 It was found earlier in optical bead based sensors to exhibit desirable lipophilicity characteristics as well, without apparent deteriorating influence on ion selectivity.25 Note that polar (yet lipophilic) plasticizers are known to result in membranes with improved selectivity characteristics for divalent ions such as calcium26 and, hence, was preferred here. The membranes also contained the lipophilic electrolyte ETH 500 to keep the resistance of the membranes as low as possible, (22) Ueberfeld, J.; Parthasarathy, N.; Zbinden, H.; Gisin, N.; Buffle, J. Anal. Chem. 2002, 74, 664–670. (23) Malon, A.; Bakker, E.; Pretsch, E. Anal. Chem. 2007, 79, 632–638. (24) Oesch, U.; Simon, W. Anal. Chem. 1980, 52, 692–700. (25) Telting-Diaz, M.; Bakker, E. Anal. Chem. 2002, 74, 5251–5256. (26) Anker, P.; Ammann, D.; Asper, R.; Dohner, R. E.; Simon, W.; Wieland, E. Anal. Chem. 1981, 53, 1970–1974.

Scheme 1. Schematic of the Thin Layer Ion-Selective Coulometric System Presented Here, Consisting of a Hollow Fiber Based Calcium-Selective Working Electrode, a High Surface Area Pt Counter Electrode, and a Double Junction Ag/AgCl Reference Electrode Immersed in 1 mM CaCl2 + 10 mM KCla

a The working electrode is a 500 µm thick Ag/AgCl wire within the 600 µm i.d. polypropylene hollow fiber doped with 10 wt % ETH 500, 10 mmol · kg-1 Ca-ionophore IV, and 30 mol % (relative to the ionophore) of ion-exchanger KTFPB in DDNPE as a calcium-selective membrane. Sample solutions pass through the tubular electrode to form a thin sample layer (b) between the wire (a) and the membrane (c), while the wire is connected to the potentiostat at (d).

in analogy to earlier work.15,16,20 The resistance was estimated with the current interrupt method (±1 µA) as 20.3 ± 0.5 kOhm. The use of the calcium ionophore ETH 5234 was explored in this initial study because of its high selectivity27,28 (higher also compared to the well-known ETH 1001) and large complex formation constant with calcium,29 which affords a wide potential window for electrochemically mediated calcium extraction. The membrane was planned to exhibit cation-exchanger properties, which was achieved with the well known tetraphenylborate derivative KTFPB. On the basis of established ion-selective electrode theory,10 the countercation of the ion-exchanger will exchange with calcium during a conditioning step, which renders the calcium activity in the membrane largely independent of the calcium activity in the contacting aqueous phase. Under these circumstances, the open circuit potential across the membrane follows the established Nernst equation, which is used here to monitor the concentration changes in the cell and to apply a correct excitation potential. A chloridized silver wire of 500 µm diameter was inserted into the polyprolyene tubing of nominally 600 µm inner diameter, leaving sufficient room for the fluidic delivery of the thin layer sample solution. As shown in Scheme 1, the cell was completed by inserting the tubing into an aqueous solution of 1 mM CaCl2 + 10 mM KCl, in which the counter and reference electrodes were placed. Note that asymmetric bathing conditions on either side of the liquid membrane may result in undesired ion countertransport processes, and a potentiometric monitoring of the transmembrane potential allows one to conveniently evaluate concentration changes in the thin layer sample with time. Figure 1 demonstrates the potentiometric response as a function of the thin layer sample composition, ranging from 10-5 to 10-2 M CaCl2 in a 10 mM KCl background electrolyte. Indeed, the near-Nernstian response slope suggests that the thin layer sample composition is sufficiently unperturbed by the outside bathing solution. The slight deviation from Nernstian behavior at high concentration can be quantitatively explained by the increasing chloride (27) Gehrig, P.; Rusterholz, B.; Simon, W. Chimia 1989, 43, 377–379. (28) Bhakthavatsalam, V.; Shvarev, A.; Bakker, E. Analyst 2006, 131, 895–900. (29) Qin, Y.; Mi, Y.; Bakker, E. Anal. Chim. Acta 2000, 421, 207–220.

Analytical Chemistry, Vol. 82, No. 11, June 1, 2010

4539

Figure 1. Potentiometric calibration curve for the hollow fiber based calcium-selective electrode with an inner thin layer sample solution ranging from 10-5 to 10-2 M CaCl2 in a 10 mM KCl background electrolyte. The outer solution is constant at 1 mM CaCl2 + 10 mM KCl. The straight line is the least-squares fit between logarithmic concentrations of -5 and -3, with a slope of 29.6 mV per 10-fold concentration change.

activity, which has an influence on the potential at the contacting silver/silver chloride electrode, as evidenced by the corrected data on the same plot. On the basis of the concentration excess of potassium ion on either side of the membrane, the Nernstian response also demonstrates a high selectivity pot to calcium ions, as expected for this ionophore (logKCa,K ) 28 -9.3). To our knowledge, this marks the first time that a hollow fiber membrane is reported as an ion-selective membrane in potentiometry. The ion-selective calcium membrane was subsequently explored for operation in a voltammetric detection mode. Earlier work with plasticized PVC membranes containing an ionexchanger salt exhibited essentially featureless ohmic response characteristics at elevated sample concentrations, according to the Nernst equation and ohmic drop.21 This result was one key reason why further work on voltammetric sensors avoided ion-exchanger based membranes.20 However, it is expected that a featureless ohmic response can only be observed when concentration polarizations in the contacting aqueous phases are negligible.21 With thin layer samples or with low sample concentrations, one will likely observe a mass transport limited current response. This is schematically shown in Figure 2, where concentration dependent featureless ohmic responses at high concentrations for a primary ion I+ and a background ion J+ give way to a current-potential curve that assumes a well-defined limiting current within the potential window given by the two mentioned ohmic response curves. This behavior was first explored with normal pulse voltammetry, with a 200 µm twisted wire in the inner tubing compartment. Here, the inner sample volume is expected to be substantially 4540

Analytical Chemistry, Vol. 82, No. 11, June 1, 2010

Figure 2. Schematic presentation of the expected current-potential behavior of ion-exchanger based ion-selective membrane electrodes. The straight lines for the primary ion I+ and the background ion J+ are expected for high sample concentrations where sample depletion is negligible. The separation between the two lines is given by the membrane selectivity and the activity of the two ions. When mass transport of the primary ion in the aqueous phase becomes rate limiting, as with low concentrations or volumes, a limiting current that is linearly dependent on concentration is expected between the two straight lines that define the potential window.

large, resulting in acceptably small sample perturbation between measurement pulses compared to a larger silver wire diameter. In normal pulse voltammetry, a correct baseline potential value needs to be applied between pulses that allows the imposed membrane concentration polarizations to essentially return back to its original, unperturbed state.14 This baseline potential, of course, must here be the open circuit potential,30 which is experimentally accessible. The normal pulse voltammetry protocol, therefore, consists of observing the open circuit potential, confirming the absence of potential drift, applying this potential for a fixed amount of time, confirming a near-zero current response, and applying 1 s potential excitation pulses at given increments relative to this baseline potential. Figure 3 demonstrates the resulting current responses for the mass transport of calcium from the inner compartment of the tubing into the membrane and outside solution. Clearly, the currents are limited by calcium mass transport, as evidenced by the linear relationship between current plateau and calcium concentration. Note that, at potentials near the baseline potential, an ohmic response to calcium can be discerned, while at large potential increments beyond the current plateau, the ohmic behavior toward the background ion potassium is observed. The potential window for the plateau is a function of the ion selectivity of the membrane and the composition of the sample solution. This behavior compares well to that schematically shown in Figure 2. To explore a thin layer coulometric operation of the tubular membrane, a thicker chloridized silver wire was inserted, as for the potentiometric data presented in Figure 1. The protocol for the coulometric detection was similar as for normal pulse voltammetry. The open circuit potential was monitored to confirm the absence of potential drifts, and subsequently, this potential was applied, confirming very low current readings. A potential incre(30) Shvarev, A.; Bakker, E. Anal. Chem. 2005, 77, 5221–5228.

Figure 3. Normal pulse voltammograms obtained for a hollow fiber based system with an increased volume of inner sample solution (smaller diameter silver wire) containing varying concentrations of CaCl2 in a constant background of 10 mM KCl, while the outer solution consists of a constant 1 mM CaCl2 + 10 mM KCl. Note the concentration dependent limiting currents for this ion-exchanger based membrane system that compare to Figure 2.

ment was then applied relative to the open circuit potential, now for a longer period to allow for the exhaustive depletion of the thin layer sample. Subsequently, the open circuit potential was again applied to return the membrane to its unperturbed state and to regenerate the silver chloride wire.14,31,32 The thin layer sample solution was replaced with a fresh sample before applying the next measurement. Figure 4 shows the observed current decays for the indicated potential increments (shown as ∆E, the applied potential minus the open circuit potential) for three different sample concentrations: (A) 1.0 × 10-5 M, (B) 2.5 × 10-5 M, and (C) 5.0 × 10-5 M. Note that the initial current responses follow Ohm’s law, similar to that shown in Figure 2, since thin layer sample depletion is negligible at short measurement times. Subsequently, the currents decay within the 30 s time period. For the 25 µM calcium sample at a ∆E of 150 mV, for example, the initial current of 27.4 µA reduces to 1.2 µA. The data also show that, with increasing potential, the current decay curves start to essentially overlap, which is consistent with an exhaustive mass transport limited process. Note, however, the increasing current amplitudes for 10 µM calcium with larger applied potentials (above 175 mV relative to the open circuit potential) in Figure 4, which do not decay to zero at long times. This is clearly due to concurrent extraction of potassium at the higher applied potential, resulting in interference. It is anticipated that extended measurement protocols that incorporate an in situ background subtraction step may distinguish to some extent between electrolyte that exhaustively depletes and more concentrated background electrolyte that does not. Such studies will be forthcoming in future work. Figure 5A illustrates the calculated charge for the individual current responses as a function of applied potential (vs open circuit potential) for different calcium concentrations. Clearly, a potential window exists where the charge is strongly concentration dependent, which confirms that thin layer coulometric readout is indeed dependent on the calcium level in the thin layer sample. The width of this potential window is dependent on the selectivity (31) Bakker, E.; Meir, A. J. SIAM Rev. 2003, 45, 327–344. (32) Zook, J. M.; Lindner, E. Anal. Chem. 2009, 81, 5155–5164.

Figure 4. Chronoamperograms obtained for a thin layer (500 µm thick Ag/AgCl wire) coulometric system for samples of (A) 1 × 10-5 M, (B) 2.5 × 10-5 M, and (C) 5 × 10-5 M CaCl2 all in a 10 mM KCl background during various 30 s potential pulses vs open circuit potential, as indicated on the graph. Other conditions as in Figure 3.

of the membrane, with higher levels of interference diminishing the potential window. In a practical protocol, the measured open circuit potential and the available potential window may likely allow one to apply appropriate potentials that can be applied to samples of variable composition. Note that there is some increase of the observed charge with applied potential within the aforementioned potential window, which is mainly explained by non-Faradaic charging currents, which have here not yet been corrected for. For comparison, the data are replotted in Figure 5B as applied potential vs Ag/AgCl, in which case the response toward the KCl background electrolyte can be better visualized. Further work will explore in situ background correction protocols to minimize this contribution. Analytical Chemistry, Vol. 82, No. 11, June 1, 2010

4541

Figure 6. Chronoamperograms of the thin layer system for the sample concentrations indicated on the graph (other conditions as in Figure 4). Here, the potential applied for individual concentrations was always 150 mV vs open circuit potential. The inset presents the calculated charge as a function of sample concentration from the chronoamperograms shown, confirming linear behavior.

squares fit of the calibration curve (0.882 µC/M) was found 10 days later as 0.868 µC/M. As a typical example, the relative deviations in this time period of the determined charge were on the order of 2% for 50 and 75 µM concentrations of calcium.

Figure 5. Calculated charge as a function of applied potential (A) vs open circuit potential and (B) vs Ag/AgCl wire for the calcium concentration indicated on the graphs. Other conditions as in Figure 4.

Repeated injections of three identical calcium samples and of samples with varying calcium concentration in a background of 10 mM KCl were performed and measured at an applied potential of 150 mV more positive than the open circuit potential. The observed current decays for the different concentrations are shown in Figure 6, while the calculated charge as a function of concentration is illustrated in the inset. No background correction was performed to best illustrate the raw performance of the system. The expected linear relationship between charge and calcium concentration is indeed observed. The nonspecific processes discussed above give a contribution of 25 µC to the background. The repeatability is very good for the initial system presented here, with relative standard deviations ranging from 0.11% for 10 µM calcium to 0.90% for 100 µM calcium (N ) 3). While the focus of this study was to introduce the underlying response principles, it is noted that higher calcium concentrations started to deviate from the calibration curve in Figure 6. For these concentrations, the electrolysis time of 30 s was not sufficiently long, as evidenced by current decays that were not exhaustive. Longer detection times, thinner membrane tubing, larger pore size of the membrane support, increased mobility within the membrane, and a thinner sample layer are all expected to increase the upper limit of detection. Higher detection limits were observed with the thin layer coulometry work of Kihara, for example, who used bulk organic solvents as the sensing phase.19 The slope of the least-

4542

Analytical Chemistry, Vol. 82, No. 11, June 1, 2010

CONCLUSIONS The study presented in this article demonstrates potential controlled thin layer coulometry with ionophore based ionselective membranes as a new technique for aqueous sample analysis. It was shown that the newly designed flow-through system consisting of a hollow fiber based ion-selective membrane and a silver/silver chloride inner electrode exhibits a potentiometric response behavior typical of a potentiometric ion-selective electrode. More importantly, it was found that an applied potential pulse forced the selective transport of the primary ion from the thin layer sample solution through the membrane and into the outer solution. The process occurs until the sample is exhausted which results in a monitored current decay. The charge calculated from the generated signal is indeed linearly related to the sample concentration. The relative standard deviation observed for several repeated sample measurements was found to be below 1% and within a time period of 10 days on the order of 2%, making this work a promising initial step toward calibration-free analysis. An effective, in situ evaluation and correction of non-Faradaic charging currents and a more thorough theoretical study are underway to further develop this promising methodology. ACKNOWLEDGMENT The authors are grateful to the Australian Research Council (DP0987851), the CSIRO through the Flagship Cluster “Sensors Systems for Analysis of Aquatic Environments”, and the National Institutes of Health (R01-EB002189) for financial support of this research.

Received for review February 26, 2010. Accepted April 20, 2010. AC100524Z