Paper-Based Thin-Layer Coulometric Sensor for Halide Determination

Jan 7, 2015 - Paper-Based Thin-Layer Coulometric Sensor for Halide ... layer, the construction of disposable halide sensors, and portability for measu...
0 downloads 0 Views 2MB Size
Article pubs.acs.org/ac

Paper-Based Thin-Layer Coulometric Sensor for Halide Determination Maria Cuartero, Gastón A. Crespo, and Eric Bakker* Department of Inorganic and Analytical Chemistry, University of Geneva, Quai Ernest-Ansermet 30, CH-1211 Geneva, Switzerland S Supporting Information *

ABSTRACT: We report on a paper-based analytical device (PAD) for the exhaustive, and therefore absolute, determination of halides in a range of diverse water samples and food supplements. A mixture of chloride, bromide, and iodide ions is assessed in a wide range of concentrations, specifically, from 10−4.8 to 0.1 M for bromide and iodide and from 10−4.5 to 0.6 M for chloride, with a limit of detection of 10−5 M. As a result of a careful optimization of the electrochemical cell, a thin layer made of cellulose paper (75-μm thickness), a cationexchange Donnan exclusion membrane (FKL), and a silver-foil working electrode were selected as optimum materials. Cyclic voltammetry (from 0 to 0.8 V) was chosen as the interrogation technique to impose the exhaustive oxidative plating and rereduction of halides on the silver element, accompanied by outward and inward counterion fluxes. The scan rate plays an important role in the ability of the technique to resolve mixtures of ions. Moderate scan rates (10 mV s−1) provide a suitable compromise between sensitivity, limit of detection, and resolution. This paper-based microfluidic device is extremely simple in terms of manipulation, cost, and contamination risk. Paper is an excellent basis for the establishment of a confined thin aqueous layer, the construction of disposable halide sensors, and portability for measuring outside the controlled laboratory environment. A discussion of the relevant analytical characteristics is presented herein, followed by a demonstration of halide assessment in water samples (sea, tap, river, and mineral waters) and food supplements enriched with iodide and chloride as early examples.

P

Although many key characteristics of paper have been considered so far in the development of analytical devices, the ability to create a well-defined thin aqueous layer has not been sufficiently exploited from an electroanalytical point of view. Indeed, the combination of confined thin-layer samples with volumes on the order of a few microliters or less in paper substrates with a coulometric readout would offer an exhaustive analyte consumption, resulting in an ideally calibration-free methodology that would only then become truly useful in lesscontrolled field situations. The basis of thin-layer coulometry for electroactive species at metal electrodes was primarily established by Bard.23 More recently, the utilization of ion-selective membranes coupled with coulometry has been the focus of attention of Kihara and co-workers as well as our group.24−28 Specifically, the thin-layer sample defined by the void between a Ag/AgCl wire element and a microporous hydrophobic tube doped with ion-selective components has been at the core of this work.29 As a result of an applied potential, the analyte ions are preferentially and exhaustively transported through the selective membrane. The integrated charge was found to vary linearly with the concentration. Subsequently, this coulometric methodology

aper-based devices have undergone rapid growth with notable implications for the analytical chemistry community.1−4 This expansion can be understood as a direct result of the unique and intrinsic characteristics of paper, which is an inexpensive material with a nearly infinite range of thickness, porosity, and chemical structure. Depending on the specific paper substrate, paper is gas permeable; has a high mechanical flexibility; and exhibits the abilities to be impregnated with liquids through capillary action, to hold and retain liquids, to be colored and printed, and to be chemically functionalized.5,6 Indeed, the use of paper substrates with the ability to transport liquids through capillary forces allowed the establishment of microfluidic paper-based analytical devices (μPADs).7−11 The groups of Henry, Whitesides, and others put forward the concept of incorporating either electrochemical (potentiometry and amperometry) or optical detection for multianalyte detection.12,13 Consequently, papers have become conducting materials after the addition of suitable compounds such as silver- or carbon-based inks.14−18 Because of the simplicity associated with a potentiometry readout, various paper-based ion-selective and reference electrodes have been reported with different geometries, such as tattoo sensors.19 This recent work indicates that such paper-based analytical devices might become an excellent platform for disposable sensors in scenarios where rapid decisions are required, as in clinical point of care20,21 and food analysis.22 © XXXX American Chemical Society

Received: November 25, 2014 Accepted: January 7, 2015

A

DOI: 10.1021/ac504400w Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

specialized technical training would be most welcome, in analogy to early μPAD devices.5,15 In this context and based on our previous experience with thin tubular-shaped layer samples, we aim here to establish an inexpensive, robust, and reliable microfluidic paper-based analytical device for the discrimination and exhaustive determination of various halides in complex matrixes. Cyclic voltammetry at moderate scan rates was chosen as the preferred electrochemical technique, which permits the sequential oxidation/plating of the halide at the silver wire, followed by subsequent regeneration with an inverted potential. Moreover, we report on the facile fabrication of a magnetically latched detection cell consisting of just two acrylic plates and four magnets. Although sample introduction is by capillary action, the force exerted by these magnets is appropriate for defining the thin-layer distance and holding the components of the electrochemical cell (papers, membrane, and Ag foils) in place during measurements. The device fabrication is significantly simpler than that in a previously reported work37 and the use of low-cost materials (Plexiglas, homemade electrodes, flat dialysis membranes, and papers) allows for disposable use. Moreover, the wicking properties of paper alleviate the need for a pumping system. Beyond the facile construction of the electrochemical cell, the novelty of this work is centered on the possibility of performing thin-layer coulometry with papers and their extraordinary analytical characteristics. As far as we know, this is the first time that a paper-based platform has been utilized as an absolute method for resolving mixtures of different compounds such as halides, and we therefore believe that this concept might become an important tool for diverse situations.

was extended to a great number of representative analytes including polyions (for anticoagulant heparin diagnostics),30 nutrients (nitrate and nitrite),31−33 and other ions (calcium and potassium).34,35 In a related design, a tubular ion-exchange Donnan exclusion membrane (IEDEM) such as Nafion was used to replace the ion-selective membrane. This configuration allows one to desalinate the sample while maintaining the concentrations of sample species that do not form insoluble silver salts, such as nutrients.36 This concept could have a significant impact on nutrient detection in seawater where the concentration of chloride (∼0.6 M NaCl) is reduced to about 1 mM. On the other hand, the same design can be used to determine various halides by cyclic voltammetry.37 Confinement of the sample between a silver element and a Nafion tubing membrane allows one to resolve a mixture of the three halides chloride, bromide, and iodide in a wide linear range and in an absolute, temperature-independent manner. As demonstrated earlier,36 the formation of a thin-layer sample is crucial to the determination of chloride, bromide, and iodide in complex samples. Chloride, bromide, and iodide are ubiquitous in different types of samples such as water and commercial food supplements. Moreover, seawater analysis is often difficult because the large amount of NaCl usually masks the determination of other ions, including other halides, nutrients, and carbonate species. Although chloride is present in a majority of natural waters, this is not the case for bromide and iodide. The maximum concentration of chloride permissible in potable waters is 250 ppm,38 which is rather associated with a tolerated flavor than health reasons. Regarding bromide and iodide, regulations for their limits are not established in all the countries and limiting values in drinking water are related to normal diets followed in the areas in question.39,40 These halides are routinely measured by either ion chromatography or volumetric titration. The first methodology is very reliable today (after years of instrument development), but disadvantages typical of benchtop devices remain. These include the cost of analysis (pumps, columns, injector, quality of solvents, etc.), the lack of portability, time-consuming sample preparation (filtering at 0.22-mm pore size), and the need for trained personnel to operate the equipment. Under optimized conditions, the analysis time for determining these halides is about 15 min for each sample without considering the calibration curve. The second methodology is based on titrimetric analysis by adding AgNO3 solution to a sample aliquot, accompanied by visual (formation of a colorful precipitate), potentiometric (Ag-based selective membrane), or conductometric sensors as end-point detectors. Even though argentometric analysis allows one to measure individual halides at relatively high concentrations (higher than 10 mM), a mixture of three halides is difficult to resolve as coprecipitation of silver halides can occur. Evidently, this type of procedure still requires sampling and splitting into aliquots and does not lend itself to detections in situ. For applications in which in situ or on-site measurements are required, these methodologies might not be the most suitable choice. Therefore, an alternative method that is able to measure halides in a few minutes is relevant for some environmental applications or food quality control. For this purpose, disposable sensing elements (emulating cartridge or screenprinted technology) operated in a manner not requiring



EXPERIMENTAL SECTION Reagents, Materials, and Equipment. Aqueous solutions were prepared by dissolving the appropriate sodium salts (Sigma-Aldrich) in deionized water (>18 MΩ/m). Silver foils with a purity of 99.97% (50 × 50 mm and 25-μm thickness) were supplied by Advent Research Materials (Oxford, U.K.). Whatman regenerated cellulose filters (RC60 membrane circles, 47-mm diameter, 1-μm pore size, 75-μm thickness), qualitative Whatman filter papers 2 (90-mm diameter, 8-μm pore size, 190-μm thickness), and MAGNA nylon supported (47-mm diameter, 5-μm pore size, 160-μm thickness) were obtained from Sigma-Aldrich. FKL cation-exchange membranes with thicknesses of 120−130 μm were purchased from Fumatech (FuMA-Tech GmbH, St. Ingbert, Germany), whereas NR-211 Nafion membranes with a thickness of 25 μm were obtained from Ion Power Inc. (New Castle, DE). FKL membranes were pretreated in deionized water for 6 h at room temperature and then in 1 M HNO3 for 1 h, to ensure that the membrane was in H+ form. Nafion and FKL membranes were placed in 1 M NaCl for 24 h before being used and were also maintained in the same solution. Nafion-coated papers were prepared by dip coating of Nafion perfluorinated resin solution (Sigma-Aldrich) on the three papers. A multivitamin complex that is administered as a food supplement was purchased at a local pharmacy. Cyclic voltammograms were obtained on a PGSTAT 302N apparatus (Metrohm Autolab, Utrecht, The Netherlands) controlled by a personal computer using Nova 1.8 software (supplied by Autolab). The same device was used for differential pulse voltammetric, square-wave voltammetric, and electrochemical impedance spectroscopy measurements. A Faraday cage was used to record the impedance spectra to B

DOI: 10.1021/ac504400w Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Figure 1. Scheme of the electrochemical cell with a closing system based on acrylic blocks secured by magnets: (a) closed, (b) opened, (c) arrangement of elements inside the cell (i, silver sheet; ii, silver/silver iodide sheet; iii, permselective membrane; iv, paper for sample chamber; v, paper for outer solution; vi and vii, electrical connections to the potentiostat), (d) electrochemical process.

that were tightly closed by means of a homemade hinge and two magnets (Figure 1). The position of each cell element was crucial to the control of the obtained signal. Figure 1 shows the correct manner of placing the elements. It is essential to always achieve the same sample volume in contact with the working electrode to obtain reproducible measurements. Note that the membrane has to completely separate the two parts of the system, namely, the working electrode in contact with one paper for sampling from the counter/reference electrode in contact with the other paper to confine the outer solution. Furthermore, the membrane must also separate the two papers outside the cell just up the contact between them and the respective liquid solutions. This cell configuration prevents the contact between the two papers and both electrodes between them. Each filter paper is dipped in a small beaker containing either the sample or the outer solution. The liquids reach the corresponding electrode surface after approximately 4 min of exposure. Both papers are immersed in the corresponding solution during the entire analysis. This allows one to avoid eventual liquid evaporation due to air flow from the outside part of the paper in contact with the environment.

protect the system from undesired noises. Potentiometric measurements were taken against a double-junction Ag/AgCl/3 M KCl/1 M LiOAc reference electrode (Metrohm AG, Herisau, Switzerland), using a 16-channel EMF interface (Lawson Laboratories, Inc., Malvern, PA). All experiments were carried out at room temperature of 22 ± 1 °C. In the case of water sample analysis, a Metrohm 761 compact IC chromatograph with an anion-exchange column (6.1006.520 Metrosep A Supp 5) was used as a reference method. The eluent was a solution composed of 1 mM NaHCO3 + 3.2 mM Na2CO3, along with 50 mM H2SO4 for regeneration of the anion suppressor (0.8 mL min−1). Preparation of the Paper-Based Electrochemical Cell. The pieces of silver foil (40 × 10 mm) were cleaned first with acetone and then with deionized water. A copper wire was soldered to one of the ends of each piece of foil to make appropriate electrical connections. One of these bare silver foils served as the working electrode. The other foil was oxidized electrochemically in a solution of either 1 M NaI or 1 M HCl, depending on the case, for 3 h at a constant anodic current density of 0.4 mA cm−2. The coated electrode foil was washed with abundant deionized water and served as the counter/ reference electrode. Two pieces (40 × 9 mm) of the corresponding paper were cut along with a piece (30 × 15 mm) of the membrane (i.e., Nafion, FKL, or Nafion-coated paper). The working electrode, one of the papers, the membrane, the other paper, and the reference/counter electrode were placed in this order between two acrylic blocks



RESULTS AND DISCUSSION This work reports on the development of a paper-based electrochemical cell, into which the sample is introduced by capillary action through a commercial filter paper, for the absolute detection of mixtures of halides. The sample is C

DOI: 10.1021/ac504400w Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

cell design (electrodes, membrane type, paper porosity, and thicknesses) and of the appropriate electroanalytical protocol (scan rate, calibration, and resolution) as described below. Initially, the optimization of the paper cell elements was accomplished by recording cyclic voltammograms of a symmetrical system composed of the same type of solution (0.1 M NaCl) as the sample and outer solution. Both Ag and Ag/AgCl foils were characterized as working electrodes against a Ag/AgCl reference electrode using a Nafion membrane. Figure 3a compares the obtained cyclic voltammograms with these two working electrodes. With the Ag/AgCl foil, the starting point of the oxidation peak was not as well-defined as that obtained with the bare Ag element. In addition, the reproducibility significantly improved for bare Ag foil compared to the oxidized foil (data not shown). In contrast to previous evidence with tubular Nafion membranes,37 no visible changes at the Ag foil or the membrane surface were observed with time, which is likely explained by variations in the chemical and mechanical internal structures of the two Nafion materials (porosity and channel forms).41−44 Because the bare Ag foil was found to be the most suitable candidate as the working electrode, subsequent experiments were carried out with this element. Additionally, different papers (normal filter paper, nylon paper, and regenerated cellulose paper; see the Experimental Section) and several types of cation-exchange membranes (Nafion, FKL, and Nafioncoated papers) were electrochemically characterized. Figure 3b shows a comparison between cyclic voltammograms obtained with papers of different thicknesses. The integrated charge of the oxidation peaks increased with paper thickness (0.234, 0.185, and 0.135 mC for normal filter paper, nylon paper, and regenerated cellulose paper, respectively) as a consequence of the increase in sample volume. The residual current after the oxidation peak transient was smallest for the 75-μm-thick paper (0.1 mA), suggesting a more exhaustive process, whereas for the thicker papers, this current was higher (0.4 and 0.8 mA for 190- and 160-μm papers, respectively). The smallest separation between the oxidation and reduction peaks was found for the regenerated cellulose paper (150 mV). Nylon and normal filter paper showed separations of 322 and 261 mV respectively, owing to the increasing thickness of the aqueous layer. On the basis of these results, regenerated cellulose paper (75-μm thickness) was chosen for further experiments. Figure 3c shows the observed cyclic voltammograms with the 75-μm-thick paper using three different membranes (Nafion, FKL, and Nafion-coated papers). Whereas Nafion and FKL are commercial membranes, Nafion-coated papers were prepared as described in the Experimental Section from liquid Nafion. Nafion-coated papers45 were fabricated with different paper thicknesses (190, 160, and 75 μm, the same papers tested as sample chamber elements). Cyclic voltammetry was utilized to characterize these Nafioncoated papers as candidates for the ion-exchange membrane. In the case of the 75-μm Nafion-coated membrane, the paper exhibited excessive fragility, whereas for 160-μm (Nylon support material), no electrochemistry was observed (see Figure 3c). This latter behavior is probably due to a lack of physical adhesion between Nafion and nylon. Suitable voltammograms for NaCl were achieved only using Nafioncoated paper of 190-μm thickness. Permselective properties were confirmed with a Nernstian response slope by potentiometry (Figure S1, Supporting Information) as previously reported for Nafion and FKL membranes.46

confined to the cavities/pores of the paper (wet paper) and brought into direct contact with the working electrode (silverfoil element) and the permselective membrane. The opposite side of the membrane is in contact with a second filter paper that contains another solution of sodium halide in contact with a silver/silver halide element part, which operates as the reference/counter electrode. All five elements (two papers, two silver foils, and the ion-exchange membrane) are squeezed together by magnetic elements embedded in the two acrylic blocks, as shown in Figure 1a,b (open and closed cell). The optimal arrangement of the mentioned elements is also displayed in Figure 1c. The operating principle is similar to that of a previous work.37 A linear potential sweep is applied to the two silver elements of the electrochemical cell. An anodic potential scan induces the oxidation of the silver foil, resulting in a current that reflects the plating rate of the halides in the confined sample as silver halide precipitates (see eq 1). Simultaneously, the halide counterion (in this case Na+) must be transported from the confined sample across the ion-exchange membrane to the outer solution to maintain charge neutrality in each compartment. This process is accompanied by reduction of the silver/silver halide element in the outer solution, which results in the release of halide ions into solution (Figure 1d). After the forward scan, a backward potential sweep is applied to regenerate the two silver elements to their previous states. The fundamental basis for the detection selectivity originates from the reduction potentials involving the associated solubility equilibria [E°(AgCl/Ag) = 0.222 V, E°(AgBr/Ag) = 0.0732 V, E°(AgI/Ag) = −0.152 V] AgX(s) + e− → Ag(s) + X−(aq)

(1)

To demonstrate the principle, a sample containing equal concentrations of the three halides I−, Br−, and Cl− was introduced into the thin-layer sample compartment by capillary action through the appropriate filter paper reaching out of the cell. A 0.1 M NaI solution was introduced in complete analogy with the second filter paper strip exposed to the outer compartment. Three peaks were resolved under optimum conditions (see below and Figure 2). These results suggest that a thin-layer sample can indeed be established with commercial paper materials, resulting in an attractive low-cost device for halide detection. This was followed by the optimization of the

Figure 2. Observed cyclic voltammogram in a mixed 10−2 M NaCl, NaBr, and NaI sample. Outer solution, 0.1 M NaI; scan rate, 10 mV s−1. D

DOI: 10.1021/ac504400w Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

the desired conditions and thus was used for further experiments. Electrical resistances were obtained by electrochemical impedance spectroscopy. (Nyquist plots are shown in Figure S2, Supporting Information.) Values of approximately 2330, 1150, and 410 Ω were obtained for cells with FKL, Nafion, and Nafion-coated paper membranes. Subsequently, the exhaustive voltammetric detection was extended to bromide and iodide anions. Cyclic voltammograms for 0.1 M NaCl, NaBr, and NaI were obtained using symmetrical systems, with 0.1 M solutions of the three sodium anion salts as inner and outer solutions (Figure S3a, Supporting Information). Well-defined oxidation peaks for chloride, bromide, and iodide were obtained at 188.4, 163.8, and −213.9 mV, respectively. However, consecutive experiments (alternating chloride, bromide, and iodide, in this order) provoked a chemical transformation of the Ag/AgCl reference/ counter electrode to the respective Ag/AgX element. Furthermore, residual oxidation peaks started to appear, even in the absence of the corresponding halide in the sample (see Figure S3a, Supporting Information). This is explained by the preference of AgI > AgBr > AgCl deposited on the electrodes (solubility products of 8.3 × 10−17, 5.4 × 10−13, and 1.8 × 10 −10 , respectively). As a consequence, an increasing irreproducibility for subsequent experiments was observed. The use of a Ag/AgI reference/counter electrode and 0.1 M NaI as the outer solution successfully overcame this undesired characteristic (see Figure S3b, Supporting Information). Note that the oxidation peak positions changed as a result of the altered reference potential (149.4, 301.9, and 510.4 mV for chloride, bromide, and iodide, respectively). Having established the appropriate configuration of the electrochemical cell, the electroanalytical characterization of the cell was performed. Panels a−c of Figure 4 show cyclic voltammograms for NaCl, NaBr, and NaI (0.1 M), respectively, at different scan rates (from 5 to 75 mV s−1). It is worth mentioning that the integrated charges from the oxidized and reduced peaks are roughly equivalent, suggesting electrochemical reversibility. For instance, the integrated charges for chloride detection at 10 mV s−1 were found as 0.1726 and 0.1731 mC from the oxidation and reduction peaks, respectively. Additionally, peak resolution improved with lower scan rates. Likewise, excellent linearity was found between peak intensity and scan rate, confirming thin-layer behavior. As shown in Figure 5a−c, separate calibration curves for Cl−, Br−, and I− were obtained with cyclic voltammetry as the readout principle. The integrated charge followed a linear relationship with concentration from 10−4.8 to 0.1 for Br− and I−. Considering a future application in samples (see below), the experimental concentration range for chloride was extended to 0.6 M (linearity from 10−4.8 M to 0.6 M was also observed). The linear regressions were charge = (1.37 ± 0.01)cCl mC M−1 + (0.0049 ± 0.0002) mC for chloride, charge = (1.32 ± 0.01) cBr mC M−1 + (0.0051 ± 0.0004) mC for bromide, and charge = (1.38 ± 0.01)cI mC M−1 + (0.0076 ± 0.0005) mC for iodide, with limits of detection of around 10−5 M. As expected, similar slopes were obtained [relative standard deviation (RSD) ≈ 2%], as expected for a coulometric readout principle. These values were calculated after subtraction from the corresponding residual peak in the absence of halide in the sample. The limit of detection of iodide was found to be slightly higher than those of the other two halides, likely due to the contamination of the thin-layer sample by an inward NaI electrolyte flux. Similar

Figure 3. (a) Comparison between observed cyclic voltammograms with bare Ag and Ag/AgCl foils as working electrodes; (b) cyclic voltammograms for different papers using a Ag/AgCl working electrode (papers are of different thicknesses); (c) cyclic voltammograms obtained with different permselective membranes, a Ag/AgCl working element, and regenerated cellulose filter papers. All measurements were carried out for 0.1 M NaCl (sample and outer solutions) at a scan rate of 10 mV s−1.

The FKL membrane displays a higher current density in Figure 3c in comparison to the Nafion and Nafion-coated papers. This suggests a more efficient transport accompanied by a larger sensitivity-to-noise ratio. Obviously, a compromise situation between sensitivity, efficiency, mechanical stability, and cost needs to be considered. The FKL membrane fulfills E

DOI: 10.1021/ac504400w Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Figure 4. Cyclic voltammograms at different scan rates of (a) 0.1 M NaCl, (b) 0.1 M NaBr, (c) 0.1 M NaI (scan rates of 75, 50, 20, 10, and 5 mV s−1 from top to bottom), and (d) 0.1 M NaCl, NaBr and NaI (scan rates of 75, 50, 20, 10, 5, and 2 mV s−1 from top to bottom). Inset: Peak height vs scan rate.

between the peak current and the scan rate for each halide (Figure S4, Supporting Information). The resolution between the three peaks can be improved with decreasing scan rates, as mentioned above. Whereas the iodide peak was always wellresolved, the bromide and chloride peaks overlapped under most conditions. A scan rate of 10 mV s−1 was found to result in optimum conditions for mixed halide samples. Lower scan rates do not substantially improve the results further. At this scan rate, oxidation peaks for iodide, bromide, and chloride appear at 212.9, 447.2, and 608.4 mV, respectively. The resolution of these three peaks was evaluated using different outer solutions (0.1 M NaI or 0.1 M NaCl, NaBr, and NaI; see Figure S5, Supporting Information) and cyclic voltammetry (CV), differential pulse voltammetry (DPV), and square-wave voltammetry (SWV) modes (see Figure S6, Supporting Information). The best resolution was obtained with a 0.1 M NaI outer solution, and therefore, this solution was maintained for the rest of experiments. DPV and SWV were applied at different potential steps (from 0.2 to 2 mV and from 0.2 to 1 mV, respectively). A potential step of 5 mV in both techniques was necessary to obtain resolved peaks (higher step values deteriorated the output signal; see Figure S6, Supporting Information) as these techniques do not apply a linear potential sweep. In contrast, cyclic voltammetry always gave three peaks, even at higher scan rates. Comparing appropriately resolved voltammograms for the different modes (potential step of 0.2 mV for DPV and SWV and 10 mV s−1 for CV), whereas the Br− and Cl− peaks were not

effects have been reported observed for other types of membranes such as Nafion36,37 and dialysis membranes.36,45,47 Specifically, iodide contamination was evaluated in our system from the residual iodide plating peaks, which appeared in samples that did not contain this halide. The iodide contamination for the conditions used here represents 1.5% of the integrated charge obtained for 10 −5 M NaI concentration. The experimental charge for halides was always smaller than calculated based on Faraday’s law without considering the porosity of the paper (i.e., 6.75 μL using a 75-μm-thickness filter paper). By comparing the calculated and experimental charges and assuming complete conversion, a linear relationship was found. In this manner, it is possible to estimate the effective sample volume from the slope of this equation. According to this calculation, the volume is around 0.2 μL, which is three times smaller than the previous one with a tubular Nafion membrane.37 This reduction factor is also in agreement with the integrated charge. For instance, 0.36 mC was obtained for the 0.1 M NaCl sample solution with the Nafion tubing cell and a charge of 0.12 mC was obtained with the paper-based cell. Moreover, a factor of 3 was also calculated from a comparison between the corresponding calibration slopes. The next step was focused on the resolution of a mixture of the three halides. Figure 4d displays cyclic voltammograms of mixed samples (10−2 M in each halide) obtained at different scan rates and using 0.1 M NaI outer solution. Thin-layer behavior was also confirmed by the observed linear relationship F

DOI: 10.1021/ac504400w Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

repeatability was found to be better than 1.5% RSD for the same cell. The reproducibility between different electrochemical cells (N = 4) was 3.5% RSD. This might indicate that the approach is sufficiently promising as a calibration-free system. Moreover, temperature fluctuations do not affect the measurements.37 Figure 6a shows cyclic voltammograms for different halide concentrations in mixed samples. Furthermore, the response for chloride concentrations from 0.1 to 0.6 M in the presence of a fixed Br− and I− concentration is visualized in Figure 6b. Linear calibration curves for the three halides were obtained when concentrations were plotted versus integrated charge (Figure 6c). The linear regressions obtained are charge = (1.38 ± 0.01) cCl mC M−1 + (0.0045 ± 0.0001) mC for chloride, charge = (1.31 ± 0.01)cBr mC M−1 + (0.0049 ± 0.0004) mC for bromide, and charge = (1.39 ± 0.02)cI mC M−1 + (0.0073 ± 0.0004) mC for iodide, with limits of detection of 10−5 M. It is noted that no significant differences were found in the obtained calibration parameters compared with individual calibrations of each halide. The linear range is broader and the limits of detection are lower for the paper-based cell than for the Nafiontubing-based cell published earlier.37 As shown below, this permits one, in principle, to analyze bromide in sea, tap, and mineral waters. This concept was also considered for measuring halides in food supplements without sample pretreatment. Cyclic voltammograms obtained for a food supplement; seawater; and river, tap, and mineral waters are shown in Figure 7. The amount of each halide was determined by external calibration. In the case of the food supplement, the presence of iodide and chloride was detected and quantified. An unfiltered solution of a tablet dissolved in 10 mL of MQ water was directly measured. The observed chloride and iodide levels (Table 1) correspond to the certified values given by the manufacturer. Concerning water analysis, Table 1 compares the obtained results with the method presented here and the reference method (ion chromatography, IC). An excellent correspondence is observed between the two methods. The new method is also able to analyze undiluted seawater without pretreatment, which is not possible in ion chromatography because of the saturation of the separation column at these high salinity levels. The complexity of the electrochemical cell, in terms of manufacturing, is dramatically reduced by the use of papers in comparison to our previous work using tubular membranes.37 The paper and the cation-exchange membrane are undoubtedly the core of this device and can be rationally considered as disposable materials because of their low costs. The remaining materials, silver foils and the acrylic cell, should be considered reusable or at least recyclable. The latter issue might be addressed with screen or inkjet printing technology, which allows one to develop inexpensive silver layers on papers. As stated above, this coulometric concept could become as simple as more established paper-based potentiometric sensors in terms of manufacturing and disposability in the near future. This detection methodology provides adequate selectivity and sensitivity for determining mixtures of halides (which cannot be achieved by potentiometry owing to cross-interference48−50) in an attractive concentration range using just a few microliters of sample in an inexpensive analysis platform.

Figure 5. Cyclic voltammograms at different concentrations of (a) NaCl, (b) NaBr, and (c) NaI at a scan rate of 10 mV s−1 (0.1, 0.05, 0.01, 5 × 10−3, 1 × 10−3, 5 × 10−4, 1 × 10−4, 5 × 10−5, and 10−5 M from top to bottom for bromide and iodide; 0.6, 0.5, 0.4, 0.3, and 0.2 M from top to bottom for chloride). Inset: Corresponding calibration curves obtained in triplicate. Scan rate = 10 mV s−1. Error bars are smaller than the data points.

sufficiently separated for SWV, similar resolution was achieved by DPV and CV. Nevertheless, CV was selected for further experiments because DPV requires longer analysis times and, importantly, does not provide a coulometric readout because the current is not continuously sampled. To obtain an accurate charge for the overlapped bromide and chloride peaks, a deconvolution Gaussian treatment was applied to the raw voltammogram.37 Using this protocol, the obtained



CONCLUSIONS We have reported on a microfluidic paper-based analytical device (μPAD) controlled by cyclic voltammetry that is able to G

DOI: 10.1021/ac504400w Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Figure 7. Voltammograms of (a) food supplement; (b) seawater; and (c) mineral, tap, and river waters. Scan rate = 10 mV s−1. Figure 6. (a) Voltammograms obtained for different concentrations of halides in equimolecular mixtures (0.1, 0.05, 0.01, 5 × 10−3, 1 × 10−3, 5 × 10−4, 1 × 10−4, 5 × 10−5, and 10−5 M from top to bottom). (b) Cyclic voltammograms using a fixed concentration of bromide and iodide (10−3 M) and different chloride concentrations (0.6, 0.5, 0.4, 0.3, 0.2, and 0.1 M from top to bottom). (c) Corresponding calibration graphs obtained in triplicate (chloride, circles; bromide, triangles; iodide, squares). Scan rate = 10 mV s−1. Error bars are smaller than the data points.

readout) followed a linear relationship with respect to halide concentrations and exhibited a limit of detection of around 10−5 M for chloride, bromide, and iodide mixtures. This concept was also applied to the detection of these three halides in sea, tap, mineral, and river waters, as well as in a food supplement. Interestingly, samples with high ionic strengths (i.e., seawater) were accurately measured without further complications, which is generally the main drawback of other methodologies. In contrast to ion chromatography, sample pretreatment such as filtration or dilution is not required for this method. We strongly believe that an inexpensive, reliable, robust, and portable analytical device controlled with dynamic techniques at our disposal will be most useful if it also functions without the need for field calibration. The work presented here

discriminate chloride, bromide, and iodide in several samples. The selected porous paper of 75-μm thickness combines two main operational functions: the physical transport of the sample into the thin-layer chamber by capillary action and the achievement of an exhaustive electrochemical process (coulometry), resulting in a potentially calibration-free methodology. The integrated charges from oxidation peaks (coulometric H

DOI: 10.1021/ac504400w Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

(13) Dungchai, W.; Chailapakul, O.; Henry, C. S. Anal. Chem. 2009, 81, 5821−5826. (14) Wang, L. B.; Chen, W.; Xu, D. H.; Shim, B. S.; Zhu, Y. Y.; Sun, F. X.; Liu, L. Q.; Peng, C. F.; Jin, Z. Y.; Xu, C. L.; Kotov, N. A. Nano Lett. 2009, 9, 4147−4152. (15) Martinez, A. W.; Phillips, S. T.; Whitesides, G. M.; Carrilho, E. Anal. Chem. 2010, 82, 3−10. (16) Novell, M.; Guinovart, T.; Blondeau, P.; Rius, F. X.; Andrade, F. J. Lab Chip 2014, 14, 1308−1314. (17) Novell, M.; Parrilla, M.; Crespo, G. A.; Rius, F. X.; Andrade, F. J. Anal. Chem. 2012, 84, 4695−4702. (18) Cui, J. W.; Lisak, G.; Strzalkowska, S.; Bobacka, J. Analyst 2014, 139, 2133−2136. (19) Bandodkar, A. J.; Hung, V. W. S.; Jia, W. Z.; Valdes-Ramirez, G.; Windmiller, J. R.; Martinez, A. G.; Ramirez, J.; Chan, G.; Kerman, K.; Wang, J. Analyst 2013, 138, 123−128. (20) Martinez, A. W.; Phillips, S. T.; Carrilho, E.; Thomas, S. W.; Sindi, H.; Whitesides, G. M. Anal. Chem. 2008, 80, 3699−3707. (21) Lankelma, J.; Nie, Z. H.; Carrilho, E.; Whitesides, G. M. Anal. Chem. 2012, 84, 4147−4152. (22) Escarpa, A. Lab Chip 2014, 14, 3213−3224. (23) Bard, A. J. Anal. Chem. 1968, 40, R64−&. (24) Sanchez-Pedreno, C.; Ortuno, J. A.; Hernandez, J. Anal. Chim. Acta 2002, 459, 11−17. (25) Yoshizumi, A.; Uehara, A.; Kasuno, M.; Kitatsuji, Y.; Yoshida, Z.; Kihara, S. J. Electroanal. Chem. 2005, 581, 275−283. (26) Bhakthavatsalam, V.; Bakker, E. Electroanalysis 2008, 20, 225− 232. (27) Amemiya, S.; Kim, Y.; Ishimatsu, R.; Kabagambe, B. Anal. Bioanal. Chem. 2011, 399, 571−579. (28) Bakker, E. Anal. Chem. 2011, 83, 486−493. (29) Grygolowicz-Pawlak, E.; Bakker, E. Anal. Chem. 2010, 82, 4537−4542. (30) Crespo, G. A.; Afshar, M. G.; Dorokhin, D.; Bakker, E. Anal. Chem. 2014, 86, 1357−1360. (31) Dorokhin, D.; Crespo, G. A.; Afshar, M. G.; Bakker, E. Analyst 2014, 139, 48−51. (32) Sohail, M.; De Marco, R.; Lamb, K.; Bakker, E. Anal. Chim. Acta 2012, 744, 39−44. (33) Ghahraman Afshar, M.; Crespo, G. A.; Dorokhin, D.; Néel, B.; Bakker, E. Electroanalysis, published online Jan 7, 2015, 10.1002/ elan.201400522. (34) Grygolowicz-Pawlak, E.; Bakker, E. Electrochim. Acta 2011, 56, 10359−10363. (35) Grygolowicz-Pawlak, E.; Bakker, E. Electrochem. Commun. 2010, 12, 1195−1198. (36) Grygolowicz-Pawlak, E.; Sohail, M.; Pawlak, M.; Néel, B.; Shvarev, A.; de Marco, R.; Bakker, E. Anal. Chem. 2012, 84, 6158− 6165. (37) Cuartero, M.; Crespo, G. A.; Ghahraman Afshar, M.; Bakker, E. Anal. Chem. 2014, 86, 11387−11395. (38) Drinking Water Contaminant Candidate List (CCL) and Regulatory Determination. U.S. Environmental Protection Agency: Washington, DC, 2014, http://water.epa.gov/scitech/drinkingwater/ dws/ccl/ (Accessed Nov. 2014). (39) Kaykhaii, M.; Sargazi, M. Spectrochim. Acta A: Mol. Biomol. Spectrosc. 2014, 121, 173−179. (40) Nabavi, S.; Alizadeh, N. Sens. Actuators B: Chem. 2014, 200, 76− 82. (41) Blake, N. P.; Petersen, M. K.; Voth, G. A.; Metiu, H. J. Phys. Chem. B 2005, 109, 24244−24253. (42) Yeager, H. L.; Steck, A. J. Electrochem. Soc. 1981, 128, 1880− 1884. (43) Zhao, Q. A.; Majsztrik, P.; Benziger, J. J. Phys. Chem. B 2011, 115, 2717−2727. (44) Okada, T.; Xie, G.; Gorseth, O.; Kjelstrup, S.; Nakamura, N.; Arimura, T. Electrochim. Acta 1998, 43, 3741−3747. (45) Agarwal, C.; Chaudhury, S.; Mhatre, A.; Goswami, A. J. Phys. Chem. B 2010, 114, 4471−4476.

Table 1. Analysis of Several Water Samples with the Proposed Methodology proposed methodology sample

chloride

food supplement

40.3 ± 0.5 mg tablet−1 (40 mg tablet−1 a) 21360 ± 70 ppm (21400 ppmb,c) 13 ± 1 ppm (13 ppmb,c) 17 ± 1 ppm (16 ppmb,c) 5 ± 1 ppm (6 ppmb,c)

sea water tap water mineral water river water

bromide nd (nda) 16 ± 1 ppm (nqb,d) 2 ± 1 ppm (