Chronopotentiometric Carbonate Detection with All-Solid-State

May 29, 2014 - (c) Schematic illustration of the all-solid-state membrane electrode mechanism. ... The relationship between the transition time (τ), c...
0 downloads 0 Views
Article pubs.acs.org/ac

Chronopotentiometric Carbonate Detection with All-Solid-State Ionophore-Based Electrodes Zdeňka Jarolímová, Gastón A. Crespo, Xiaojiang Xie, Majid Ghahraman Afshar, Marcin Pawlak, 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 present here for the first time an all-solid-state chronopotentiometric ion sensing system based on selective ionophores, specifically for the carbonate anion. A chronopotentiometric readout is attractive because it may allow one to obtain complementary information on the sample speciation compared to zero-current potentiometry and detect the sum of labile carbonate species instead of only ion activity. Ferrocene covalently attached to the PVC polymeric chain acts as an ion-to-electron transducer and provides the driving force to initiate the sensing process at the membrane−sample interface. The incorporation of a selective ionophore for carbonate allows one to determine this anion in a background electrolyte. Various inner electrolyte and all-solid-state-membrane configurations are explored, and localized carbonate depletion is only observed for systems that do not contain ion-exchanger additives. The square root of the transition times extracted from the inflection point of the chronopotentiograms as a function of carbonate specie concentration follows a linear relationship. The observed linear range is 0.03−0.35 mM in a pH range of 9.50−10.05. By applying the Sand equation, the diffusion coefficient of carbonate is calculated as (9.03 ± 0.91) 10−6 cm2 s−1, which corresponds to the established value. The reproducibility of assessed carbonate is better than 1%. Additionally, carbonate is monitored during titrimetric analysis as a precursor to an in situ environmental determination. Based on these results, Fc-PVC membranes doped with ionophores may form the basis of a new family of passive/active all-solid-state ion selective electrodes interrogated by a current pulse.

S

aims to cover applications that cannot be easily achieved with traditional sensor designs. Single-use determinations for clinical samples (blood, plasma, and urine) as well as in situ environmental monitoring are especially promising targets for such all-solid-state sensor systems. Recently, some of these materials were applied to develop a new family of sensors controlled with dynamic electrochemistry techniques.3,11−14 Although ion transfer stripping voltammetry at ion-selective membranes is a most promising technique for trace level analysis, chronopotentiometry may be most useful in speciation analysis, where information on the labile fraction of an analyte can be obtained. These dynamic electrochemistry techniques may therefore become complementary to a zerocurrent potentiometric readout. Here, the challenge is even more complex because a defined current or potential is applied to the membrane for a considerable period of time, requiring a high redox capacity to provoke a significant concentration change at the membrane−sample interface. Moreover, both electrochemical processes (ion-to-electron transducer and iontransfer) must be coupled to establish a reliable sensor. A recent example with conducting polymers such as poly(3,4-ethylenedioxythiophene) (PEDOT) or poly(n-octylthiophene) (POT) was used by Amemiya et al. as a solid

olid contact ion-to-electron transducers have been extensively characterized during the last two decades in the potentiometric sensing community.1,2 Such transducers are mainly grouped in conducting polymers, active redox monolayers, and nanostructured materials.3−6 Most of them displayed adequate transducer characteristics in zero-current measurements. The involved ion-to-electron transducing mechanism for conducting polymers depends on the specific material and fabrication process, but there is electrochemical evidence for a significant capacitive contribution demonstrated by impedance and chronopotentiometric analysis.7,8 As was reported by Bobacka, the low-frequency impedance response of a conducting polymer exhibits nearly ideal capacitance behavior as a consequence of a reversible doping process.7 In subsequent work, a linear increase of equilibrium potential with time was observed in the presence of constant current by using currentreversal chronopotentiometry, indicative of a dominant capacitive charging process.8 On the other hand, spectroscopic evidence of other polymeric systems suggests a localized faradaic process at the interface between conducting polymer and overlaid ion-selective membrane.6 Nanostructured materials such as 3D carbon, on the other hand, are purposely used for capacitive charging as the main transduction mechanism.9 Substantial practical and fundamental improvements were demonstrated when the inner solution was replaced by a solid transducer in ion-selective membrane electrodes (see references for more details).2,3,6,10 Solid contact technology © 2014 American Chemical Society

Received: January 29, 2014 Accepted: May 29, 2014 Published: May 29, 2014 6307

dx.doi.org/10.1021/ac5004163 | Anal. Chem. 2014, 86, 6307−6314

Analytical Chemistry

Article

Table 1. Compositions of the Investigated Membranes (in mg) M1 7% FcPVC Ionophore Ion-Exchanger DOS DOA PVC ETH 500

5.15 2.90 (100 mM) 37.5

M2 5.00 1.40 (50 mM) 37.5

M3 4.90 0.68 (25 mM) 37.5

4.80 0.13 (5 mM) 37.5

M5

M6

4.78

7.5 5.74

37.5

M7 8.30 2.00

37.5 92.50 60.00

5.05

4.90

4.80

4.75

4.70

5.69

tetrakis(4-chlorophenyl)borate tetradodecylammonium salt (ETH 500), bis(2-ethylexyl) sebacate (DOS), bis(2-ethylexyl) adipate (DOA), poly(vinyl chloride) (PVC, high molecular weight), tridodecylmethylammonium chloride (TDMAC), citric acid, boric acid, monosodium phosphate, sodium phosphate dibasic, hydrochloric acid, tris(hydroxymethyl)aminomethane (Tris), sodium hydroxide, sodium azide, ethynylferrocene, ascorbic acid, and copper sulfate pentahydrate were purchased from Sigma-Aldrich with analytical grade (used without further purification). Nitrogen gas (99.9 vol %) was purchased from PanGas in Switzerland. Aqueous solutions were prepared by dissolving the appropriate salts in Milli-Q purified water. Electrochemical Equipment. A double-junction Ag/ AgCl/3 M KCl/1 M LiOAc reference electrode was used in potentiometric and chronopotentiometric measurements (Metrohm Autolab, Utrecht, The Netherlands). pH values during volumetric titration were determined using a Metrohm 827 pH meter (Metrohm Autolab, Utrecht, The Netherlands). A platinum working rod (3.2 cm2 surface area) was used as a counter electrode. The total potential differences (electromotive force, EMF) were measured with a 16-channel electrode monitor EMF-16 interface from Lawson Laboratories, Inc. at ambient temperature. Chronopotentiometric measurements were performed with an Autolab PGSTAT302N (Metrohm Autolab, Utrecht, The Netherlands). Glassy carbon rods of 3.0 ± 0.1 mm electrode diameter (7.06 mm2 surface area) were used as working electrodes and were purchased from Metrohm Autolab (model no. 6.1204.300). The thin membrane film was prepared using the rotating disk electrode Autolab RDE-2 (Metrohm Autolab, Utrecht, The Netherlands). Electrode bodies (Oesch Sensor Technology (Ostec), Sargans, Switzerland) were used to mount the polypropylene membranes. A Faraday cage was used to protect the system from undesired noise. The measurements were carried out in a special and closed vessel to avoid dissolution of carbon dioxide from the atmosphere. Synthesis of Poly(vinyl chloride) Covalently Modified with Ferrocene Groups (Fc-PVC). The procedure of preparation of Fc-PVC was described by Pawlak et al.18 7% Fc-PVC was prepared by the click chemistry reaction between azide-modified PVC and ethynylferrocene (1114.29 mmol kg−1) with CuSO4.5H2O and ascorbic acid as catalysts in THF medium. Ferrocene modified PVC was filtered and washed with methanol and then dissolved again in THF. This solution was filtered to remove insoluble impurities. The product was precipitated with methanol, filtered and dried under reduced pressure. Composition of Membrane Cocktails. The final composition of membrane cocktails is shown in Table 1. All components were dissolved in 1 mL of THF. THF was used to enhance the solubility of the solid compounds into the

contact ion-to-electron transducer for the detection of small hydrophilic ions by ion-transfer voltammetry.15,16 Very low limits of detection (ca. 1 nM) were obtained for potassium and perchlorate with all-solid-state thin-layer stripping ion voltammetry at a rotating disk electrode. As an interesting alternative to the use of conducting polymers, Samec et al. used freely dissolved dimethylferrocene (DMFc), a well-established electroactive material that exhibits excellent redox capacity. This was embedded in a polymeric nonperm-selective film containing PVC, a nonpolar plasticizer (DOS), and a lipophilic salt. The oxidation of DMFc at the electrode created anionexchanging groups at the metal−membrane interface during the amperometric measurement period, allowing one to detect the polyanionic anticoagulant drug heparin in that example.17 Pawlak et al. introduced ferrocene groups covalently attached to the PVC backbone by “click chemistry” (Huisgen cycloaddition).18 This prevents the possible loss of the ferrocene moiety due to leaching into the sample solution and provides a materials approach to confine the ferrocene functionality to a layer close to the contacting metal electrode on the inner side of the sensing film. It was shown that pulsed chronopotentiometry could be used to interrogate membranes on the basis of Fc-PVC as the ion-to-electron transducer.19 The observed calibration curves exhibited near-Nernstian behavior, in analogy to open-circuit potentiometry. This concept was recently extended and characterized in more detail with flash chronopotentiometry. Jarolimova et al. reported a fundamental study on an initially nonperm-selective membrane that contained the reduced form of Fc-PVC, a plasticizer, and the lipophilic electrolyte ETH 500.20 Membranes were interrogated with a galvanostatic pulse of 5 s duration, and a range of anions was depleted at the membrane surface that allowed one to calculate ion diffusion coefficients according to the Sand equation. As expected, a Hofmeister selectivity pattern was obtained owing to the absence of ionophore in the sensing membrane. To explore new analytes, especially those of environmental relevance, selective chemical recognition helps to provide adequate membrane selectivity. The oxidation of Fc-PVC generates positive charges, thereby mimicking an anionexchanger compound. This can be coupled to anion transfer and depletion at the membrane−sample interface. An ionophore for carbonate was incorporated into the membrane as a proof of concept, and a comparative study between traditional ion-selective and all-solid-state membrane electrodes is provided.



M4

EXPERIMENTAL SECTION

Reagents and Solutions. Sodium fluoride, sodium chloride, anhydrous sodium carbonate, N,N-dioctyl-3α,12αbis(4-trifluoroacetylbenzoyloxy)-5β-cholan-24-amide (carbonate ionophore VII), anhydrous tetrahydrofuran (THF), 6308

dx.doi.org/10.1021/ac5004163 | Anal. Chem. 2014, 86, 6307−6314

Analytical Chemistry

Article

Figure 1. Illustration of sensor mechanism in the membrane phase and interaction between the carbonate selective trifluoroacetophenone moieties and CO32−.

Potentiometric Experiments. Electromotive forces of the membranes (M2, M5) between the electrodes and the doublejunction Ag/AgCl reference electrode were measured under zero-current conditions in magnetically stirred and unstirred solution. The experiments were carried out in 0.1 mM Na2CO3 and 0.01 M NaF. The dynamic response curves were obtained by adding calculated aliquots of standard solution to magnetically stirred background electrolyte (0.1 M Tris-H2SO4 pH 8.0 or 0.1 M phosphate buffer pH 10.0) every 120 s to vary the carbonate concentration stepwise from 10−8 to 10−2 M. The observed EMF values after stabilization were used for the calibration curves with respect to total carbonate concentration. The pH response of the carbonate-selective electrodes was determined by adding aliquots of NaOH to a solution of universal buffer containing 0.005 M boric acid, citric acid, and NaH2PO4 to vary the concentrations of OH− stepwise from 10−11 to 10−1 M. Freely dissolved CO2 was removed before potentiometric measurements using nitrogen gas and a homemade capping seal. The potential measurements were acquired every 0.6 s at room temperature.

plasticizer and thus was removed by evaporation before casting the membranes, with the exception of the membranes containing PVC (M6 and M7), where the cocktails were cast as THF solution (see below). Porous polypropylene (PP) membranes (Celgard, 22.1 mm2 surface area and 25 μm thicknesses, provided by Membrana, Wuppertal, Germany) were used as supporting material for the configuration with inner solution (membrane M1-M5). The PP membrane was washed with THF for 5 min in order to remove any possible contaminants before using. When the membrane was completely dry, an excess volume of 3 μL of the cocktail solution was deposited on it (see cocktail preparation below). The membranes (M1−M5) were allowed to remain in the Petri dish for ∼10 min to ensure a homogeneous and reproducible impregnation of the pores. The pore filling solution composition is assumed to remain identical to the initial THF-free cocktail. Afterward, the membranes were conditioned in 1 mM Na2CO3 for 1 h. Finally, the membrane was mounted in the Ostec electrode body. The inner compartment was filled with 0.01 M NaCl and 0.01 M Tris buffer (pH = 7.4). All-solidstate electrodes were prepared by the drop-casting technique using 13.5 μL of membrane cocktail M6. The surface of the GC electrode was oriented upward and was spin-coated at 2000 rpm to achieve a thin membrane layer. The membrane cocktail M7 was poured into a glass ring (22 mm in diameter) placed on a slide glass and dried overnight at room temperature under a dust-free environment.21 Small disks were punched from the cast film and mounted in Ostec electrode bodies. The membranes were placed into 0.01 M Tris-H2SO4, pH = 7.4, for several hours. The inner compartment was filled with 0.01 M NaCl, 0.1 M NaH2PO4, and 0.1 M Na2HPO4. Inert lipophilic salt ETH 500 was added to provide counterions for the extracted analyte in the membrane and also keeps the electrical resistance of the membrane low. The THF was evaporated under ambient conditions. Electrochemical Experiments. The electrochemical measuring cell consisted of reference, counter, and working electrode. Chronopotentiometric experiments were performed in 0.01 M NaF as a background electrolyte with a galvanostatic pulse of 5 s duration. The potential was recorded as a function of time during the constant current pulse. After each chronopotentiometric determination, the membrane was regenerated either by a galvanostatic pulse (with opposite sign) or potentiostatic pulse at open circuit potential (OCP). Cyclic voltammetry experiments were performed at 25 mV s−1 scan rate between 0 to 0.8 V potential range using 100 mM Na2CO3/NaNO3.



RESULTS AND DISCUSSION Three different strategies were explored with the aim of developing an ionophore-based membrane electrode controlled with dynamic electrochemical techniques. As with previous reports by our group,20 the first explored configuration was a perm-selective membrane doped with a selective ionophore toward carbonate anion (Figure 1). The membrane was composed of a neutral ionophore (L), ion-exchanger (R+), lipophilic salt ETH 500 (R+R−), and DOS as plasticizer (M1− M4 in Table 1). Figure 2a illustrates the expected working principle of such perm-selective membranes when interrogated with the flashchronopotentiometry protocol. Because of the presence of ionexchanger, the primary analyte (i.e., carbonate) must ideally be transported from the sample to the inner element when an anodic pulse is applied to the electrochemical cell. Eventually, at a given time, the flux of the primary analyte cannot sustain the imposed current, and thus, the ion concentration drops to zero at the membrane interface. Particularly, in chronopotentiometry, this phenomenon is visualized as a transition (inflection) in the recorded signal over time. Surprisingly, however, carbonate depletion was not observed using this perm-selective membrane configuration (membrane M1). Note that other ions such as calcium, hydrogen ion, and even protamine had earlier been successfully determined in this manner.11,22,23 6309

dx.doi.org/10.1021/ac5004163 | Anal. Chem. 2014, 86, 6307−6314

Analytical Chemistry

Article

membrane must be coupled to the extraction of a cation from the inner solution side to compensate for the extracted charge. According to our results, the current was limited by mass transport in the aqueous phase rather than in the membrane phase, as desired. Figure 3 shows the resulting chronopotentiometric data (left axis) and corresponding observed pH change (right-axis) for

Figure 3. Titration of 1 mM Na2CO3 with 0.1 M HCl monitored by chronopotentiometry at 15.5 μA (0.70 μA mm−2), see left axis along with the ideal linear decrease of carbonate with added HCl shown as black trace. Right axis, experimental pH values, and comparison to theory is shown as solid blue line. Figure 2. (a) Schematic illustration of working mechanism in classical configuration for ISEs with inner solution (IS). The membrane contains carbonate ionophore (L), ion-exchanger (R+Cl−), lipophilic salt (R+R−), and a nonpolar plasticizer (DOS). (b) Schematic illustration of second configuration without ion-exchanger. The membrane contains carbonate ionophore (L), lipophilic salt (R+R−), and nonpolar plasticizer (DOS). The applied current provokes the transfer of carbonate to the membrane and simultaneously a cation (M+) from the inner solution to the membrane to compensate for the extracted charge. (c) Schematic illustration of the all-solid-state membrane electrode mechanism. The membrane contains carbonate ionophore (L), lipophilic salt (R+R−), attached Fc to PVC, and nonpolar plasticizer (DOS). An applied current provokes an oxidation of Fc at the electrode surface (GC), and a high number of positives sites (Fc+) are generated at the metal-membrane junction. The accumulation of positive charge during the pulse causes the diffusion of R− inside the membrane and is stabilized by ion pairing. As a result, the carbonate ion is transferred from the sample to the membrane.

the titration of 1 mM Na2CO3 with 0.1 M HCl at 15.5 μA (0.70 μA mm−2). The initial concentration of carbonate at pH 10.5 was around 0.65 mM. This concentration was slightly too high to be observable by chronopotentiometry, and the first depletion started at 0.35 mM (upper limit) at pH 10.05 and finished at 0.15 mM (lower limit) at pH 9.50 (see Figure 2S in Supporting Information for raw and processed data). The relationship between the transition time (τ), current density, concentration, and the diffusion coefficient of carbonate is expressed by the Sand equation in the absence of electrical migration effects. The obtained diffusion coefficient of carbonate for this configuration was smaller ((5.6 ± 1.2) × 10−6 cm2 s−1) than the one reported in the literature (9.20 × 10−6 cm2 s−1).24 The potential change before and after the transition was around 200 mV (see Figure 2S in Supporting Information), which corresponds to the expected selectivity of about 6 orders of magnitude (6 times the Nernstian slope of 29.6 mV for carbonate).25 The results for these two membrane configurations suggest that the presence of ion-exchanger does not allow us to determine carbonate in the sample. Instead, if the membrane sensing mechanism is altered by avoiding the ion-exchanger, carbonate depletion is observed. In this second case, the inner electrolyte solution is a source of ions that compensates for the charge imbalance during the sensing and the regeneration steps in the membrane phase. It also plays the role of ion-to-electron transducer, together with the Ag/AgCl wire. We postulated that the lack of carbonate depletion with perm-selective membranes may originate in a possible kinetic limitation of the decomplexation reaction at the inner membrane side, because in the second membrane configuration, no such reaction takes place. In contrast to traditional ionophores that work on the basis of ion pair interactions, such carbonate ionophores are believed to form covalent bonds (see Figure 1 for interaction chemistry between two trifluoroacetophenone moieties and carbonate). The kinetics of ion-

In order to explain these results, different ion-exchanger concentrations were evaluated (compositions M1 to M4). At higher ion-exchanger levels such as 100 mM (ratio 1:1 respect to 100 mM of ionophore), 50 mM (1:2), and 25 mM (1:4), no transitions were observed, both in the raw chronopotentiogram and in the processed data (see Supporting Information, Figure 1Sa−c). A wide range of carbonate concentration from 0.01 M to 1 mM was explored. Some transitions were only observed when reducing the ion-exchanger concentration to 5 mM (membrane M4), see Figure 1S in Supporting Information. However, there was no clear dependence with respect to the added concentration of carbonate. On the basis of these results, we explored a second inner solution configuration without ion-exchanger in the membrane (membrane M5). Figure 2b shows the chemical composition of this initially non-perm-selective membrane composed of ionophore (L), lipophilic salt ETH 500 (R+R−), and DOS as plasticizer (see M5 in Table 1). In contrast to the first example (Figure 2a), the current-driven transfer of carbonate to the 6310

dx.doi.org/10.1021/ac5004163 | Anal. Chem. 2014, 86, 6307−6314

Analytical Chemistry

Article

stirring was switched off, the EMF response with M2 was found to increase (see Figure 3S in Supporting Information), suggesting a local depletion, presumably due to spontaneous diffusion of carbonate from the sample into the membrane phase. No significant potential changes were obtained in different stirring mode for the second membrane configuration, M5. As established earlier,28 a stir effect is likely to indicate a spontaneous carbonate ion flux. The presence of a spontaneous flux makes the control of the flux by an applied current unreliable, providing an explanation for the lack of visible carbonate depletion with perm-selective membranes. We then reasoned that the inner liquid solution might be replaced with another electrochemical couple such as reduced and oxidized ferrocene (Fc°, Fc+) immobilized in the PVC polymer in order to suppress undesired spontaneous carbonate fluxes. Figure 2c illustrates the third, all-solid-state electrode configuration that was explored (M6). The membrane contained Fc chemically attached to the PVC backbone, lipophilic salt ETH 500, and plasticizer DOS, deposited on a glassy carbon electrode with a defined area (membrane M6). When a galvanostatic pulse is applied to the all-solid-state electrode, Fc is oxidized to Fc+, and positive charges were thereby injected at the electrode/membrane interface. In order to compensate this unbalanced charge, R− (from ETH 500) migrated inside the membrane to form the counterions of the oxidized Fc+ species. Consequently, an excess of quaternary ammonium (R+) in the outer membrane/sample interface (together with an excess of ionophore) facilitated the transfer of carbonate from one phase to another. The concentration of ionophore has to be sufficiently high to obtain a concentration polarization in the aqueous phase instead of in the membrane. Figure 5 illustrates the chronopotentiometric carbonate determination with a membrane that contained chemically bound Fc-PVC and ionophore (L) at 5 μA. No carbonate depletion is observed for the membrane without ionophore (M6, see Figure 4S in Supporting Information). This is explained with the high hydration energy for carbonate, making carbonate transfer into the membrane energetically unfavorable. Figure 5a shows the chronopotentiograms for carbonate depletion. A potential change of 250 mV was observed in the chronopotentiogram before and after transition. This variation is related to the transport of a second ion (mainly chloride). The time derivative with respect to time is plotted in Figure 5b for better visualization. The transition time was then extracted from the maximum peak values. Subsequently, a linear relationship between the square root of the transition time and carbonate concentration at 5 μA (0.70 μA mm−2) was observed [τ1/2 = 7.17 (±0.13); ccarbonate = 0.23 (±0.16)]. The observed linear range was 0.1−0.35 mM (Figure 5c). Several calibration curves for carbonate at different current amplitude and at fixed pH are shown in Figure 6a, and all calibration curves display a linear behavior in the range of 0.1− 0.34 mM. The maximum depleted concentration could, in principle, be extended with higher current densities. However, in doing so, no significant extension of linear range was observed. Instead, the membrane exhibited a short lifetime at higher current densities (higher than 0.70 μA mm−2). The observed diffusion coefficient is ((9.03 ± 0.91) × 10−6 cm2 s−1) and corresponds to the one reported in the literature (9.20 × 10−6 cm2 s−1). Figure 6b represents the square root of transition time multiplying by the applied current as a function of carbonate concentration, in accordance to the Sand equation.

transfer were evaluated by cyclic voltammetry in a polymeric thin membrane layer (membrane M6, see further below for details). Figure 2c shows a schematic illustration of anion extraction into and out of the membrane layer during the oxidation−reduction processes. Once a sufficiently large anodic potential is applied, the Fc-PVC is oxidized to Fc+-PVC, and the anion is extracted from the aqueous to organic phase due to an excess of quaternary ammonium (R+). An applied backward sweep causes reduction of Fc+-PVC and back-extraction of the anion into the aqueous phase. An incorporated ionophore (L) in the organic phase assists the anion transfer due to the complexation reaction. As a result, the transfer of carbonate is obtained at milder potentials in comparison with other anions due to lower free energy of transfer and binding affinity between ionophore and analyte. If the rate of decomplexation and free ionophore regeneration would be kinetically hampered, the carbonate should not be easily released to the inner solution, the membrane should become saturated, and the carbonate could not be depleted from the aqueous phase. As shown in Figure 4, cyclic voltammograms of carbonate uptake

Figure 4. Cyclic voltammograms of the ion transfer of CO32− or NO3− from aqueous phase into a thin sensing layer containing the carbonate ionophore at a 25 mV s−1 scan rate in the potential range of 0 to 0.8 V. Blue line shows carbonate ion transfer (Ip = 5.46 × 10−7 A), black line corresponds to the transfer of nitrate (Ip = 7.96 × 10−7 A).

and release are characterized in the forward and backward scan of the cyclic voltammograms and compare qualitatively well to those for the nitrate ion, which is not expected to bind to the ionophore in any significant way. Although the voltammogram for carbonate is shifted owing to the extraction selectivity, the two curves are of similar shape, suggesting that an irreversible back-reaction for carbonate can be excluded (see Figure 4). However, the separation of anodic and cathodic peaks was found to be ∼150 mV, which is larger than the theoretical value for a one-electron-transfer reaction and for freely dissolved ferrocene in low viscous media (∼56 mV in acetonitrile).26 This peak separation may point to an increased activation barrier to ion transfer, but this could also be partially caused by the ohmic resistance in the PVC membrane.18,27 To find another explanation for the lack of carbonate response with membrane M2, membranes M2 and M5 with inner solution (M2 and M5) in stirred and unstirred solution of 0.1 mM of Na2CO3 were measured by potentiometry. Once the 6311

dx.doi.org/10.1021/ac5004163 | Anal. Chem. 2014, 86, 6307−6314

Analytical Chemistry

Article

Figure 6. Observed linear calibration curve of the square root of the transition time τ1/2 as a function of carbonate concentration at the indicated current amplitudes (2.5 to 7.5 μA, 0.35−1.06 μA mm−2) at pH 9.20. (b) Normalized response curve of the square root of transition times multiplied by the applied current as a function of the carbonate concentration with error bars (relative standard deviation).

A linear relationship was obtained and indicates the absence of other transport processes such as migration or convection. Error bars are relative standard deviations determined from triplicates. The ionophore concentration in the membrane was varied to extend the linear range. Figure 5S in Supporting Information displays several calibration curves performed at 22.5, 50, 100, and 150 mM of carbonate ionophore. As expected, with a higher concentration of ionophore, there is a higher observable carbonate concentration. However, after 100 mM, no further significant increase was found. Figure 7 shows the chronopotentiometric (left axis) and potentiometric (rightaxis) data for the titration of 1 mM Na2CO3 with 0.1 M HCl (see Figure 6S in Supporting Information for the raw data). The initial concentration of carbonate at pH 10.50 was around 0.65 mM. Membrane electrodes based on trifluoroacetophenone derivatives respond to carbonate only at alkaline pH where they may be susceptible to hydroxide interference. To evaluate the influence of sample pH to the sensor response, ion-selective

Figure 5. Chronopotentiometric determination of carbonate in the third configuration at 5 μA (0.70 μA mm−2). (a) Observed potential changes during galvanostatic pulse for different carbonate levels. (b) Observed time derivatives of the chronopotentiometric response after successively increasing the final carbonate concentration from 0.12 to 0.34 mM. (c) Linear calibration curve of the square root of the transition time τ1/2 as a function of carbonate concentration at 5 μA (0.70 μA mm−2). 6312

dx.doi.org/10.1021/ac5004163 | Anal. Chem. 2014, 86, 6307−6314

Analytical Chemistry

Article

application in environmental samples, because the upper detection limit is not yet sufficiently high. After each chronopotentiometric determination, the membrane must be regenerated in order to produce repeatable transition times: the membrane carbonate concentration must be reduced to their initial values after each measurement period. Different strategies based on potentiostatic or galvanostatic pulses were evaluated. The membrane was first returned to its original condition by applying a potential equal to the open circuit potential determined before the current excitation. However, we believe that regeneration was not complete, because subsequent chronopotentiograms displayed a significant variability, as the signal became smoother and became hidden by the sharp decay function, resulting in a less reliable determination of the transition time (see Figure 11S in Supporting Information). These details point to a significantly less reliable potential controlled regeneration with Fc-PVCbased solid contact electrodes, likely indicating a change of the inner redox potential with time. Inspired by recent work by Lindner’s group,10 we subsequently explored a controlledcurrent regeneration protocol, which is independent of the redox state at the solid inner electrode. Different galvanostatic pulse durations of 2.5, 5, and 10 s (Figure 11Sb−d in Supporting Information) of opposite current sign but at the same amplitude were characterized. We found that 5 s was the optimum value to successfully regenerate the membrane for the protocol used in this work.

Figure 7. Obtained chronopotentiometric (left axis) data and potentiometric (right axis) data for the titration of 1 mM Na2CO3 by additions of 0.1 M HCl in third configuration at 5 μA (0.70 μA mm−2). Left axis shows the ideal linear decrease of carbonate with added HCl (black trace), and right axis displays experimental pH values and comparison to theory (solid blue line).

electrodes based on M7 were placed into a universal buffer solution, and aliquots of NaOH were added to increase the sample pH. As expected, the membrane exhibited a response toward pH (see Figure 7S in Supporting Information). The electrodes exhibited a near-Nernstian slope for hydroxide ion of 57 mV. The selectivity coefficient logKCO3,OHpot = 2.2 was determined with the separate solutions method.29 The potentiometric response properties of the carbonate-selective membrane (M7) were evaluated in 0.1 M Tris-H2SO4 at pH 8 and 0.1 M Na2HPO4 at pH 10. The electrodes exhibited Nernstian behavior with response slopes ∼29.0 mV decade−1 and a detection limit of 10−6 M at pH 8. As can be seen in Figure 8S in Supporting Information, the carbonate slope and detection limit deteriorated slightly as the pH increased. These potentiometric experiments demonstrate that the electrodes exhibit selective behavior to carbonate at pH 8, whereas the depletion of carbonate is not observed in a chronopotentiometric mode at a pH lower than 8.70 (see Figure 9S in Supporting Information). Inorganic carbon is present in natural water as molecular carbon dioxide, carbonic acid, bicarbonate, and carbonate, and their relative abundance depends on pH (see Figure 10S in Supporting Information). Under environmental conditions (pH 8.0−8.5), the total concentration of inorganic carbon in water is approximately 2 mM, and bicarbonate is the most abundant form (see Figure 10S in Supporting Information). The bicarbonate−carbonate species form a buffer system that give us an explanation for the lack of carbonate depletion. The reason for a lack of chronopotentiometric response for carbonate at pH 8 lies in the large excess of bicarbonate (∼98% of total inorganic carbon) over carbonate (∼1% of total inorganic carbon) at this pH. Any carbonate lost from the ion transfer process is readily replenished by dissociation of bicarbonate (at constant pH), and electrochemical depletion cannot be observed. The relatively high pH required for carbonate detection is therefore not due to a selectivity limit but an intrinsic characteristic of the method. Unfortunately, the limited pH range (9.5−10.5) and buffering properties make the membrane electrode not yet suitable for a direct practical



CONCLUSIONS We developed a new sensing concept that is based on the detection of carbonate using all-solid-state membrane electrodes controlled with short galvanostatic pulses. The membrane contains Fc-PVC in the reduced state. Once the current is applied, the oxidized Fc+-PVC state is generated and thus acts as an ion-to-electron transducer at the inner interface (glassy carbon/membrane). Because the ion-to-electron transducer and the ion-transfer processes are directly coupled, carbonate ion depletion at the aqueous phase side was observed. Additionally, we demonstrated how the ionophore concentration levels modified the linear range of the sensor. The linear range of the sensor was about 0.03−0.35 mM, and the pH range was 9.50−10.05, which is not suitable for direct environmental analysis at a natural pH of 8.0 to 8.5. The working mechanism of the sensor was illustrated by monitoring the concentration of carbonate during a titrimetric analysis. Besides these analytical characteristics, we successfully adapted the membrane configuration from an inner liquid permselective membrane to an all-solid-state membrane electrode. A perm-selective membrane did not work for carbonate sensing, owing to the occurrence of spontaneous carbonate ion fluxes, making the use of current to impose a precise flux unreliable. On the basis of this work, we now aim to develop a potentially disposable platform of all-solid-state sensors for the detection of other ions, including polyions, for use in clinical and environmental analysis.



ASSOCIATED CONTENT

S Supporting Information *

Additional data and information as noted in the text. This material is available free of charge via the Internet at http:// pubs.acs.org. 6313

dx.doi.org/10.1021/ac5004163 | Anal. Chem. 2014, 86, 6307−6314

Analytical Chemistry



Article

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS The authors thank EU FP7 (SCheMA), the Swiss National Foundation, and the University of Geneva for supporting this research. Z.J. thanks the Erasmus program for supporting her scientific stay in Geneva.



REFERENCES

(1) Bobacka, J.; Ivaska, A.; Lewenstam, A. Chem. Rev. 2008, 108, 329−351. (2) Michalska, A. Electroanalysis 2012, 24, 1253−1265. (3) Bobacka, J. Electroanalysis 2006, 18, 7−18. (4) Crespo, G. A.; Macho, S.; Rius, F. X. Anal. Chem. 2008, 80, 1316−1322. (5) Grygolowicz-Pawlak, E.; Plachecka, K.; Brzozka, Z.; Malinowska, E. Sens. Actuators, B 2007, 123, 480−487. (6) Veder, J. P.; De Marco, R.; Patel, K.; Si, P. C.; GrygolowiczPawlak, E.; James, M.; Alam, M. T.; Sohail, M.; Lee, J.; Pretsch, E.; Bakker, E. Anal. Chem. 2013, 85, 10495−10502. (7) Bobacka, J. Anal. Chem. 1999, 71, 4932−4937. (8) Bobacka, J.; Lewenstam, A.; Ivaska, A. J. Electroanal. Chem. 2001, 509, 27−30. (9) Fierke, M. A.; Lai, C.-Z.; Buhlmann, P.; Stein, A. Anal. Chem. 2010, 82, 680−688. (10) Lindner, E.; Gyurcsanyi, R. E. J. Solid State Electrochem. 2009, 13, 51−68. (11) Afshar, M. G.; Crespo, G. A.; Bakker, E. Anal. Chem. 2012, 84, 8813−8821. (12) Grygolowicz-Pawlak, E.; Bakker, E. Anal. Chem. 2010, 82, 4537−4542. (13) Kabagambe, B.; Izadyar, A.; Amemiya, S. Anal. Chem. 2012, 84, 7979−7986. (14) Kim, Y.; Rodgers, P. J.; Ishimatsu, R.; Amemiya, S. Anal. Chem. 2009, 81, 7262−7270. (15) Guo, J. D.; Amemiya, S. Anal. Chem. 2006, 78, 6893−6902. (16) Perera, H.; Fordyce, K.; Shvarev, A. Anal. Chem. 2007, 79, 4564−4573. (17) Langmaier, J.; Olsak, J.; Samcova, E.; Samec, Z.; Trojanek, A. Electroanalysis 2006, 18, 1329−1338. (18) Pawlak, M.; Grygolowicz-Pawlak, E.; Bakker, E. Anal. Chem. 2010, 82, 6887−6894. (19) Pawlak, M.; Grygolowicz-Pawlak, E.; Bakker, E. Pure Appl. Chem. 2012, 84, 2045−2054. (20) Jarolimova, Z.; Crespo, G. A.; Afshar, M. G.; Pawlak, M.; Bakker, E. J. Electroanal. Chem. 2013, 709, 118−125. (21) Xie, X. J.; Bakker, E. Anal. Chem. 2013, 85, 1332−1336. (22) Crespo, G. A.; Afshar, M. G.; Bakker, E. Anal. Chem. 2012, 84, 10165−10169. (23) Crespo, G. A.; Afshar, M. G.; Bakker, E. Angew. Chem., Int. Ed. 2012, 51, 12575−12578. (24) Buffle, J.; Zhang, Z.; Startchev, K. Environ. Sci. Technol. 2007, 41, 7609−7620. (25) Xie, X.; Pawlak, M.; Tercier-Waeber, M.-L.; Bakker, E. Anal. Chem. 2012, 84, 3163−3169. (26) Tsierkezos, N. G. J. Solution Chem. 2007, 36, 289−302. (27) Feldberg, S. W. Anal. Chem. 2014, 86, 2268−2268. (28) Malon, A.; Bakker, E.; Pretsch, E. Anal. Chem. 2007, 79, 632− 638. (29) Bakker, E.; Pretsch, E.; Buhlmann, P. Anal. Chem. 2000, 72, 1127−1133.

6314

dx.doi.org/10.1021/ac5004163 | Anal. Chem. 2014, 86, 6307−6314