Membranes for Improved All-Solid-State Ion-Selective Sensors

Dec 7, 2007 - of Cambridge, Tennis Court Road, Cambridge, CB2 1QT, UK. A novel type of self-plasticizing polyacrylate-based mem- brane was developed ...
0 downloads 0 Views 126KB Size
Download PDF
Anal. Chem. 2008, 80, 321-327

Composite Polyacrylate-Poly(3,4ethylenedioxythiophene) Membranes for Improved All-Solid-State Ion-Selective Sensors Anna Rzewuska,† Marcin Wojciechowski,† Ewa Bulska,† Elizabeth A. H. Hall,‡ Krzysztof Maksymiuk,† and Agata Michalska*,†

Department of Chemistry, Warsaw University, Pasteura 1, 02-093 Warsaw, Poland, and Institute of Biotechnology, University of Cambridge, Tennis Court Road, Cambridge, CB2 1QT, UK

A novel type of self-plasticizing polyacrylate-based membrane was developed for all-solid-state ion-selective potentiometric electrodes. The membrane composition contains a conducting polymer (CP): poly(3,4-ethylenedioxythiophene) end capped with methacrylate groups, chemically grafted with the membrane during the photopolymerization step. This composition results in ionselective membranes with the following advantages: lower electrical resistance compared to the CP-free membrane, facile ion-to-electron transduction between the membrane and the electrode support, controlled low activity of analyte ions, and high concentration of interferent ions (incorporated with the CP) within the membrane, potentially resulting in improved analytical parameters. Ca2+and K+-selective membranes were chosen as model systems to study the effect of pretreatment and CP content on the potentiometric sensor’s characteristics. For Ca2+ sensors, reproducible and stable Nernstian characteristics were obtained within the range from 0.1 to 10-9 M CaCl2, without a time-consuming preconditioning step. For K+selective sensors, the influence on Nernstian response range was observed for varying KCl concentrations in the conditioning solution, with the lowest detection limit found close to 10-8 M KCl. Mass spectrometry coupled with laser ablation studies of the membranes revealed that in this case the detection limit is not related to primary ion content in the membrane contacting a sample solution, but is affected by interfering ion concentration close to the membrane surface. Ion-selective electrodes are established electroanalytical tools allowing determination of many analytes, including clinically or environmentally important ions, in concentrations reaching trace levels.1 An interesting modification, leading to sensors that could be potentially easier to manufacture and handle, is replacement of the internal solution with a conducting polymer ion-to-electron transducer, thereby creating all-solid-state sensors2,3 (ASS-CP* To whom correspondence should be addressed. E-mail: agatam@ chem.uw.edu.pl. † Warsaw University. ‡ University of Cambridge. (1) Bakker, E.; Pretsch, E. Trends Anal. Chem. 2005, 24, 199-207. (2) Bobacka, J. Electroanalysis 2006, 18, 7-18. 10.1021/ac070866o CCC: $40.75 Published on Web 12/07/2007

© 2008 American Chemical Society

ISEs). A conducting polymer layer included in the sensor construction has been reported to contribute to significant increase in stability of recorded potentials2,3 compared to other internal solution free arrangements. ASS-CP-ISEs are usually produced in at least two steps, with initial deposition of the conducting polymer transducer layer, which is next covered by an ion-selective membrane, most often plasticized poly(vinyl chloride) (PVC).2,3 From a practical point of view, simplicity in sensor construction is desirable. The approach used for poly(vinyl chloride)-based membranes takes advantage of solubility of some of conducting polymers in solvents that can be added to the plasticized poly(vinyl chloride)-based ionselective membrane formulation.4-9 Unfortunately, solubility can be achieved only for some of the available conducting polymers: undoped (semiconductor form) poly(3-octylthiophene),4 protonated (conducting) form of polyaniline (PANI) and its derivatives,4-6,9 or doped poly(pyrrole).5,8 Calcium or lithium sensors obtained with polyaniline or poly(3-octylthiophene) included in the membrane up to a few weight percent (single piece electrodes4,6) have resulted in a performance that is slightly worse than for traditional internal solution containing ion-selective electrodes; e.g., detection limits achieved were close to 10-4 M.4,6 Slightly lower detection limits were also obtained for planar miniature potassium sensors containing PANI in the ion-selective membrane phase.7,8 On the other hand, plasticized PVC is sometimes not an optimal choice of membrane material, since leakage of plasticizer from the membrane compromises the performance, especially if long-term usage of the sensor is required.10,11 Self-plasticizing polyacrylate membranes seem to be an interesting alternative, e.g., (3) Michalska, A. Anal. Bioanal. Chem. 2006, 384, 391-406. (4) Bobacka, J.; Lindfors, T.; McCarrick, M.; Ivaska, A.; Lewenstam, A. Anal. Chem. 1995, 67, 3819-3823. (5) Lindfors, T.; Bobacka, J.; Lewenstam, A.; Ivaska, A. Analyst 1996, 121, 1823-1827. (6) Lindfors, T.; Sjo¨berg, P.; Bobacka, J.; Lewenstam, A.; Ivaska, A. Anal. Chim. Acta 1999, 385, 163-173. (7) Zachara, J. E.; Toczylowska, R.; Pokrop, R.; Zago´rska, M.; Dybko, A.; Wro´blewski, W. Sens. Actuators, B 2004, 101, 207-212. (8) Toczylowska, R.; Pokrop, R.; Dybko, A.; Wro´blewski, W. Anal. Chim. Acta 2005, 540, 167-172. (9) Grekovich, A. L.; Markuzina, N. N.; Mikhelson, K. N.; Bochenska, M.; Lewenstam, A. Electroanalysis 2002, 14, 551-555. (10) Daviers, M. L.; Tighe, B. J. Sel. Electrode Rev. 1991, 13, 159-226. (11) Armstrong, R. D.; Horvai, G. Electrochim. Acta 1990, 35, 1-7.

Analytical Chemistry, Vol. 80, No. 1, January 1, 2008 321

refs 12-20, especially as sensors with these membrane materials have shown some improved analytical parameters compared to PVC (e.g., refs 16 and 17). A polyacrylate matrix also enables grafting of ionophore13,14,21 or ion-exchanger20 within the membrane phase. However, polyacrylate membranes usually have high resistivity (e.g., refs 16 and 19), which can cause noisy, unstable potentiometric readings. It has been shown (e.g., ref 19) that adding a plasticizer to the polyacrylate membrane formulation helps to decrease resistivity, but obviously reduces the advantage of using self-plasticizing media. To date, all the proposed ASSCP-ISEs seem to have required rigorous conditioning protocols (e.g., refs 17, 22-27) to achieve detection limits and selectivities that are comparable with the best internal solution electrodes28,29 (regardless of the ion-selective membrane matrix used). Alternative engineering of the transducer composition30 or even galvanostatic polarization31 has been tried to overcome this, but simplifying or eliminating such procedures would clearly be beneficial. An alternative remedy for this problem is explored in the work reported herein, whereby a conducting polymer is combined with the polyacrylate to form a composite membrane. In contrast to single piece electrodes4,6 (obtained by mixing of the conducting polymer with ion-selective membrane components), this approach explores for the first time chemical grafting of the conducting polymer within the poly(n-butyl acrylate) matrix containing the usual additives (ionophore and ion-exchanger) during a single photopolymerization step. The conducting polymer explored in this context was poly(3,4-ethylenedioxythiophene) end capped with methacrylate groups, used in the form of a suspension stabilized by p-toluenesulfonate anions. Thus, the electrical resistivity of ion-selective membranes was decreased and additional benefits arise, since together with the conducting polymer suspension, p-toluenesulfonate countercations (sodium or potassium ions) are also introduced to the membrane, at the polym(12) Heng, L. Y.; Hall, E. A. H. Anal. Chim. Acta 1996, 324, 47-56. (13) Heng, L. Y.; Hall, E. A. H. Anal. Chem. 2000, 72, 42-51. (14) Malinowska, E.; Gawart, Ł; Parzuchowski, P.; Rokicki, G.; Brzo´zka, Z. Anal. Chim. Acta 2000, 421, 93-101. (15) Heng, L. Y.; Toth, K.; Hall, E. A. H. Talanta 2004, 63, 73-87. (16) Michalska, A. J.; Appaih - Kusi, C.; Heng, L. Y.; Walkiewicz, S.; Hall, E. A. H. Anal. Chem. 2004, 76, 2031-2039. (17) Chumbimuni-Torres, K. Y.; Rubinova, N.; Radu, A.; Kubota, L. T.; Bakker, E. Anal. Chem. 2006, 78, 1318-1322. (18) Grygolowicz-Pawlak, E.; Wygladacz, K.; Sek, S.; Bilewicz, R.; Brzo´zka Z.; Malinowska, E. Sens. Actuators, B 2005, 111-112, 310-316. (19) Wygladacz, K.; Durnas, M.; Parzuchowski, P.; Brzo´zka Z.; Malinowska, E. Sens. Actuators, B 2003, 95, 366-372. (20) Qin, Y.; Bakker, E. Anal. Chem. 2003, 75, 6002-6010. (21) Qin, Y.; Peper, S.; Radu, A.; Ceresa, A.; Bakker, E. Anal. Chem. 2003, 75, 3038-3045. (22) Michalska, A.; Dumanska, J.; Maksymiuk, K. Anal. Chem. 2003, 75, 49644974. (23) Sutter, J.; Radu, A.; Peper, S.; Bakker, E.; Pretsch, E. Anal. Chim. Acta 2004, 523, 53-59. (24) Sutter, J.; Lindner, E.; Gyurcsa´nyi, R.; Pretsch, E. Anal. Bioanal. Chem. 2004, 380, 7-14. (25) Michalska, A.; Maksymiuk, K. J. Electroanal. Chem. 2005, 576, 339-352. (26) Michalska, A.; Konopka, A.; Maj-Zurawska, M. Anal. Chem. 2003, 75, 141144. (27) Konopka, A.; Sokalski, T.; Michalska, A.; Lewenstam, A.; Maj-Zurawska, M. Anal. Chem. 2004, 76, 6410-6418. (28) Bakker, E.; Pretsch, E. Anal. Chem. 2002, 74, 420A-426A. (29) Sokalski, T.; Ceresa, A.; Zwickl, T.; Pretsch, E. J. Am. Chem. Soc. 1997, 119, 11347-11348. (30) Michalska, A.; Skompska, M.; Mieczkowski, J.; Zago´rska, M.; Maksymiuk, K. Electroanalysis 2006, 18, 763-771. (31) Michalska, A. Electroanalysis 2005, 17, 400-407.

322

Analytical Chemistry, Vol. 80, No. 1, January 1, 2008

erization step. The sensors obtained are expected to be characterized by high stability of the potential readings, low detection limits, and high selectivities. EXPERIMENTAL SECTION Apparatus. In the potentiometric experiments, a multichannel data acquisition setup and software, Lawson Labs. Inc. (3217 Phoenixville Pike, Malvern, PA 19355) was used. In other electrochemical measurements (electrochemical impedance spectroscopy (EIS), chronopotentiometry), galvanostat-potentiostat CH-Instruments model 660A (Austin, TX), and a conventional three-electrode cell, with platinum sheet as a counter electrode, was used. The pump systems 700 Dosino and 711 Liquino (Metrohm, Herisau, Switzerland) were used to obtain sequential dilutions of calibrating solution. The double junction silver/silver chloride reference electrode with 1 M lithium acetate in the outer sleeve (Mo¨ller Glasbla¨serei, Zu¨rich, Switzerland) was used. The recorded potential values were corrected for the liquid junction potential calculated according to the Henderson approximation. An inductively coupled plasma mass spectrometer ELAN 9000 (Perkin-Elmer) (LA-ICPMS) equipped with the laser ablation system LSX-200+ (CETAC) was used.32 The applied laser energy was 3.2 mJ, repetition rate was 5 Hz, and spot size was 100 µm. The changes in distribution of elements (23Na, 39K, 44Ca, 32S) within the ion-selective membrane were followed. A comparative measure of their amounts in different membranes was achieved; the full quantitative analysis of the membrane components was not the aim. Each LA-ICPMS analysis was performed in duplicate (for at least two different points on the ISM surface). Reagents. Calcium-selective ionophore N,N-dicyclohexylN′,N′dioctadecyl-3-oxapentanediamide (ETH 5234), potassiumselective ionophore valinomycin, and sodium tetrakis[3,5-bis(trifluoromethyl)phenyl]borate (NaTFPB) were from Fluka AG (Buchs). 1,6-Hexanedioldiacrylate (HDDA), 2,2-dimethoxy-2diphenylacetophenone (DMPP), and n-butyl acrylate were from Aldrich. Poly(3,4-ethylenedioxythiophene) (PEDOT) tetramethacrylate end-capped solution (mean molar mass ∼6000 g/mol) in nitromethane was obtained from Aldrich (Germany). This solution was concentrated prior use by slow, mass change controlled, evaporation of nitromethane at room temperature. The concentrated conducting polymer (CP) composition obtained was used in the membrane preparation. Doubly distilled and freshly deionized water (resistance 18.2 MΩ cm, Milli-Qplus, Millipore) was used throughout this work. All salts used were of analytical grade and were obtained from POCh (Gliwice, Poland). Electrodes. Glassy carbon (GC) electrodes of area 0.07 cm2 were used. The substrate electrodes were polished with Al2O3, 0.3 µm, and rinsed well in water. Ion-Selective Membrane. Calcium composite membrane cocktails contained the following (by weight): 1.1% NaTFPB, 1.4% ETH 5234, 0.2% HDDA, 1.4% DMPP, 6.4 or 8.2% CP, and n-butyl acrylate. Potassium composite membrane cocktails contained the following (by weight): 1.3% NaTFPB, 1.7% valinomycin, 0.2% (32) Michalska, A.; Wojciechowski, M.; Wagner, B.; Bulska, E.; Maksymiuk, K. Anal. Chem. 2006, 78, 5584-5589.

HDDA cross-linker, 1.4% DMPP, 6.4% CP, and n-butyl acrylate. For both membranes, the ratio of numbers of acrylate groups originating from CP and n-butyl acrylate is below 1%. Membranes for electrochemical experiments contained 0.2% HDDA, 1.4% DMPP, and n-butyl acrylate (blank polyacrylate membrane); 0.2% HDDA, 1.4% DMPP, CP, n-butyl acrylate or 1.1% NaTFPB, 1.4% ETH 5234, 0.2% HDDA, 1.4% DMPP, and n-butyl acrylate (CP-free membrane used to obtain coated disk electrodes). A 10-µL sample of the composite membrane cocktail was applied on the top of a glassy carbon electrode placed in up-side down position. Photopolymerization was carried out using a UV lamp (360 nm) for 5 min under argon. The same membrane cocktails were used to prepare samples for (LA-ICPMS) analysis; 5 µL of the membrane cocktail solution was pipetted on an acetate transparency and photopolymerized as above. Conditioning of the Composite Membranes. Before EIS and chronopotentiometric measurements, the membranes were conditioned for 20 h in 0.1 M KCl (potassium sensors) or 0.1 M CaCl2 (calcium sensors). For potentiometric measurements, calcium-selective sensors were conditioned for 3 days in 10-3 M CaCl2 and then kept inbetween measurements in EDTA buffer of constant and low Ca2+ ion activity (0.05 M Na2EDTA, 3.5 × 10-2 M CaCl2, 0.146 M NaCl, adjusted with NaOH to pH 10.3, calculated aCa2+ ) 2.3 × 10-9 M, calculated aNa+ ) 0.175 M), alternatively the sensors were conditioned in the above buffer only, without pretreatment. Potassium sensors were conditioned and were kept in-between measurements in 1 or 10-3 M KCl. Samples for LA-ICPMS analysis were conditioned in 10-3 M CaCl2 (calcium sensors) or 1 or 10-3 M KCl (potassium sensors) for 20 h. RESULTS AND DISCUSSION Resistance and Capacitance of Composite Membranes. Poly(n-butyl acrylate) membranes are typically characterized with a relatively high resistance, determined from EIS data corresponding to a phase angle close to zero. In this instance, for the blank poly(n-butyl acrylate) membrane (free from conducting polymer, ion-exchanger or ionophore addition), the frequency range corresponding to a purely resistive response was 0.1 Hz ÷ 10 Hz, and the recorded resistance was close to 1010 Ω. However, introducing CP into the membrane led to a substantial decrease in the resistance. For example, adding ∼2% CP to the membrane results in a resistance close to 107 Ω (determined in the frequency range from 0.1 to 100 Hz), whereas further increase in CP content in the membrane to ∼6% reduces the resistance to a limiting value of ∼106 Ω. This value does not change further for higher amounts of CP in the poly(n-butyl acrylate) phase. These resistance values are also largely comparable with those reported previously for classical plasticized PVC ion-selective membranes (e.g., ref 33). The chronopotentiometric curves obtained for composite calcium-selective membranes containing poly(n-butyl acrylate), ionophore, ion-exchanger, and 6.4 or 8.2% w/w CP are presented in Figure 1. The resistance estimated in this experiment, from (33) Pawlowski, P.; Michalska, A.; Maksymiuk, K. Electroanalysis 2006, 18, 1339-1346.

Figure 1. Chronopotentiograms (applied current (10-8 A) for (A) composite membrane calcium-selective sensors containing (1) 6.4 and (2) 8.2% conducting polymer and (B) for calcium-selective membrane, free from CP additive, recorded in 0.1 M CaCl2 following conditioning in 10-3 M CaCl2 (note the potential scale difference for plots A and B).

potential jump upon current direction change, is slightly higher (compared to EIS data) and is close to 107 and 106.9 Ω for the membrane containing 6.4 and 8.2% w/w CP, respectively. For anodic and cathodic polarization, potentials recorded were (irrespective of amount within the studied range) only slightly affected by 10-8 A galvanostatic polarization for both compositionss potential drift reaching a few millivolts for 60 s. Therefore, according to previous reports,34 in the presence of CP within the membrane, significantly higher potential stability than in the case of coated disk arrangement is suggested for both cases. The capacitance (determined from the linear potential versus time dependence34) is then estimated to be above 10-3 F/cm2, much higher than a typical double layer capacitance of GC electrode (∼10-6 F/cm2). This confirms the transducing properties of the CP and points to the absence of a blocked interface at the membrane/electrode substrate. The stabilizing effect on electrode potential can be explained by the amount of redox-active polymer present in the membrane, resulting from photopolymerization of 10 µL of cocktail. It is in the range of 10-3 mmol, i.e., comparable (34) Bobacka, J. Anal. Chem. 1999, 71, 4932-4937.

Analytical Chemistry, Vol. 80, No. 1, January 1, 2008

323

with a typical amount of electropolymerized PEDOT in a film, used for solid-state contact sensors, where the potential stabilizing effect is well documented.34 In a chronopotentiometric experiment conducted for composite potassium-selective membranes containing 6.4% CP (results not shown) in 0.1 M KCl, the recorded potentials were also practically unaffected by the applied current (single mV drift observed), regardless of whether the sensor was conditioned in 1 or 10-3 M KCl. The resistance of the composite potassium-selective membranes, estimated from this experiment was close to 107.4 Ω. In contrast, in a parallel experiment conducted for the ionselective membrane, free from CP additive, coated directly on a GC support (coated disk arrangement), a pronounced potential drift was observed, Figure 1. Using the procedure proposed by Bobacka34 for the calcium-selective sensor, the estimated capacitance of this arrangement is close to 10-6 F/cm2, corresponding to the typical value for a noncoated GC electrode. Calcium-Selective Composite Membranes. Calcium-selective composite membranes were used as model systems to study the effect of the CP content. For sensors containing 6.4% CP in the membrane, linear Nernstian responses were obtained within the CaCl2 activity range from 10-9 to 0.1 M; the reproducibility of potentials obtained within a one-week period (n ) 10) is presented in Figure 2. These sensors were characterized by a wide linear response range and high stability of the measured potential values: the slope of the line presented in Figure 2A, within the activity range from 10-9 to 0.1 M, is 24.9 ( 0.5 mV/decade (R2 ) 0.997). Although the sensor was repeatedly tested both in highand low-activity solutions, the potentials recorded for activities of g10-7 M were reproducible within a few millivolts. With lowering activity, within the range from 10-7 to 10-9 M, increasing fluctuations in potential values were observed. However, even in this activity range, the variability obtained within 1 day did not exceed a few millivolts. It is particularly noteworthy that these highly reproducible, low detection limit responses of the calciumselective electrodes were obtained even without applying timeconsuming, complicated pretreatment procedures.17,27,35 Also noteworthy is that, for Ca sensors prepared without CP in the membrane phase (i.e., coated disk arrangement), preconditioned in the same way, super-Nerstian responses were obtained just after pretreatment. This supports the thesis of the beneficial role of conducting polymer and its components when included in the membrane phase. In line with the considerations above, when the activity range was restricted to 10-3-10-9 M CaCl2, a linear responses were obtained with slope 31.1 ( 1.0 mV/decade, R2 ) 0.992 with SD reaching 7 mV for an activity range from 10-7 to 10-9 M and close to 3 mV for higher activities for data obtained over two weeks, Figure 2B. Increasing the CP content in the composite membrane to 8.2% w/w had little effect on both slope and reproducibility of the potential values recorded for calcium ion activities higher than 10-6 M. Within the activity range from 10-6 to 0.1 M, slopes close to Nernstian were obtained (for this range, the slope of the line presented in Figure 2C is 29.9 ( 0.8 mV/decade, R2 ) 0.997). For activities lower than 10-6 M, a greater variation in the recorded

Figure 2. Reproducibility of potentiometric responses of calciumselective composite membrane electrodes containing (A) 6.4% CP in the membrane within one week, error bars SD, n ) 10, (B) 6.4% CP in the membrane for activities ranging from 10-3 to 10-9 M within two weeks, error bars SD, n ) 9, and (C) 8.2% CP in the membrane within one week, error bars SD, n ) 10. Before measurements electrodes were conditioned in 10-3 M CaCl2 for 3 days, in-between measurements, electrodes were kept in EDTA buffer of constant and low activity of Ca2 ions.

(35) Konopka, A.; Sokalski, T.; Lewenstam, A.; Maj-Zurawska, M. Electroanalysis 2006, 18, 2232-2242.

potentials was observed compared with the 6.4% CP preparation. In all cases, the overall potential change in the range from 10-6

324 Analytical Chemistry, Vol. 80, No. 1, January 1, 2008

to 10-9 M was greater than Nernstian; however, the magnitude of the super-Nernstian potential change and its onset on the activity scale was variable. Furthermore, no trend could be elucidated from the observed changes, but the variability recorded for calcium activities lower than 10-6 M is reflected in the higher SD presented in Figure 2C. The variability of potentials recorded within the range from 10-7 to 10-9 M was close to 30 mV; i.e., close to a decade equivalent change in concentration. It is clear, therefore, that comparison of potentiometric responses from composite membranes containing 6.4 and 8.2% w/w CP, pretreated in the same way and tested in parallel, leads to the conclusion that CP present in the composition significantly affects the properties of the resulting phase. To the authors’ best knowledge, the addition of a CP component to the membrane has not been previously characterized by improved detection limits. In this instance, we attribute the effect to the ion-exchange properties of CP doping anions to the presence of sodium/ potassium cations in the CP composition. The dispersion of poly(3,4-ethylenedioxythiophene) methacrylate end capped in nitromethane is stabilized by the presence of p-toluenesulfonate anions. The ICPMS analysis of the conducting polymer suspension has revealed the presence of a high amount of sodium and potassium (p-toluenesulfonate counterions) and the presence of sulfur. The estimated sodium content is close to 1 M, and potassium to 0.15 M. The presence of sulfur, sodium, and potassium (and the presence of calcium in the conditioned composite membrane) was also confirmed in an LA-ICPMS experiment. Therefore, both sodium and potassium are introduced to the poly(acrylate) composite ion-selective membrane together with the CP suspension. This implies that for the first time ionselective membranes with a high loading of interferent ions can be obtained by polymerization, without necessitating pretreatment procedures. The p-toluenesulfonate ions affect the free calcium ion activity within the membrane and in this way help to prevent unwanted leakage of Ca2+ ions from the membrane in a low-activity range. On the other hand, sodium and potassium ions introduced to the membrane together with CP composition result in high interfering ions loading of the composite membrane, which is also an essential prerequisite to obtain improved sensor responses (e.g., ref 1). Therefore, the CP composition applied in this work not only has resulted in enhancement of charge transfer between the membrane and the support but also has significantly contributed to improvement of analytical parameters of the sensor. As seen in Figure 2, changing the CP content in the composite membrane leads to change in the sensor characteristics from broad range linear Nernstian response to super-Nernstian in a limited low-activity range. Interestingly, this change affects the stability of the potential values recorded in these low-activity solutions, without influence on the reproducibility of potentials values recorded for high activities of Ca2+ solutions. At high activities, leakage of primary ions from the membrane or depletion of the boundary region due to incorporation of ions into the membrane is not significant compared with the high activity of primary ions in solution. In this range, the stability of the potentiometric response is determined mainly by charge-transfer processes at the membrane/substrate interface, which can be studied by means of chronopotentiometry.34 As shown earlier,

Table 1. Mean Potentiometric Selectivity Coefficients, log Kpot Ca,J ( SD and Slope (mV/decade) Obtained within the Activity Range from 10-2 to 10-4 mol/dm3, Separate Solution Method, for Ca-Selective Electrodes with Composite Membranes Following Conditioning in EDTA Buffer log Kpot Ca,J ( SD

J Li+ K+ Na+ Mg2+ Ba2+

6.4 % PEDOT

8.2 % PEDOT

lit. data37

-3.2 ( 0.1 (51.2) -3.0 ( 0.1 (52.6) -3.1 ( 0.2 (53.7) -7.6 ( 0.2 (19.1) -2.9 ( 0.2 (30.6)

-3.2 ( 0.3 (53.2) -3.0 ( 0.2 (54.1) -3.0 ( 0.3 (55.3) -7.7 ( 0.1 (24.5) -2.5 ( 0.3 (33.2)

-4.9 ( 0.6 -5.6 ( 0.8 -6.4 ( 0.3 -8.6 ( 0.3 -3.1 ( 0.1

currents in the range of nanoamperes, to tens of hundreds of nanoamperes, do not affect ion fluxes at the ion-selective membrane/solution interface for solution activities higher than 10-3 M.22,33,34 On the other hand, at activities lower than 10-5-10-6 M, the ion fluxes at the membrane/solution interface give rise to significant potential changes, especially when super-Nernstian slopes are observed. The magnitude and direction of primary ion flux can be affected by many factors related to the membrane itself, internal solution if used, sample, etc.,36 and higher fluctuations in potential are observed even for well-defined charge transfer at the back side.37 Thus, composite membranes containing a higher amount of CP and higher amount of p-toluenesulfonate, and sodium/potassium ions, with a tendency to incorporate calcium ions into the membrane phase (as proven by a superNerstian slope at low activities37), are characterized with apparently lower stability. Table 1 presents selectivity coefficients determined for composite calcium membrane sensors. Notwithstanding the ionexchange properties introduced with the CP preparation, similar values were obtained regardless of the amount of CP introduced pot pot to the membrane. Log KCa,Mg and log KCa,Ba were comparable with values reported earlier for tailored internal solution electrodes with PVC membranes containing the same ionophore.37 However, composite membrane selectivity toward Na+, K+, and Li+ ions was lowered compared to the tailored internal solution electrode with PVC membranes, but still, selectivity coefficients were comparable with those obtained for electrodes with this type of membrane and “conventional” (10-2 M CaCl2) internal solution.37 The values of the selectivity coefficients for these monovalent cations are not unexpected taking into account that high amounts of sodium and potassium are introduced to the membrane with the CP composition. Nevertheless, contrary to the “conventional” calcium-selective electrodes,37 the sensors with composite membranes gave slopes recorded in the presence of interfering ions that were close to Nernstian, with the exception of Mg2+ interference. Introduction of a higher amount of CP to the ion-selective membrane can sometimes result in increased redox sensitivity of the membrane.4 Therefore, the effect of a redox couple (Fe(CN)63-/4-) present in the sample was studied with oxidized (36) Ceresa, A.; Sokalski, T.; Pretsch, E. J. Electroanal. Chem. 2001, 501, 7076. (37) Sokalski. T.; Ceresa, A.; Fibioli, M.; Zwickl, T.; Bakker, E.; Pretsch, E. Anal. Chem. 1999, 71, 1210-1214.

Analytical Chemistry, Vol. 80, No. 1, January 1, 2008

325

Figure 3. Reproducibility of potentiometric responses of potassiumselective composite membrane electrodes containing 6.4% CP in the membrane within three weeks, error bars SD, n ) 15 conditioned before (for 20 h) and in-between measurements in (b) 10-3 and (O) 1 M KCl. For easy comparison, experimental lines were shifted to give the same value at 10-3 M KCl.

to reduced form concentration ratios varying from 1/10 to 10/1, spiked with KCl to achieve constant K+ concentration equal to 1 M. It was found for the composite membrane, containing either 6.4 or 8.2% w/w CP, that the recorded potentials were practically independent of logarithm of [Fe(CN)63-/Fe(CN)64-] (slope of the dependence