Anal. Chem. 2000, 72, 3236-3240
Improved Detection Limits and Unbiased Selectivity Coefficients Obtained by Using Ion-Exchange Resins in the Inner Reference Solution of Ion-Selective Polymeric Membrane Electrodes Wei Qin, Titus Zwickl, and Erno 1 Pretsch*
Department of Organic Chemistry, Swiss Federal Institute of Technology (ETH), Universita¨tstrasse 16, CH-8092 Zu¨rich, Switzerland
By using a high concentration of an interfering ion and a low one of the primary ion in the inner reference solution of polymeric membrane ion-selective electrodes (ISEs), the lower detection limit may be improved and unbiased thermodynamic selectivity coefficients may be obtained. To this purpose, a cation-exchange resin is used here to keep the low concentration of the primary cation constant. Different compositions of the internal solution are required for obtaining optimal lower detection limits and unbiased selectivity coefficients. All ISEs studied here, i.e., for K+, Ca2+, and NH4+, based on valinomycin, ETH 5234, and nonactin/monactin, respectively, show improved lower detection limits in the range of 10-7.6 (NH4+) to 10-8.8 M (Ca2+). Nernstian responses and, therefore, unbiased selectivity coefficients are obtained with the K+ISE for the discriminated ions, Na+, Mg2+, and Ca2+. The measuring range of conventional ion-selective membrane electrodes (ISEs) lies between 1 and 10-6 M.1-3 Recently, it was shown that this lower detection limit is not an inherent property of ISEs but determined by a small flux of primary ions from the organic sensing membrane into its aqueous surface layer.4-6 As a consequence, a constant low activity, ai,lim, of the primary ion (Izi) is maintained in this surface layer even when the activity of Izi in the sample, ai, is lower than ai,lim.7-9 This small flux of primary ions has been successfully eliminated by choosing an internal solution with a constant low activity of Izi and a high one of an (1) Umezawa, Y. Handbook of Ion-Selective Electrodes: Selectivity Coefficients; CRC Press: Boca Raton, FL, 1990. (2) Bu ¨ hlmann, P.; Pretsch, E.; Bakker, E. Chem. Rev. 1998, 98, 1593-1687. (3) Bakker, E.; Bu ¨ hlmann, P.; Pretsch, E. Electroanalysis 1999, 11, 915-933. (4) Sokalski, T.; Ceresa, A.; Zwickl, T.; Pretsch, E. J. Am. Chem. Soc. 1997, 119, 11347-11348. (5) Mathison, S.; Bakker, E. Anal. Chem. 1998, 70, 303-309. (6) Maj-Zurawska, M.; Erne, D.; Ammann, D.; Simon, W. Helv. Chim. Acta 1982, 65, 55-62. (7) Sokalski, T.; Zwickl, T.; Bakker, E.; Pretsch, E. Anal. Chem. 1999, 71, 12041209. (8) Sokalski, T.; Ceresa, A.; Fibbioli, M.; Zwickl, T.; Bakker, E.; Pretsch, E. Anal. Chem. 1999, 71, 1210-1214. (9) Morf, W. E.; Badertscher, M.; Zwickl, T.; de Rooij, N. F.; Pretsch, E. J. Phys. Chem. 1999, 103, 11346-11356.
3236 Analytical Chemistry, Vol. 72, No. 14, July 15, 2000
interfering ion, Jzj.4 Originally, internal solutions with ion buffers (e.g., EDTA) were used for this purpose. Obviously, this is not feasible with cations such as alkali metal and ammonium ions as well as many anions for which no appropriate complexing agents are available. Another method to avoid leaching of primary ions from the membrane into the sample makes use of an externally applied current that generates a steady flux of cations toward the inner compartment of the ISE.10 Although this method does not require any complexing agents, it needs more sophisticated instrumentation and measurement procedures and is, therefore, less appropriate for routine applications. Ions leaching from the membrane into the sample not only influence the lower detection limits of ISEs but also bias the selectivity coefficients since they may be potential-determining even if measurements are taken in pure solutions of a strongly discriminated ion.3,11 It was found that selectivity coefficients thus obtained represent upper limits rather than true (thermodynamic) values, which may be better by many orders of magnitude than those determined by conventional methods.11 This bias was first avoided by Bakker, who showed that Nernstian response curves and thermodynamic selectivity coefficients for discriminated ions are obtained if conventional ISE membranes have never been in contact with primary ion solutions before measuring the interfering ions.12,13 Another possibility of determining unbiased selectivity coefficients is to use internal solutions with ion buffers,4,8 which prevent the leaching of primary ions. However, so far, this method has been limited to primary ions for which appropriate complexing agents are known. In this work, a cation-exchange resin is added to the internal solution of ISEs to keep the primary ion activity at a constant low level. It is shown that the lower detection limit is improved and true selectivities are obtained for a valinomycin-based K+-ISE. The general applicability of the method is demonstrated by improving the lower detection limits of NH4+- and Ca2+-ISEs. (10) Lindner, E.; Gyurcsa´nyi, R. E.; Buck, R. P. Electroanalysis 1999, 11, 695702. (11) Bakker, E.; Pretsch, E.; Bu ¨ hlmann, P. Anal. Chem. 2000, 72, 1127-1133. (12) Bakker, E. J. Electrochem. Soc. 1996, 143, L83-L85. (13) Bakker, E. Anal. Chem. 1997, 69, 1061-1069. 10.1021/ac000155p CCC: $19.00
© 2000 American Chemical Society Published on Web 06/10/2000
EXPERIMENTAL SECTION Reagents. All salts, NaOH, the ionophores (valinomycin, N,Ndicyclohexyl-N′,N′-dioctadecyl-3-oxapentanediamide (ETH 5234), nonactin), sodium tetrakis[3,5-bis(trifluoromethyl)phenyl]borate (NaTFPB), 2-nitrophenyl octyl ether (o-NPOE), bis(2-ethylhexyl) sebacate (DOS), poly(vinyl chloride) (PVC), and tetrahydrofuran (THF) were puriss. p.a. or Selectophore from Fluka AG (CH-9471 Buchs, Switzerland). The cation-exchanger resin (Dowex C-350, H+ form, 30-80 mesh, Fluka AG) was chosen because of its high capacity and fast ion-exchange kinetics. Suprapur Mg(NO3)2‚6H2O and Ca(NO3)2‚4H2O (maximum K+ content, 1 and 2 ppm, respectively) were from Merck (D-64293 Darmstadt, Germany). Aqueous solutions were prepared with freshly deionized water (18.0 MΩ cm specific resistance) obtained with a NANOpure reagent-grade water system (Barnstead, CH-4009 Basel, Switzerland). Preparation of Ion Exchanger and Determination of Its Selectivities. The cation-exchange resin (H+ form, 5 g) was equilibrated with 1 M NaOH (50 mL) for 12 h under constant stirring. The resulting Na+ form was then washed with deionized water and dried overnight at 100 °C. Its cation-exchange capacity was determined as 5.2 mequiv g-1 resin by titrating the dry H+ form (0.5 g) with 0.1 M NaOH.14 To determine the selectivities of the cation-exchange resin (Km,Na ) mM,resinmNa,aq1/zM/mNa,resin1/zMmM,aq with M for K+, Ca2+, or NH4+), the Na+ form (0.10 g) was stirred for 12 h with 50 mL of either 10-3 M KCl, 10-3 M CaCl2, or 10-2 M NH4Cl, and the K+, Ca2+, or NH4+ concentrations in the respective aqueous phases were measured with the corresponding ISEs yielding 5.0 × 10-5 M K+, 1.0 × 10-7 M Ca2+, and 4.6 × 10-3 M NH4+. This gave KK,Na ) 1.9, KCa,Na ) 33.6, and KNH4,Na ) 1.3 as cation-exchanger selectivity coefficients. Membranes and ISEs. Membranes contained (in wt %) valinomycin (1.1), NaTFPB (0.5), DOS (65.7), and PVC (32.7) for K+-ISEs, ETH 5234 (1.3), NaTFPB (0.7), o-NPOE (65.4), and PVC (32.6) for Ca2+-ISEs, and nonactin/monactin (75%/25%) (1.0), NaTFPB (0.6), DOS (65.7), and PVC (32.7) for NH4+-ISEs. The components of each membrane (totaling 240 mg) were dissolved in THF (2.5 mL) and poured into a glass ring (i.d. 28 mm) fixed on a glass plate. Overnight evaporation of the solvent yielded transparent membranes of ∼200-µm thickness. For each ISE, a disk of 3-mm diameter was punched from the membranes and glued to a plasticized PVC tubing with THF/PVC slurry. For the conventional ISEs (type A), the internal filling and conditioning medium were 10-2 M chloride solutions of the respective primary ions. For the other ISEs, an appropriate amount of the dry Na+ form of the cation-exchange resin was added to 1 mL of the internal filling solutions. Their compositions and calculated activities are given in Table 1. For calculating the compositions, a swelling water content of 40% of the cation-exchange resin was assumed (manufacturer data). In two cases, the activity of the primary ion was measured (see Table 1). The results agreed with the calculated values within experimental error. The ISEs with ion exchanger in the internal filling were conditioned in 10-3 M chloride solution of the primary ion for 1 day. Emf Measurements. All measuring solutions were obtained by successive automatic dilution of stock solutions with a Metrohm (14) Harland, C. E. Ion Exchange: Theory and Practice, 2nd ed.; Royal Society of Chemistry: Herts, U.K., 1993.
Table 1. Compositions and Calculated Activities of Internal Solutions Used K+-ISEs composition
activities (M)
type
KCl (M)
NaCl (M)
resin (g)
K+
Na+
A B C D
10-2 10-3 10-4 0
0 0 10-3 10-3
0 0.2 0.5 0.5
9.0 × 10-3 5.5 × 10-7 a 3.1 × 10-8 0
0 1.1 × 10-3 1.5 × 10-3 1.3 × 10-3
Ca2+-ISEs composition
activities (M)
type
CaCl2 (M)
NaCl (M)
resin (g)
Ca2+
Na+
A B C
10-2 5 × 10-4 5 × 10-5
0 10-2 10-2
0 0.5 0.5
5.3 × 10-3 6.2 × 10-9 5.1 × 10-10
0 1.4 × 10-2 1.3 × 10-2
NH4+-ISEs compositionb
activities (M)
type
NH4Cl (M)
resin (g)
NH4+
Na+
A B C
10-2 10-3 10-3
0 0.05 0.5
9.0 × 10-3 3.0 × 10-6c 4.2 × 10-7
0 9.7 × 10-4 1.3 × 10-3
a Measured: cMeasured: 4.3
5.9 × 10-7 M. × 10-6 M.
b
The internal solution had pH 6.3.
Liquino 711 and two Metrohm Dosino 700 instruments equipped with 50-mL burets. The experiments were performed under full computer control in a 100-mL polyethylene beaker. Potentials were measured with a custom-made 16-channel electrode monitor at room temperature (20-21 °C) in the galvanic cell: Ag/AgCl/1 M LiOAc/sample solution/ISE membrane/inner filling solution/ AgCl/Ag provided with a double-junction reference electrode (type 6.0729.100, Metrohm AG, CH-9101 Herisau, Switzerland). All emf values were corrected for the liquid-junction potential according to the Henderson equation. Activity coefficients were calculated by the Debye-Hu¨ckel approximation.15 Selectivity coefficients were determined by the separate solution method in metal chloride solutions.11,16,17 RESULTS AND DISCUSSION Recently, it was established that leaching of primary ions from the sensing membrane can influence the lower detection limit and the measured selectivities of ISEs. This ion release is a consequence of zero-current ion fluxes generated by concentration gradients within the membrane caused mainly by two processes: (1) coextraction of primary ions, Izi, with their counterions from the inner electrolyte solution into the inner surface layer of the membrane5 and (2) ion exchange of Izi in the outer surface layer of the membrane by an interfering ion, Jzj, from the sample.7,9,18 The latter process, eventually, causes potentiometric interference when ∼50% of Izi is replaced by Jzj (for zI ) zJ). However, as even (15) Meier, P. C. Anal. Chim. Acta 1982, 136, 363-368. (16) Guilbault, G. G.; Durst, R. A.; Frant, M. S.; Freiser, H.; Hansen, E. H.; Light, T. S.; Pungor, E.; Rechnitz, G.; Rice, N. M.; Rohm, T. J.; Simon, W.; Thomas, J. D. R. Pure Appl. Chem. 1976, 48, 127-132. (17) Buck, R. P.; Lindner, E. Pure Appl. Chem. 1994, 66, 2527-2536. (18) Zwickl, T.; Sokalski, T.; Pretsch, E. Electroanalysis 1999, 673-680.
Analytical Chemistry, Vol. 72, No. 14, July 15, 2000
3237
Table 2. Potentiometric Selectivity Coefficients for a K+-Selective Electrode (Type D) with Cation-Exchange Resin in the Internal Solution, Compared with Literature Values Nernstian response slope ion J range (M) (mV dec-1) K+ Na+ Mg2+ Ca2+
10-1-10-6 10-1-10-6 10-3-10-6 10-3-10-6
61.0 58.7 29.7 29.2
a log Kpot KJ
b log Kpot KJ
c log Kpot KJ
0.0 0.0 0.0 -4.2 ( 0.1 (n ) 6) -4.5 ( 0.1 -4.1 ( 0.1 -7.6 ( 0.1 (n ) 4) -7.5 ( 0.1 -5.2 ( 0.1 -6.9 ( 0.1 (n ) 4) -6.9 ( 0.1 -5.0 ( 0.1
a Mean values obtained from n corresponding pairs of concentrations of K+ and the respective interfering cation in the Nernstian response range ( SD. bReference 12 (conditioning in 10-2 M NaCl). c Reference 12 (conditioning in 10-2 M KCl).
Figure 1. Emf responses of a K+-ISE (type D, see Experimental Section), whose internal solution consists of 10-3 M NaCl and 0.5 g of cation-exchange resin (Na+ form), to K+ and main interfering cations.
very small concentration differences between the two membrane sides induce relevant ion fluxes, leaching due to ion exchange occurs at a much lower activity of the interfering ion than its direct potentiometric interference. The lower detection limits and measured selectivity coefficients of polymeric membrane ISEs have been improved by using an internal solution with a high concentration of an interfering ion, Jzj, and a low one of the primary ion, Izi.4 Since in such a system, a steady flux of primary ions from the sample toward the inner compartment is maintained, their activity must be kept at a constant low level by chemical means, e.g., by adding to the internal solution a complexing agent4 or a counterion that forms a sparingly soluble salt. Clearly, this is limited to electrodes selective for ions for which such agents are available, but is not practicable with numerous analytes, including alkali metal cations and various anions. The topic of this work is to extend the method to such cations by using cation exchangers and to demonstrate its feasibility with several cation-selective membrane electrodes. In Figure 1, the response to different cations is shown for a valinomycin-based K+-ISE whose internal solution contains NaCl and the Na+ form of a cation exchanger (type D, see Experimental Section). Close to theoretical responses are obtained not only for the primary ion but also for the discriminated interfering ions, Na+, Mg2+, and Ca2+ (Table 2). So far, such a response has only been obtained by Bakker when taking the calibration curves for interfering ions before the K+-ISE, with 10-2 M NaCl as internal solution, had ever been in contact with K+, but not after.12,13 In 3238 Analytical Chemistry, Vol. 72, No. 14, July 15, 2000
Figure 2. Emf responses of four K+-ISEs with identical membranes but different internal solutions (see Experimental Section): type A (conventional), 10-2 M KCl; type B, 10-3 M KCl, 0.2 g of cationexchange resin (Na+ form); type C, 10-3 M NaCl, 10-4 M KCl, 0.5 g of cation-exchange resin (Na+ form); type D, 10-3 M NaCl, 0.5 g of cation-exchange resin (Na+ form).
contrast, the curves shown here were determined after conditioning the K+-ISE in 10-3 M KCl. The apparently super-Nernstian response to Mg2+ and Ca2+ at higher activities is probably due K+ impurities of the nitrates used.13 According to the calibration curves (Figure 1), the K+ contents of 0.1 M solutions of these nitrates are 1 × 10-7 and 4 × 10-7 M, respectively, and correspond thus to ∼25% of the maximum values specified by the manufacturer (see Experimental Section). Deviations from Nernstian response of Mg2+ and Ca2+ were not observed by Bakker.12 This can be explained by the fact that, in his case, the ISEs were not in contact with K+ before using them in Mg2+ and Ca2+ solutions
Figure 3. Time-dependent emf response traces of K+-ISEs, types A-D (cf. Figure 2) in KCl solutions. Sample concentrations are indicated on the upper x axis.
so that the uptake of K+ impurities from the sample was much stronger. Theoretical response to Na+ with a K+-ISE based on another ionophore (BME-44) has also been obtained, though within a smaller range, by Lindner,10 using the method of exponential dilution in a wall-jet cell.19 Such a setup ensures a high sample flow rate that effectively reduces the thickness of the diffusion layer and, thereby, the influence of leaching of primary ions on the lower detection limit and selectivities, but cannot fully eliminate it.10 The potentiometric selectivity coefficients obtained from the Nernstian part of the calibration curves11 are also given in Table 2, together with those obtained with K+ISEs conditioned in NaCl (column 5) or in KCl (column 6).12 As expected from the calibration curves, our values agree with the unbiased ones of Bakker12 (column 5). ISEs with nearly complete replacement of the primary ion by an interfering one at the inner membrane side are optimal for giving unbiased selectivity coefficients but do not exhibit the best possible lower detection limits. This is due to the fact that, at low sample activities, the strong ion fluxes through the membrane induce a significant concentration polarization in the aqueous Nernst layer of the membrane surface. Therefore, at low sample activities, the ISE is no longer sensitive to primary ions and, in the intermediate activity range, the electrode function is superNernstian.7,8 The response of various K+-ISEs having the same membrane but different compositions of the internal solution is shown in Figure 2. The optimal response range with a lower detection limit of 10-8.3 M is obtained with type B, for which the activity of Na+ in the internal solution is larger by ∼4 orders of magnitude than that of K+. The apparently improved sensitivity below 10-6 M K+ (super-Nernstian slopes) of types C and D, which have higher ratios of aNa+/aK+ in the internal solution, is of no practical use because, as seen from Figure 3, both ISEs show strong potential drifts in this activity region. If steady state had been achieved, the response at the critical activity should be steplike.9 On the other hand, a good stability of the response down to 10-8 M K+ is observed with the optimal type B (Figure 3). The addition of a cation exchanger to the internal solution of an ISE in order to maintain a low constant concentration of the (19) Horvai, G.; To´th, K.; Pungor, E. Anal. Chim. Acta 1976, 82, 45-54.
Figure 4. Emf responses of three Ca2+-ISEs with identical membranes but different internal solutions (see Experimental Section): type A (conventional), 10-2 M CaCl2; type B, 10-2 M NaCl, 5 × 10-4 M CaCl2, 0.5 g of cation-exchange resin (Na+ form); type C, 10-2 M NaCl, 5 × 10-5 M CaCl2, 0.5 g of cation-exchange resin (Na+ form).
primary ion is a convenient alternative to that of a complex-forming agent. The basic equivalence of the two methods is demonstrated in Figure 4, where the responses of three Ca2+-ISEs again with the same membrane but different internal solutions (see Experimental Section) are compared. The improvement of the lower detection limit from 10-6.8 (type A) to 10-8.8 M (type B) and the apparently super-Nernstian slope observed for a 10-fold decrease in the free Ca2+ activity of the internal solution (type C) is in analogy to our recent results obtained with EDTA as ion buffer in the internal solution.8 It must be noted that the binding selectivity of the ion exchanger is low (see Experimental Section), whereas EDTA complexes are very strong with Ca2+ and negligible with Na+. Although the ion exchanger used lacks selectivity, it is capable of keeping the aCa2+/aNa+ ratio in the internal solution fairly constant because of the high total ion concentrations involved. Experiments with the NH4+-ISEs show that the method described here is also applicable in cases where the ion exchanger has practically no binding selectivity for the primary ion. This is because a high total amount of an interfering ion is sufficient to maintain the low activity of the primary ion in the internal solution even if the selectivity of the ion-exchange resin toward these ions is poor. Figure 5 again demonstrates that, by successively reducing the NH4+ activity of the internal solution, the lower detection limit is first improved from 10-6.9 (type A) to 10-7.6 (type B) but then, a super-Nernstian response appears for sample Analytical Chemistry, Vol. 72, No. 14, July 15, 2000
3239
obtained with a flow-through system and ISEs having 10-4 M NH4Cl as internal solution.21 In principle, by choosing an internal solution of high pH, NH4+ diffusing through the membrane should be successfully removed. Since, however, NH3 thus generated diffuses through the membrane, the detection limit did not improve with 10-3 M NaCl in 10-2 M NaOH as internal solution (results not shown). The lifetime of ISEs having a constant flux of primary ions toward the inner compartment clearly depends on the flux rate and the ion-exchange (or buffer) capacity of the internal solution. No attempt was made here to optimize the ISEs in this respect, although various possibilities are feasible, including changes in the membrane composition and thickness as well as in cell design.18 Upon continuous contact of K+-ISEs of type D with 10-3 M KCl for one week, the selectivities did not change significantly. The lifetime of polymeric membrane ISEs used here is only influenced by their contact with primary ion solutions. Hence, for a long lifetime, such ISEs should preferably be stored in a solution of an interfering ion, ideally that present in the internal solution.
Figure 5. Emf responses of three NH4+-ISEs with identical membranes but different internal solutions (see Experimental Section): type A (conventional), 10-2 M NH4Cl; type B, 10-3 M NH4Cl, 0.05 g of cation-exchange resin (Na+ form); type C, 10-3 M NH4Cl, 0.5 g of cation-exchange resin (Na+ form). In the structural formula, R1 ) CH3 (75%) and CH2CH3 (25%), R2 - R4 ) CH3.
activities below 10-6 M (type C). The lower detection limit of ISE B is not the best possible one since it is presumably caused by NH3 in ambient air.20 Similar detection limits have also been (20) Meyerhoff, M. E. Anal. Chem. 1980, 52, 1532-1534. (21) Hara, H.; Matsumoto, S. Analyst 1994, 119, 1839-1842.
3240 Analytical Chemistry, Vol. 72, No. 14, July 15, 2000
CONCLUSIONS Polymeric membrane ISEs can be improved with a view to obtaining better response ranges and/or unbiased selectivity coefficients. Both aims can be achieved by carefully controlling the ionic composition of the internal solution. It is shown that ion-exchange resins added to the internal solution can be used for this purpose. This extends the applicability of the principle to ions for which no adequate complexing agents are available. ACKNOWLEDGMENT This work was supported by the Swiss National Science Foundation, the World Laboratory (Lausanne), and Orion Research Inc. (Beverly, MA). We thank Prof. M. E. Meyerhoff for helpful discussions and Dr. D. Wegmann for careful reading of the manuscript. Received for review February 7, 2000. Accepted May 1, 2000. AC000155P
