Anal. Chem. 1999, 71, 5304-5308
Emulsion-Based Fluorosensors for Potassium Featuring Improved Stability and Signal Change Christian Krause, Tobias Werner,* Christian Huber, and Otto S. Wolfbeis
Institute of Analytical Chemistry, Chemo- and Biosensors, University of Regensburg, 93040 Regensburg, Germany
A novel kind of potassium optode is presented which is based on the use of lipophilic droplets containing valinomycin and entrapped in a structured hydrogel. A positively charged solvatochromic dye located near the surface of the droplets responds to the valinomycinassisted extraction of potassium from the sample by dramatic decrease of fluorescence intensity. The dynamic range is from 5 to 100 mM potassium, with negligible cross sensitivity to ionic strength. Cross sensitivity to pH is negligible too in the pH range from 6.5 to 7.3. The effect of interfering lipophilic anions is discussed. Response times within the dynamic range are less than 3 min on going from low to high potassium ion concentrations but about 10 min in the reverse direction. Response is fully reversible with only small drifts in baseline. The measurement uncertainty of determining 5 mM potassium is better than 0.2 mM. Analysis of potassium in blood is of tremendous interest in medical intensive care.1 Conventional analytical methods such as flame spectroscopy are time-consuming and require skilled personnel.2 Optical sensors (“optodes”) for potassium have gained growing interest due to the ease of handling and the availability of disposable test strips.1 Sensors based on so-called potential- or polarity-sensitive dyes (PSDs) are reported to show no response behavior according to the mass action law.3 This is in contrast to ion-exchange,4,5 coextraction,6-8 or chromoionophore-based optodes.9,10 They also do not require complete mass transport of the analyte from the sample into the interior of the sensing layer. Therefore, they do not suffer from slow response times which are a disadvantage of bulk optodes.11 A cationic PSD located at the surface of the sensor indicates the adsorption of the analyte. (1) Ng, R. H.; Sparks, K. M.; Stateland, B. E. Clin. Chem. 1992, 38, 13711372. (2) Gibb, I. J. Clin. Pathol. 1987, 40, 298-301. (3) Wolfbeis, O. S. Sens. Actuators, B 1995, 29, 140-147. (4) Suzuki, K.; Ohzora, H.; Thoda, K.; Miyazaki, K.; Watanabe, K.; Inoue, H.; Shirai, T. Anal. Chim. Acta 1990, 237, 155-164. (5) Wang, K.; Seiler, K.; Morf, W. E.; Spichiger, U. E.; Simon, W.; Lindner, E.; Pungor, E. Anal. Sci. 1990, 6, 715-720. (6) Charlton, S. C.; Fleming, R. L.; Zipp, A. Clin. Chem. 1982, 28, 1857-1861. (7) Krause, C.; Werner, T.; Wolfbeis, O. S. Anal. Sci. 1998, 14, 163-167. (8) Krause, C.; Werner, T.; Huber, C.; Leiner, M. J. P.; Wolfbeis, O. S. Anal. Chem. 1999, 71, 1544-1548. (9) Garcia, R. P.; Moreno, F. A.; Diaz-Garcia, M. E.; Sanz-Medel, A.; Narayanaswamy, R. Clin. Chim. Acta 1992, 207, 31-40. (10) Wolfbeis, O. S.; Offenbacher, H. Chem. Monthly 1984, 115, 647-654. (11) Janata, J. Anal. Chem. 1992, 64, 921A-927A.
5304 Analytical Chemistry, Vol. 71, No. 23, December 1, 1999
The variety of such sensors has been reviewed.3 Besides Langmuir-Blodgett films,12,13 which proved to be mechanically labile, plasticized poly(vinyl chloride) (PVC) was often used as the sensor material.3,14 The carrier-containing PVC membrane was dipped in a dye solution to improve the ratio of bulk dye/surface dye.15 In other research work, PVC layers thinner than a few micrometers were used.3,14 In the case of valinomycin-based potassium optodes, the use of PSD sensors is limited by improper dynamic ranges and small signal changes. Additionally, the response to the analyte can hardly be adjusted to the desired dynamic range because variations in membrane composition do not substantially alter the turning point of the calibration curve.14 In this work, a PSD-based optode is presented that makes use of the high inner surface of a oil-in-water emulsion. This will combine the advantage of a fast response time of surface-based sensors, e.g., Langmuir-Blodgett systems, with the advantage of bulk optodes possessing high signal changes. Furthermore, the use of such a system consisting of a two-phase membrane offers a higher degree of freedom in changing the dynamic range than in single-phase sensors. Solid oil-in-water emulsion-based membranes were previously used for pH-insensitive coextraction-based optodes.8 A highly solvatochromic, lipophilic cationic dye is dissolved together with a lipophilic anion and the lipophilic cation carrier valinomycin in plasticizer droplets entrapped in a hydrogel. EXPERIMENTAL SECTION Chemicals and Solutions. Analytical grade solvents, potassium tetrakis(chlorophenyl)borate (PTCPB) and valinomycin were obtained from Fluka (Buchs, Switzerland). Polyester foil (Mylar, 125 µm) was from Goodfellow (Cambridge Ltd., Cambridge, U.K.). The hydrogel HN80 (a partially hydrolyzed polyacrylamide) was from Hymedix (Dayton, OH), and CPDDE (2-cyanophenyl dodecyl ether) was synthesized as reported previously.16 4-(4-(Dihexadecylamino)styryl)-N-methylpyridinium iodide (DIA) (Catalog No. D-3883) was from Molecular Probes (Eugene, OR). Phosphate buffers were used throughout. All buffer compositions (12) Wolfbeis, O. S.; Schaffar, B. P. H. Anal. Chim. Acta 1987, 198, 1-12. (13) Shimomura, M.; Honma, A.; Kondo, S.; Shinohara, E.; Tajima, N.; Koshiishi, K. Sens. Actuators, B 1993, 13-14, 629-631. (14) Murkovic, I.; Lobnik, A.; Mohr, G. J.; Wolfbeis, O. S. Anal. Chim. Acta 1996, 334, 125-132. (15) Kawabata, Y.; Tahara, R.; Kamichika, T.; Imasaka, T.; Ishibashi, N. Anal. Chem. 1990, 62, 1528-1531. (16) Krause, C.; Werner, T.; Huber, C.; Klimant, I.; Wolfbeis, O. S. Anal. Chem. 1998, 70, 3983-3985. 10.1021/ac9907383 CCC: $18.00
© 1999 American Chemical Society Published on Web 10/23/1999
Table 1. Cocktail Composition of Examined Membranes membrane HN 80 CPDDE valinomycin DIA PTCPB DMSO no. (mg) (mg) (mg) (mg) (mg) (µL) M1 M2 M3 M4 M5 M6 M7 M8 M9 M10 M11 M12
10 10 10 10 10 10 10 10 10 10 10 10
0.53 0.25 1.6 0.53 0.53 0.53 0.53 0.53 0.53 0.53 0.53 0
0.1 0.05 0.3 0.025 0.01 0.2 0.1 0.1 0.1 0.1 0.1 0.1
0.014 0.007 0.042 0.014 0.014 0.014 0.007 0.042 0.014 0.014 0.014 0.014
0.0086 0.0043 0.0258 0.0086 0.0086 0.0086 0.0043 0.0258 0.0172 0.0014 0 0
350 350 350 350 350 350 350 350 350 350 350 350
were adjusted to constant ionic strength with NaCl as an “inert“ salt according to Perrin and Dempsey.17 The total buffer concentration (i.e. [HPO42-] + [H2PO4-]) was 20 mmol/L. Buffers and inorganic salts were of analytical grade (Merck, Darmstadt, Germany). Water was double distilled. Apparatus. All optical measurements were performed on an Aminco AB2 fluorometer (SLM Aminco, Rochester, NY). Time traces and calibration curves were recorded at an excitation wavelength of 500 nm and an emission wavelength of 605 nm. Membranes were placed in a self-made flow-through cell, and fluorescence was measured using an arrangement described earlier.7 The flow-through cell was machined from Teflon and had a volume of about 900 µL. A Minipuls 3 pump (from Gilson, Villiers-le-Bel, France) with a typical flow about 1 mL/min was used. All tubes consisted of silicone. pH was measured using a WTW 638 pH meter (WTW GmbH, Weilheim, Germany). Experiments were performed at 22 ( 1 °C. Membrane Preparation. A 150 µL aliquot of the respective membrane cocktail (see Table 1) was spread onto a dust-free 24mm-diameter spot of polyester foil. Membranes were left in a water-saturated atmosphere for at least 12 h. During water uptake membranes become turbid due to formation of an emulsion. To remove all dimethyl sulfoxide (DMSO) membranes were rinsed extensively with water and afterward conditioned by passing a potassium-free buffer over it followed by 100 mM potassium and a potassium-free buffer again.
Figure 1. Cross-sectional view of the sensor membrane composed of a hydrogel layer containing lipophilic droplets (dark), and schematic of the response mechanism (V, valinomycin).
Figure 2. Time drive of freshly prepared membrane M1 at constant ionic strength of 131 mM and pH 7.4 but varying potassium ion concentration.
RESULTS AND DISCUSSION Response to Potassium. Potassium ions diffuse from the sample into the hydrogel and adsorb at the lipophilic droplets in a way similar to that of common PVC based-PSD sensors. The droplet surface becomes charged and the dye, located in the droplets and especially near their surface, lowers its fluorescence drastically by moving out of the droplet. Because the hydrogel is hydrophilic, the dielectric constant of the environment of the DIA is dramatically changed even by small displacement of the dye. A schematic view of the membrane and its response to potassium is given in Figure 1. Solutions of constant ionic strength and pH but varying analyte concentrations were pumped through the flow-through cell where
the sensing layer was placed. Due to the spatial position of the dye, increasing the analyte concentration leads to a decrease of fluorescence. The largest spectral changes were observed at an emission wavelength of 605 nm and an excitation wavelength at 500 nm. A typical time trace is shown in Figure 2. Response time t90 (time to reach 90% of the final signal) was found to be less than 3 min within dynamic range. This is in the same order of magnitude as the time needed for exchanging the sample in the flow-through cell. The response is fully reversible, but a drift of 0.1%/min for freshly prepared membranes can be observed. This small drift most likely originates from the slow washout of potassium and iodide from PTCPB and the dye. The accumulation of the dye near the surface (DIA+ is a surfactant-like molecule) also requires some time. Furthermore, TCPB- is reported to decompose in PVC membranes.18 Therefore, aging of the membrane due to decomposition of TCPB- must be considered. Despite this limitation, due to the high photostability of the dye, the lifetime of operation of such PSD emulsion optode is much longer as can be found for coextraction-based emulsion membranes.8 Choice of Material. The sensing scheme requires (a) a lipophilic phase immiscible with water and (b) a hydrophilic phase. The fluorescence quantum yield of the dye is expected to be high in the lipophilic phase, and the ion carrier must be soluble in it. Most plasticizers are useful as lipophilic phase. CPDDE was used in this work. The hydrophilic phase has to be highly ion permeable to allow fast diffusion of the analyte cation to the lipophilic droplets.
(17) Perrin, D. D.; Dempsey, B. Buffers for pH and Metal Ion Control; Chapman and Hall Laboratory Manuals: London, 1974; Chapter 5, pp 62 ff.
(18) Dinten, O.; Spichiger, U. E.; Chaniotakis, N. A.; Gehrig, P.; Rusterholz, B.; Morf, W. E.; Simon, W. Anal. Chem. 1991, 63, 596-603.
Analytical Chemistry, Vol. 71, No. 23, December 1, 1999
5305
The fluorescence quantum yield of the dye should be very low in this hydrogel to observe significant signal changes during small displacements of the dye at the surface of the lipophilic droplets. Therefore, hydrogels with high water content are the preferred matrix materials. HN80, a partially hydrolyzed polyacrylonitrile with a water content of 80% was used. This structured hydrogel was found to stabilize lipophilic droplets without the addition of emulsifiers.8 Valinomycin is a lipophilic, highly selective potassium carrier with pH-independent binding properties and therefore the preferred ionophore for most potassium sensors. Its neutrality and outstanding potassium selectivity over sodium, which is higher than 10 000, avoids cross-sensitivity to ionic strength. The high lipophilicity ensures that valinomycin is placed in the apolar plasticizer phase. A wide variety of such ionophores were developed for ion-selective electrodes and optodes.19-21 The lipid-soluble polarity indicator DIA was chosen because of its lipophilicity and its large Stokes shift of nearly 100 nm. Its fluorescence quantum yield is high in a hydrophobic environment, but very low within a hydrophilic phase. Therefore, it is widely used as a potential-sensitive probe in cell biological investigations and also very useful for sensor applications where the analyte causes a relocation of the dye between the polar/apolar environment.14,22 Lipophilic anionic additives improve sensor characteristics, with respect to signal change and cross-sensitivity to anions.14 PTCPB is the most common lipophilic anionic additive in PVCbased ion-selective optodes23 and electrodes.24 It does not quench the fluorescence of DIA. Membrane Composition and Response Behavior. Membrane parameters were varied in order to estimate the best cocktail composition for measuring potassium in the clinical relevant range of 1-10 mM. The characteristics of the PSD emulsion membrane can be varied at least to some extent by cocktail composition in the same way as reported for our recently described coextraction membranes.7,8 Table 2 summarizes the different response characteristics according to different membrane compositions. Increasing the plasticizer fraction C (C ) mCPDDE/mHN80; M1M3, at constant concentrations of valinomycin, DIA, and PTCPB in the plasticizer) did not significantly alter the response. This is in contrast to coextraction-based sensors.8 The only small influence of plasticizer fraction is in correspondence to a surfacecontrolled process, where only the size of the droplet (and therefore the ratio bulk/surface) but not the total amount is important. An increase of the plasticizer fraction does not result in bigger droplets, but in a higher number of droplets, if the plasticizer fraction is less than 0.15. It was found that a plastizicer fraction of more than 0.10 causes mechanical instability of the membrane; however, compositions with a fraction of 0.05 turned out to be useful for preparing sensing layers with an improved stability compared to the recently described coextraction membranes8 where plastizicer fractions of about 0.15 are required. (19) Bakker E.; Bu ¨ hlmann, P. Pretsch, E. Chem. Rev. 1997, 97, 3083-3132. (20) Lehn, J. M. Supramolecular Chemistry Concepts and Perspectives; VCH: Weinheim, 1995; Chapter 2. (21) Takeda, Y. Top. Curr. Chem. 1984, 121, 1-37. (22) Schwarz, C.; Thier, P. J. Neurosci. 1995, 15, 3475-3489. (23) Rosatzin T.; Bakker, E.; Suzuki, K.; Simon, W. Anal. Chim. Acta 1993, 280, 197-280. (24) Eugster, R.; Spichiger, U. E.; Simon, W. Anal. Chem. 1993, 65, 689-695.
5306 Analytical Chemistry, Vol. 71, No. 23, December 1, 1999
Table 2. Response Characteristics and Membrane Compositions of the Sensor Membranesa
M1 M2 M3 M4 M5 M6 M7 M8 M9 M10 M11 M12
mDIA/mCPDDE (×104)
mvalinomycin/mCPDDE
mCPDDE/mHN80
nDIA/nPTCPB
R3/5
R0/100
14 14 14 14 14 14 7 42 14 14 14 ∞
0.19 0.19 0.19 0.05 0.02 0.4 0.19 0.19 0.19 0.19 0.19 ∞
0.053 0.025 0.16 0.053 0.053 0.053 0.053 0.053 0.053 0.053 0.053 0
1 1 1 1 1 1 1 1 0.5 6 ∞ ∞
3.7 4.6 3.5 3.5 1.3 4.5 5.5 2.4 2.9 5.8 2.4 0
62 63 56 43 22 60 55 23 35 57 35 0
a R 3/5 ) (F3mMK+ - F5mMK+)/F5mMK+ × 100, representing the slope of the calibration curve at 5 mM K+. R0/100 ) (F0mMK+ - F100mMK+)/ F0mMK+ × 100, representing the total signal change. F0mMK+, F3mMK+, F5mMK+, and F100mMK+ are fluorescence intensities at concentrations of 0, 3, 5, and 100, mM potassium ion, respectively.
Without any plasticizer, no response to potassium was observed. A response to potassium is also expected without plasticizer if the sensing mechanism is mainly governed by aggregation/deagregation processes or electrochromic processes. The lipophilic, water-insoluble valinomycin is stabilized by surfactant-like dyes in hydrogels. The dye forms a micelle including valinomycin. Potassium can be extracted into this micelle and the micelle gets positively charged.25 An electrochromic dye changes its fluorescence properties during such a process. Solvatochromic effects or effects on quantum yield in dependence of solvent polarity are minor because the charged fluorophore is mainly located in the hydrophilic hydrogel phase in the absence as well as in the presence of potassium. Therefore, the observed lack of a decrease in fluorescence in the presence of potassium without plasticizer droplets is a hint for a mainly solvatochromic and quantum yield governed response mechanism in this approach. Increasing the fraction of valinomycin B (B ) mvalinomycin/ mCPDDE; M1, M4-M6), and therefore increasing the quantity of sites for adsorption of potassium ions, leads to a shift in the dynamic range to lower potassium concentrations, larger slope at 5 mM potassium (R3/5 ) (F3mMK+ - F5mMK+)/F5mMK+) and to higher signal changes (R0/100 ) (F0mMK+ - F100mMK+)/F0mMK+). Valinomycin weight fractions (with respect to plasticizer) of more then 0.2 do not result in a further shift of dynamic ranges because of a saturation effect. Calibration curves of membranes containing different concentrations of valinomycin are given in Figure 3. Variations of the dye fraction A (A ) mDye/mCPDDE) at constant dye/borate ratio (M1, M7, M8) caused no change in the response characteristic as long as the dye/plasticizer weight fraction did not exceed 2 × 10-3. Higher dye concentrations lead to smaller relative signal changes. This finding can be attributed to the higher background fluorescence of the dye located in the bulk of the droplet and to self-quenching of the dye located at the surface of the droplets. The ratio between PTCPB and DIA, D ) nDIA/nPTCPB, was varied (M1, M9-M11) and found to be the most effective way to (25) Charlton, S. C.; Makowski, E.; Michaels, A. A. Miles Sci. J. 1997, 12-17.
Figure 4. Cross-sensitivity of M1 to pH at constant ionic strength of 131 mM in the absence and presence (5 mM) of potassium.
Figure 3. Calibration curves of membranes with different valinomycin content in plasticizer [M1, 20% (w/w); M4, 5% (w/w); M5, 2% (w/w); M6, 40% (w/w)] at ionic strength of 131 mM and pH 7.4.
alter sensor characteristics. Anionic sites in the sensing phase retain the dye in the interior of the lipophilic droplets and facilitate diffusion of potassium ion into the interior. If PTCPB is used in excess, the dynamic range is shifted to higher concentrations. This is because increasing quantities of anionic sites in the droplets hold back the dye in the bulk of the lipophilic droplet and more potassium is needed for displacement of the dye. Membranes possessing an excess of DIA are more suitable for measuring potassium in the clinical range; certainly equimolar quantities of dye and PTCPB are preferable, because otherwise ion-exchange processes must be considered as a source of crosssensitivity. In the complete absence of anionic sites, signal changes become smaller and conditioning times of more than 2 h in the flow-through cell are required to wash out the iodide originating from DIA. The possibility to adjust the dynamic range as well as the cross-sensitivity by altering the membrane composition is a distinct advantage over bulk-based optical sensors. Cross-Sensitivity to pH. In the presence of potassium ions, a small decrease of fluorescence with increasing pH above 7.3 was observed. This decrease was fully reversible. If the dye was dissolved in hydrogel without being dissolved in plasticizer droplets, no effect of pH was observed. Therefore, this response to pH does not originate from an interaction of the dye with protons or hydroxide ions. The response mechanism of PSD-based optodes is virtually independent of pH. Anyway, some PSD systems described in the literature show a slight cross-sensitivity to pH.26 In the emulsion system, this pH independence is even more relevant. Changes occur in the partition coefficient of DIA in the hydrogel/plasticizer system due to alteration of the hydrogel corresponding to pH. In the absence of potassium ions, no effect of changing the pH within the physiological range could be found. In the absence of potassium, the dye is located mainly in the interior of the lipophilic droplet and therefore changes in the hydrogel do not affect partitioning of the dye. A typical response to pH changes in the absence and presence of potassium ions is shown in Figure 4. Cross-Sensitivity to Ionic Strength. Variation of ionic strength affects most optical sensors. One reason is the different optical properties, such as refractive index (and therefore dielectric (26) Wolfbeis, O. S.; Kovacs, B.; He, H. Proc. SPIE 1994, 2085-2093.
Figure 5. Cross-sensitivity of M1 to ionic strength at constant pH of 7.4 in the absence and presence (5 mM) of potassium ions.
Figure 6. Cross-sensitivity of M1 to lipophilic anions at constant ionic strength of 131 mM and pH of 7.4 in the absence and presence (5 mM) of potassium ions.
constant) and optical pathway, of the matrix materials, another is a different analyte activity, the third is the cross-sensitivity to the “inert” salt. Furthermore, the analyte concentration near a charged surface depends on ionic strength. Our PSD membranes show a surprisingly low cross-sensitivity to ionic strength. Neither in the presence nor in the absence of potassium ions was a measurable signal change by variation of the ionic strength from 20 to 200 mM observed. The lack of signal change due to variations in ionic strength is shown in Figure 5. Cross-Sensitivity to Lipophilic Anions. It is known that lipophilic anions cause the dye to move in the interior of the lipophilic droplets and therefore increase fluorescence intensity.14 Surprisingly, lipophilic anions dramatically lowered the fluorescence in the presence of potassium, while in the absence of potassium no effect was observed. Therefore, fluorescence intensity decrease is not due to quenching by the anion. Results are summarized in Figure 6 and Table 3. Analytical Chemistry, Vol. 71, No. 23, December 1, 1999
5307
Table 3. Cross-Sensitivity to Lipophilic Anions of Membranes with Different PTCPB/DIA Ratiosa (F0mMI- - F10mMI-)/ F0mMI-
(F0mMSali- - F10mMSali-)/ F0mMSali-
membrane
at 0 mM K+
at 5 mM K+
at 0 mM K+
at 5 mM K+
M1 M9 M10
0.0 0.0 0.0
0.05 0.03 0.03
0.0 0.0 0.0
0.2 0.08 0.5
a CS ) (F I 0mMI- - F10mMI-)/F0mMI- represents the signal change on going from 0 to 10 mM iodide. F0mMI- are F10mMI- are the fluorescence intensities at 0 and 10 mM iodide, respectively. CSS ) (F0mMSali- F10mMSali-)/F0mMSali- represents the signal change on going from 0 to 10 mM iodide. F0mMSali- and F10mMSali- are the fluorescence intensities at 0 and 10 mM salicylate, respectively.
Iodide and salicylate showed very different cross sensitivity (CSI ) (F0mMI- - F10mMI-)/F0mMI- and CSS ) (F0mMSal- - F10mMSal-) /F0mMSal-) although their lipophilicity is not very different. If salicylate adsorbs at some preferred sites in HN80,8 the resulting negative charge outside but near the surface of the lipophilic droplets moves out DIA molecules of the plasticizer and therefore lowers fluorescence. The cross-sensitivity to iodide or salicylate is higher if the dye is in excess relative to borate. If borate is
5308 Analytical Chemistry, Vol. 71, No. 23, December 1, 1999
used in excess, the cross-sensitivity becomes lower but is still not negligible. One can assume, that some of the borate is not located in the plasticizer droplets but occupies the adsorption sites in HN80. Anionic sites in the hydrogel reject anions from the sample. Emulsion-based PSD membranes proved to be superior to PVC- or Langmuir-Blodgett-based sensors in respect to signal change, dynamic range, and mechanical stability. PSD emulsion systems offer an alternative to the use of ion-exchange or coextraction-based systems. Especially with respect to pH crosssensitivity, they are superior to ion-exchange systems. Up to now, the distinct cross-sensitivity toward lipohilic anions limits their use for potassium ion detection in biological fluids. ACKNOWLEDGMENT The authors thank Ms. Hannelore Brunner for the perfect technical assistance.
Received for review July 7, 1999. Accepted September 21, 1999. AC9907383
