Anal. Chem. 1999, 71, 3673-3676
A Chronoamperometric Method To Estimate Ionophore Loss from Ion-Selective Electrode Membranes Bradford D. Pendley*
Department of Chemistry, Rhodes College, 2000 North Parkway, Memphis, Tennessee 38112 Erno Lindner
Department of Biomedical Engineering, University of Memphis, Memphis, Tennessee 38152
A novel chronoamperometric method was developed to estimate the concentration of a neutral ionophore in fixedsite, dioctyl sebacate plasticized, poly(vinyl chloride)based, ion-selective electrode membranes. The membranes contained between 0.5 and 16 mmol/kg valinomycin. The chronoamperometric technique was used to estimate the valinomycin concentration in freshly prepared membranes and after extraction of some of the ionophore from the membranes using dioctyl sebacate. Replicate measurements indicated a relative standard deviation in the calculated valinomycin concentration of less than 10%, and these values accurately represented the true concentration of valinomycin within 10%. The method permitted an estimate of the valinomycin concentration after valinomycin was leached from a membrane. The results of preliminary experiments using heparinized dog blood suggest that blood protein adsorption does not interfere qualitatively or quantitatively with the analysis. Solvent polymeric membranes containing neutral ionophores such as valinomycin have found widespread use in analytical chemistry, particularly in clinical applications.1-3 Although potentiometric sensors are applied routinely for diagnostic purposes, they have only recently begun to be used for acute and chronic in vivo measurements.4,5 A critical issue associated with the miniaturization as well as the chronic implantation of ion-selective sensors is their lifetime. In addition to catastrophic failure of the sensor, the loss of analytical performance with time is a concern. Dinten et al.6 defined the lifetime of a potentiometric sensor as “the time interval between the conditioning of the membrane and the moment when at least one parameter of the functionality characteristics of the device changes detrimentally”. For neutral-carrier-based membrane sensors, Oesch and Simon7 showed that the leaching of the ionophore from the membrane results in a loss of selectivity, thereby limiting the lifetime. The loss of selectivity is related to the change in the optimal ionophore to site (fixed or mobile) (1) Bakker, E.; Buhlmann, P.; Pretsch, E. Chem. Rev. 1997, 97, 3083-3132. (2) Buhlmann, P.; Pretsch, E.; Bakker, E. Chem. Rev. 1998, 98, 1593-1687. (3) Oesch, U.; Ammann, D.; Simon, W. Clin. Chem. 1980, 32, 1448-1459. (4) Cosofret, V. V.; Lindner, E.; Johnson, T. A.; Neuman, M. R. Talanta 1994, 41, 931-938. (5) Lindner, E.; Cosofret, V. V.; Ufer, S.; Johnson, T. A.; Ash, R. B.; Nagle, H. T.; Neuman, M. R.; Buck, R. P. Fresenius’ J. Anal. Chem. 1993, 346, 584588. (6) Dinten, O.; Spichiger, U. E.; Chaniotakis, N.; Gehrig, P.; Rusterholz, B.; Morf, W. E.; Simon, W. Anal. Chem. 1991, 63, 596-603. 10.1021/ac990137b CCC: $18.00 Published on Web 07/20/1999
© 1999 American Chemical Society
ratio,8-10 but it is also accompanied by increased membrane resistance, increased detection limit, more noise, etc. To estimate the lifetime of the ionophore-based solvent polymeric (potentiometric or optical) membranes experimentally, the ionophore concentration in the membrane or the dissolved amount in the bathing solution should be determined as a function of time and bathing solution composition. The determination of the amount of ionophore in an ion-selective membrane has been a difficult analytical problem to solve. Dinten et al.6 followed the decrease in absorbance of plasticized poly(vinyl chloride) membranes containing a chromoionophore upon exposure to diluted serum samples. This is the only report to our knowledge of the direct determination of ionophore loss from an ion-selective membrane. Since many potentiometric sensors do not use chromoionophores, high-sensitivity absorbance measurements in the visible spectral range are limited in their application. The indirect determination of ionophore loss7 into complex biological matrixes such as whole blood, blood serum, or plasma is also very complex due to the extremely low concentrations and the high background. However, none of the above methods are appropriate for estimating the residual “use lifetime” of a microsensor built into a flowthrough analyzer cartridge or a chronically implanted device. We are interested in developing a relatively rapid means to estimate sensor lifetime in situ that can be used for all ionophores. Earlier work by Iglehart et al.11 reported that the chronoamperometric current transients of fixed-site, valinomycin-loaded membranes showed an initial nearly constant current region followed by an abrupt decrease in the current. This break in the current occurred at a characteristic time, τ, and was indicative of the total depletion of ionophore at one interface of the membrane. This break time, τ, was then modeled to follow the equation
τ1/2 )
FACcarrierR(Dcarrierπ)1/2 2V
(1)
where τ is the break time in the current transient, A is the (7) Oesch, U.; Simon, W. Anal. Chem. 1980, 52, 692-700. (8) Meier, P. C.; Morf, W. E.; La¨ubli, M.; Simon, W. Anal. Chim. Acta 1984, 156, 1-8. (9) Bakker, E.; Xu, A.; Pretsch, E. Anal. Chim. Acta 1994, 295, 253-262. (10) Eugster, R.; Gehrig, P. M.; Morf, W. E.; Spichiger, U. E.; Simon, W. Anal. Chem. 1991, 63, 2285-2289. (11) Iglehart, M. L.; Buck, R. P.; Horvai, G.; Pungor, E. Anal. Chem. 1988, 60, 1018-1022.
Analytical Chemistry, Vol. 71, No. 17, September 1, 1999 3673
membrane area, F is the Faraday constant, Ccarrier is the concentration of ionophore in the membrane, R is the ohmic resistance of the membrane, Dcarrier is the diffusion coefficient of the ionophore in the membrane, and V is the applied voltage. It was pointed out by Lindner et al.12 that a calibration plot 1/2 (τ versus Ccarrier/I, where I ) V/R) might be constructed for membranes with different concentrations of carrier. We hypothesized that such a calibration plot can be used to estimate the concentration of the carrier in the membrane after ionophore has been lost due to exposure to aqueous electrolyte solution or blood samples. This study seeks to test this hypothesis. EXPERIMENTAL SECTION Reagents. Valinomycin, high molecular weight poly(vinyl chloride) (PVC), and bis(2-ethylhexyl) sebacate (dioctyl sebacate, DOS) were purchased from Fluka. Tetrahydrofuran (ACS reagent grade) was purchased from Aldrich. Cyclohexanone (Certified) and potassium chloride (ACS Certified) were purchased from Fisher. All other chemicals were of at least reagent grade and were used as received. Water was purified using a Milli-Q Gradient A10 system (Millipore Corp.). Apparatus. An EG&G Princeton Applied Research model 283 potentiostat/galvanostat interfaced to a Dell Optiplex GX-1 computer was used to perform the chronoamperometric studies. EG&G model 250 research electrochemistry software was used to control the potentiostat. The PAR 283 potentiostat was used to control the potential across the cell. When the applied voltage exceeded (4 V, a locally built voltage divider reduced the voltage sensed by the reference electrode circuit of the potentiostat. The voltage divider was built using precision film resistors (1 and 10 MΩ ( 0.1% resistors, Caddock Electronics). The system (potentiostat with the voltage divider) was tested, and the actual voltage applied between the reference/counter and working/sense leads was measured using a Keithley 175 autoranging multimeter over a range of potentials from (1 to (33 V. These values agreed with the calculated values within 0.1%. For applied voltages greater than (4 V, the measured current was corrected for the supply current passing through the voltage divider. Electrodes and Cells. A locally built cell was used to house the membrane for all chronoamperometric experiments. The cell consisted of two electrolyte compartments separated by the membrane. The cell was machined from Teflon and consisted of two separate pieces, each containing a single compartment approximately 2 cm × 2 cm × 3 cm. Two 1-cm-diameter holes were located in opposite ends of the two compartments to accommodate two silver electrodes (approximate geometric area was 1.5 cm2 with a thickness of 1 mm). The back ends of the silver electrodes were soldered to a copper lead that screwed into the back of the Teflon compartment. The front ends of the silver electrodes faced the inside of the compartment and were sealed with a Teflon O-ring. The silver electrodes were electrolytically coated with a AgCl layer. The membrane was placed between the two compartments that were brought together and secured to each other by means of six screws. The screws were tightened to ensure no leakage between compartments. When the cell was assembled, the overall dimensions of the cell body were ap(12) Lindner, E.; Cosofret, V. V.; Nahir, T. M.; Buck, R. P. In Diagnostic Biosensor Polymers; Usmani, A. M., Akmal, N., Eds.; ACS Symposium Series 556; American Chemical Society: Washington, DC, 1994; pp 149-157.
3674 Analytical Chemistry, Vol. 71, No. 17, September 1, 1999
proximately 5.3 cm × 3.5 cm × 3.5 cm. Membranes. Membranes comprising approximately 33 wt % poly(vinyl chloride), 65-67 wt % dioctyl sebacate, and 0.05-2 wt % (0.5-16 mmol/kg) valinomycin were prepared in the following manner. A mixture of tetrahydrofuran and cyclohexanone (1:1, v/v) and poly(vinyl chloride)/dioctyl sebacate was prepared. A portion of this mixture was mixed vigorously with an appropriate amount of freshly prepared valinomycin solution (in 1:1 tetrahydrofuran/cyclohexanone) to yield a clear, colorless mixture. A thick-walled glass or Teflon ring (diameter approximately 3 cm) was secured to a glass or Teflon plate by means of a rubber band. The membrane cocktail was transferred to the inside of the ring, and the solvent was evaporated slowly overnight at 60 °C. A 1.75cm-diameter ring of polymer membrane was cut and weighed. The thickness of the membrane was estimated from its weight, the density of the poly(vinyl chloride)/dioctyl sebacate mixture, and the dimensions of the cut membrane. Membrane thicknesses were approximately 100 µm. Procedures. A membrane was placed inside the transport cell, and approximately 9 mL of potassium chloride solution of the same concentration (usually 1 mM) was placed in each compartment. The membrane was soaked in the potassium chloride solution for at least 1 h prior to commencing the experiment. A stir bar was placed into each compartment, and the solutions in both compartments were stirred throughout the experiment. The counter/reference electrode lead was connected to one Ag/AgCl electrode while the other Ag/AgCl electrode was connected to the working/sense lead. Voltage steps between -2 and -15 V were applied, and the resulting current was measured. For replicate measurements, membranes were allowed to stand at open circuit for a minimum of 1 h. Extraction Experiments. For experiments involving the extraction of valinomycin from the ion-selective membrane, the membrane was placed in a glass vial. A portion of either dioctyl sebacate or heparinized dog blood was pipetted into the vial and shaken gently for a few minutes. The membrane was soaked in the dioctyl sebacate or blood from several hours to 2 days. For multiple extraction experiments, the membrane was removed from the vial, blotted with a KimWipe (dioctyl sebacate extraction) or rinsed (phosphate buffer saline solution, 1 mM KCl, and deionized water in the case of the blood), and placed in a vial with a fresh portion of either dioctyl sebacate or heparinzed blood. RESULTS AND DISCUSSION Figure 1 shows a current-time transient obtained from a membrane containing 4.6 mmol/kg valinomycin after application of a -15 V potential step. As observed by Iglehart et al.11 and Nahir and Buck,13 the current rises rapidly to a value dependent upon the applied voltage and the ohmic resistance of the membrane. This initial, ohmic current is relatively stable until the carrier has been depleted at one interface, at which time the current decays in a Cottrell fashion and then eventually approaches a limiting value.11,13 The τ value is determined from the intersection of the extrapolated ohmic current and the extrapolated best line through the current just after the break region (see insert of Figure 1). (13) Nahir, T. M.; Buck, R. P. J. Electroanal. Chem. Interfacial Electrochem. 1992, 341, 1-14.
Figure 1. Current-time transient of a membrane containing 4.6 mmol/kg valinomycin. Vapplied ) -15 V; bathing solution 1 mM KCl. Insert: Expanded region near break illustrating determination of the break time, τ.
The selection of the voltage used in the experiment is an important consideration. The break is most pronounced when the initial ohmic current is much larger than the final limiting current. Since the initial ohmic current is proportional to the applied voltage but the final limiting current is not related to the applied voltage,11 a larger applied voltage makes the break more pronounced. However, if too large of a voltage is applied, Donnan exclusion failure can occur, leading to a breakdown of membrane permselectivity.11 Similarly, membrane thickness also plays a role in how pronounced the break is. The impedance of the membrane is proportional to its thickness, meaning that the initial ohmic current is inversely proportional to the membrane’s thickness. Likewise, the final limiting current is inversely proportional to the membrane’s thickness. It would seem that, since both currents are inversely proportional to membrane thickness, membrane thickness would not play a role. However, as discussed previously, there is an upper limit on the voltage that can be applied without Donnan exclusion failure, and this limits the thickness of the membrane. On the basis of our observations, membranes with thicknesses greater than 200 µm lead to significantly less pronounced breaks at applied voltages under the Donnan exclusion failure breakdown voltage. To test our hypothesis that a calibration plot could be constructed and used to calculate the concentration of valinomycin in the membrane, several membranes of known concentrations of valinomycin (0.5-16 mmol/kg) were prepared and the current
Figure 2. (A) Current-time transient of a membrane containing 9.1 mmol/kg valinomycin. (B) Current-time transient of the same membrane after extracting about 63% of the valinomycin with dioctyl sebacate. Vapplied ) -15 V; bathing solution 1 mM KCl. Insert: Calibration plot for valinomycin membranes.
transients were measured and used to construct a calibration plot. The results are shown in the insert of Figure 2. The resulting line through the origin gave a reasonable correlation between τ1/2 and Ccarrier/I (r ) 0.98) and a slope of 9.42. From this slope, a diffusion coefficient of (1.8 ( 0.2) × 10-8 cm2/s (95% confidence limit) was calculated, in close agreement with a value of 1.7 × 10-8 cm2/s measured by Iglehart et al.11 Replicate transients of the same membrane showed that the calculated concentration of valinomycin in replicate trials agreed with each other well (relative standard deviations were less than 10% in all cases) and were in reasonable agreement with the actual concentration of valinomycin (average error 10%) in the membrane. The concentration of KCl in the bathing solution has no effect on the calculated valinomycin concentration for KCl concentrations between 10-4 and 10-2 M. This result is expected since the flux of K+ depends on the concentration profile of potassiumvalinomycin complex (Kval+) within the membrane. This profile is independent of the bathing concentration of K+. To test our hypothesis that chronoamperometry could be used to estimate loss of ionophore from the membrane, valinomycin was extracted from a membrane containing 9.1 mmol/kg (1 wt %) valinomycin. Dioctyl sebacate, the membrane plasticizer, was used for the extraction. Prior to the extraction, the variation in the calculated concentration of valinomycin was 9.7% (N ) 4; current transients obtained over 6 days). A typical current transient Analytical Chemistry, Vol. 71, No. 17, September 1, 1999
3675
of the membrane before extracting valinomycin is shown in Figure 2A. A volume of dioctyl sebacate equal to the volume of dioctyl sebacate in the membrane was placed in contact with the membrane for 18 h. After equilibration, the membrane was removed from the dioctyl sebacate, blotted dry, and soaked in 1 mM KCl, and a transient was obtained (see Figure 2B). As is evident in Figure 2, the break time decreased dramatically and, using this new break time, the initial ohmic current, and the calibration plot (insert of Figure 2), the concentration of valinomycin calculated was 37% of the original concentration. This new concentration is in reasonable agreement with that calculated (50%) by assuming valinomycin has a distribution coefficient of 1.0 between the dioctyl sebacate and the membrane. In another experiment, valinomycin was extracted from a membrane containing 11.5 mmol/kg valinomycin. Three successive extractions were conducted using 0.5 mL portions of dioctyl sebacate each time. The resulting current transient showed Cottrell behavior immediately following application of a -15 V potential step, indicating that practically no valinomycin remained in the membrane. This agreed with our estimate that 99.9% of the valinomycin was extracted (again assuming valinomycin has a distribution coefficient of 1.0 between dioctyl sebacate and the membrane). An initial assessment of whether this method could be used to estimate the loss of ionophore after the membrane had been exposed to blood was conducted. A membrane containing 4.8 mmol/kg valinomycin was exposed to 1 mL of heparinized dog blood for 5 min. This time was long enough that significant protein adsorption would occur14,15 but short enough that no significant loss of valinomycin was expected.6 The resulting current transient in 1 mM KCl (see Figure 3) looked nearly identical to the current transients obtained prior to exposure to dog blood, and the calculated valinomycin concentration agreed with that calculated from the transients obtained prior to exposing the membrane to blood. After four extractions over a period of nearly 1 week and using 1 mL portions of blood, the resulting shape of the current transients in 1 mM KCl also looked nearly identical to those obtained prior to exposing the membrane to blood. However, the calculated mean valinomycin concentration decreased 30%. Furthermore, current transients obtained in whole, heparinized dog blood showed qualitatively similar behavior to those obtained in KCl solution, suggesting that it might be possible to use this chronoamperometric method in vivo. CONCLUSIONS We have presented evidence that supports our hypothesis that chronoamperometry can be used to estimate the concentration of a neutral ionophore in fixed-site, dioctyl sebacate plasticized, poly(vinyl chloride)-based, ion-selective electrode membranes. Experimental results indicate that the method can be used to quantitatively follow the loss of ionophore from the membrane. The results of preliminary experiments suggest that blood protein adsorption does not interfere qualitatively or quantitatively with the analysis. Therefore, it should be possible to use this method to estimate ionophore after the membrane has been exposed to (14) Horbett, T. A.; Cheng, C. M.; Ratner, B. D.; Hanson, S. R.; Hoffman, A. S. J. Biomed. Mater. Res. 1986, 20, 739-772. (15) Horbett, T. A. Cardiovasc. Pathol., 2nd Suppl. 1993, 137S-148S. (16) Nahir, T. M.; Buck, R. P. J. Phys. Chem. 1993, 97, 12363-12372.
3676 Analytical Chemistry, Vol. 71, No. 17, September 1, 1999
Figure 3. Current-time transient of a membrane containing 4.8 mmol/kg valinomycin after a 5 min exposure to heparinized dog blood. Vapplied ) -15V; bathing solution 1 mM KCl.
blood and perhaps to use the method to estimate ionophore loss in vivo. Work by Nahir and Buck16 has demonstrated that the current transient of a membrane containing an ionophore and a lipophilic additive (e.g., tetraphenylborate compounds) shows a similar break in the current transient. We hypothesize that this chronoamperometric technique can be extended to membranes containing mobile sites, and work is currently underway in our laboratory to test this hypothesis. ACKNOWLEDGMENT The authors appreciate the efforts of Dr. Robert A. Malkin, Department of Biomedical Engineering at the University of Memphis, who provided us with the dog blood used in this work, and Mr. Barry Wymore, Herff College of Engineering at the University of Memphis, for assistance with the design and construction of the transport cell used. B.D.P. is grateful to Dr. Michael R. Neuman, Department of Biomedical Engineering at the University of Memphis, for his financial support and for allowing him to work in his laboratory while on sabbatical leave and to Rhodes College for the financial support of this sabbatical leave. This work was partly supported by the U.S.-Hungarian Joint Fund (Grant JF568).
Received for review February 10, 1999. Accepted May 27, 1999. AC990137B
