Exchange kinetics at ion-selective membrane electrodes

(6) F. Aurich and E. Lippert, Spectrochim. Acta, 22, 1073 (1966). (7) G. Weber and B. Bablouzian, J. Biol. Chem., 241, 2558 (1966). (8) B. Witholt and...
1 downloads 0 Views 806KB Size
P = polarization, defined by p = -AV - Ah Av Ah

+

LITERATURE CITED (1) W. 8. Dandliker and V. A. deSaussure, Immunochemistry, 7, 799 (1970). (2) W. B. Dandiiker. R. J. Kelly, J. Dandliker, J. Farquhar, and J. Levin, Immunochemistry, I O , 219 (1973). (3) W. B. Dandliker, H. C. Schapiro, R. Alonso, and D. E. Williamson, San Diego Symp. Biomed. Eng., 3, 127 (1963). (4) L. Monnerie and J. Neel, J. Chim. fhys., 62, 504 (1965). (5) D.Deranleau, Anal. Biochem., 16, 438 (1966). (6) F. Aurich and E. Lippert, Spectrochim. Acta, 22, 1073 (1966). (7) G. Weber and B. Bablouzian, J. Bo/.Chem., 241, 2558 (1966). (8) 8. Witholt and L. Brand, Rev. Sci. Instrum., 39, 1271 (1968). (9) R. H. McKay, Arch. Biochem. Biophys.. 135, 218 (1969). (10) C. Rosen, Acta Chem. Scad., 24, 1849 (1970).

(1 1) S. Claesson and H. Odani, Discuss. Faraday Soc., 49, 268 (1970). (12) J. Lavorel, C. Vernotte, B. Arrio. and F. Rodier, Biochimie, 54, 161 (1972). (13) R. D. Spencer, F. B. Toledo, B. T. Williams and N. L. Yoss, Clin. Chem. ( Winston-Salem, N.C.), 19, 838 (1973). (14) R. E. Curry, H. L. Pardue, G. E. Mieling, and R. E. Santini, Clin. Chem. ( Winston-Salem, N.C.), 19, 1259 (1973). (15) J. E. Wampler and R. J. De Sa, Anal. Chem., 46, 563 (1974). (16) W. B. Dandliker, H. C. Schapiro, J. W. Meduski, R. Alonso, G. A. Feigen, and J. R. Hamrick, Immunochemistry, 1, 165 (1964). (17) P. Pringsheim, "Fluorescence and Phosphorescence," Interscience, New York, 1949, Table 69.

RECEIVEDfor review August 21, 1975. Accepted February 9, 1976. Supported by Cordis Corporation, Miami, Fla., Research Grant No. GB-31611 from the National Science Foundation, and Contract No. N01-CB-43905 from the National Cancer Institute.

Exchange Kinetics at Ion-Selective Membrane Electrodes Karl Cammann' and G. A. Rechnitz* Department of Chemistry, State University of N e w York, Buffalo, N e w York 74274

Mechanlstlc information concerning the operation and selectivity of several ion-selective membrane electrodes Is obtained from exchange current measurements carried out under varying solution conditions. Particularly high exchange current densities were found for AgZS-based membrane electrodes. The potentiometric behavior of LaF3 type F--selective electrodes in the presence of OH- or La3+ can be successfully explained on the basis of the exchange current density estimates obtained in this study.

This study attempts to examine the kinetics of chargetransfer processes occurring a t phase boundaries of ion-selective membrane electrodes in order to test whether or not thermodynamic equilibrium is established a t these boundaries. Traditionally, the heterogeneous kinetics of charge-transfer reactions are described electrochemically in terms of the standard exchange current density, ioo ( I ) . For metal electrodes, the determination of exchange current densities from current-voltage curves is not difficult, but the determination and evaluation of quasi-stationary current-voltage curves for the phases of ion-selective membrane electrodes involves severe experimental and interpretive difficulties. In the case of metallic electrodes, both the chemical composition and electrostatic potential remain constant in the metal phase and the potential difference between the electrode phase and the solution arises mainly between the metal phase and the outer Helmholtz plane. This is not so in the case of ion-selective membrane electrodes which have high ohmic resistances. Since the precise details of the chemical and electrostatic potential distribution across ion-selective membrane electrodes are not known, an exact theoretical treatment of electrode kinetics is not attempted here; instead, the concepts and methods of electrode kinetics developed for metal electrodes are employed as a first approximation and are used to estimate exchange current densities which should be viewed strictly as apparent exchange current Visiting Researcher from the U n i v e r s i t y of M u n i c h , Germany.

856

ANALYTICAL CHEMISTRY, VOL. 48, NO. 6, MAY 1976

densities, i&. With this qualification in mind, the methods employed in this study are based on the Butler-Volmer equation, which describes the stationary current-voltage curve, e.g., for one redox reaction (with 2 = 0) a t a metal electrode ( 2 , 3 )

where i is the current density a t overpotential 7, io is the exchange current density a t 7 = 0, and p is the transfer coefficient. Since the absolute potential difference between two phases is not measurable, one should compare individual electrode reactions according to their kinetics a t 7 = 0. For ion-selective membrane electrodes the exchange current density also is more characteristic of electrode kinetics a t the equilibrium potential than the absolute heterogeneous rate constant a t zero field strength. Jaenicke and Haase ( 4 ) have already employed principles of metal-electrode kinetics for the dissolution of '!salt electrodes" consisting of thin (-10 bm) silver halide layers on silver metal and developed individual expressions for the exchange current densities of cations and anions, respectively. In case of ion electrodes, of course, the exchange current density of the counterion can usually be neglected by proper selection of the electrolyte. There are basically two methods for determination of the exchange current density from equation 1. A t low overpotentials ((10 mV) equation 1 reduces t o

.

F =io-? RT and the transfer coefficient drops out. Assuming the validity of the simplified equivalent circuit shown in Figure 1,we can define a charge-transfer resistance Rt, for the behavior of a phase boundary under current flow, as 1

(3) with knowledge of Rt we can then calculate io from io = RT/FRt

(4)

Alternately, io can be determined a t high overpotentials (>lo0 mV) where equation 1 reduces to tl=-

RT

(1 - P)FIn io - In i

-2iR T t

RTC -ti R T [ Z3F3(Co0)2D

5

m e mbrone

(5)

and io is obtained from the well-known Tafel plot of log i vs. 7. In order for this method to be valid, the electrode reaction must be exclusively kinetically controlled. In order to minimize interpretive complications, we also employed in this study the galvanostatic step method ( 5 , 6 ) with consideration of diffusion polarization. In this method, the voltage-time relationship, after a short time delay, is described by

‘= ,1/2Z2F2CoOD1/2

solution 1

’1

- io

(6)

where t is the time, Coo the bulk concentration, C the double layer capacitance, D the diffusion coefficient of the ion involved, and Z the charge on the ion. If the general assumptions are valid, a plot of the voltage arising during the galvanostatic measurements vs. t1l2should be linear and the diffusion coefficient can be evaluated from the slope. If the double layer capacitance is known, io can be evaluated from extrapolation to t = 0. In practice, then, this technique is equivalent to a resistance measurement with correction for diffusion. The resistance obtained by this extrapolation is, of course, the sum of all the resistances in the cell; for the present purpose, only the charge-transfer resistance at the outer solution-membrane phase boundary is of interest. In this connection, the simplified equivalent circuit shown in Figure 1 is useful even though no circuit may give a complete description of the total electrochemical behavior ( 7 ) . The circuit shown would be appropriate (8) for thick membranes with high ohmic resistances. For high rate constants the quantity of interest, Rt, would be quite small compared to the high ohmic resistance in series. However, since the exchange current density is concentration dependent, Rt can be increased by reducing the concentration of the potential determining ion; extrapolation of the log concentration vs. log R t plot to 1 M then yields the apparent standard exchange current density for this ion, provided all other resistances are held constant. The measured changes in the initial resistance can then be correlated to the desired charge-transfer resistance and related to the overvoltage via equation 3. Because R t changes over two orders of magnitude, the larger resistance changes will be in effect equal to Rt within the overall accuracy of the method.

EXPERIMENTAL Both laboratory-constructed and commercially available electrodes were employed in this study. Figure 2 shows schematic diagrams of the electrodes constructed in the laboratory, e.g., a solid state Ag2S membrane electrode fitted with an internal silver metal connection (A) and mounted in the Beckman 188551 rotating electrode assembly, and a 70911AgzS/30% PbS membrane electrode (B) with 0.1 M AgN03 internal solution. Corning 476137 LaF3 crystal membrane electrodes were also employed. AgzS for the solid state electrodes was prepared by bubbling H2S into a lo-’ M AgN03 solution until no further precipitate was formed. After filtration, the precipitate was washed three times with dilute “ 0 3 and ten times with hot distilled water. The dried Ag2S was treated with CSZ to remove any excess elemental sulfur. The black powder was then washed with ethanol, dried a t 110 OC, and formed into membranes of -3 mm thickness using a pellet press under applied pressures of 1000 Kpa/cm2 for 10 min and 8000 Kpa/cm2 for 2 h under vacuum. In the case of the AgZS/PbS electrode the membrane material was coprecipitated from AgN03Pb(N03)z solution. Membrane surfaces were fine polished with di-

Figure 1. Simplified equivalent circuit

for membrane electrodes

= solution resistances (conductivity); RM = membrane resistance; Rtj.t2= charge-transfer resistances at membranelsolution interfaces; Cdl,,dl2 = double-layer capacitances at rnembranelsolution interfaces Ral,sp

A92S area :0.28crn2

A

‘epoxy

resin



A g 2 S )PbS a r o o : 6 . 4 ern2

B

Figure 2. Schematics of membrane electrode construction (areas are geometric)

amond dust; surface areas were calculated from the geometrical dimensions. A silver metal counter electrode of large surface area was used for the potentiostatic measurements with a Princeton Applied Research Model 173 potentiostat/galvanostat. The potential in the vicinity of the membrane electrode surface was measured with a Luggin capillary and Ag/AgCl reference electrode in the high-impedance loop of the instrument. It should be noted that this arrangement only controls the overall potential difference between the working electrode and the tip of the Luggin capillary. Thus, corrections for the iR drop inside the membrane phase and for diffusion polarization are necessary (9) before equations 4-6 can be applied. EMF, current, and galvanostatic transition or resistance measurements were made using either a Princeton Applied Research Model 134 or Keithley Model 610 B electrometer with built in constant current source. Reference electrodes were chosen to have low resistance and junction potential; when necessary, corrections were applied. All measurements were made in shielded vessels a t room temperature. Unless otherwise specified reagent grade materials were used. Current-voltage curves were recorded using the techniques of pulse polarography. A Princeton Applied Research Model 174 pulse polarograph was used to control the potentiostat to yield voltage pulses of 57 ms duration with 5-s intervals a t the equilibrium voltage (7 = 0). To minimize charging effects, the last 17 ms of the voltage pulse were used for the actual measurement. Oscillographic measurements were performed with a Solatron CX 1448 dual channel oscilloscope connected to the recorder output of the potentiostat or electrometer. The current voltage curves were plotted with a Hewlett-Packard X-Y recorder which was also used in the X-t mode for voltage or current measurements of longer duration. Required voltage ramps and square waves were produced using a Hewlett-Packard function generator. ANALYTICAL CHEMISTRY, VOL. 48, NO. 6, M A Y 1976

857

~

Table I. Conductance Behavior of AgZS/Ag Electrode at 25 "Cas a Function of Current Density Current density (at 2 kHz)f FA/cm* 3 6 15 30 150

Total

Ionic

Electronic

Ionic transference no. ti

7.2 7.8 7.6 7.8 8.0

5.2 5.2 5.6 5.6 5.4

2.0 2.6 2.2 2.2 2.6

0.72 0.67 0.74 0.72 0.68

Conductance X 10-4, Q-' cm-l

20

40

60 t

Table 11. Conductance Behavior of AgzS/Ag Electrode at 25 "Cas a Function of Frequency Current density ( f 1 2 pA/cm*), Hz

Conductance X

Q-l

Total

Ionic

Electronic

Ionic transference n. ti

10 20 50 150 250 500

6.8 6.8 6.8 6.8 6.8 6.2

2.2 2.6 3.2 3.8 3.8 4.4

4.6 4.2 3.6 3.0 3.0 2.8

0.32 0.38 0.47 0.56 0.56 0.71

cm-'

Table 111. Conductance Behavior of the AgzS/PbS Electrode at 25 "Cas a Function of Frequency Current density ( f 6 bA/crn*), Hz

5 10 20 160

Total

Ionic

Electronic

Ionic transference no. ti

0.32 0.32 0.32 0.32

0.22 0.22 0.22 0.28

0.10 0.10 0.10 0.04

0.69 0.69 0.69 0.88

Conductance X I 0-4, 0-l cm-'

RESULTS The Ag2S and AgZS/PbS Crystal Membrane Electrodes. Since AgzS is known (10) to be a mixed conductor, we examined the extent of electronic and ionic conductance of the electrode a t room temperature using the method of Jaenicke et al. (9).This method is based on the assumption that the total conductance is measured during a transitory current a t t = 0 while only the ionic conductance is involved with a steady state current after some time. The ionic conductance would persist after the electronic pathway is blocked a t the AgzS/electrolyte phase boundary when the mobile electrons in the membrane phase are exhausted during anodic polarization. The ionic transference number, t i , can then be obtained by dividing the ionic conductance by the sum of the ionic and electronic conductances. Tables 1-111 summarize the results obtained from galvanostatic square-wave measurements carried out using various current amplitudes and frequencies in 0.08 M AgN03 and 0.2 M KNOB solutions. The results agree well with previous measurements ( 1 1 ) . The observed Ohmic behavior for the ionic and total conductance is typical of materials having a high degree of disorder. As expected, the electronic conductivity for the Ag2S membrane in contact with silver metal is especially high because of the high interdiffusion coefficient of Ag+ in Ag2S; Rickert (10) found this coefficient to be -0.5 cm2/s a t 200 "C, a value several orders of magnitude higher than diffusion coefficients in solution, and probably involves an electron jumping mechanism. The frequency dependence of the ionic transference 858

-

ANALYTICAL CHEMISTRY, VOL. 48, NO. 6, M A Y 1976

80

100

120

140

160

msec

Figure 3. Voltage-time behavior recordings. Galvanostatically controlled square-wavecurrent at 5 Hz Dashed line: Ag metal electrode in 0.08M AgN03 4- 0.2 M KN03 (f40FA current amplitude). Solid line: AgzS/PbS electrode with 0.08M AgN03 0.2 M KN03 (+4 PA current amplitude)

+

number for the metal connected AgzS electrode suggests that the AgJAgzS phase boundary becomes strongly polarized when current flows and, thus, disturbs the measurement. This effect is clearly shown below by the currentvoltage curves obtained for this electrode. Figure 3 compares the voltage-time relationship obtained for the AgZS/PbS electrode with electrolytic solution on both sides of the membrane with that of a silver metal electrode. For the silver metal electrode, considering a galvanostatic current step in a parallel array of a resistance Rt and a capacitance C, one can calculate a charge-transfer resistance of -30 Q cm and a double-layer capacitance of -2000 pF/cm2; the resistance and the capacitance are remarkably high. Budewsky (12) explained the high-chargetransfer resistance at silver metal with low overpotentials as being due to a hindrance of the crystallization step and found R t values as high as 2000 R a t overpotentials less than 7 mV. This may explain why the silver metal electrode is not an ideal Nernstian sensor for silver ions a t very low concentrations, i.e., the establishment of an equilibrium potential involves steps with very low overpotentials. Gerischer (13) proposed that the high capacitance of the silver metal electrode is due to a pseudocapacitance from adatoms. By extrapolation from higher overpotentials, Figure 4 shows that the standard current density a t the Ag/Ag+ boundary could be as high as 1 A/cm2 for = 0.73. In order to study the polarization a t the internal AgzS/ Ag phase boundary in more detail, current-voltage curves were also recorded using a pulse technique in 0.1 M AgN03 and 0.1 M KZS solutions, respectively. The results and corresponding log i vs. V plots are shown in Figure 4. As has been pointed out above, the application of conventional electrode kinetic treatments to ion-selective membrane electrodes should be regarded as tentative (14), but the results of Figure 4 show that the basic relationships are valid a t least as a first approximation. Although only the overall voltage, V, between the internal silver metal phase and the external reference electrode can be controlled in these experiments, it can be seen that the log i vs. V plot is linear a t overvoltages less than 0.3 V and that apparent exchange current densities can be obtained by extrapolation to V = 0. The results obtained are not drastically altered by correcting for the membrane iR drop using the previously measured ionic conductance and the currents involved. In view of the low values obtained, we conclude that the exchange current densities are those a t the internal AgzSlAg phase boundary. This view is supported by the fact that the exchange current densities found are not concentration de-

: $50

II

10-

mA

$+?j0

Volts

I

I f

I

I

..

i

''

0.1

0.3

0.2

0.4

0.5

0.7

O.b

0.8

0.9

I

1 -2

1.0

2 5 x 10

+v vo 10

M La"

30

20 I

40

sec

Figure 5. Evaluation of galvanostatic current step measurements: LaF3 electrode: 1 pA current steps

,.onodic

-3.01

"

I

"

"

0.2

0.1

0.3

0.5

0.4

0,6

07

'

08

"

0.9

tV

1.0

Volts

Figure 4. Current-voltage curves (top) and Tafel plots (bottom) for various electrodes. (x)

silver electrode in

M AgNO3

M AgN03 -I-0.1 M

+ 0.1 M K N 0 3 + lo-*

M "03;

(.) Ag2S electrode in (0)Ag2S electrode in 0.1 M "03;

+

K2S; (A)Ag&PbS electrode in 0.05 M AgN03 0.5 M KNOJ or 0.05 M K2S 0.5 M KN03; (A)Ag2S/PbS electrode in 0.05 M KI 0.5 M KNOJ

+

+

pendent. The decrease of exchange current density values in the 0.1 M K2S media is reasonable in view of Ag+ binding by sulfide and the attendant reduction of available free Ag+ in the Ag2S phase, as shown by the accompanying 20fold decrease in conductivity (15). The increase in current at greater overvoltages is due to decomposition of AgZS. Since our primary goal was to study the phase boundary AgzS/solution, Figure 4 also shows measurements taken on the AgZSPbS electrode with electrolyte on both sides of the membrane and with varying Ag+ concentrations in the outer solution. T o exclude from the measurement any polarization a t the auxiliary silver wire, the voltage was measured with an additional silver wire in the outer solution connected to the high impedance input of the potentiostat. The premise that the overpotentials a t the membrane/ solution phase boundaries are small is supported by the linearity of the current-voltage plots obtained. The effect of the overpotentials cancels when the inner and outer solutions are the same. It is remarkable that the 0.05 M AgN03 and 0.05 M KzS solutions give identical total resistances. This suggests that sulfide ions are also involved in the charge transfer and contribute toward determination of the potential. One possibility is that the sulfide ions react with interstitial Ag+ and become part of the lattice (9, l l ) , e.g.9 2Ag0+ + S2-

F*

Ag81attice

(7)

Such a mechanism would explain the rapid response time of the AgzS electrode in sulfide solutions where the free Ag+ concentration is extremely low and where it is difficult to justify the dynamic behavior of the electrode on the

basis of Ag+ as the sole potential determining ion. In 0.05 KI solution, on the other hand, the membrane resistance increases by 200 ohms (Figure 4). This suggests that iodide ions do not enter the lattice to an appreciable extent. If one interprets the observed resistance increase as an increase in charge-transfer resistance corresponding to a decrease in silver ion concentration in the solution, one can estimate (using (? = 0.5) an apparent standard exchange density of -200 A/cm2 a t the membrane/solution interface. The corresponding exchange current densities for the silver metal electrode calculated from Figure 3 at overpotentials of less than 7 mV are of the order of IOM3A/cm2. This high ratio of exchange current densities is in agreement with other measurements (11) and may explain why the AgzS membrane is superior to the silver metal electrode as a sensor for Ag+ a t low concentrations. The Lanthanum Fluoride Crystal Membrane Electrode. The fluoride ion selective membrane electrode, first introduced by Frant and Ross (16), is not only very widely used but is one of the most selective membrane electrodes presently known. The only serious interference for the fluoride electrode is that due to OH- when present in considerable excess over F-. This interference could be the result of OH- entering the LaF3 lattice, since OH- and F- have the same charge and are similar in size, or from surface reactions of the type LaF3

+ 30H-

+

La(OH)3 3F-

(8)

Exchange current density measurements may serve to distinguish between these possible mechanisms. Figure 5 shows typical voltage-time curves obtained with LaF3 membrane electrodes as well as the method used for evaluation of the resistance. Basically, a plot of V vs. t1/2is extrapolated to t = 0 to show the validity of equation 6 (e.g., reasonable diffusion coefficients) and to obtain the diffusion polarization. The fluoride ion concentration is then raised until the initial total resistance undergoes no further change; a t this point, the charge-transfer resistance will be small compared to the pure ohmic resistance of the membrane and the latter is taken as that being equal to the initial resistance. Resistance values are obtained by shorting the circuit with a known resistance or by passing a small ( 10-8-10-6 A) current through the cell. Double-layer charging could not be observed because the electrometer used was rate limiting, but such a charging effect would be a constant contribution of no consequence here since only resistance differences are employed. Earlier studies by Brand and Rechnitz (8) showed that the capacitance of a LaFs/solution interface is many orders of magnitude smaller than those a t metals. Buck and Krull (17) found similar ANALYTICAL CHEMISTRY, VOL. 48, NO. 6, MAY 1976

859

Table IV. Resistance of LaF3 Membrane Electrodea Resistances

Initial, kR

Electrolyte EMF, mV (measured) 0.7 M KzSO4 t 1.4 M K F -530 148 0.7 M K2SO.j + 1.4 X 10-1 M K F -475 150 0.7 M K2SO4 + 1.1 X M KF -410 160 0.7 M + 5.5 X M K F -340 187 0.7 M KzS04 + 1.7 X M KF -260 265 2.5 X lo-' M K F 0.18 M -430 144 2.5 X M KF + 0.25 M NaCl -375 167 2.5 X M KF 0.25 M acetate buffer 0.10 M EDTA -270 239 -170 280 Pure water -180 275 0.7 M KzS04 -360 180 0.7 M K2S04 + 2 X lo-' M KOH -320 185 0.7 M K2S04 + 2 X lo-' M KOH -280 195 0.7 M + 2 X M KOH 2.5 X 10+ M La(N03)3 -50 330 a 25O, Hgl HgS04 reference electrode, values corrected for diffusion polarization.

1

low capacitances at glass membrane/solution interfaces. Table IV shows the results obtained for various fluoride concentrations in different supporting electrolytes. Changes in the total measured resistance should correspond to changes in the charge-transfer resistance at the outer LaFJsolution phase boundary. Because the chargetransfer resistance in 1 M F- solution is negligible compared to the ohmic resistance of the membrane, the measured resistance value in this medium was taken as a summation of all the constant resistances involved in the cell and subtracted from the values obtained in more dilute solutions to obtain, as a first approximation, the desired charge-transfer resistances. The charge-transfer resistance values so extrapolated are plotted as log R t vs. log F- concentration in Figure 6, which, when extrapolated to 1 M F-, corresponds to an apparent standard exchange current density of -10-5 A/cm2 with a transfer coefficient of -0.65. The lower initial resistance value obtained in the presence of EDTA could be the result of La3+ complexing. However, the accuracy of the measurements is no better than f 2 % and, since differences between two large values are taken, the effect observed is not significant compared to the overall uncertainty. If the estimated exchange current density is correct, then the fluoride electrode should function as a Nernstian sensor even when relatively large currents are drawn. Figure 7 shows some results obtained with shorted circuits. Although it is clear that some concentration polarization occurs, e.g., plots of t112 are linear, Nernstian behavior persists as shown by the linear relationship between the initial current and the logarithm of the fluoride concentration. Indeed, voltages obtained by multiplying the currents by the proper membrane resistance show Nernstian behavior for currents as high as A/cm2. It is interesting to note that the initial currents are higher for a dry electrode surface than for a wet surface. This effect is probably due to the charging current required to establish the double layer and indicates, in view of the parallel shift of the plot, that the double layer capacitance does not change with changes in fluoride concentration in the presence of excess supporting electrolyte. In order to investigate the mechanism of OH- interference, apparent standard exchange current densities were also measured for OH- solutions with the results shown in Table IV. Since the estimated standard exchange current 880

ANALYTICAL CHEMISTRY, VOL. 48, NO. 6. MAY 1976

Constant, kR (extrapolated)

Variable, kR (extrapolated)

146 146 146 146 146 144 144

2 4 14 41 119

144 146 146 146 146 146 146

95 134 129 34 39 49 184

23

2.0

1.5

log

Rt,kn 1.0

fluoride electrode

0.5

I

IO-'

IO-^ concentration

I

I

16' M

Figure 6. Log Rt vs. log F- concentration plot for LaF3 electrode (0)KF in K2S04 electrolyte: (0) KF in 0.18 M K2S04 t 0.25 M NaCl M EDTA (acetate buffer, pH 5.5); (A)KOH electrolyte

+ 0.01

density is 8 A/cm2) anion exchange current densities for the dissolution of AgBr layers in cyanide solutions; this finding is interesting in view of the fact that AgBr membrane electrodes can be used to measure cyanide ion levels. The observed polarization a t the AgIAgzS phase boundary has important implications for the design of Ag2S-based crystal membrane electrodes. In particular, if there is any risk of appreciable currents being drawn in the measure-

Volts

Figure 8. Log Rt vs. EMF plot for LaF3 electrode (0)varying F- concentration in K2SO4 (saturated): (0) varying F- concentra0.24 M NaCl 0.01 M EDTA (acetate buffer, pH tion in 0.18 M K2S04 5.5); (A)varying KOH concentration: (x) KpS0, (saturated); (0) 2.5 X lo-' M h(N03)3

+

+

ment, the Ag2Slelectrolyte interface is to be preferred to AgZSlAg internals. The results obtained for the LaF3 type fluoride electrode support the view that F- is the only ion able to enter the crystal lattice to any appreciable extent. Since the estimated standard exchange current density at the LaF3/F- interface is relatively high, equilibrium potentials are rapidly established except a t very low fluoride concentrations. The electrode behavior in the presence of OH- or La3+ can be explained by chemical equilibria of the type shown in equation 8 and by the existence of surface films also observed in impedance measurements (8). The experiments performed in this study demonstrate the importance of electrode kinetics for membrane electrodes where thermodynamic equilibrium is not necessarily achieved in the phase boundary reaction. Lyklema et al. (20) demonstrated several aspects of nonequilibrium behavior a t glass electrodes. In view of the success of kinetic theories in treating corrosion processes, it seems that the relationships developed for mixed potentials at zero current by various workers (21-23) could be fruitfully applied to ion-selective membrane electrodes. We have demonstrated that relationships originally developed for metal electrodes are approximately valid for ion-selective membrane electrodes as well. In this model of parallel electrode reactions, the mixed potential observed is primarily determined by the equilibrium potential of the electrode reaction having the higher exchange current density. Thus, the behavior of ion-selective membrane electrodes can be interpreted on the basis of virtual current-voltage curves for the individual electrode reactions.

LITERATURE CITED (1) J. 0'. M. Bockris and A. K. N. Reddy, "Modern Electrochemistry", Vol. 2, Plenum Press, New York, N.Y.. 1970. (2) J. A. V. Butler, Trans. Faraday SOC.,19, 729 (1924). (3) T. Erdey-Gruz and M. Volrner, 2. Phys. Chem. (Leipzig), 150, 203 (1930). (4) W. Jaenicke and M. Haase, 2.Elekfrochem.,63, 521 (1959). (5) P. Delahay and T. Berzins. J. Am. Chem. SOC., 75, 2486 (1953); 77, 6448 (1955). (6) D. inrnan, J. 0'. M. Bockris, and E. Blomgren, J. Electroanal. Chem., 2, 506 (1961). (7) E. E. Conway. "Theory and Principles of Electrode Processes", Ronald

ANALYTICAL CHEMISTRY, VOL. 48, NO. 6, MAY 1976

861

Press, New York. N.Y.. 1965. M. J. D. Brand and G. A. Rechnitz, Anal. Chem., 41, 1185 (1969); 42, 478 (1970). W. Jaenicke, E. M. Khairy. and W. Schaefer, 2.Elekhochem., 70, 421 (1966). H. Rickert, “Einfuhrung in die Elektrochemie fester Stoffe”. Springer Verlag, Berlin, 1973. M. Koebel, Dissertation No. 4853, Zurich, 1972. E. Budewski. Electrochim. Acta, 11, 1697 (1967). H. Gerischer, Z.Elektrochem.,54, 366 (1950). K. Camman, Dissertation, University of Munich, 1975. C. Tubandt and H. Reinhold, Z.Elektrochem.,37, 589 (1931). M. S. Frant and J. W. Ross, Science, 154, 1533 (1966). R. P. Buck and I. Krull, J. Electroanal. Chem., 18, 387 (1968). J. W. Ross, Natl. Bur. Stand. (U.S.), Spec. Pub/.,No. 314 (1969).

(19) H. Gerischer, Z.Phys. Chem. (Frankfurt am Mein). 26, 223 (1960): 27, 48 (1961). (20) T. F. Tadros and J. Lyklema, J. Electroam/. Chem., 22, 91 (1969). (21) C. Wagner and W. Trand. Z.Elektrochem., 44, 391 (1938). (22) G. Kimball and A. Glassner. J. Chem. Phys., 8,815 (1940). (23) J. 0’.M. Bockris, “Modern Aspects of Electrochemistry”, Vol. 1, J. 0’. M. Bockris and B. E. Conway, Ed., Butterworths,London, 1954.

RECEIVEDfor review November 10,1975. Accepted February 2, 1976. We thank the National Science Foundation and the Deutsche Forschungsgemeinschaft for financial support of this research.

Highly Selective Enzyme Electrode for 5’-Adenosine Monophosphate D. S. Papastathopoulos and G. A.

Rechnitz’

Department of Chemistry, State University of New York, Buffalo, N. Y. 142 14

A 5’-AMP sensing electrode Is devised using a highly seiective deaminase enzyme in conjunction with an ammonia gas sensing membrane electrode. The resulting nucleotide sensor is very highly Selective for 5’-AMP with good operating sensitivity and dynamic response. Optimum conditions for the concentration and immobilization of the enzyme are explored in terms of electrode operating requirements.

Wide attention has recently been given to the development of membrane electrode sensors selective for various biological substrates using immobilized enzymes in conjunction with potentiometric ion or gas sensing membrane electrodes. Various analytical and biochemical aspects of this field have been the subject of recent reviews and articles (1-7) especially in regard to electrodes for major body fluid constituents and pharmaceuticals. In one of these reviews ( 8 ) , attention was called to the possibility of devising such electrodes as sensors for nucleotides with particular emphasis on an electrode for 5’-adenosine monophosphate (AMP). We now report on the construction and evaluation of the AMP sensor in detail. It will be seen that the AMP sensing electrode has good sensitivity and excellent selectivity characteristics provided optimum enzyme activity and solution conditions are achieved. The new electrode employs a layer of suspended 5’-adenylic acid deaminase enzyme (AMP deaminase), classification number EC 3.5.4.6, in conjunction with an ammonia gas sensing membrane electrode. The substrate is selectively deaminated to inosine 5’-monophosphate (5’-IMP) via the reaction 5’-AMP

+ HzO

Enzyme

5’-IMP

+ NH3

(1)

producing NH3 in stoichiometric quantities and gives rise to a steady-state potential reading reflecting the AMP concentration in the sample to be measured. Use of the gas sensing electrode as a component of the sensor ensures freedom from ionic interferences.

EXPERIMENTAL Apparatus. An Orion Model 95-10 ammonia gas sensing electrode was employed in construction of the enzyme electrode. No 862

ANALYTICAL CHEMISTRY, VOL. 48, NO. 6, MAY 1976

external reference electrode is required with this design. Potential measurements were carried out with a Corning Model 1 2 meter and were recorded using a Heath-Schlumberger SR-255B recorder. Measurements were made in a thermostated cell held a t 27 f 0.2 “C. A Beckman BD-G spectrophotometer, with thermostated cell compartment, was employed for the optical enzyme activity determinations. Reagents. The AMP deaminase enzyme (grade IV, from rabbit muscle) was obtained from Sigma Chemical Co., St. Louis, Mo., as were the substrates 5’-AMP, 3’,5’-cyclic AMP, 5’-ADP, 5’-ATP, adenosine, and adenine used for selectivity studies. The enzyme is received as a suspension in 66% glycerol, containing 0.33 M KC1 a t pH 7.4, with an enzyme activity of 30 unitdml. Working substrate solutions were prepared in 0.05 M Tris-HC1 buffer, pH 7.5, and stored under refrigeration. 5’-AMP solutions were also prepared in 0.1 M citrate buffer, pH 6.50, 0.1 M succinate buffer, pH 6.40, and in 0.05 M Tris-HC1 buffer a t pH 7.00 and pH 8.4. All substrates were tested for possible ammonia contamination with the Orion 95-10 electrode. Both the 5’-AMP (sodium salt) and 3’,5’-cyclic AMP were found to contain appreciable ammonia background levels and were purified by recrystallization and ion exchange (Baker, ANGC-101 resin), respectively, to negligible ammonia levels prior to use. Enzyme Concentration Procedure. Since the enzyme, as received, had an activity of only 30 units/ml, it was thought desirable to raise enzyme activity by concentration. For this purpose, a molecular filtration procedure was employed (9).One hundred units of the enzyme suspension were passed through a PSED Pellicon molecular filter (25 000 molecular weight cut off) a t 4 OC under 50 psi nitrogen gas pressure over a 16-18 h time period. Enzyme activity was measured spectrophotometrically a t 265 nm using the procedure recommended by the supplier (IO). The concentrated enzyme preparation obtained after 16 h has an activity of -90 units/ml. This preparation was stored at 4 OC and periodically tested for enzyme activity. No significant loss of activity was observed over a 2-month period. Electrode Preparation. The 5’-AMP enzyme electrode was assembled using the general techniques previously described for the urea electrode (11). In the present case, 10 pl of the concentrated enzyme preparation (corresponding to -0.9 unit) was placed between a circular cellophane dialysis membrane and the gas permeable membrane of the ammonia electrode. The resulting electrode was preconditioned by soaking for at least 3 h in 0.05 Tris-HC1 buffer, pH 7.50, and was also stored in this buffer when not in use.

RESULTS AND DISCUSSION Figure 1 shows a schematic representation of the phases comprising the enzyme electrode for 5’-AMP and identifies some of the key steps involved in the overall response of the electrode system to the substrate in the sample solu-