J. Chem. SOC. Faraday Trans. 1 1986,82 1105-1116 Analytical Applications of Gas Membrane Electrodes Stanley Backenstein* and James S. Symanski Department of Chemistry State University of New York at Buffalo Buffalo New York 14214 U.S.A. Porous and permeable membrane electrode structures are now widely exploited in sensors for species in aqueous and gas phases. Potentiometry and amperometry have beenapplied frequently in conjunction withmembrane electrode methodology. Only recently has the classic technique of conduc- tometry been used in tandem with membrane electrodes. Ambient-level sulphur dioxide and carbon dioxide conductance sensors based on structures involving porous membranes and thin water layers are described and the principles governing their response are discussed.An electrode separated from an analyte-containing phase by a membrane through which the analyte must pass to reach it is known as a membrane electrode. The electrode may be separated from the membrane by a thin layer of aqueous solution. Alternatively it may be deposited on or partially into the membrane. On reaching the electrode the analyte is determined by a suitable electrochemical technique. Membrane electrodes are widely used as sensors in gases and in aqueous solutions. Gas membrane electrodes are particularly convenient to determine the concentration of species present in the gas phase. They are readily incorporated into small hand-held monitors that meet critical design requirements. These requirements include low power consumption read-on-demand display fast response analytical selectivity response compensation over a wide range of ambient temperatures and use by non-technical personnel.Various kinds of structures for analytical gas membrane electrodes have been designed. The membranes are conveniently classified into two types permeable or porous. Gas membrane electrodes based on these membranes have electrocatalysts (metals) deposited on the membrane side not contacting the gas phase. Also traditional fuel-cell structures may be used. Regardless of the electrode structure a variety of electrochemical techniques have been employed including potentiometry amperometry and conductometry . Membranes A permeable membrane behaves as a homogeneous phase so that the transport of a species into through and out of it can be described readily.This description includes the distribution equilibria involving the membrane and the species in the two phases contacting the membrane the transport of species from one face to the other and the mass-transport situation existing in the boundary layer adjacent to the membrane faces. Classical diffusion and boundary-layer theory are usually sufficient in modelling such an electrode structure. A porous membrane behaves as a two-phase structure. Those used in gas membrane electrodes are hydrophobic and frequently are made from Teflon which also provides the mechanical structure of the membrane. One common type of membrane can be visualized as being a sintered mat of Teflon particles with tortuous interconnected channels filled with gas between the two faces.There are two unique problems in describing transport of species through a porous membrane. The first involves the area 1105 1106 Gas Membrane Electrodes of the membrane that is effective in the transport process. It is normally considered to be the gas-phase area in the membrane surface. The second problem is the uncertainty in the effective diffusional path length between the two membrane faces. If a gas phase contains the analyte the pores within the membrane are filled with this gas and the analyte. Thus no partitioning of analyte occurs at the membrane face in contact with the gas phase. However if the analyte-containing phase is a liquid selective partitioning into the gas-filled membrane pores may occur.The differences between permeable and porous membranes have important consequences. First it is possible to obtain selectivity for the analyte with respect to other species by using a permeable membrane. This selectivity results from differences in partition coefficients of the various species and the membrane phase as is the case of sulphur dioxide with respect to both nitric oxide and nitrogen dioxide at a polyethylene membrane.l However the diffusion coefficient of the analyte in a permeable membrane is orders of magnitude smaller than it is in the gas pores of a porous membrane. Thus usually it is necessary to use very much thinner and more fragile permeable membranes to approach the fluxes of analyte found using porous membranes.On the other hand a large value of the analyte’s partition coefficient can increase the flux through the permeable membrane and minimize this difficulty. Secondly a hydrophobic porous membrane shows no selectivity for the analyte with respect to other species present in a gas phase. However it is selective with respect to species present in an aqueous phase. This selectivity is governed by the partial pressures of the dissolved species. Hence selectivity in gas-phase analysis problems can only be obtained with porous membrane-based gas membrane electrodes by using selective electrochemical techniques and/or chemical procedures. The simplest of the chemical techniques frequently are satisfactory. For example a porous mat filled with sodium bicarbonate interposed between the gas phase and the membrane prevents a wide variety of acid and basic gases from reaching the membrane.Porous Electrodes Porous electrodes may be deposited on a membrane face in several ways. The main techniques are vacuum evaporation sputtering and thermal decomposition of dissolved metal compounds usually from organic phases.2 The latter technique is convenient for laboratory-scale studies as it is simple and requires no expensive equipment. Vacuum evaporation and sputtering are more versatile where a wide variety of metals are needed.3 Thicker metal deposits are most conveniently prepared by the first technique which makes use of commercially available ‘metal inks’ and other materials widely used in the electronics and ceramics industries.? Fuel-cell-type electrodes are made from a mixture of the electrocatalyst and Teflon particles by applying heat and pressure.A wide variety of such structures have been fabricated and the details of their fabrication are usually not disclosed by the manufacturer. Elect roc hemical Techniques The most frequently used membrane-based electrochemical techniques are potentiometry and amperometry. It is only recently that conductometry has been combined with membrane methodology. This paper deals primarily with application of membranes in conductometry but the distinctive characteristics of all the techniques will be summarized briefly. 7 Engelhard Industries East Newark New Jersey U.S.A. manufactures a wide variety of such materials.S. Bruckenstein and J. S. Symanski 1107 Potentiometry The major analytical field of ion-selective electrode methodology is founded on the response of selective ‘membranes’ to dissolved ionic species. These membranes can be solid-state or liquid and exhibit ion specificity. Existing ion-selective electrodes have been combined with permeable or porous membranes to produce a unique class of aqueous sensors. The principle involves separating an external aqueous phase from the ion-selective electrode by a membrane. In one variation the membrane is affixed to the electrode so that a very thin layer of electrolyte solution separates the membrane and the electrode. The analyte in the external aqueous phase diffuses through the membrane reacts with something present in the electrolyte solution to produce a species that is then detected and quantitated by the electrode.Ordinarily the potential of the electrode is logarithmically related to the analyte’s concentration in the external aqueous phase. Some recent examples of this approach are potentiometric sensors for CO which involve the measurement of P H . ~ ~ Transport to the membrane is usually ill-defined and depends on natural convection in the external aqueous phase. Transport through the membrane or the thin electrolyte film or rate of reaction within the electrolyte film is usually rate-limiting in determining the potentiometric response of the electrode. It is important to note that the electrode is not a sink for the diffusing species rather the chemical or biochemical reaction within the thin electrolyte layer causes the layer to function as a sink.High sensitivity and selectivity is possible with this technology when appropriate reaction chemistry can be coupled to an ion-selective electrode. The analyte can be caused to react within the membrane if a reagent is incorporated into the membrane. Enzymes are used in membranes in this way and a variety of molecules of biochemical interest are now determined in this way. Examples of this approach coupled to integrated-circuit technology are summarized elsewhere.6 Amperometry Amperometry is widely used to determine volatile species present in aqueous and gas phases. The Clark electrode for dissolved oxygen is a classic example of the former while there exist many illustrations of the latter.Among the latter are sensors for hydrogen s~lphide,~? hydrogen ~yanide,~q carbon mon~xide,~? lo nitric oxide’ll nitrogen dioxide1’ and chlorine.12 In all sensors the indicator electrode potential is held at a potential which results in the electrolysis of the analyte species. The resultant current is proportional to the analyte’s concentration in the phase contacting the membrane. Various schemes have been used to establish the indicator electrode’s potential. A three-electrode potentiostat is the most versatile since it places the least electrochemical restraints on the choice of reference electrode. Two-electrode circuits using large-area reference electrodes are common. An artificial distinction has been made between two-electrode cells in which a voltage is applied between the reference and the indicator electrode to establish the desired potential (electrolysis cell) and those in which the reference electrode is shorted to the indicator electrode through a current-measuring device (galvanic cell).There are a large number of publications in this area and patents make up a substantial portion of this literature. Conduc tometry One concept involved in these sensors is based on the transport of a species through the membrane to a pure water solution followed by measurement of the conductivity. Thus any species which reacts with water to produce ions may be determined with high sensitivity. However this conductometric method does not provide selectivity and 1108 A 6 C Fig.1. Schematic representation of the sensing region of a conductometric cell (A) mixed-bed ion exchanger (B) cell body (Plexiglass) (C) thin-layer chamber for deionized water (D) conductance electrodes; (E) porous Teflon membrane (F) filter for removing interferents (activated G Gas Membrane Electrodes carbon). H2O reservoir 1' I 1 i filter gas I Fig. 2. Diagram of thin-layer sensor. The arrows show the direction of water flow when the pulse pump is actuated manually. The one-way check valves A and B control the direction of water flow. additional techniques must be combined with conductometry if a selectivity problem exists. Van Kempen and Kreuzer constructed a conductometric sensor from a double lumen catheter whose tip was covered with a membrane permeable to carbon di0~ide.l~ A set of conductance electrodes situated in each lumen measured the conductances of the solution in each lumen.Water flowed from one lumen to the other passing over the membrane. The conductance difference between the water in the two lumens yielded the carbon dioxide concentration in the gas phase contacting the permeable membrane. Control of water flow rate is necessary to assure reproducible results in this sensor as it is in all non-equilibrium sensors based on steady-state hydrodynamic conditions. 1109 26 27 dioxide 165 5 1 10 S. Bruckenstein and J. S. Symanski Table 1. Response time (in s) plotted against gas flow rate and partial pressure for carbon dioxide gas flow rate/dm3 min-l 0.5 0.2 0.05 0.5 24 25 Table 2.Response time (in s) plotted against gas flow rate and partial pressure for sulphur gas flow rate/dm3 min-l 300 210 270 145 120 40 120 120 110 40 50 55 Himpler et aZ.14 described another carbon dioxide sensor in which carbon dioxide diffused through a dialysis membrane into a thin water layer whose conductance change was determined. Their design provided no convenient way to replace the water in the thin layer and the sensor exhibited a slow increase in conductance with time. Martinchek concurrently described a sensor using a porous Teflon membrane in a thin water layer geometry that included an integral mixed-bed ion-exchange column and a simple means for replacing the water behind the membrane.2* l5 This approach provided a sensor with long time zero stability.Further details about this kind of methodology as implemented in a portable sensor are given below. Portable Conductometric Gas Membrane Electrode for Carbon Dioxide and Sulphur Dioxide Fig. 1 and 2 illustrate one way we have applied membrane methods to conductometric determination of gaseous species. The manually operated pulse pump in fig. 1 forces water from the reservoir through the mixed-bed ion-exchange column into the thin water layer C of fig. 2. Water initially present in this layer is collected in the water reservoir for re-use. An analyte present iri the gas phase diffuses through the filter F (designed to remove interferents) and the porous Teflon membrane into the thin water layer.Conductance electrodes deposited on the water side of the membrane monitor the conductance. After a sufficient time has passed the conductance becomes constant. The equilibration time required varies with the analyte gas flow rate and the cell geometry. The thin layer cell described previously2v l5 for the determination of atmospheric carbon dioxide was tested from ambient levels to high partial pressures of carbon dioxide. A steady-state conductance was attained in ca. 25 s independent of gas flow rate as seen in table l.29 l5 However times of ca. 5 min were required to attain conductance steady state for ppm levels of sulphur dioxide.16 Therefore the thin-layer cell was modified to 1110 shorten this time response.The modified cell was capable of determining 20 ppb SO with a response time of 110 s as seen in table 2. Even this cell showed a marked dependence of response time on the gas flow rate and Pso2. The reasons for the differences in the behaviour between carbon dioxide and sulphur dioxide are discussed below. Gas Membrane Electrodes Ka (2) XO,(aq)+H,O,H+(aq)+HXO;(aq). Reactions (1) and (2) describe the equilibrium of gaseous carbon dioxide and sulphur dioxide with dissolved CO and SO,. Raman,17 infrared1* and ultravioletlg absorption studies indicate that aqueous solutions of sulphur dioxide consist almost exclusively of uncombined SO molecules. No evidence has been found to support the presence of the H,SO molecule in solution. Also the primary undissociated species in aqueous carbon dioxide solutions is CO,(aq) although a small amount of H,CO exist at equilibrium.The rates of both reactions (1) and ( 2 ) O are fast compared to the timescale of mass-transport-controlled equilibration of the thin water layer for both carbon dioxide and sulphur dioxide. Thus the sum of these reactions the overall equilibrium between gaseous XO and the ions formed in the water layer can be used to discuss the mass-transport-controlled transient conductance response of both species. The overall equilibrium constant K (= Kp K,) describes the equilibrium K Theory The conductometric sensor’s response to both CO and SO is described using the following equilibria where X represents either C or S At equilibrium (3) where K is the ‘apparent’ dissociation constant of XO,(aq) and Kp is the Henry’s law constant describing the equilibrium between XO in the gas phase and XO,(aq).(4) (6) In eqn ( 5 ) and (6) square brackets signify equilibrium concentrations,f-terms are activity coefficients and A-terms are equivalent ionic conductances of the designated species. The equilibrium concentrations of XOi- (and HS,OT)~~ are negligible because the first ‘ apparent’ dissociation constant of XO,(aq) is many orders of magnitude larger than the second., Also for the range of studied partial pressures the ionic strength is so small that the activity coefficients of H+ and HXO; approach unity and the equivalent ionic conductances do not vary. Substituting eqn (4) and (5) into eqn (6) yields the expression for the specific conductance of the water layer at equilibrium (8) XO,(g)+H,OeH+(aq)+HXO;(aq) [H+] = [HXO,] and the specific conductance IC is given by IC = 10-3(AH+[H+]+;1HSO;[HX0~]).The experimental cell conductance S is proportional to IC. Therefore (Aw++;lrrxo;). S2 = (A/O)’ KPxo2 where 8 is the cell constant (cm-l) and A is equal to 1111 s2,,/p,as KP species I 1.20 1.29 x lo- 42 I 395 27 1 so2 co2 NH3 H2S 3.39 x 10-2 54.9 0.1 4.45 x 10-7 1.75 x 10-5 1.26 x 10-7 8 . 6 ~ 2.6 x lo- 6.6 x lo-' 378 a Constants from ref. (25H31). Eqn (8) applies to sulphur dioxide and carbon dioxide. Martinchek has verified this expression for CO for partial pressures near ambient to 100% carbon dioxide and Symanski has confirmed it for SO,l6 from 0.02 to 50 ppm.Eqn (8) also holds for any gas that dissolves in water to form an acid or base which is weakly dissociated. For example NH also obeys eqn (8).16 The ratio of the square of the equilibrium conductance for another gaseous species divided by it partial pressure with respect to the same quantity for SO is given by Table 3 presents the calculated results for various gases obtained by substituting the appropriate thermodynamic constants into eqn (9). The fourth column of the table compares the analytical sensitivity of these gases relative to SO,. For routine determi- nations of SO in the range of 0.1-5 ppm both hydrogen sulphide and carbon dioxide will contribute negligibly to the SO response at their ambient levels.Ambient CO (400 ppm) yields a response equivalent to ca. 3 x lo-* ppm SO,. Ammonia which is unlikely to be encountered in SO determinations would yield a false SO response of 0.03 ppm ifpNH3 was 1 ppm. These calculations show that an atmospheric gas membrane conductance sensor for sulphur dioxide does not need a filter for ambient levels of interferents. On the other hand an atmospheric carbon dioxide sensor needs an appropriate filter15 to remove interfering levels of sulphur dioxide and other gases that dissolve to produce relatively strongly dissociated acids or bases. S. Bruckenstein and J. S. Symanski /mol dme3 atm-l l2-l cm2 K,/mol dm-3 Table 3. Thermodynamic constants for SO and various gases comparison of sensitivities us.s 0 a s2,s/Pso2 (10) Discussion The transient conductance responses for carbon dioxide and sulphur dioxide are markedly different. Carbon dioxide transients are virtually independent of partial pressure and of gas flow rate whereas the opposite behaviour is found for sulphur dioxide.16 Conductance equilibration in the water layer is controlled by diffusion of XO through it. For carbon dioxide the diffusion process is not perturbed significantly by the dissociation step reaction (3). This simple diffusion model has been treated by Crank,, and the time required to reach 98 % of the equilibrium uptake of XO after a water pulse is t = 1.50 L2/D where L is the thickness of the thin water layer and D is the diffusion coefficient of the dissolved species in water.Using experimental values of L = 0.004-0.005 in.? and t 1 in. = 2.54 x lop2 m. 1112 Gas Membrane Electrodes 0.2 0.5 Table 4. Values of 2 for sulphur dioxide and carbon dioxide at different partial pressures 0.05 0.966 0.897 0.873 9.66 6.13 4.34 1 .o 0.858 0.1 0.25 0.5 1 .o 5.0 3.08 1.39 0.44 5.5 cm2 s-l yields t = 15-24 s; this result agrees satisfactorily with the data for carbon dioxide but not for the traces for sulphur dioxide. A gas stream containing 1 ppm sulphur dioxide requires ca. 2.5 min to equilibrate the water layer at a flow rate of 0.5 dm3 min-l.16 At the same flow rate only 40 s are required for equilibration with 55 ppm sulphur dioxide.ls This phenomena was also observed by Terraglio and Manganelli.24 They reported that the rate of solution of sulphur dioxide into water over a range of atmospheric concentrations of 0.31-3.3 ppm was a function of the partial pressure of the gas with saturation being reached more rapidly at higher concentrations.D = The shorter response time for sulphur dioxide as its partial pressure increases is a consequence of the very substantial dissociation of sulphurous acid occurring at these low partial pressures. In general the total solubility of XO depends on the sum of all the aqueous undissociated forms and the ionic forms produced by dissociation. Taking the relevant equilibria into account yields where the first term on the right-hand side arises from the undissociated form and the second term from HXO,.Table 4 summarizes solubility calculations based on eqn (1 1) for relevant partial pressures of sulphur dioxide and carbon dioxide. The parameter 2 [ = S/[xO,] (g)] compares the molar concentrations of the gas in the water layer and in the gas phase at equilibrium. It measures the accumulation of XO in the aqueous phase. Zso2 and Zco2 are calculated to be 1.39 x lo3 and 0.858 respectively for the partial pressures listed. Sulphur dioxide is concentrated more than a thousand-fold in the thin water layer from the gas phase. Therefore the slow equilibration time of SO results from the longer time necessary to deliver enough gaseous sulphur dioxide to the thin water layer through the membrane.In the case of carbon dioxide no such accumulation occurs. Furthermore the decrease of 2 as the partial pressure of sulphur dioxide increases explains the accompanying faster response time. The dependence of equilibration time on concentration is the result of the significantly large value of the ‘apparent’ dissociation of SO,(aq) (K = 1.3 x which controls the total solubility of sulphur dioxide in water at ppm levels. When the thin water layer is in equilibrium with 5 ppm sulphur dioxide ca. 98% of the dissolved species is in the form of HSO,. The second term of eqn (1 1) is significantly larger than the first one for sulphur dioxide. This is not true for carbon dioxide. Table 4 shows that over a twenty-fold range of Pco2 2 varies by no more than 13%.In contrast for a twenty-fold range in Pso (0.25-5.0 ppm) 2 decreases by a factor of 4.4. intercept (s.d.) /n-2 species s 0 a 1113 0.11 (0.078) x lo-" -0.03 (0.03) x 10-l' 0.09 (0.08) x s 0 a c 0 a S. Bruckenstein and J . S. Symanski Table 5. Least-squares calibration data for carbon dioxide and sulphur dioxide pressure (pprn) 0.0-1 .oo 0.00.20 &10000 slope (s.d.) /Q-2/(PPm)-1 6.39 (0.337) x 10-l' 6.75 (0.219) x lo-" 10.6 (0.15) x a Strip cell 120 s data gas flow rate = 0.5 dm3 min-l. Circular cell 16 s data natural diffusion. Analytical Consequences The sensor described previous1yl5 functioned by capturing the conductance of the thin water layer precisely 16 s after the pulse pump in fig. 3 refilled the thin layer with fresh ion-exchanged water.The conductances obtained this way obeyed eqn (8) from ambient carbon dioxide levels to pure carbon dioxide. However even after 5 min the conductances obtained for flowing gas streams (0.5 dm3 min-l) containing sulphur dioxide in the range 1-10 ppm did not agree with theory. Therefore although the thin-layer conductometric cell used for carbon dioxide has considerable sensitivity when used with low levels of sulphur dioxide its slow response time and deviation from equilibrium theory are drawbacks. Geometry of Membrane Cell The membrane in the cell used for the carbon dioxide sensor was circular and placed in the cell so that mass transport could only occur normal to both of the membrane's faces. A new cell was designed in which the thin layer was a very narrow and shallow channel milled into a Plexiglas block.ls The porous membrane with electrodes on the water side was placed across this channel.This geometry provided gas transport to one membrane face over an angle of 180" rather than just normal to the membrane. Transport through and into the thin water layer was still constrained to be normal to the membrane face. As a consequence of the increased sulphur dioxide flux to the membrane the flux of sulphur dioxide into the thin water layer increased and the equilibration time decreased considerably for the new cell design the strip cell. Strip Cell Behaviour A comparison of calibration curves obtained using the circular cell design with carbon dioxide and the strip cell design with sulphur dioxide is given in table 5.The flow rate of gas to the cell was 0.5 dm3 min-l for sulphur dioxide a velocity which was high enough to make the results independent of the gas flow rate. The time required to reach 95% of the equilibrium conductance for various values of Pso2 ranged from 60 s (1.0 ppm) to I10 s (0.02 ppm). These response times are a marked improvement over those listed in table 2 obtained in the circular cell geometry for much higher concentrations of sulphur dioxide. The signal equivalent of the N blank at 120 s is ca. 3 x lo-* ppm sulphur dioxide. The difference in the slopes of the two sets of sulphur dioxide data is only 6% well within the experimental error introduced by the gas-proportioning system we used.1114 L Gas Membrane Electrodes Fig. 3. Continuous conductometric sensor (exploded view) (A) water (B) glass or plastic tube (C) conductivity electrodes on porous Teflon membrane (D) gas phase containing carbon dioxide (E) mixed bed of ion-exchange beads. Conclusions A thin water layer conductance cell that has a gas-porous membrane wall affords a simple accurate and convenient sensor for the determination of atmospheric SO for concentrations as low as 20 ppb. The relatively large value of the aqueous partition coefficient of sulphur dioxide in water and its acid dissociation constant produce a significant increase in sensitivity as compared with other gases which dissolve in water to form ions and which are likely to be found under ambient conditions.Continuous Carbon Dioxide Sensor The sensors described above operate in a read-on-demand mode. They provide values for carbon dioxide or sulphur dioxide concentration on manually operating a water pulse pump. Fig. 3 illustrates a design for a continuous conductometric analyser which we evaluated for carbon dioxide determinations. It consists of a tube B with porous Teflon membrane dividing the tube into two parts. The gas phase containing carbon dioxide D contacts one side of the membrane and the other side contacts water A. This side of the membrane has deposited conductance electrodes C separated from a mixed bed ion of ion-exchange beads E by a thin fine screen (not shown). If the carbon dioxide concentration in the gas phase is stepped from zero to some finite value a transient change in aqueous carbon dioxide concentration occurs.This transient change in carbon dioxide concentration inside the water phase in the region of the screen is shown in fig. 4. The concentration of carbon dioxide is constant as a good approximation from the bulk of the gas phase through the membrane to the boundary between the membrane and the water. There is a decrease in concentration of carbon dioxide on the water side of the membrane because of the carbon dioxide's partition coefficient. Its gradient is initially very high on the aqueous side of the membrane and the gradient rapidly becomes constant because the ion-exchange beads act as a sink for the diffusing aqueous carbon dioxide. This situation is analogous to the constant 1115 A S.Bruckenstein and J. S. Symanski water-layer screen B ,-> -distance Fig. 4. Transient linearized concentration profiles (A-B) gas phase (B-C) inside membrane pores ; (El E and Em) gradients for carbon monoxide in order of increasing time after a step change in carbon dioxide concentration at the gas phase-membrane boundary. potential generation of a species at a rotating-disc electrode or any other electrode with a steady-state diffusion layer thickness. As the gradient becomes constant lines El E and E in order of time after a step chany in pco2 so does the conductivity. Under these conditions the conductivity varies as P&, (ambient to 1%) just as it does for the equilibrium read-on-demand sensors. The response time for such a sensor was of the order of 95 s for 1% carbon dioxide at a sample gas flow rate of 0.5 dm3 min-l.The sensitivity of this device is larger than the read-on-demand sensor because the effective thickness of the thin water layer is larger and is related to the size of the ion-exchange beads. This sensor configuration has wide application and is readily miniaturized. This work was supported by the United States Air Force Office of Scientific Research under Grant no. AFOSR 83-0004. 3 I. Bergmann Amperometric Gas Monitors. A New Generation paper given at the 30th International References 1 J. W. Harrison D. L. Gilbert P. A. Lawless and J. H. White Development Strategy for Pollutant Dosimetry (1975) p. 74. 2 G. Martinchek Ph.D. Thesis (State University of New York at Buffalo N.Y.1983). Congress of Pure and Applied Chemistry Section 8.B.5 New Electrochemical Sensors September 10 1985). 4 R. K. Kobos S. J. Parks and M. E. Meyerhoff Anal. Chem. 1982,54 1976. 5 M. A. Jensen and G. A. Rechnitz Anal. Chem. 1979 51 1972. 6 T. Morizzumi and Y. Miyahara Transducers '85 1985 International Conference on Solid-state Sensors and Actuators. Digest of Technical Papers. IEEE Catalog no. 85CH2127-9 p. 148. 7 F. Opekar and S. Bruckenstein Anal. Chem. 1984 56 1206. 8 F. Opekar and S. Bruckenstein Anal. Chim. Acta in press. 9 G. L. Holleck J. L. Bradspies S. B. Brummer and L. L. Nelsen Final Report contract no. NAS8-2903 1 December 1973. 10 S. Bruckenstein and W. Sherwood US. Patent 4052478 (1977); U.S.Patent 4166775 (1979). 1 1 J. Kosek Ph.D. Thesis (State University of New York at Buffalo N.Y. 1980). 12 H. Tataria A. A. Schneider and L. E. Martin U.S. Patent 4,184,937 (1980). 13 L. H. Van Kempen and F. Kreuzer Respir. Physiol. 1975 24 89. 14 H. A. Himpler S. F. Brand and M. J. D. Brand Anal. Chem. 1978 50 1627. 15 J. S. Symanski G. A. Martinchek and S. Bruckenstein Anal. Chem. 1983,55 1152. 1116 16 J. S. Symanski Ph.D. Thesis (State University of New York at Buffalo N.Y. 1985). 17 A. Simon and A. Pisctschan Z. Anorg. Alfgem. Chem. 1961 313 281. 18 L. H. Jones and McLaren E. J. Chem. Phys. 1958,28,995. 19 P. A. D. J. DeMaine Chem. Phys. 1957 26 1036. 20 D. M. Kein J. Chem. Educ. 1960 37 141. 21 D. J. Spedding and P. Brimblecombe Atmos. Environ. 1974 8 1063. 22 I. M. Kolthoff Treatise on Analytical Chemistry ed. P. J. Elving and E. B. Sandell (Wiley Interscience New York 1959) part I vol. I p. 432. 23 J. Crank Mathematics of Diffusion (Oxford University Press London 1st edn 1956) chap. 4. 24 R. M. Terraglio and F. P. Manganelli J. Air. Pollution Control 1967 403. 25 International Critical Tables (McGraw-Hill New York 1st edn 1928) vol. 111 p. 259. 26 International Critical Tables (McGraw-Hill New York 1st edn 1929) vol. VI pp. 260-261. 27 A. E. Rabe and J. F. Harris J. Chem. Eng. Data 1963,8 334. 28 H. S. Harned and B. B. Owen Physical Chemistry of Electrolyte Solutions (Reinhold New York 2nd edn 1950) pp. 617 and 589-591. 29 R. A. Robinson and R. H. Stokes Electrolyte Solutions (Butterworths London 2nd edn 1959) p. 463. 30 T. Shedlovsky and D. A. MacInnes J. Am. Chem. Soc. 1935,57 1705. 31 W. Stumm and J. J. Morgan Aquatic Chemistry (Wiley Interscience New York 1970) p. 102. Gas Membrane Electrodes Paper 512010; Received 15th November 1985