Electrochemical oxygen gas sensors
AE/AEo values of 0.99. Consequently measurements were made at sufficiently long times that gas-phase diffusion was, presumably, no longer a controlling factor.
Some of the results of Winnubst et a1 (1985) are shown in figure 2. Response times were determined for 10-90% response and showed several interesting features: in particular, under the conditions of the work, the response of electrodes on 25-02- based electrolytes was faster than on Bi,03-based materials; also a Au electrode showed superior performance to Pt.
1000 I T (K-' )
Figure 2. Response time as a function of temperaturefor stabilised Zr02- and BizOpbased electrolytes.Composition (mol %): A. 83 ZrO~-17YOis:0.079Zr02-21 YOi.s;8,80Bi:03-20Er20,; m,70 Bi203-30 Er203; A,60 Biz03-40 Erz03. Pt electrodes were used for all cases except 0,where Au was used (Winnubstet a1 1985). ( ~ ( 0 2 ) 21-100 kPa. (Diagram courtesy Chapman and Hall Ltd. London.)
employing the AC technique Bauerle (1969) postulated that the electrode semicircle (figure l(b)) resulted from the resistance to
( R I ) and
that diffusional processes requiring the addition of a
were influencing the Warburg impedance to
(1983) refined the
analysis and made the based electrolytes the explained in terms of the
following observations. On zirconia- impedance spectra could be fully dissociative adsorption of oxygen onto
Pt electrodes and the diffusion of oxygen atoms phase boundary. With Bi203-basedelectrolytes,
hand, a associated
dominant Warburg impedance appeared with a diffusional process within the bulk
electrolyte. It was speculated that the involved the transport of holes (h').
These findings aid the understanding of the slow response of electrodes on Bi203-basedelectrolytes (8 2.3.3.(i)) and the following is a possible explanation. In response to a change in oxygen partial pressure the stoichiometry of the Bi203adjusts (5 2.3.1.(ii)) and this involves the diffusion of holes (and complementary transport of Vi') throughout the bulk of the electrolyte. This retards the response because the achievement of a stable electrode potential requires the double layer region to come to equilibrium with the gas phase. The diffusion of holes to
the surface may also induce a significant overpotential on the electrode.
(iii) Influence of electrode material. The work of Fouletier et a1 (1974) and of Winnubst et a1 (1985) has demonstrated that the sensor response can be improved by employing either Au or Ag rather than Pt as electrode material. Fouletier et a1 (1976) recorded that Kleitz (1968) had found the overpotential on Ag electrodes to be lower than on Pt electrodes. Issacs and Olmer (1982). on the other hand, reported lower currents on Au compared with Pt for a given overpotential. Badwal et a1 (1984) found the relaxation tinies for urania-scandia electrodes to be one or two orders of magnitude shorter than those for Pt electrodes under similar condit:ms.
Clearly there is scope for improving the response of oxygen sensors by optimising the electrode composition and structure. Such developments will benefit from further work aimed at improving our understanding of the role of the electrode in the oxygen reaction and identifying rate-determining steps and sources of capacitance. The importance of this work lies in developing sensors that retain acceptable response times while operating at lower temperatures,
2 , 4 . Reactiue gases
2.4.1. General formulation
of the model. Only situations
involving non-reactive review (e.g. 0 2 in Ar).
gases have In practical
been discussed so far in this conditions reactive gases, i.e.
gases not in thermodynamic equilibrium, are and the response of gas sensors to such
often encountered mixtures must be
considered. Reactive gases active components, but such satisfactorily. Fortunately,
sometimes contain a plethora of mixtures are too complex to model a relatively small number of
components tractable yet
generally has an overriding influence meaningful analysis to be achieved.
The scheme described below is based upon that presented by Anderson and Graves (198 1) but ha5 been generalised to apply to a three-component system containing gases A, B and C that
interact according to the reaction
xA +yB +ZC
The species reach the electrode surface from the bulk gas by diffusion and because the diffusion layer may be poorly defined or complex it is referred to as a boundary layer. In order to model the system so that qualitative and possibly quantitative predictions may be made, the transport and reaction equations must be set up as shown below. In this model it is assumed that homogeneous reaction in the gas phase does not occur. The gases A, B and C may be monatomic, diatomic etc or monomeric, dimeric etc and thus are written as A,, B b , C, where a, b and c represent the number of the components A, B and C in each molecule. The first three equations quantify surface reactions and state that the rate of change in fractional surface coverage of a species i, B,, depends upon the adsorption and desorption and the forward and reverse rates of reaction. In all
cases isothermal Langmuir-type kinetics are assumed __
r akApb.6: -XkR616$
The next three equations quantify rates of transport from the bulk gas across the boundary layer to the electrode surface
dpa/dt= -kl(PA -PA)-
kAPAB? + k2B;
d p ~ / d=
k 3 ( p -pb) -kBpB6e + k46bB
(24) (25) (26)