Electrochemical oxygen gas sensors
3.2. Response time The theoretical current-time characteristic, derived from data presented by Crank (1956), for the case where the limiting process is diffusion in the barrier adjacent to the cathode is shown in figure 9. A 90% response to a step change in px, would be predicted for t90=0.3 12/D.Substituting in the typical values 1=2 mm, D= 100 mm2s - l reveals 90 = 12 ms. Thus the response is rapid and not likely to be rate-determining.
Figure 9. The theoretical response of a limiting-current sensor to a change in sample gas pressure when the diffusion barrier is controlling the behaviour. I , l o and lc are the measured current at time t, at t =0 and t= c respectively, D is the diffusion coefficient and 1the pore length.
3.3. Reactive gases Very little has been reported in this area. Dietz, using Pt electrodes with 02-CO mixtures (containing sufficient O2 to allow conversion of all the CO to COz), found that the sensor measured the equilibrium oxygen value. However, it was not clear whether the diffusion barrier or the Pt electrode had catalysed the reaction.
When reducible gases other than O2 are present in the sample gas then these may also contribute to the reduction current provided a sufficiently large overvoltage is applied to the cathode (Dietz 1982, Saji et a1 1985). The reduction curves for HzOand CO2 occur in the same potential range and cannot be separated.
3.4. Degradation Additional modes of deterioration compared with potentiometric devices relate to the diffusion barrier and result in progressive changes in characteristics. Two possible mechanisms that result in such changes are the following.
Blocking of pores in the barrier by dust in the sample gas.
Structural changes of a porous ceramic. Dietz observed
changes in devices employing a thin layer of ceramic (20-30 pm thick) applied over an electrode to produce a barrier with very low porosity.
3.5. Operating region The amperometric oxygen sensor can only be used where the
oxygen-reductant This is a serious
(or air/fuel) ratio is limitation and arises
lean of because
stoichiometry. potentials are
essentially referred to that of the anode in the passing through the stoichiometric point the
sample gas. On anode potential
changes by more than 500 mV (see figure 8 of Maskell and Steele 1986). Typically the amperometric sensor is operated at 0.8 V. On passing through stoichiometry to ,< 1 (rich region) the overvoltage applied to the cathode is then approximately 1.3 V (relative to an air reference) and this is now sufficiently cathodic to reduce CO2and HzOresulting in current flow (Saji et a1 1984). However, there is no indication to the user that the system is now sub-stoichiometric and the current might be interpreted as indicating an oxygen surplus (i.e. A > 1). This problem can be solved by using the pump-gauge devices discussed in Q 5 .
4. Coulometric sensors In coulometry a given volume of gas is quantitatively converted by electrolysis and the partial pressure determined from the charge passed. There are many ways of performing such a measurement but the one described below is sufficient to explain the principle. Heyne (1976) described the device shown schematically in figure 10. It comprises an electrochemical pump and a leak into an enclosed volume. Initially a constant current is applied to the pump to remove oxygen rapidly from the cavity much faster than its leakage rate through the aperture.
Enclosed volume V,
Figure 10. Coulometric sensor (Heyne 1976).
When almost all the oxygen is removed the voltage on the pump rises sharply. The current is then discontinued and a period follows during which oxygen leaks into the enclosed volume from the surrounding gas. This period is long enough so that the inner and outer gases closely approach equilibrium. The pump current is then re-applied and the cycle repeated. The oxygen partial pressure in the sample gas can be calculated by
invoking Faraday's law as follows =RTq/4FV1 PI
where q is the charge passed and VI the enclosed volume. If the
current applied is constant ( I )thenp, is given by
The measured value t , is directly proportioned to the partial pressure, p , , which is an advantage in the lean region (compare the logarithmic response of the potentiometric sensor, equation (4)).Interestingly, unlike the amperometric sensor,the calibration is independent of the capillary characteristics and depends only upon I, VI and T. The sole requirement is that the waiting time is greater than the diffusion time which is much greater than