WO2004031758A1

WO2004031758A1 - Electrochemical oxygen sensor

Info

Publication number: WO2004031758A1
Application number: PCT/GB2003/004308
Authority: WO
Inventors: Anthony Desmond Shand Tantram; Martin Jonathan Kelly; Andrew Peter Millar
Original assignee: City Technology Limited
Priority date: 2002-10-07
Filing date: 2003-10-06
Publication date: 2004-04-15
Also published as: AU2003267669A1; GB0223202D0

Abstract

An electrochemical oxygen sensor comprises a sensing electrode (6) and a consumable counter electrode (3), a porous separator (4) between the electrodes, and liquid electrolyte absorbed in the separator, all mounted in a housing (10). The housing (10) is sealed except for a gas phase diffusion barrier (7) to allow gas access to the sensing electrode (6), and a vent (1) opening into a part of the housing on the opposite side of the separator to the sensing electrode. A membrane (2) allows gas to pass into and out of the housing (10) via the vent (1) but prevents the passage of liquid.

Description

ELECTROCHEMICAL OXYGEN SENSOR

This invention relates to liquid electrolyte fuel cell oxygen sensors which employ a gas phase diffusion barrier, typically in the form of a capillary or highly porous membrane material. Such sensors have been commercially available for some years, and the beneficial properties of this type of diffusion barrier have been described previously (eg GB1571282, and ^{^}Liquid Electrolyte Fuel Cells ' , Ch 6 in ^{^}Techniques & Mechanisms In Gas Sensing ' , Eds Moseley, Norris & Williams, Adam Hilger, 1991) .

Such oxygen sensors typically employ a precious metal cathode to reduce oxygen which diffuses into the sensor, whilst the balancing reaction is provided by the oxidation of a consumable anode component, for example formed of high surface area lead. The .individual and overall cell reactions can therefore be summarised as follows;

Sensing electrode 0₂ + 2H₂0 + 4e^~ → 40H^~ Counter electrode 2Pb + 40H^" - 2PbO + 2H₂0 + 4e^~

Overall 2Pb + 0₂ → 2PbO

The liquid electrolyte is commonly a concentrated ionically conducting aqueous solution, for example of potassium hydroxide. A key aspect of the sensor design to permit full utilisation of the consumable anode is to provide sufficient free volume within the sensor housing so that the expansion of the anode during oxidation can be accommodated. When using aqueous electrolytes, the free volume also acts as an expansion space allowing the accommodation of volume increases resulting from operation for extended periods in moist conditions where water vapour will gradually diffuse into the cell through the diffusion barrier.

When a sensor is initially manufactured and primed with electrolyte, electrolyte wicks up into the separator which covers the anode and provides electrical isolation between the electrodes whilst permitting ionic transport between them. The free space inherent in the design is present as one or more bubbles in the region below the separator, but the precise position is likely to depend on sensor orientation and other considerations.

Since the separator material has a significantly lower pore size than the lead wool mass, it will become completely flooded with electrolyte and therefore produce a sealed, constant volume system in the lower part of the sensor. As a result, reductions in temperature will cause a reduction of the gas pressure in this lower region which will continue until either

(a) the outer sensor casing or seals fail and allow the external atmosphere to equalise the pressure, or

(b) the differential pressure across the separator is sufficient to force a bubble of gas from the upper region of the sensor into the lower part.

Given the relative pressures involved, scenario (b) is much the more likely to occur in a well sealed and properly constructed device. For example, a device filled at 20C and 1 atmosphere pressure when cooled to -15C will experience a change in internal pressure of -0.12 atmospheres. Experience shows that this is sufficient to exceed the so- called "bubble pressure' of many commonly used separator materials when flooded with electrolyte. When such a bubble is forced through the separator from the upper portion of the sensor, gas is drawn in suddenly from the outside atmosphere in order to maintain equal pressures on either side of the diffusion barrier. This process occurs despite the presence in most designs of a so-called bulk flow membrane, which is intended to prevent unwanted influences upon the capillary action caused by minor eddies and other flow effects. However, the pressure differentials generated by temperature changes of the type considered here are sufficient to overcome the protection afforded by this membrane, which is not intended to overtly restrict the communication of the sensor with the outside atmosphere under normal conditions and as such is comparatively porous. It will be clear that the problem of such bulk flow is limited to sensor designs incorporating gas phase diffusion barriers such as capillaries or highly porous membranes, where intermolecular collisions dominate. Types employing solid state diffusion through dense membranes, or Knudsen membranes where wall collisions dominate over intermolecular collisions, do not allow bulk flow per se .

The sudden ingress of air results in a rapid rise in the oxygen concentration in the vicinity of the sensing electrode, which is normally maintained very close to zero by the combined -action- of -the electrode and the diffusion^' limiting barrier. In practise, a significant transient current may be generated which leads to undesirable fluctuations in the output of the device, not representative of the true oxygen content of the environment. Once the bubble has been forced through, the separator will re-flood with electrolyte in the affected areas. Normally, such breakthrough would be expected to occur through relatively small areas of the separator having larger pores .

The whole process as described is intimately dependent upon the local conditions across the membrane. The pressure required to force a bubble through will depend on the pore size of the system and this in turn may vary with the packing between the separator and the lead anode. Orientation changes and even the volume changes of the anode during operational conversion to the oxide can alter these features of a particular sensor during its life. As a result, the transient phenomenon described here is rather difficult to reproduce reliably, but has been observed in a significant minority of sensors and as such is a considerable disadvantage in operational conditions.

Analogous problems do not occur in oxygen sensors of broadly the same type which use a solid membrane as the diffusion barrier, since such designs do not permit any bulk movement of gas in the same way as a capillary. In

- these sensors, the oxygen moving into the cell must first dissolve in and percolate through' the membrane, a process which slows the equalisation process sufficiently that there are no significant problems with transient signals.

However, it is well known that such sensors provide an output proportional to the partial pressure of oxygen in the environment rather than the concentration signal of the capillary type, and in some circumstances, concentration is a greatly preferred measurand.

In principle, a sudden increase in ambient pressure can produce similar conditions within the sensor and result in transient outputs. The reverse process, caused by a sudden drop in pressure or by moving into an environment at a significantly higher temperature than the one in which the sensor has equilibrated, has also been observed. For example, a sensor taken from 20C to 55C will experience an internal overpressure of 0.12atm. However, in this case, any gas escaping from within the sensor by bubbling through the separator comprises, by virtue of the electrochemical action of the system, almost entirely of nitrogen. Although some disruption to the normal activity occurring on the sensing electrode is to be anticipated, this is usually comparatively minor and the effect of transient signals so generated is not as great a concern as the reverse effect.

However, increased internal cell pressures may have other undesirable effects, for example by contributing to electrolyte leakage through any weak points in the housing. In extreme circumstances, the top cap of the sensor case may be pushed off if sufficiently high pressures build up. It must also be borne in mind that many sensors will experience cycling through the temperature (or pressure) conditions leading to such internal pressure changes many times in their operational life (which can be 2 years or more) . The repeated stressing which thus occurs can also lead to failure via a variety of modes. It is therefore clear that a solution to the problems caused by internal pressure changes within such capillary sensors must be sought .

One approach is to employ a separator whose "bubble pressure' is sufficiently high that the pressure differences across it can be maintained for periods long enough to allow pressure equalisation via other mechanisms (eg by the transport of dissolved gas) . A material which has been employed in this application is cellophane, which, once saturated with electrolyte, allows ionic and diffusive flow to occur but prevents bulk gas flow. In sensors where the separator is well sealed around its perimeter, the problem -of ^■ transient signals caused by internal pressure drops can generally be overcome. However, there are considerable difficulties in providing a reliable and complete seal between the membrane and the outer housing of the sensor which ensures that no apertures capable of transmitting bubbles as described above are present.' Gaps as small as 20 microns in diameter may be sufficient to allow the passage of bubbles generated by the 20C to .-15C step considered earlier, and in some operational circumstances, sensors of this type may be required to operate through even larger temperature transients without producing erroneous outputs which derive from the transient ingress of oxygen. Furthermore, it has been observed that materials like cellophane which offer relatively high bubble pressures may not have the long term stability which is essential in order to maintain constant sensor performance throughout the product life. For example, degradation and cracking can occur which degrades the quality of the pressure seal. This compromises the extent to which the pressure differentials can be resisted and may also lead to other unwanted effects, manifested as output instabilities.

In any event, even when carefully and properly implemented, this solution fails to address the fundamental issue of pressure differentials being established across the separator. An alternative approach is to incorporate some form of pressure relief into the design of the sensor. For example, US3767552 describes a solid membrane oxygen sensor incorporating a flexible member. The internal volume of the cell is completely filled with electrolyte, and the purpose of the flexible component is to allow hydrostatic pressure relief by increasing the available internal volume. A small vent connects an expansion chamber behind the flexible member to the external atmosphere, but the system would operate effectively as a means of pressure relief if the flexible portion formed part of the outer casing. -In principle, such an approach could be adopted for capillary sensors containing a mixture of lead, electrolyte and free space as described earlier, but the reliable engineering and sealing of a flexible member as described is extremely challenging for long life sensors designed to operate across a wide range of ambient conditions in rugged environments. Therefore, it is not a favoured means of addressing the problem. In principle, a much simpler solution would be- to provide a vent in the rear of the sensor so that facile pressure equalisation across the separator could occur without the passage of gas from one side to the other. Vents are known in some liquid electrolyte fuel cell toxic gas sensor designs to allow equalisation of the pressure within the electrolyte reservoir. (See, for example, GB-A- 2235050.) Here, free space is incorporated to allow for electrolyte expansion by dilution due to moisture absorption from the atmosphere and other environmental changes. Were this not done, then hydrostatic pressure increases could force electrolyte out through the weakest point in the assembly, for example through the sensing electrode backing tape. However, in such systems the only connection from the reservoir to the active electrode regions is generally via the wick transporting electrolyte to the inter electrode space. There is no requirement to incorporate free space between the electrodes as there are no consumable components which may undergo volume changes over the life of the sensor. The use of an oxygen reduction counter electrode and Pt-air reference systems requires a certain amount of oxygen access, but this is very small and is irrelevant from the point of view of pressure build up and release. (See GB2049952.)

Such vents in toxic gas sensors are not employed for the purpose of pressure release from the largely anaerobic free space which exists within the electrode structure of oxygen sensor anodes, and in capillary barrier oxygen sensors the venting approach would appear to be • -unfavourable - (and against accepted- custom and practise) for two main reasons .

(a) The target sensed species is oxygen and therefore all access of oxygen to the sensor should ideally be controlled through the capillary diffusion barrier. If significant amounts of oxygen which have not traversed the diffusion barrier are able to reach and react on the sensing electrode, it is not possible to unambiguously relate the electrode current to the external conditions.

(b) Access of oxygen to the lead anode would allow parasitic consumption of the material, thereby shortening the operational life of the device. Oxygen is capable of being reduced at some sites on the lead surface and this process is balanced by an equivalent degree of lead corrosion elsewhere in the anode mass. The rate of such anode self discharge is governed in the first instance by the rate of access of oxygen to the surface and this will be quite different in regions where the lead is covered by electrolyte (which transports oxygen relatively slowly) and in regions where the it contacts the gas voids and free space within the structure (where diffusion will occur much more rapidly) .

In accordance with the present invention, an electrochemical oxygen sensor comprises a sensing electrode and a consumable counter electrode, a porous separator between the electrodes, and liquid electrolyte absorbed in the separator, all mounted in a housing, the housing being sealed except for a gas phase diffusion barrier to allow gas access to the sensing electrode, and a vent opening into a part of the housing on the opposite side of the separator to the sensing electrode; and means for allowing gas to pass into and out of the housing via the vent but preventing the passage of liquid. In practice, we have found that the size of vent required to reduce the pressure differential problems to manageable proportions (optionally when used in conjunction with an appropriate separator material offering appropriately high bubble pressure) can be employed without seriously compromising the overall sensor performance.

Oxygen accessing the sensor through such a vent must traverse a relatively long and tortuous path which does not offer an easy diffusion route to reach the sensing electrode. Therefore, any interference with the signal generated by oxygen diffusing into the cell via the capillary barrier is minimal . The parasitic consumption of the (lead) anode by self discharge as a result of oxygen ingress through the vent certainly occurs at a far slower rate than the reduction of oxygen at the (precious metal) cathode and hence of the lead oxidation by the intended counter electrode reaction. Although there will be a minor effect of such parasitic effects upon the overall cell life achieved, this can readily be compensated (if necessary) by increasing the volume of the anode. Typically, the sensing electrode will be located above the counter electrode, the vent being located in a bottom wall of the housing. However, the vent could be located in a side wall of the housing but it is preferred that the vent is spaced from the counter electrode.

Some examples of electrochemical oxygen sensors according to the invention will now be described with reference to the accompanying drawings, in which :-

Figure 1 is a schematic cross-section through a first example;

Figures 2 and 3 illustrate graphically the response of oxygen sensors of the type shown in Figure 1 but without and with a vent respectively; and,

Figure 4 is a view similar to Figure 1 but of a second example .

The sensor shown in Figure 1 comprises a housing 10 having a capillary aperture 7 defining a gas phase diffusion barrier which is covered with a bulk diffusion membrane 8, made of porous PTFE . A sensing electrode

' --"catalyst β of a precious- metal -known -in- the art is mounted on a PTFE backing tape 5 and is located within the housing

10 adjacent the capillary 7. Beneath the sensing electrode catalyst 6 is provided a conventional separator 4 typically made of a glass fibre type material with good wicking properties. The excellent wicking properties of the separator 4 ensure that a continuous meniscus of liquid exists around the perimeter of the component, effectively producing a complete seal. In this example, the separator

4 is not sealed to the casing of the sensor, allowing very much simpler manufacturing arrangements than those required when relying solely on a high bubble pressure separator to prevent pressure related effects of the type discussed earlier.

Although not shown in the diagram for clarity, the separator 4 will contact the electrode catalyst 6.

Beneath the separator 4 and in contact with it is provided a consumable lead anode 3 defining a counter electrode. Finally, a vent 1 is provided in the bottom wall of the housing 10, the vent 1 being covered on its internal face by a highly porous PTFE membrane 2 or the like which allows easy diffusion of gas for pressure relief purposes 5 but which has a high water initiation pressure to prevent electrolyte being forced through the aperture as the internal cell pressure rises. Tapes of the type commonly used to back the gas diffusion sensing electrode have been found to be quite adequate for this purpose and may be heat 0 sealed into position using conventional methods. We have additionally found that the vent is preferably placed in an area of the base which is unlikely to be compressed by the anode mass, since this can compromise the facile passage of gas through the aperture and so inhibit the pressure relief 5 performance. In more severe cases, the membrane covering the vent might even be punctured by the lead anode, leading

- ■ --to electrolyte leakage and-cell failure. Such compression may occur on assembly of the sensor or as a result of the anode expansion which occurs as oxidation proceeds. 0 Therefore, in Figure 1, the vent is shown as off centre as the anode assembly tends to bottom out in the central region of the housing in this design.

By way of example, we have constructed sensors with capillary 7 diameters of the order of 100 microns and vent 5 1 diameters of 1mm. The use of relatively large vents 1 may lead to some operational drawbacks, for example when running in hot dry environments over extended periods . We have shown that greatly reduced vent sizes, for example in the range 140 to 100 microns in diameter, are still capable 0 of providing an effective solution to the overpressure problem for devices with 100 micron diameter capillary access, even in cases where the additional membrane is omitted from the sensor assembly. Furthermore, these reduced diameter vents do not lead to water balance 5 difficulties under likely sensor operating conditions.

It has been confirmed that, as expected, the effect of oxygen reaching the cathode 6 via the vent 1 is insignificant in comparison with the flux produced by ambient air through the capillary access. Sensors of this type built without capillaries (so that the only oxygen route into the device is via the vent) generate currents of ~400nA as compared with ~0.26mA obtained in a device with a standard capillary access.

To demonstrate the effectiveness of the invention, the traces in Figures 2 and 3 show the response of similar sensors with (Figure 3) and without (Figure 2) the addition of the vent when subjected to temperature cycling between approximately -32C and 50C over a period of 15000 seconds.

The noisy and comparatively long lived transient currents generated on cooling the sensor are clearly shown in Figure

2, but the additional incorporation of the vent in the device tested in Figure 3 has eliminated the problem, leaving only the anticipated steady state temperature

- coefficient- inherent in such sensors before compensation is applied.

Figure 4 illustrates a modified form of the Figure 1 example in which an additional, high bubble pressure cellophane separator 9 has been incorporated between the separator 4 and the electrode catalyst 6. In an alternative arrangement, the separators 4 and 9 could be reversed. However, the preferable arrangement is that shown where the separator 4 is retained to provide extra protection for the cellophane 9 which may become brittle after extended use and could be punctured by the lead anode 3. As with the separator 4, the separator 9 is not sealed to the sensor housing which again leads to simpler manufacturing arrangements .

In practice, the trace of Figure 3 will be substantially the same for a sensor constructed according to Figure 4.

Therefore, both vents and additional membranes can contribute to the solution of the overpressure problem. Vents with diameters much greater than those of the main capillary barrier size do not compromise the diffusion control of the device. But much smaller vents (with diameters approaching the size of the main diffusion barrier) also offer effective pressure relief even in the absence of additional membrane and do not lead to water balance problems. The incorporation of such vents is therefore the simplest and favoured design improvement to address the issue.

Claims

1. An electrochemical oxygen sensor comprising a sensing electrode and a consumable counter electrode, a porous separator between the electrodes, and liquid electrolyte absorbed in the separator, all mounted in a housing, the housing being sealed except for a gas phase diffusion barrier to allow gas access to the sensing electrode, and a vent opening into a part of the housing on the opposite side of the separator to the sensing electrode; and means for allowing gas to pass into and out of the housing via the vent but preventing the passage of liquid.

2. A sensor according to claim 1, wherein the sensing electrode is located above the counter electrode, the vent being located in a bottom wall of the housing.

3. A sensor according to claim 1 or claim 2, wherein the vent is-spaced-from the counter • electrode..-•

4. A sensor according to any of the preceding claims, wherein the vent has a diameter in the range 50-150μm, preferably 100-140μm.

5. A sensor according to any of claims 1 to 3 , wherein the vent has a diameter up to substantially 1mm.

6. A sensor according to any of the preceding claims, wherein the gas phase diffusion barrier is a capillary.

7. A sensor according to claim 6, wherein the capillary has a diameter in the order of lOOμm.