The good resistance of 20/25 steel to high tempera­ture oxidation in CO2 was one of the main reasons why it was chosen for use in the Windscale AGR. In high temperature (>500°C) CO2 the steel forms a protective oxide, the composition of which ranges from Fe304 on the outside to mixed spinels closest to the metal substrate. The rate of scale formation is controlled by the diffusion of chromium from the metal through the oxide so that the kinetics are es­sentially parabolic, with a rate of scale thickening which gradually decreases as the scale thickens. Since the process is diffusion-controlled it is markedly tem­perature dependent, the rate constants having activa­tion energies of about 240 kj/g or more (Simpson and Evans, 1985 [13]). As the chromium diffuses into the scale, the underlying metal becomes corre­spondingly depleted. Since it is the chromium which largely promotes formation of the protective spinel oxide, it follows that local disruption of the scale, by mechanical interaction or temperature cycling, can expose a chromium depleted layer to the oxidising coolant. Oxidation will now result in the formation of a non-protective oxide and, as this grows, the oxide/metal interface will move into the body of the metal until the chromium level reaches about 18.5% when a protective oxide will once again form (Lobb and Evans, 1983 [14]). This, in outline, is the me­chanism for the formation of pits.

As with the oxides of uranium, the thermal expan­sion coefficient for the clad oxide is quite different from that of the cladding steel. Consequently, it will be appreciated that once an oxide has formed on the surface of the clad, large decreases in pin temperature will produce differential thermal contraction between

the two which will cause the oxide to have very high compressive stresses. If the temperature decrease is large and/or the oxide is thick, the elastic energy associated with these stresses can be sufficient to overcome the interfacial bonding energy between the oxide and the substrate, causing the oxide to spall. Laboratory tests commonly show such effects and, because of this, devices (central inertial collectors — CICs) which collect such spalled oxide were installed in ail the UK’s CAGRs. The main reason for doing this was to minimise contamination of the boilers by the highly radioactive oxide dust.

Operation of the reactors to date has shown little need for the CICs. The main reason for this seems to be that thermal contraction of the clad in the most affected positions is limited by its permanent contact with the fuel pellets. Since the UCb pellets and the clad oxide have similar coefficients of thermal expan­sion, the oxide is not so highly stressed as might have been suggested by laboratory tests (on unsupported steel coupons). Even so, it is quite possible that if, in future, peak clad temperatures are allowed to rise, the greater oxide thicknesses will allow the spalling threshold to be crossed.