Fh. Stott et A. Atkinson, THE MODELING OF GROWTH STRESSES DURING HIGH-TEMPERATURE OXIDATION, Materials at high temperatures, 12(2-3), 1994, pp. 195-207
The development of stresses in the scale during the high-temperature o
xidation of metals can have important consequences for the long-term p
rotection of components in service, particularly if this leads to scal
e failure, allowing access of the environment to the metal surface. Su
ch stresses may result from externally applied deformation of the scal
e/substrate system, from thermal effects due to differential thermal c
ontraction/expansion between the scale and the metal, from geometrical
effects or from the intrinsic scale-growth process itself. In this pa
per, a review is presented of some of the models that have been publis
hed to account for scale-growth stresses and thermal stresses, with em
phasis on quantitative estimation of such stresses and the resulting s
trains in the scale, in the metal and in the metal/scale interface. Al
though most models of intrinsic scale-growth stresses are based on the
volume change as metal is converted to oxide in a confined location w
ithin the scale or at the scale/metal interface, there is little conse
nsus on how these stresses develop. Several quantitative, but incompat
ible, models have been proposed for outward-growing scales in which ox
ide is assumed to form in the scale grain boundaries or at the base of
these boundaries following inward transport of oxidant, although othe
r qualitative models have stressed the importance of pores or cracks a
s paths for such species. Most models for the development of growth st
resses use elastic analysis and neglect plasticity and creep effects,
which may not be justified for a slow-growing scale. A qualitative mod
el has been suggested that can account for the presence of a stress in
such a scale without invoking formation of oxide within a constrained
location. Rather, it results from climb of a fraction of the intrinsi
c misfit interfacial dislocations into the metal to annihilate vacanci
es at the scale/metal interface, followed by adjustment of the spacing
of the remaining dislocations to maintain epitaxy. The importance of
relaxation has been demonstrated in the model of stress generation dur
ing the oxidation of silicon. Reasonable quantitative models are now a
vailable to describe the development of geometrically induced stresses
and thermal stresses. The latter are generally based on elastic analy
sis, which is justified for the high strain rates induced on rapid coo
ling or heating cycles.