Abstract : Reactions of decomposition of solids implicate two distinct processes : nucleation and growth. Bibliographic study shows that kinetic modelling is based on three type of model: either the nucleation is instantaneous, or the nucleation takes place in the bulk (Avrami's laws), or the nucleation takes place at the surface. Microscopic observation showing that for decomposition of solid the nucleation occurs at the surface, both first models are not allowed. In the case of models involving surface nucleation and isotropic growth (Mampel-like model) or anisotropic growth, two kinetic constants are useful : the areic frequency of nucleation and the areic reactivity of growth , which are supposed to be independent of time in isothermal and isobaric conditions. Previous studies of reactions of decompositions or between a solid and a gas have shown that the nucleation process is more difficult to quantify than the growth process.
In order to have a better understanding of the nucleation process, we have chosen a “model” reaction which is the dehydration of Li2SO4,H2O.
Experiments in thermogravimetry, at 80°C under controlled water vapour pressure, on single crystals, show that kinetic curves α(t) have a sigmoïdal shape and present an induction period. Observations with optical and electronic microscopes allow to visualize both surfaces and bulk of single crystals during dehydration, and show that the nucleation occur at the surface and that the nuclei grow isotropically toward the centre of the crystal.
A kinetic model based on surface nucleation and isotropic growth allows to account for the experimental curves. The calculations, using Monte-Carlo simulations, take into account the real shape and dimensions of each single crystal. From the good agreement between the experiments and the model, we obtained the values of and for each single crystal. We have shown that the areic frequency of nucleation is very dependant on the surface state of the single crystal.
A specific model for nucleation, based on the appearance of water vacancies and their agglomeration up to the formation of a nucleus, allows to obtain distributions of the date of appearance of the first nucleus. These calculated distributions can be qualitatively compared to experimental distributions of induction periods. These comparisons together with in situ imaging let think that edges and faces have different reactivities in the nucleation process.
Due to the induction periods experimentally measured and a kinetic model describing more precisely the beginning of the (t) curves (for ≤ 0,004), it is possible to determine a relation between the date of appearance of the first nucleus and the areic frequency of nucleation . This theoretical relation has been validated by the experiments.