Annealing experiments have revealed a significant distinction in the amorphous-to-crystalline transition between bulk and micrometre-sized glassy Mg-silicates on the one hand and nanometre-sized Mg-silicate smoke particles on the other hand. In detail, micrometre-sized particles and bulk sheets of glassy transformed into orthoenstatite whereas the and smokes changed into forsterite, tridymite and amorphous silica. Weak indications of forsterite and tridymite formation in the glass powder have been observed in the XRD spectra only after annealing at 1000 K up to 50 hours. The timescale of crystallization of the silicate glass powder exceeds that of the smoke-like samples - showing the grain size dependence of heterogeneous nucleation.
At 1000 K, the smoke of composition changed from the amorphous state into forsterite in approximately one day. The excess appeared as crystalline tridymite and amorphous . Because there had been a very Mg-deficient small particle species, almost fully amorphous particles are still contained in the smoke.
smoke has been found to evolve more quickly. At 1000 K, the annealing time was proved to be about 1 h. In contrast to the Mg-poorer smoke, the diffusion rate of Mg-atoms does not limit the growth rate of the nanocrystals because of the near forsterite stoichiometry of most of the particles. Amorphous and crystalline as well as MgO are also present in the smoke, because the smaller nanoparticles had been a Mg/Si-ratio below 2.
For amorphous nanoparticles, extrapolations proved the annealing time to surmount years at temperatures below 1000 K. At 1220 K, crystallization takes place after about 4.5 h. As was already noted above that might be true only for pure silica whereas Mg-silicates incline to tridymite formation.
Calculations of AGB star outflows showed that the temperature of silicate dust grains exceed 1000 K for mass losses higher than M _ (Sogawa & Kosaza 1999; Gail & Henning 1999). The outflowing gas decreases its temperature from the condensation point (maximum of the nucleation rate) in an outflow expanding with v on the timescale of years (Sogawa & Kosaza 1999). Using the activation energy obtained for smoke ( K), the annealing times can be estimated for lower temperatures. Even at 800 K, homogeneous nanoparticles are expected to develop a short-range order in about 3 yr. With decreasing Mg/Si-ratio, the activation energy and, therefore, the annealing times increase. Pure silica condensates with an activation energy of K may only at mass loss rates exceeding develop an at least short range order. Silica may be a candidate to identify a feature at about 21 m that recently has been observed in infrared spectra of evolved oxygen rich stars (Molster et al. 1999a).
© European Southern Observatory (ESO) 2000
Online publication: December 15, 2000