Résumé
J. Phys. Chem. B 103, 4104 (1999) We investigate to what extent the specific heat of amorphous silica can be
calculated within the harmonic approximation. For this we use molecular
dynamics computer simulations to calculate, for a simple silica model (the BKS
potential), the velocity autocorrelation function and hence an effective
density of states g(nu). We find that the harmonic approximation is valid for
temperatures below 300K but starts to break down at higher temperatures. We
show that in order to get a reliable description of the low frequency part of
g(nu), i.e. where the boson peak is observed, it is essential to use large
systems for the simulations and small cooling rates with which the samples are
quenched. We find that the calculated specific heat is, at low temperatures
(below 50K), about a factor of two smaller than the experimental one. In the
temperature range 200K <= T <= T_g, where T_g=1450K is the glass transition
temperature, we find a very good agreement between the theoretical specific
heat and the experimental one.