The measurements of the specific heat of silicon were carried out using the
DSC method. The single-crystal silicon, containing 5.5x10(17) cm(-3) of ox
ygen, was the subject of our study. The measurement error was 5%. The tempe
rature range of the measurements was 350-770 K. The scan rates were 4 or 16
KI min. Slow heating at 4 K/min shows an abnormality in temperature depend
ence of the heat capacity in the range 550-680 K. The recorded curve resemb
les a lambda -shaped curve characteristic of 2nd-order phase transitions. A
t a higher scan rate, C-p (T) is a monotonic function, which is identical t
o the data recommended in reference books. We assumed a possible mechanism
of the observed phenomenon. When a crystal of silicon is heated, stressed z
ones arise along subgrain and domain boundaries due to the anisotropy of th
ermal expansion. The new high pressure phase of silicon with the structure
of white tin (or Si-II) can be formed in these zones. The appearance releas
es stress and stops the further growth of precipitates. According to our es
timate, the concentration of the newly-formed phase is about 5 vol.%. Heati
ng at higher rates does not cause similar phenomena.