M. Gauch et G. Quentel, EXPERIMENTAL ROUGHNESS EXCITATION OF SURFACE ELECTROMAGNETIC-WAVES AND THEIR DETECTION BY ELLIPSOMETRY, Thin solid films, 233(1-2), 1993, pp. 53-57
The dynamic aspects of the growth process of the (0001) basal face of
a large cadmium single crystal have been carefully investigated by rea
l-time ellipsometry, over a large range of crystal temperature and sup
ersaturation, during in situ vapour growth experiments. At the same ti
me, the ellipsometric response can be correlated with the morphologica
l evolution of the growing face observed with a differential interfere
nce optical system that is very sensitive to local surface slope varia
tions. From numerous experiments, it has been proven that the face gro
ws at low supersaturation by the birth and spread of spiral layers ori
ginating at the outcropping of screw dislocations. The behaviour of th
e resulting step-trains looks like an evolutive periodic roughness whi
ch induces a very sensitive ellipsometric response. Analysis of the fl
uctuating signals reveals growth instabilities at different scales (py
ramid or step interaction) but some abnormal large variations (deltaDE
LTA almost-equal-to 20-degrees), often observed even at constant super
saturation and temperature, are interpreted in terms of a surface roug
hness excitation of surface electromagnetic waves (SEWs). It has been
shown that an optical grating can be used for SEW production and the e
llipsometric detection consists in plotting the parameters (cos DELTA,
tan PSI) vs. the wavelength lambda at a fixed incident angle THETA(i)
. Such spectra exhibit a very large variation in the phase DELTA and a
deep minimum for the ellipticity tan PSI (resonant response). In our
case, a step-train looks like an optical grating surface with a variab
le periodicity and thickness depending on the growth conditions (tempe
rature, supersaturation, step interaction, ...); in this case, it is t
hen possible to obtain the resonant condition with a fixed wavelength
and a critical interstep distance because of the very large possible a
djustment of the interstep distance.