Kj. Bachmann et al., HETEROEPITAXY OF GAP ON SI(100), Journal of vacuum science & technology. B, Microelectronics and nanometer structures processing, measurement and phenomena, 14(4), 1996, pp. 3019-3029
In this article, we analyze the kinetics of heteroepitaxial growth of
GaP on Si(100) by pulsed chemical beam epitaxy on the basis of results
obtained by real-time optical process monitoring. In view of the larg
e barrier to epitaxial growth on oxygen or carbon contaminated silicon
surface elements and the low stacking fault energy for CaP, residual
contamination of the silicon surface contributes to defect formation i
n the initial phase of GaP heteroepitaxy on Si, and requires special m
easures, such as surface structuring, to Limit the propagation of defe
cts into the epitaxial film. The control of the supersaturation during
the first 10-20 s of film formation is essential for the quality of s
ubsequent epitaxial growth and is limited to a narrow process window b
etween three-dimensional nucleation and overgrowth at low Ga supersatu
ration and gallium-cluster formation at high Ga supersaturation. Stead
y state heteroepitaxial growth is described by a four-layer stack subs
trate/epilayer/surface reaction layer (SRL)/ambient and, depending of
the source vapor flux, allows for more than monolayer coverage. Under
this special condition of low-temperature CBE, the kinetics of chemica
l reactions in the SRL is composed of homogeneous reactions creating/c
onsuming intermediates that participate in surface reactions including
the incorporation of Ga and P atoms into the epitaxial film. For the
homogeneous part of the surface kinetics the dipole approximation prov
ides an adequate correlation of the changes in the dielectric function
of the SRL to the activities of randomly distributed reactants and pr
oducts. No adequate correlation of the dielectric function exists to t
he concentrations of strongly bonded surface atoms and surface molecul
es. Therefore, quantitative assessments of the heterogeneous kinetics
on the crystal surface cannot use real-time optical monitoring as a re
liable basis. (C) 1996 American Vacuum Society.