The interaction of mechanistic experiments and detailed models are greatly
improving our understanding of the mechanism of diamond growth by chemical
vapor deposition. Methyl-radical models typically predict growth rates on (
111) planes that are much smaller than experiments, unless contributions fr
om acetylene in nucleating new layers are included. These models predict ra
ther different contributions of methyl radicals and acetylene to growth on
(100) vs. (111) planes. On the other hand, other models predict rapid inter
-conversion of adsorbed hydrocarbons and surface migration, and equivalence
of the behavior of methyl radicals and acetylene (apart from a sticking co
efficient) might be expected. We have nucleated and grown mum-sized diamond
particles at 800 degreesC in a flow-tube apparatus that permits growth fro
m only methyl radicals or acetylene in atomic hydrogen, in contrast to the
complex mixture of species found in a normal reactor. Growth from methyl ra
dicals only produced cubo-octahedral crystals with an a value (root3 x the
ratio of growth rates in the [100] and [111] directions) near 1.8, indicati
ng that the absence of acetylene is not a significant impediment in nucleat
ing new (111) planes. Diamond growth from pure acetylene produced octahedra
(alpha = 3), indicating that (100) growth is much more facile than (111) g
rowth in the absence of methyl radicals, and the (111) facets had a high co
ncentration of contact twins. The implications of these results for the mec
hanism of diamond growth are discussed. (C) 2001 Elsevier Science B.V. All
rights reserved.