Recently we proposed a model for diamond nucleation by means of the su
bstrate bias pretreatment method. In order to verify the effectiveness
of this model, the nucleation process was directly observed by transm
ission electron microscopy (TEM). The Si (100) substrate was treated w
ith methane (concentration 40%) and biased at -100 V; then a cross-sec
tion of this sample was observed using TEM. The growth of tree-like no
n-crystalline carbon from convex parts of the substrate with diamond m
icrograins within them was observed. After the etching of this substra
te with hydrogen plasma for 5 min, the diamond component increased. Fu
rthermore, when this substrate was treated with methane (concentration
0.5%) for 5 min to allow the growth of diamond, 5-10 nm diamond cryst
als with a clearly defined shape grew. The results of evaluation of th
e nucleation process on the basis of this TEM observation were not con
tradictory to the conclusions drawn using our previous model.