Angle-resolved x-ray photoelectron spectroscopy (XPS) and laser-induce
d thermal desorption (LD), combined with laser-induced fluorescence (L
IF) detection, were used to study the etching of polycrystalline Si (p
oly-Si) and single crystal Si(100) in high density (1-2 x 10(11) ions/
cm(3)), low pressure (0.5-10 mTorr) Cl-2/HBr-containing, helical reson
ator plasmas. The XPS measurements on both unmasked Si(100) and fine-l
ine masked poly-Si samples were performed after the sample was etched
and then transferred under high vacuum from the plasma reactor to the
ultrahigh vacuum (UHV) analysis chamber. The LD-LIF measurements on un
masked Si(100) samples were performed in-situ during etching. In these
experiments, XeCl excimer laser pulses rapidly heat the Si surface to
near the melting point, causing thermal desorption of SiCl. The tail
of the same laser pulse excites SiCl to the (B-2 Sigma(+)) state in th
e gas-phase near the surface. The subsequent fluorescence signal from
this state is proportional to Cl-coverage, verified by XPS. In HBr-con
taining plasmas, analogous LD-LIF detection was used for SiBr, providi
ng a measure of Br coverage. The major findings of these studies are t
hat Si surfaces rapidly become covered with a stable (in vacuum), satu
rated layer of about 2 monolayers of halogens during plasma etching. T
he layer consists of silicon mono-, di-, and tri-halides. In Cl-2 plas
mas, the Cl coverage increases with increasing ion energy, but is near
ly independent of pressure (0.5-20 mTorr). Chlorination occurs rapidly
with respect to the time required to etch one monolayer, at pressures
as low as 0.5 mTorr. Consequently, the etching rate is limited by the
ion flux, and not the neutral flux under these conditions. In mixed C
l-2/HBr plasmas, the coverages of Cl and Br are simply proportional to
the total respective halogen content of the feed gas. Other implicati
ons for etching mechanisms are discussed.