A lab-scale nonflowing reactor was built to study chemical vapor depos
ition reactions. Mass spectrometry is used to follow reaction pathways
and to determine instantaneous reaction rates throughout film growth.
In each experiment, the kinetic rate dependence on concentration for
a wide range of concentrations is observed as reactants convert to pro
ducts. This method of obtaining kinetic data is efficient in terms of
sample loading, gas usage, and time, since over 200 instantaneous rate
/composition pairs can be determined from one 30-min deposition. Becau
se the rate is determined from gas-mass balance, rather than film-thic
kness measurements, an unlimited number of rate studies can be made on
one sample. As a test case, the SiH4 reduction of WF6, used to deposi
t tungsten during integrated-circuit production, was investigated in t
he 0.64-L nonflowing laboratory reactor. Gas compositions were measure
d 2 mm fi om the growing surface, throughout time, with a mass spectro
meter equipped with a capillary sampling tube. Tungsten was deposited
on the 95 degrees C surface and SiH3 was the primary silicon fluoride
reaction product for most tested conditions. A multiple-regression ana
lysis of 1,975 instantaneous composition/rate pairs gives orders of 1.
22 in silane, 0.27 in hydrogen, and -2.17 in WF6. The ratio of SiF4 to
SiHF3 stays low and constant until the gas becomes silane-rich. The e
volution of the instantaneous rate over time implies that a minimal le
vel of thermal activation of the reactive gases is necessary for the d
eposition to be surface-rate-limited. Preliminary heat-transfer models
of the wire substrate imply that heat transfer to the gas phase is in
the knudsen regime.