Paper PS1-WeA2
Effect of Cu Contamination on Recombination of O Atoms on Plasma Conditioned Surfaces

Wednesday, October 22, 2008, 2:00 pm, Room 304

Advanced dual-damascene integration schemes require patterns to be transferred through organic masks and anti-reflection coatings as well as through inorganic dielectric materials. During plasma etching of dielectrics down to underlying Cu layers, sputtered Cu may coat the reactor walls, causing process drift. In a CCP etch reactor, it was found that the photoresist (PR) etch rate drops for an O₂-based plasma process, after the plasma reactor has been exposed to Cu-containing wafers. Also the local PR etch rate is depressed opposite a Cu coupon attached to the upper electrode. The decrease in etching rate suggests that Cu could cause a decrease in the etchant concentration in the plasma, perhaps due to an increase in the heterogeneous atom recombination rate on the chamber walls. We have therefore studied the effects of traces of Cu on O recombination on an oxygen plasma-conditioned surface, using the spinning wall technique. With this method, a cylindrical spinning substrate (in this study, stainless steel coated with oxygen, as well as silicon from etching of the discharge tube) is rotated through differentially pumped chambers, allowing the surface to be periodically exposed to a 5mTorr, 600W O₂ plasma, an Auger spectrometer, and a Cu PVD source. With no Cu on the surface, a pressure rise was observed in the Auger chamber, due to desorption of recombined O₂. This pressure rise was converted into an absolute desorption flux through calibrations and was measured as a function of substrate rotation frequency. With separate measurements of absolute O-atom impingement fluxes, a Langmuir-Hinshelwood recombination coefficient of γ_O = 0.10 was derived for the steady-state, Cu-free surface, coated with a layer with an atomic composition of Fe:[Al+Si]:O ~ 1:2:9. This surface was then exposed to a Cu dose of ~10¹¹ cm^-2, depositing a small fraction of a monolayer (~5 x 10¹⁴cm^-2), which is well below the detection limit by Auger analysis (~0.1 monolayers). This trace amount of Cu caused a 13% increase in γ_O The surface was further exposed to the Cu doses of ~2, 3, and 8 x 10¹¹ cm^-2. Each added dose causes γ_O to increase accordingly. At 8 x 10¹¹cm^-2 γ_O = 0.15. Much larger doses (~3 x 10¹³ cm^-2) resulted in detectable Cu on the surface and a γ_O of ~0.3.