r d

con dso ure del roa 56 rec te L © T DowExperimental Electroplating and cyclic voltammetry ~CV! studies were per-formed in an acid copper sulfate electrolyte containing 17.5 g/L copper metal, 175 g/L sulfuric acid, and 50 mg/L chloride ion added as HCl. For filling tests, the electrolyte contained 1.0 mL/L of a proprietary organic accelerator additive Ultrafill A2001 ~Shipley Company, Marlborough, MA! of the sulfide type ~similar to SPS!, and 25.0 mL/L of a proprietary organic additive suppressor Ultrafill S2001 ~Shipley! of the polyglycol type ~similar to PEG!. Voltammetry was performed using a PAR273 potentiostat with a standard calomel electrode ~SCE! reference electrode. Each scan was initiated at the rest potential ~;150 mV vs. SCE! of copper in the acidic copper sulfate electrolyte and cycled first to a more ca-thodic voltage and then back to rest potential. The electrode used was a Pt disk of approximately 0.16 cm2 area. shown. The large features show that deposition at the bottom cor-ners of the trenches is accelerated relative to the rest of the surface. The enhanced deposit thickness in the bottom corner is more than would be expected from a ‘‘geometric leveling’ ’ effect that would be obtained by conformal deposition. For the smaller features, a bump is seen to emerge above the center of the features. Theory Diffusion-adsorption mechanisms2-11 do not predict the void-free deposits of Fig. 2, the bumps, or the significantly accelerated depo-sition at the bottom corner. We propose an alternative mechanism that is consistent with these observations. Specifically, we hypoth-esize a model based on a transition from an inhibited to a noninhib-ited state at the feature bottom to explain the superfilling effect. The key aspect of the model is that a decrease in the amount of surface area available for additive adsorption leads to a temporary decreaseA Superfilling Model that