Additive-assisted nucleation and growth of thin films: Controlled experiments and multiscale simulations

The present work has served to develop a computational and experimental methodology for developing an understanding of the role of additives in the electrochemical nucleation, growth, and coalescence of thin metal films on wafers coated with a thin resistive substrate consisting of a second metal. The technological application which was chosen to drive the fundamental investigations of this thesis is the electrochemical deposition of on-chip copper interconnects on a wafer coated with a thin metal barrier layer which prevents diffusion of copper into the wafer components. One such barrier layer material is gold; for the copper-on-gold system, simulations indicated that additives that served to slow the deposition rate (high chloride and high poly-(ethylene glycol) (PEG)) promoted uniform thickness distributions along a wafer-scale substrate, but lower copper coverage at the center of the wafer. An additive plating regime that optimized the performance of uniform film thickness and rapid film coalescence was found at medium to high chloride concentrations (120 to 200 ppm) and medium PEG concentrations (750 to 2500 ppm). Chloride concentrations less than 50 ppm yielded poor performance of both deposition goals. Another such technologically-relevant barrier layer material is ruthenium; the copper-on-ruthenium system followed additive trends similar to gold; high chloride and high PEG concentrations yielded more uniform films, and low chloride and PEG concentrations yielded more rapid film coalescence. The best additive mixture for forming thin metal films, however, may be different from the best mixture needed for other electrodeposition operations that occur simultaneously. An additive regime that optimized the performance of uniform film thickness, rapid film coalescence, and trench infill was found at low chloride (< 5 ppm) and both low (< 100 ppm) and medium PEG (250 to 325 ppm).;An electron-beam-induced-deposition (EBID) method was used to fabricate a region of uniformly repeated surface features of Pt on a Au film. Such controlled experiments allowed for the statistical analysis of how Pt seed cluster spacing affected the trade-off between seed cluster growth and wild nucleation during copper electrodeposition. As the center-to-center seed spacing was increased from 200 to 400 nm, the Cu nuclei density on the Au substrate increased from 2.4 x 107 to 1.5 x 108 cm-2 and the average seed cluster radius decreased from 51 to 26 nm.;An electron-beam-lithography (EBL) method was used to fabricate a region of uniformly repeated surface features of Cu on a Au film. Experiments were performed to evaluate how the (1) additives, (2) applied potential, and (3) seed spacing affect the trade-off between seed cluster growth and wild nucleation during copper electrodeposition. Additive compositions that served to suppress the deposition rate (high chloride (10 ppm) and high PEG (3000 ppm)) slowed the deposition rate on the Cu seeds and increased the level of wild nucleation on the Au substrate. Decreased levels of PEG (3 ppm) and chloride (0.1 ppm) served to decrease the level of wild nucleation and also to slightly increase the deposition rate on the Cu seeds. Elevated levels of SPS served to increase the deposition rate slightly and thus increase the Cu seed growth rate, while decreased levels of SPS depressed the rate of Cu seed growth. The SPS concentration had little impact on the level of wild nucleation. The level of chloride concentration was found to be able to mediate the extent of seed growth versus wild nucleation. Low levels of chloride (0.1 ppm) yielded growth only at the Cu seeds (average seed size 72.5 +/- 7.7 nm after 1 second of deposition) with zero wild nuclei. Progressively higher levels of chloride, 1 and 10 ppm, yielded lower deposition rates at the Cu seeds (average seed size 65.1 +/- 7.1 and 24.4 +/- 5.1 nm for 1 and 10 ppm Cl-, respectively) and proportionally higher levels of wild nucleation (6.6 +/- 0.1 x 10 9 and 9.8 +/- 0.4 x 109 cm-2 (200 nm seed spacing) for 1 and 10 ppm Cl-, respectively). Island dynamic simulations for copper deposition onto a patterned gold substrate were compared to experimental observations. The computational results were qualitatively similar to the additive trends observed experimentally. However, the simulations indicated that the additive composition had little impact on the seed growth rate, which contrasts with experimental observations. The integration of the current experimental data with improved mathematical models for predicting nucleation and growth dynamics holds promise for improved engineering procedures for the design of additive systems. (Abstract shortened by UMI.)...