Novel slurry formulations and associated mechanisms for chemical mechanical polishing of polysilicon, silicon dioxide and silicon nitride films in microelectronic applications.

摘要

In the first part of this thesis, four abrasive-free aqueous polycationic solutions---poly(diallyldimethyl ammonium chloride) ( PDADMAC), poly(dimethylamine-co-epichlorohydrin-co-ethylenediamine) , poly-(allylamine), and poly(ethylene imine) (PEI)---at only 250 ppm concentration were developed for polishing poly-Si at a removal rate (RR) of > 500 nm/min with negligible SiO2 and Si3N4 RRs (∼ 0 nm/min). The pH-dependent interactions of these and two other polycations with poly-Si, SiO2, and Si 3N4 films and IC1000 polishing pads were extensively studied using zeta potentials and contact angle measurements. The variation in the RR magnitude and dependence on pH among the different polycations is related to the relative charge density of the polycations as well as the films being polished. Based on the zeta potential data and relative bond strengths, it is suggested that the strong polycation-mediated bridging interactions between the polarized and weakened Si-Si bonds of the poly-Si surface and the polyurethane IC 1000 pad are responsible for the high poly-Si RRs. Addition of silica/ceria abrasives to the PDADMAC solution had minimal impact on poly-Si, SiO2, and Si3N4 RRs.;The dramatic differences in the effects of two different polishing pads (IC1000 and Politex) on the RRs of three dielectric films using aqueous abrasive-free solutions of PDADMAC were identified and investigated. For example, with a 250 ppm of aqueous PDADMAC solution, poly-Si RR is 1 nm/min with a Politex pad but is about 500 nm/min using an IC1000 pad. The difference in the RRs is attributed to differences in the strengths of PDADMAC-mediated bridging interactions between the pads and the substrates.;The poly-Si wafer pattern response was also investigated using abrasive-free aqueous solutions of 250 ppm PDADMAC and PEI. Marked differences have been observed between these two solutions as well as when compared to conventional particle-containing slurries. These differences arise from the differences in the molecular weights of the two polymers, the differences between removal mechanisms, as well as the reduction in mechanical interaction due to the removal of particles from the system. Finally, it was shown that post-CMP cleaning of adsorbed PEI from the three dielectric surfaces and the polishing pads can be achieved by water adjusted to pH=2 but clearing the PDADMAC adsorption appears to be more challenging.;Part II. Achieving excellent planarity with no silicon nitride loss, excellent within-die thickness variations and within-wafer uniformity and eliminating the defects caused by the abrasives are still unresolved and remain critical for 32 nm and smaller STI features. To achieve these stringent challenges at these geometries requires a CMP slurry which can polish the overburden silicon dioxide at a rate of 100 nm/min or higher and the underlying silicon nitride at a rate that is much less than 1 nm/min. Potentially, these challenges can be achieved by understanding and adjusting the slurry chemistry while simultaneously minimizing abrasive loading. In the second part of this work, I will describe nearly abrasive-free ceria-based slurries containing various additives that can achieve silicon dioxide removal rates (RRs) of > 200 nm/min while suppressing the silicon nitride RRs to less than 1 nm/min.;Furthermore, a generalized mechanism was developed for explaining the observed suppression of the silicon nitride RRs in the presence of these additives as well as the observed selectivities between silicon dioxide and silicon nitride polish rates. This is more detailed than various mechanisms already documented in literature. These additives strongly adsorb on the silicon nitride surface and hinder the silicon nitride hydrolysis thereby suppressing the silicon nitride removal rates. The relative strength of the adsorption on a nitride surface is shown to be much stronger than flow-induced shear. In contrast, these additives weakly bind to the silicon dioxide surface and are easily removed by the polishing pad and by the ceria abrasives even at very low concentrations and, hence, do not affect silicon dioxide polish rates. (Abstract shortened by UMI.)