Modeling of Pattern Dependencies for Multi-Level Copper Chemical-Mechanical Polishing Processes
- PDF / 41,740 Bytes
- 6 Pages / 612 x 792 pts (letter) Page_size
- 15 Downloads / 157 Views
MODELING OF PATTERN DEPENDENCIES FOR MULTI-LEVEL COPPER CHEMICAL-MECHANICAL POLISHING PROCESSES Tamba Tugbawa, Tae Park, Brian Lee, Duane Boning, Microsystems Technology Laboratories, MIT Rm 39-567, Cambridge, MA; Paul Lefevre, International SEMATECH, Austin, TX; Lawrence Camilletti, Conexant Systems, Newport Beach, CA. Abstract We propose an integrated contact mechanics and density-step-height model of pattern dependencies for the chemical-mechanical polishing (CMP) of multi-level copper interconnects, and show preliminary comparisons with experimental data for the overburden copper removal stage. The model uses contact mechanics to correctly apportion polishing pressure on all sections of an envelop function that reflects the long-range thickness differences on the chip, or region of interest. With the pressure over the entire envelop known, the density-step-height part of the model is then used to compute the amount of material removed in the local “up-areas” and “down-areas”. This model shows promise in accurately and efficiently predicting post CMP copper and dielectric thicknesses across an entire chip. Introduction Copper CMP is now a multi-step polish process where different polish process settings and different consumables are used in each step [1]. For any model to accurately predict the post CMP thicknesses across a chip, it must accurately predict the evolution of the thicknesses in each polish step. This means that such a model should take into account the incoming long range thickness variation caused by copper plating technology, and must also account for pressure redistribution caused by density induced thickness variation as the polishing progresses. Previously, we developed a density-step-height model for copper CMP processes [2]. The model identifies three intrinsic stages in copper CMP processes, and expresses the removal rates in each stage in terms of effective density and step-heights. It is capable of rapidly simulating thickness evolution on an entire chip, for a given calibated CMP process. However, it does not account for long range thickness variation introduced by the plating technology used, and does not properly redistribute pressure due to changing long range thickness variation as the polishing progresses. Hence it can lead to inaccurate prediction of thickness evolution during CMP and can thus give inaccurate post CMP thicknesses. Contact mechanics in combination with Preston’s equation has been used to model dielectric and copper CMP processes [3-4]. These models account for the initial long range thickness variation caused by copper plating technology, and they also redistribute the pressure to take into account the changing long range thickness differences as the polishing progresses. However, they can be computationally prohibitive or inaccurate when simulating an entire chip or a small section of a chip, depending on the discretization size used. In copper CMP, the feature sizes on the lower metal levels are very small. Hence the discretization size for a full chip simulation m
Data Loading...
