Chemical mechanical planarization (CMP) is a semiconductor process which is necessary for multi-layer interconnection structure. CMP pad is a consumable used in the process and with numerous asperities on the surface that wear out by the load applied from the contact with the wafer. Also, it has a patterned wafer, the step height is gradually removed by contact of the asperities with upper and lower layers. The contact state would be different according to the step height reduction. Likewise, depending on the pattern size at the specific step height, the maximum radius of the asperity curvature differs whether it reaches the down area. In this study, the height distribution of asperities was expressed as a function of time and asperity height taking into account the wear of asperities, and based on the Greenwood-Williamson theory, a mathematical model for material removal rate considering pattern size was derived. The consistency of the novel model is verified with the CMP experiment conducted using oxide patterned wafers, and the experimental data were compared with the residual step height using theoretical removal rate. The root mean square error of the step height reduction was 19.84 nm.

Chemical Mechanical Planarization (CMP) is an indispensable process of forming multilayer integrated circuit. However, it is necessary to understand the pattern in order to achieve global planarization. Material Removal Rate (MRR) depends on the pattern density in the actual CMP process and is required to predict the MRR according to density of the pattern. Based on the Preston equation (CMP governing equation), the MRR can be expressed as a product of pressure, relative velocity, and the Preston`s coefficient. Therefore, understanding of pressure distribution acting on the patterned wafer is essential. Pressure distribution depends on contact area between pad asperity and wafer surface. In this study, pressure distribution according to contact mode between asperity and wafer surface where step height exists was analyzed, and the planarization model presented. Finally, a comparison was done between the mathematical model and the experimental data, and the planarization model was verified.

Finite element analysis of CMP process was studied to understand uneven pressure distribution between polishing pad and wafer. Since WIWNU (Within wafer non-uniformity) is mainly influenced by dynamic viscoelastic properties of CMP polishing pad, the dynamic property of the polishing pad has to be understood first for dynamic finite element analysis of the process. To measure viscoelasticity of the polishing pad, time-dependent strain data by load were obtained using a viscoelasticity measurement system capable of measuring deformation by periodic load. Primary and secondary elastic modulus and relaxation time could be achieved for the behavior of the polishing pad by load. Finite element analysis was carried out under the same conditions as viscoelastic measurement. Material properties of the polishing pad were assumed based on results of experiments. By comparing experimental results with analytical results, material properties in the analytical model were modified and FEA was carried out again. It was confirmed that the behavior of the polishing pad by load in the experiment and FEA according to modified material properties were well matched. Through this process, viscoelastic properties of polishing pad were well defined for dynamic analysis of CMP process.