An Approach to Modeling Particle-Based and Contact-Based Wear in CMP
- PDF / 1,115,524 Bytes
- 21 Pages / 612 x 792 pts (letter) Page_size
- 1 Downloads / 162 Views
0991-C06-04
An Approach to Modeling Particle-Based and Contact-Based Wear in CMP Elon Jahdal Terrell, Michael Kuo, and C. Fred Higgs Mechanical Engineering, Carnegie Mellon University, Scaife Hall, 5000 Forbes Ave., Pittsburgh, PA, 15213 Abstract A volume-pixel, or ìvoxelî-based wear model was developed in order to predict featurescale wear in chemical-mechanical polishing (CMP), and was compared to the measured evolution of lithographically-patterned features during full-scale CMP tests. In order to conduct this study, a lithographic technique was used in order to pattern a set of raised square features into Cu-coated silicon wafers.
A two-dimensional (2D) contact
profilometer was used to measure the topography of an isolated feature on the wafer both before polishing and at various intervals throughout the polishing process. For wear modeling, the voxel modeling framework involved the conversion of each of the prepolished feature topographies into a square matrix of cuboids, which allowed for contact mechanics and wear modeling to take place between the interacting cuboids of the sample and a simulated polishing pad. After the predicted wear of the Cu feature was then compared to experiment.
Introduction Chemical mechanical polishing (CMP) is widely used as an intermediate step in the fabrication of integrated circuits and hard disk drives. In regards to integrated circuits fabrication, CMP is commonly used for shallow trench isolation (STI) of electrically conductive films that have been deposited into the trenches of a lithographicallypatterned dielectric layer. Because many integrated circuit devices are multiple layers of features, the CMP process has become crucial towards maximizing the complexity of the device. CMP involves the sample that is to be polished to be mounted onto a rotating wafer carrier and pressed against a rotating pad that is flooded with slurry, as depicted in Fig. 1. The slurry itself is a chemically reactive fluid with suspended abrasive particles. During the CMP process, the chemicals in the slurry serve to weaken the surface of the sample,
while the mechanical interaction between the sample, the polishing pad, and the particles cause abrasive wear of the sample surface.
Down Force
Slurry
Pad Wafer Carrier
Wafer Surface
Figure 1: Diagram of CMP process The desire for predictive capability in CMP has given rise to a number of models that predict the various physical phenomena behind the CMP process. A number of studies have focused on analyzing the slurry fluid hydrodynamics [1] during CMP, assuming that the sample and the polishing pad are fully separated by a layer of slurry film. A few studies have also analyzed the motion and effects of the abrasive particles in the slurry during CMP, using both statistical [2] and Lagrangian-tracking [3] approaches. The material removal rate (MRR) of the sample surface during CMP has also been studied extensively, from both a modeling and experimental standpoint. In regards to CMP wear modeling, studies have been performed that analyze CM
Data Loading...
