Multiscale modeling of the slurry flow and the material removal in chemical mechanical polishing.

摘要

Rapid advances in the microelectronics industry demand continuously a decrease in the device sizes to produce faster and more functional processors, which results in a stringent requirement of global planarization across the die. Chemical mechanical polishing (CMP) is widely adopted in achieving excellent local and global planarization for microelectronic device manufacturing. It has been demonstrated experimentally that the polishing performance is a result of the synergetic effect of both the chemicals and the particles involved in CMP. However, the fundamental mechanisms of material removal and the interactions of the chemical and mechanical effects are not well understood, which limits a better control and improvement of the CMP process.; In this study, the CMP process is modeled in different length scales based on a systematic study of the pad-particle-wafer interactions during polishing. One of the main problems in CMP is polishing non-uniformity caused by wafer-scale slurry transport. In this investigation, therefore, the delivery of the slurries by the polishing pad onto the to-be-polished wafer surface is first studied by solving the Navier-Stokes equation between two eccentrically rotating disks. It is shown that a similar rotational speed of both the wafer and the pad is necessary for a uniform delivery of the polishing slurry. In the feature scale, the shearing effect of the slurry flow on different shapes of features is simulated using the finite element method. Although stress concentration is found always in the corners of the features, no direct evidence about feature breakage off the wafer surface by the slurry flow is seen. To investigate the mechanism of material removal, a micro contact model is developed by combining the chemically formed surface layer with mechanical abrasion. Most of the variables involved in the CMP process, such as solid loading, particle size and distribution, pad modulus, asperity size and distribution, down pressure and rotational velocity, are all included in the proposed model. It provides a reliable tool in predicting the effects of these variables, particularly the properties of the surface layer. Furthermore, the model shows that a balance between the chemical effect and the mechanical effect has to be achieved for optimal slurry performance. Finally, the correlation between the material removal and the coefficient of friction is quantified. The results of this development show that a relatively higher friction coefficient between the abrasive particle and the wafer surface is needed for higher polishing rate. The reported findings in this study provide basic understandings of the CMP process in multiple length scales. A slurry design criterion can be developed to achieve optimal polishing performance.