A Model of Chemical Mechanical Planarization to Predict Impact of Pad Conditioning on Process Performance

  • PDF / 568,346 Bytes
  • 7 Pages / 612 x 792 pts (letter) Page_size
  • 27 Downloads / 209 Views

DOWNLOAD

REPORT


A Model of Chemical Mechanical Planarization to Predict Impact of Pad Conditioning on Process Performance G. Bahar Basim1 and Serkan Kincal2 1 2

Department of Mechanical Engineering, Ozyegin University, Istanbul, 34794, Turkey. Department of Chemical Engineering, Middle East Technical University, Ankara, 06800,

Turkey. ABSTRACT This study presents an effort to couple a wafer removal rate profile model based on the locally relevant Preston equation to the change in pad thickness profile which reflects to post polish profile of the wafers after Chemical Mechanical Planarization. The result is a dynamic predictor of how the wafer removal rate profile shifts as the pad ages. These predictions can be used to fine tune the conditioner operating characteristics without having to carry out high cost and time consuming experiments. The accuracy of the predictions is demonstrated by individual confirmation experiments in addition to the evaluation of the defectivity performance with the varied pad conditioning profiles. INTRODUCTION The purpose of the Chemical Mechanical Planarization (CMP) process is to provide a planarized wafer surface to subsequent photolithography steps. Variations in film thickness can result in local critical dimension (CD) variations which limits on the valuable CD allowances as the dimensions get smaller [1]. These variations are eventually responsible for end-of-the line fails and loss of revenue. Post CMP thickness is driven by various factors that include incoming patterns and film thickness profiles as well as operating conditions during the CMP process. The incoming factors are generally more controlled than the CMP related process variations. Some preventive methods such as printing dummy fields near the edge of the wafer to improve edge thickness profile are standard to most CMP process flow [2]. The difficulty in dealing with CMP process variability is the dependence on the consumable aging. A number of wafer-by-wafer [36] and run-to-run [4,5] controllers have been outlined in the literature but these schemes typically deal with wafer or lot level average thickness issues and be oblivious to shifting removal rate (RR) profiles with consumable life. In order to account for shifts in RR profiles, one needs to have more complex polishing equipment (i.e. tools with multi-zone polish heads) and can employ multivariable control strategies to adjust pressure distribution within the multiple zones to achieve optimum thickness range [7,8]. However, such methods also have limitations, particularly near the very edge of the wafer as well as being quite complex both for implementation and equipment maintenance. This work focuses on the optimization of the conditioner geometry and the sweep profile with the objective of optimizing the pad thickness profile to achieve a desired RR profile. A physical model for pad wear is presented followed by experimental verification of the model through pad profile characterization and conditioner parameter optimization as well as defectivity analyses.

EXPERIMENTAL