Slurry Admittance and Its Effect on Polishing

  • PDF / 209,200 Bytes
  • 6 Pages / 612 x 792 pts (letter) Page_size
  • 19 Downloads / 211 Views

DOWNLOAD

REPORT


F5.1.1

Slurry Admittance and Its Effect on Polishing David R. Evans SHARP Laboratories of America, Inc., Camas, WA 98607, USA ABSTRACT Conventional polishing process variables such as down force, platen speed, slurry flow rate, etc. tend to be both coupled together and indirect. Within this context, one might argue that the net rate of slurry admission or concomitantly, effective volume of slurry between the substrate and pad is a direct process parameter, which is affected by polishing pressure, substrate-pad relative velocity, pad wear and texture, etc. In this work, the role of slurry admittance is investigated by direct control of interference between the pad surface and the carrier retainer ring. Both fixed and dynamic implementations are investigated. Observed patterns of material removal on substrates polished with a fixed orientation are correlated and discussed. INTRODUCTION It should come as no surprise that the language of tribology provides a particularly convenient description of chemical mechanical polishing (CMP). Tribology and wear have long been concerns of conventional mechanical engineers and, indeed, early work by Runnels and Eyman1 treated wafer polishing essentially as a classical slider bearing gliding on a lubricated, rigid surface. The fundamental behavior of this system was established many years ago and predicts that a positive hydrodynamic pressure will appear in the fluid layer causing separation of the relatively moving surfaces. Of course, this is nothing more than conventional lubrication. Unfortunately, when applied to CMP this approach predicts a slurry layer thickness of the order of fifty or sixty micrometers, which seems quite unrealistic in view of actual feature size and vertical relief characteristic of modern integrated circuit manufacturing. In addition, the angle of attack for a classical slider bearing is positive, which when applied to CMP predicts that the wafer carrier will slope downward from leading to trailing edge, i.e., a convergent fluid flow produces a positive fluid pressure. Subsequently, Tichy et al2 developed a tribological description of CMP that accounted for the compressibility of the pad and/or pad asperities. This model predicts that a negative, suction pressure should appear beneath the wafer carrier near the leading edge during polishing. This behavior has, indeed, been observed experimentally. Of course, the existence of negative hydrodynamic pressure is in sharp contrast to the simple slider bearing picture of polishing which requires that any hydrodynamic pressure existing between the wafer carrier and pad surface should be positive. Within this context, positive hydrodynamic pressure does sometimes appear near the trailing edge of the wafer carrier even in the case of a compressible pad. However, this is not always observed. Moreover, suction pressure is consistent with a divergent fluid flow, which is also consistent with a negative angle of attack, i.e., the wafer carrier will slope upward from leading to trailing edge. This behavior is also obs