Real-time coefficient of friction ͑COF͒ analysis is used to identify the tribology of the process and to determine the useful life of the polishing pad as a function of key processing parameters. Results indicate that minor COF changes affect interlayer dielectric ͑ILD͒ removal rate and pad life. Slurry flow rate is shown to modulate average COF without altering the tribological mechanism associated with a particular range of Sommerfeld numbers. This finding is used to isolate and study the effect of COF on pad life. Regardless of process tribology, results indicate that pad life is increased with increasing slurry flow rate. Due to the effect of COF on removal rate, the absolute magnitude of ILD removal rate is shown to be highly dependent on the tribological mechanism. The work illustrates the complexities involved in reaching a compromise between removal rate, slurry flow rate, high volume manufacturing, and pad life.In ILD chemical mechanical planarization ͑CMP͒, it is critical to establish potential correlations between pad life, slurry flow rate, coefficient of friction ͑COF͒, removal rate, and the tribological mechanism. With recent improvements in semiconductor equipment, today's cost of ownership focus has shifted to consumables such as slurries and pads. Increased frequency of pad changes adversely affects tool availability because pad change, pad break-in, and pad requalification can take away hours of otherwise useful production time from a polishing tool. In addition, large amounts of slurry and monitor wafers are used every time pads are changed, thus increasing total cost. Because pads account for roughly 15% of the overall cost of ownership ͑COO͒, 1 increasing pad life would significantly increase tool utilization and decrease the overall COO for a given process. In this paper, tests were conducted to examine how process tribology affects polyurethane pad life.
ExperimentalA scaled version of a Speedfam-IPEC 472 polisher was used for all tests. The polisher and its associated accessories are described in detail elsewhere. [2][3][4][5] To measure shear force between the pad and the wafer, a sliding table consisting of a bottom plate bolted to the ground and an upper plate bolted to the polisher was used. As contact was made between the wafer and the pad, the upper plate would slide relative to the bottom plate due to friction generated between the pad and the wafer. The degree of sliding was quantified by coupling a load cell to the two plates that would output voltage to a data acquisition board. The apparatus was calibrated to report the force associated with the particular voltage reading. Tribological and removal rate data was collected using Freudenberg FX-9 perforated pads. Prior to data acquisition, the pad was conditioned for 30 min using Fujimi's PL-4217 fumed silica slurry. In all cases, the abrasive content for pad conditioning and experimentation was held constant at 12.5% silica by weight. Conditioning consisted of using a 100-grit diamond disk at a pressure of 0.5 psi, rotational velo...