In this study we determined the relationships that exist between the coefficient of friction (COF) and platen motor current (PMC), and also between shear force and PMC at highly non-steady-state conditions during chemical mechanical planarization (CMP). For the 12 cases studied (8 for copper, and 4 for dielectric CMP including one as a cobalt buff step), we found that real-time PMC data closely mimicked shear force data as evident from high average values of correlation coefficient and coefficient of determination for all 12 cases (0.913 and 0.835, respectively). As for COF vs. PMC, average correlation coefficient and coefficient of determination for Cases A through D were 0.949 and 0.900, respectively. However, for the remaining 8 cases, correlation coefficients ranged from zero to 0.425, with coefficients of determination ranging from zero to 0.180. The reason behind the strong correlations seen in the first 4 cases, and the lack of any correlation seen in the remaining 8 cases, were explained based on the particulars of the dominant tribological mechanism for each case.
Correlations between shear force and platen motor current (PMC), as well as those between coefficient of friction (COF) and PMC were investigated for various tungsten and interlayer dielectric (ILD) chemical mechanical planarization (CMP) cases where the processes were highly non steady-state. We chose to initially focus on non-steady-state conditions because we believed the relationships among shear force, COF and platen motor current to be clearer as opposed to steady-state conditions. Shear force, normal force and PMC data were collected from twelve different Stribeck+ curves at an acquisition frequency of 1,000 Hz and analyzed in order to determine any emerging trends. For the 12 cases, involving 8 pre-polished blanket CVD tungsten and 4 silicon dioxide blanket wafers, it was discovered that PMC closely mirrored shear force as evidenced by a high average correlation coefficient (0.955) and coefficient of determination (0.916) obtained from all runs. For COF vs. PMC, the average correlation coefficient and coefficient of determination for all cases were 0.758 and 0.608, respectively. These average values were dragged down by 5 cases in which the dominant tribological mechanism was found to be "boundary lubrication" where COF changed minimally with pseudo-Sommerfeld number.
Following up our earlier work that highlighted the relationship between shear force (SF) and platen motor current (PMC), in parallel with the relationship between coefficient of friction (COF) and PMC for various tungsten and interlayer dielectric (ILD) chemical mechanical planarization (CMP) cases at non-steady-state conditions, we explored whether or not PMC could be used as a reliable indicator instead of SF and COF at steady-state conditions. For the 12 cases studied, 72 distinct steps were analyzed. It was determined that PMC somewhat mirrored SF and PMC for long time (i.e. 10 s or longer) intervals after data averaging and applying a trend matching algorithm. SF and PMC trends matched only about 64% of the time (ranges between 45% to 85%) for all 72 steps, while PMC and COF trends matched 62% of the time ranging between 42% and 85%. PMC-SF and PMC-COF correlations were fairly poor at 1-sec time intervals as evidenced by much lower percent match values. Such poor correlations proved that at small time intervals, PMC was not sensitive enough to capture important information regarding myriad fluid dynamics and tribological phenomena and the instantaneous stick-slip occurrences encountered in CMP.
In a follow up from our earlier work that focused on highlighting the relationship among platen motor current (PMC), shear force (SF), and coefficient of friction (COF) in non-steady-state cases, we examined whether PMC was a reliable indicator that could be used in place of SF and COF data at steady-state conditions. For the 12 cases studied, we examined 60 distinct steady-state steps from their associated Stribeck+ curves. Data averaging, coupled with a trend matching algorithm, showed that for large time spans at steady-state, PMC was a reasonably good indicator of both SF and COF, as 68.5% of the time, PMC and SF trends matched (ranges between 62% to 86%) for all 60 cases. Regarding PMC and COF, we observed trend matching for 68.8% of the time (ranges between 62% and 86%) for all 60 cases. However, correlations between SF and PMC, and also between COF and PMC were found to be very poor (in most cases, non-existent) since at small timescales PMC was not sensitive enough to capture the instantaneous stick-slip occurrences and other important tribological and fluid dynamics phenomena present in CMP.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.