This study investigated the effect of different pad surface micro-textures on the tribological, thermal and kinetic attributes during copper chemical mechanical planarization (CMP). Different micro-textures were generated by two different chemical vapor deposited (CVD) diamond-coated conditioner discs (i.e. Disc A and Disc B). Results showed that while pad temperature and removal rate increased with polishing pressure and sliding velocity on both discs, Disc B generated consistently lower removal rates and coefficients of friction (COF) than Disc A. To fundamentally elucidate the cause(s) of such differences, pad surface contact area and topography were analyzed using laser confocal microscopy. The comparison of the pad surface micro-texture analysis indicated that Disc A generated a surface having a smaller abruptness (λ) and much more solid contact area which resulted in a higher removal rate. In contrast, Disc B generated less contact areas and COF. A two-step modified Langmuir–Hinshelwood model was employed to simulate copper removal rates as well as chemical and mechanical rate constants. The simulated chemical to mechanical constant ratios indicated that Disc A produced a more mechanically limited process under all of the polishing conditions tested.

The effect of pad surface characteristics on the thermal, tribological and kinetic attributes of copper CMP was investigated. Three CMC D100 pads with very different surface micro-textures were generated using three very different CVD-coated diamond conditioning discs. Pad samples were collected after polishing and analyzed for their surface contact area and topography using confocal microscopy. The contact area, contact density, and asperity height increased with increasing conditioner aggressiveness.Copper removal rates and pad surface temperatures increased with increasing polishing pressure and sliding velocity for all pads albeit in a non-Prestonian manner. The pad generated by the most aggressive disc caused the highest removal rates yet it showed the lowest overall coefficient of friction (COF). The tall asperities, open pores, and adequate contact of this pad produced a greater removal rate than pad surfaces with shorter asperities and glazed pores produced by the less aggressive discs. Trends in COF, temperature and removal rate were successfully simulated using a two-step modified Langmuir-Hinshelwood model which also yielded values for the chemical and mechanical rate constants. The simulation results indicated that the process was chemically limited for all polishing conditions, and that the process became even more chemically limited as P × v increased.

In chemical mechanical planarization (CMP), Stribeck curves are normally constructed by plotting average coefficient of frictions (COF) against the average Sommerfeld number. Consequently, traditional Stribeck curves fail to provide a full explanation of the lubrication phenomena simply because COF and polishing downforce can fluctuate significantly due to stick-slip phenomena and transient instabilities caused by polishing kinematics and consumables. This study introduces a new method for rapidly generating an "improved" Stribeck curve (i.e. Stribeck+ curve) that shows a more complete tribological picture of the process. The method significantly reduces the consumables and time required to obtain the curve compared to traditional means. Results of the Stribeck+ curve are consistent with individual tests using several different consumables combinations. All copper CMP Stribeck+ examples clearly indicate the lubrication mechanism and transitions thereof between different polishing conditions. Variability in COF as well as a much wider range in v/P are also explored.

The effect of two different commercially-available CVD-coated diamond conditioning discs on copper CMP was investigated tribologically and kinetically with the intent of correlating pad surface micro-texture to polish performance. Data analyses were particularly focused on in-situ shear and normal force measurements taken during polishing at several pressures (P) and sliding velocities (V), including Fast Fourier Transform (FFT) analysis of the data. One of the two discs consistently resulted in greater variances of shear force with increasing P and V indicative of increased pad-slurry-wafer stick-slip events. This trend correlated to higher values of coefficient of friction (COF) and removal rate (RR). By dividing the shear force variance by the normal force variance, a new parameter, termed directivity, was used to further emphasize the improved performance of the above-mentioned disc. Lastly, the tribological and kinetic data, combined with data regarding the pad's micro-texture and its extent of contact with the wafer, showed that at all values of P and V, higher values of directivity correlated to more contacting pad asperities, a greater asperity density and a thinner micro hydrodynamic lubrication layer, thus highlighting the potential importance of directivity in future studies.

The Stribeck+ curve was successfully applied to silicon dioxide chemical mechanical planarization processes to characterize the tribology of such processes under different process conditions and consumables. Results showed that the Stribeck+ curve was capable of rapidly determining and differentiating the tribological mechanism among all cases studied in this manuscript. The Stribeck+ curve could indicate process stability as shown by the spread of the COF vertical clusters. The Stribeck+ curve also confirmed a previously known effect that the greater the ratio of pad's up-features to the total pad area, the greater the probability of wafer hydroplaning. This work underscore the importance of a new method for determining an “improved” Stribeck curve (referred as the “Stribeck+ curve”) while dramatically reducing the amount of consumables and time required to obtain the curve through traditional means.