Effect of pad break-in time and in-situ pad conditioning duty cycle for porous and non-porous pads in CMP

This paper explores shear force spectral fingerprinting to understand the effect of break-in time and in-situ pad conditioning duty cycle during copper CMP. Polishing is carried out on a system that has a unique ability to measure shear force in real-time. Using Fast Fourier Transformation, shear force data is converted from time domain to frequency domain. In the first set of experiments, porous and non-porous pads are 'broken-in' for different durations. Results show that, under identical polishing conditions, porous and non-porous pads have opposing effects on removal rate and coefficient of friction. Furthermore, shear force variance increases with pad break-in time and unique and consistent spectra emerge which show increasing fundamental peaks for longer break-in times. In the second set of experiments, pad conditioning is performed during copper CMP for 0, 25, 50, 75 and 100 percent of the total polishing time using the non porous pad. Results show that removal rate and the coefficient of friction are not affected b? conditioning duty cycle while shear force variance is found to increase with duty cycle up to 50 percent, after which saturation is reached. Further investigation of shear force spectra indicates spectral similarities among the 50, 75 and 100 percent duty cycles thus suggesting similar conditioning outcomes for these three processes. Furrow density Monte Carlo simulations of various break-in and conditioning methods noted above are performed. Results indicate that the fundamental peaks seen in the experimental results are most likely generated by the pad conditioner. This work underscores the importance of pad break-in to achieve early steady-state polishing through optimum break-in time and to extend pad life through an optimum conditioning duty cycle.