Abstract:

Particles in some CMP slurries tend to agglomerate when the slurry is exposed to chemicals and gases, foreign materials and mechanical forces. These agglomerates my adversely impact surface roughness and increase defects during CMP. The impact of many components and handling conditions on slurry health has been investigated extensively in previous studies. The goal of this paper is to quantitatively compare the impact of various components and conditions on large particle counts (LPC) and to identify the key elements that influence slurry health. Another goal is to correlate the changes in LPCs to defects for various CMP processes.

The slurry handling elements considered in this comparison are shipping containers, tanks, valves, different kinds of recirculation pumps, dispense pumps, loop filters, point of use filters and variations in the loop design. Recirculation pumps, loop filters, point of use filters, shipping containers and tanks can have a significant impact on LPC. In a widely used fumed silica slurry, the impact on particles > 2?m varied from a 95% reduction in LPCs after installation a 0.5 um point of use filter to an increase of 570% in LPCs after 100 turnovers with a bellows pump. Loop design can also have a significant impact on slurry health. Strong negative suction pressure in a test circuit resulted in significant increases of LPCs and must therefore be avoided. Shipping containers and daytanks can strongly influence slurry health, too. It is important to use proper materials and minimize gas headspace in the container to reduce LPCs. Other components like valves and dispense pumps demonstrated relatively low impact on LPCs. There are several other factors besides LPCs such as tendency of clogging, service intervals or dispensing precision that influence the selection of these components.

A key question is: Do increased LPC levels (due to these slurry handling elements) correlate with higher CMP defectivity and reduced surface quality? Several independent studies involving different slurries and test substrates have attempted to address this question. In all studies, CMP slurries were extensively handled with different types of pumps before polishing the substrates. Since the test conditions were quite different, it is difficult to directly compare the results. In this paper, the data from various studies have been compiled and normalized to enable their comparison.