Comparison Of Instruments Used For Measuring Concentrations of Large Particles (>1um) in CMP Slurry
BOC Edwards, 322 Lake Hazeltine Drive, Chaska, MN 55318 2
Mark Litchy and Donald C. Grant
CT Associates, Inc., 10777 Hampshire Ave. S., Bloomington, MN 55438
The “health” or “quality” of chemical mechanical polishing (CMP) slurry is an important factor in main- taining CMP process consistency. The health of a slurry is defined by a number of factors, including its inherent stability and level of external contamination. Several common properties can be measured to assess slurry health, including specific gravity, pH, weight percent solids, ionic contamination level, zeta potential and particle size distribution [1-2]. One of these properties, the particle size distribution (PSD), was evaluated in this study. There are several commercially available instruments capable of measuring PSDs in CMP slurry. Some instruments characterize the distribution over the entire range of particle sizes. Other instruments measure only the large particle tail of the PSD, which was the focus of this evaluation.
The presence of “large” (? 1 ?m) or agglomerated slurry particles may result in scratches on wafers, thereby reducing yields [1-2]. Several factors may alter the concentration of large particles in CMP slurry, such as delivery of the slurry through pumps, tanks, valves, elbows and other piping components. In addition, pressure differences, flow rates, stagnant regions, and exposure to air can affect the PSD . Inevitably, CMP slurries are exposed to many or all of these factors during IC manufacturing processes. Since slurry quality is sensitive to many variables, the need for on-line evaluation of slurry health, espe- cially the particle size distribution, is critical. Two analyzers capable of making these measurements were evaluated in this study.
There are many issues involved with integrating a PSD analyzer into an on-line slurry metrology system. Some of these issues are ease of integration for software and hardware, chemical compatibility and cross- contamination, and optimization of parameters for both the analyzer and final metrology unit. In addition, there is an issue with how to interpret the data. Should one look at comparative results, absolute results or both? This question is answered by evaluating different analyzers and interpreting their data.
CTA Publication #36: Proceeding of the Fifth International Conference on CMP Planarization for ULSI Interconnection (CMP-MIC), Santa Clara, CA, March, 2000