Characterization of chemical polishing materials (monomodal and bimodal) by means of acoustic spectroscopy
It is shown that acoustic spectroscopy can sense the presence of a small sub-population of large particles in a concentrated dispersion of much smaller particles. The detection limit can be as low as a single one micron particle per 100 000 particles of 100 nm size. In order to achieve this high sen...
Saved in:
Published in | Colloids and surfaces. A, Physicochemical and engineering aspects Vol. 158; no. 3; pp. 343 - 354 |
---|---|
Main Authors | , |
Format | Journal Article |
Language | English |
Published |
Elsevier B.V
1999
|
Subjects | |
Online Access | Get full text |
Cover
Loading…
Summary: | It is shown that acoustic spectroscopy can sense the presence of a small sub-population of large particles in a concentrated dispersion of much smaller particles. The detection limit can be as low as a single one micron particle per 100 000 particles of 100 nm size. In order to achieve this high sensitivity the acoustic spectrometer must be able to measure ultrasound attenuation with a precision of 0.01 dB/cm/MHz. It is shown that DT-1200 Acoustic Spectrometer (Dispersion Technology, NY, USA) meets this requirement over a frequency range of 3–100 MHz. A model dispersion with a known bimodal particle size distribution (PSD) was created by adding a small amount of larger particles to a stable slurry containing only small particles. Dupont Ludox™ (silica 30 nm) and Cabot SS25 (silica 63 nm) were used to represent typical chemical-mechanical polishing (CMP) slurries. Two samples of Silica Geltech silica (0.5 and 1.5 micron) were used to model the target aggregate particles. It is shown that the attenuation spectra measured with the DT-1200 has sufficient sensitivity that it can detect the larger particles at concentrations as low as 2% relative to the total solid content of the slurry (12% wt). Moreover, the bimodal PSD calculated from the attenuation spectra are consistent with the known composition of these mixed model dispersions. Importantly, a software error analysis can correctly select either a bimodal distribution or a lognormal representation of the test samples. |
---|---|
ISSN: | 0927-7757 1873-4359 |
DOI: | 10.1016/S0927-7757(99)00155-7 |