Chemical mechanical planarization (CMP) slurry quality...

Details Chemical mechanical planarization (CMP) slurry quality... Chemical mechanical planarization (CMP) slurry quality...

1999-04-22

2001-08-14

Pham, Hoa Q. (Department: 2877)

Optics: measuring and testing

For size of particles

Reexamination Certificate

active

06275290

ABSTRACT:

FIELD OF THE INVENTION
The invention pertains to the field of measurements performed on slurries to determine the slurry particle size distribution. More specifically, the measurements concern a use of instrumentation to determine particle concentration as a function of particle size in substantially opaque slurries, such as chemical mechanical planarization (“CMP”) slurries currently used in semiconductor manufacturing. The invention further relates to quality control processes used to improve semiconductor manufacturing processes.
BACKGROUND OF THE INVENTION
CMP processes are used in the semiconductor and optics industries to provide ultra-smooth surfaces. CMP process slurries typically consist of SiO
2
or Al
2
O
3
particles suspended in an acid or base solution at a concentration of 4% to 18% solids by weight. SiO
2
slurries are referred to in the art as “oxide” slurries, and Al
2
O
3
slurries are referred to as “metal” slurries. It is difficult to check the quality the particle size distributions within these slurries due to the sub-micron sizes of the particles and the substantially opaque nature of the slurry.
CMP slurries facilitate the deposition of uniform planarized layers in multiple layer wafers, resulting in ultra-smooth surfaces that enhance the resolution of embedded integrated circuit microfeatures. Particles having dimensions that exceed a delimiting value for a particular application are analogous to sandpaper having grit that is too large, and disadvantageously score or scratch the surface that is being smoothed. Thus, it is an essential quality control process to eliminate the use of slurries having particles that are too large.
The use of CMP slurries in semiconductor manufacturing has risen sharply over the last five years. It has emerged as the preferred method of planarization for manufacture of multiple layer semiconductor wafers having feature sizes less than or equal to 0.35 micron. It has been observed that semiconductor wafers can be scratched and thereby damaged if a significant concentration of large particles appear in the slurry through either contamination or agglomeration. The size threshold for particles that are large enough to damage wafers is believed to be in the range of 0.5-3.0 microns. CMP slurry manufacturers attempt to produce slurries that consist predominantly of particles less than 1.0 or even 0.5 micron in size.
Commercially available sensor devices are presently unable to meet the needs of those who wish to measure the particle size distribution of CMP slurries. It is desirable to perform continuous measurements of the CMP slurry particle size distribution in real-time, in order to eliminate the risk of using slurries having particles or agglomerated particles that are too large. This enhanced process control, if available, would allow early detection and cure of slurry problems. The use of the term “real-time” in this discussion means that the measurement results are available within a few seconds after sampling. It is also desirable to measure the particle size distribution of undiluted slurry because dilution and the subsequent change in pH can alter the distribution. Furthermore, dilution combined with continuous sampling creates large volumes of waste slurry. These needs characterize the present state of the art in measuring and/or detecting the particle size distribution of CMP slurries.
Existing commercial particle size sensors include those based on measurement of angular light scattering, dynamic light scattering or photon correlation spectroscopy, ultrasonic transmission, and capillary hydrodynamic fractionalization. These measurement techniques are problematic because they require (a) substantial dilution of the optically dense CMP slurries, or (b) discontinuous batch sampling of the slurry, or they have insufficient sensitivity to detect small changes in the particle size distribution over the critical size range of 0.5 to 3.0 microns.
The need to dilute CMP slurries for particle size measurements creates large amounts of waste that cannot be recovered into usable CMP slurry. According to the data of Bare et al.,
Monitoring slurry stability to reduce process variability
, Micro. Vol. 15, No. 8, pp. 53-63 (1997), oxide slurries typically have 2×10
5
/cm
3
particles greater than one micron, and metal slurries typically have 7×10
8
/cm
3
particles greater than one micron. This data was obtained using a Particle Measurement Systems LiQuilaz SO5 particle size detector, which is specified for a maximum particle concentration of 12,000/cm
3
to keep coincidence errors less than ten percent. The SO5 detector is typical of commercially available single particle light scattering devices. Thus, a minimum dilution factor of seventeen is required to reduce coincidence errors for oxide slurries, and a minimum dilution factor of 58,350 is required for metal slurries. These dilution factors represent significant amounts of process slurry waste, and the dilution itself is suspected of altering the size distribution through agglomeration.
U.S. Pat. No. 5,710,069 (the '069 patent) to Farkas et al. discusses an optical particle counter that detects only one partide at a time in CMP slurries. The single particle must flow through a sample volume consisting of the intersection of a light beam and a detector field of view. The '069 patent does not discuss the difficulty in requiring the light beam to penetrate the slurry towards the measurement area (sample volume), nor in achieving detection of one particle at a time in slurries which typically contain of 10
13
-10
14
particles per cm
3
. The idea of being able to measure only one particle at a time is unsupported by any calculations, numerical arguments, or design details. It is unclear whether the '069 patent uses Mie scattering calculations or empirical correlations to calculate particle size distribution based upon the number of single particles that are counted. The technique of “photocorrelation” is said not to work, but no description is provided of a technique that does work.
U.S. Pat. No. 5,616,457 (the '457 patent) to Garcia-Rubio teaches an apparatus for detecting the presence of a microorganism in a liquid sample. A Twomey linear inversion with a smoothing constraint is used to calculate particle size distribution for the organism. A standard commercial spectrophotometer having a one centimeter cell path length is used to perform the measurements. Additional detail regarding the Twomey linear inversion can be found in Twomey,
Comparison of Constrained Linear Inversion and an Iterative Nonlinear Algorithm Applied to the Indirect Estimation of Particle Size Distributions
, J. Comp. Phys. Vol. 18, No. 2, 188-200 (1975). The '457 patent does not require dilution because it addresses solutions that are much less optically dense than CMP slurries.
Examples of the present state of the art in measuring particle size distribution in optically dense mixtures of sub-micron particles suspended in a liquid solution include two presentations at a recent American Chemical Society symposium: Kourti et al.,
Particle Size Determination Using Turbidimetry
, Particle Size Distribution II—Assessment and Characterization, pp. 35-63, Amer. Chem. Soc. Symposium No. 472 (1991); and Brandolin et al.,
On-line Particle Size Distribution Measurements for Latex Reactors
, Particle Size Distribution II—Assessment and Characterization, pp. 65-85 (1991). These authors typically utilize measurements at two to three wavelengths in the range of 0.2-1.0 microns. Conventional sample cells on the order of one centimeter in thickness were apparently utilized. The limited wavelength range and conventional sample cell dimensions force significant sample dilution, which in turn results in generation of a large waste stream of diluted product. An off-line batch sampling system may also be used, but this type of system has an unacceptably slow time response.
There remains a need for a real-time probe for use in obtaining continuous particle size distribution measurements

Affiliated with

Also associated with

Rating

Chemical mechanical planarization (CMP) slurry quality... does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Chemical mechanical planarization (CMP) slurry quality..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Chemical mechanical planarization (CMP) slurry quality... will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-2488305