frames for particles, and the way we do
that is based on the gray-scale or color
values,” Gislao said.
If the filter is white, a black particle will
show up readily against it. But because a
whiter particle will have a much different
gray scale, the system must be trained to a
given application. Typically, the particles
of interest measure a few microns or larger
and are easily visible under the microscope
using bright-field imaging. Additional scattering, and therefore composition, information can be gleaned by dark-field inspection, but at a loss of size data.
Tom Calahan, product materials microscopy marketing manager at Carl Zeiss
USA of Thornwood,
N. Y., noted that the
trend has been for
analysis to be done
on smaller and
smaller particles.
Beyond a simple
count of those
above a certain size,
the analysis may
involve classifying
them as to shape, with, for example, long
and narrow particles indicating a different
problem than would be the case for spherical particles. Another distinction might
be to categorize particles as metallic or
nonmetallic.
But this push toward analysis of smaller
particles comes at a cost. It may require
imaging at a higher resolution, which also
will force the use of a smaller field of
view. Because the area of the filter is not
changing, shrinking the field size means
that more fields must be scanned, and the
scanning time hit can mount rapidly.
“If you double the resolution, you
quadruple the time, effectively,” Calahan
said.
Vendors are aware of this and are
attacking many areas to eliminate bottlenecks. The list includes camera capture,
image transfer, autofocus and stage movement times. In shaving time off, however,
any gains must be balanced against performance hits.
There has been a tenfold increase in
speed over the past few years, and the
expectation is that further increases are
in the offing, said Wayne A. Buttermore,
marketing manager for industrial microscopy at Leica Microsystems AG in
Bannockburn, Ill. In part, this is because
greater computing power, along with
cheaper memory, will make image processing faster.
He also noted that there will be increas-
ing automation, which can show up in a
number of ways. One is through wizards,
software that guides end users through the
steps needed to implement a given task.
Another is through taking actions that en-
sure reproducible results. These go beyond
setting illumination, camera gain and other
parameters to some predefined value.
The techniques that are good
at composition or molecular
analysis ... tend not to be good
at finding particles and
pinpointing their location.
