World's Mightiest Microscope
Chicago firm tops
in precision machining
John J. Kendrick
Gailieo didn't know what he'd see when he built
the first telescope. Likewise, researchers don't know what they might uncover
with the world's most powerful microscope.
Built at the University of Chicago, it is 12 feet
long and weighs two tons. It will provide the most intimate look yet at
the rare world of atomic particles. Scientists-for the first time- will
peer inside dense clusters of atoms whose composition, until now, has remained
a mystery.
University of Chicago physicist Albert Crewe, former
director of Argonne National Laboratory, headed the three-year project to
build the world's most powerful electron microscope. The instrument consists
of hundreds of ultra-precise components with mandated manufacturing tolerances
of 50 millionths of and inch and less.
Crewe, has been advancing the state of the art
in electron microscopy for 20 years. In 1964 he invented the high-resolution
scanning transmission electron microscope. In 1970 he used it to take the
first photographs of isolated atoms. The new microscope is possible because
two years ago Crewe discovered a way to correct aberrations in the magnetic
lenses used in electron microscopes. He says the new device should resolve
objects measuring as small as one half of an angstrom. Existing microscopes
can directly resolve objects only if they are about tow angstroms in size.
The new microscope is expected to have a resolution of 0.5 angstroms. Currently,
the world's most powerful microscope has a resolution of 1.6 angstroms.
Crewe intends to observe how atoms interact with
their neighbors. Sharper pictures of how they behave in small groups could
lead to important advances in electronics, metallurgy, ceramics and life
sciences.
Primarily funded by the National Science Foundation,
the project has received a $1 million computer equipment donation from IBM
and $500,000 donation in precision metalworking services from the Chicago-based
Tool & Die Institute.
"For this microscope to function according
to design specifications, every mechanical and electrical component part
must be machined to the most exacting tolerance levels," Crewe said.
The design called for a margin of error of no more
than 50 millionths of an inch, but the institute members were able to reduce
that error margin to 20 millionths of an inch, said Greg Panek, head of
the institute's voluntary effort.
One firm, Surface Finishes of Addison, IL
(Chicago suburb) surpassed all expectations for dimensional tolerances on
flat surfaces. They held flatness to one quarter of a wavelength of a helium
neon laser, equal to five millionths of an inch.
The University of Chicago provided the specifications
and materials. The tool and die makers determined how to machine and measure
the components. Despite machining at what both scientists and manufacturers
termed a "ridiculous tolerance," Gregg Panek, president of Panek
Precision Products, said the tolerances were easy compared to other problems,
like difficult-to-machine metals and machining intricate shapes. Materials
included consumet iron (more than 99.5 percent pure), beryllium copper,
silicon aluminum bronze, stainless steel 304 and platinum. In some cases,
measuring the part afterward was more difficult than machining it, he said.
Some manufacturers working on the project said
they were able to machine intricate shapes with electrical discharge machines
(EDM). In other cases, the solution turned out to be in the design of complex
fixtures and jigs.
Mark Drzewiecki, president/owner of Surface Finishes (SFI), said that consumet iron posed the most problems in his machining.
He said efforts to machine it with carbide and ceramic cutting tools failed.
Carpenter Technology Inc., producers of the iron, solved the problem by
using high-speed steel cutting tools. On other parts, Drzewiecki said, they
used free abrasive machines that are usually used to manufacture glass lenses.
Because they provide low stress, low heat and close tolerances, they were
helpful in machining the metal also. The disadvantage is that they are extremely
slow, he said.
To comply with the measurement accuracy needed,
Drzewiecki said they used several types of interferometry technology (light
interference). Panek said participating members were using a wide variety
of both contact and noncontact gaging methods. Lasers were used for many
flat components while three-axis coordinate measuring machines solved other
measurement problems.
Much of the most exacting work was performed by
Surface Finishes, a small firm with 25 employees. The company produces
air bearings for semiconductor wafer manufacturing and laser mirrors for
aircraft reconnaissance systems accurate to within two millionths of and
inch. For the microscope, the firm donated more than 350 hours of labor
to produce magnetic lens housings.
"When Professor Crewe first came to us with
his drawings, he didn't think we could reach even 50 millionths accuracy,"
said Drzewiecki. "When I told him we could reach 20 millionths with
no problem, he was surprised and delighted."
"Finishing to accuracies in millionths, with
surface condition to less than one microinch, is in itself a problem. Even
more critical is the problem of developing equipment and technology to measure
results," Drzewiecki said.
"Tenths" accuracy has become "coarse"
measurement. The growing trend over the past 10 years is to specify flatness,
straightness and roundness requirements in millionths. Squareness needs
of less than one arc second are common. As for surface finishes, demand
for less than one microinch is becoming standard for many applications.
SFI uses a Cartrilamp, flatness measuring, monochromatic inspection light, which they manufacture, to show flatness accuracy to millionths of
and inch. Parts are placed on top of an optical flat and checked in a viewing
mirror. This proves to be faster than methods that require placing the flat
on top of the part, checking the part, and then lifting the flat off the
part again. It is also more accurate because it eliminates the distortion
or erroneous readings caused by the weight of the flat on the part, according
to Drzewiecki.
The Cartrilamp standard for measuring the workpiece
is calibrated against a master standard that has an accuracy 10 times greater
than the tolerance being measured on the Cartrilamp. A calibration report
from the National Bureau of Standards (NBS) shows a total deviation of less
than one millionth of an inch on the master flat (Figure 4).
|
