Are today’s test-and-measurement tools up to the demands of high-volume small-tech manufacturing?
BY JO MCINTYRE
High-volume production is necessary for commercialization of small-tech products-both nano and micro. Testing, an important production step, is typically time-consuming, so you may ask the following questions: Can today’s test-and-measurement tools keep up with high-volume pressures, or are they draining profits by creating bottlenecks?
To learn the answers, we spoke with some of the industry’s most successful producers. We also heard from a couple of the toolmakers themselves to find out about customer trends and what’s on their horizon.
Trends among tool suppliers
The list of small-tech test-and-measurement equipment vendors is long and growing. It includes suppliers of MEMS test-and-production tools such as SUSS Microtec and EVG, and microscopy manufacturers targeting nanotechnology.
Nano has a way of segueing into micro, according to experts at FEI and Hyphenated Systems, both of which supply metrology tools. These companies report working with device- and materials-development pioneers, hoping theirs will be the instruments of choice when experiments turn into commercially viable products and businesses.
“We work in nano prototyping,” says Bruno Janssens, vice president and general manager for the nano research and industry marketing division of microscope manufacturer FEI Co., in Hillsboro, Ore. “Researchers use our tools to design specific experiments to check out devices people have in mind. We are much more proactively working with pioneers in prototyping than we were three years ago, because we believe at some moment we will be connected to a breakthrough. We want to be very closely connected at a very early process,” adds Janssens.
Terence Lundy, vice president and managing director for metrology toolmaker, Hyphenated Systems, in Burlingame, Calif., says a major trend in the field of microscopy is the creation of systems in which traditional metrology devices are being “stacked,” for instance, to combine light microscopy with probe microscopy, microscopy with spectroscopy, or microscopy with interferometry.
Most inspection systems are compromised because Z-axis information is not sufficiently accurate, too complicated to extract, or too slow to acquire, he explains.
Although Hyphenated Systems is currently focused on nanotechnology, Lundy sees an opportunity for his company in MEMS design and production. “We’re finding the MEMS industry as a whole is a fairly large marketplace for some of our equipment,” he says.
Zyvex Corp., based in Richardson, Texas, produces test equipment as well as nanomaterials. The company measures the raw multi-walled carbon nanotubes that arrive from its supplier, Arkema, in a powder form-expecting 90% of it will be actual nanotubes. “We use tools internally here to verify quality of nanomaterials as they are coming in the door. We typically use SEMs. That doesn’t yield specific numbers: It’s more of a visual check,” says Lance Criscuolo, business manager in Zyvex’ NanoSolve Materials division.
Zyvex treats the tubes to make them functional and puts them into resins and composites. Working with sports equipment makers, the company supplies resins to strengthen bicycle frames, baseball bats, hockey sticks, and golf clubs.
Arkema has a total capacity to produce 50 tons of nanotubes per year. That is a larger volume than most people realize, Criscuolo says. “Single-wall carbon nanotubes, to my knowledge, are not supplied in that volume at this point. That’s probably more of a cost and manufacturing problem than a test-and-measurement issue,” he adds.
Zyvex has a supply-chain certification process that also involves some testing, but it is a relatively tedious process to go through, Criscuolo says. “We look through SEMs to get an idea of small sample sizes’ characteristics,” he says.
A Fresnel zone plate pattern fabricated by direct FIB milling is an example of advanced prototyping using FEI Co.’s Helios NanoLab.
Raymor Industries, of Quebec, is among the companies working to incorporate nanotubes in different manufactured products and production lines. Through its subsidiary, AP&C Advanced Powders and Coatings, the company makes single-walled carbon nanotubes (SWCNTs), metallic powders, thermal spray coatings, and a custom net-shape forming process useful for creating specialized high-tech components.
Raymor CEO Stephane Robert and Frédéric Larouche, director of production at Raymor Nanotech, say a large number of well-designed test-and-measurement tools are available. “We usually obtain the quantity [of tools] required, even if time delays [for delivery of equipment] can be long. Obviously, we also need to develop our own tools to satisfy the time and process constraints.”
Robert and Larouche confirm Criscuolo’s hunch about SWCNT production, saying cost is an important factor impeding commercial sale. “However, we believe we are close to a breakthrough with the massive use of single-walled carbon nanotubes, given the arrival of efficient, large-scale production processes as Raymor has developed,” they note.
Another nanomaterials producer, Houston-based Carbon Nanotechnologies Inc., which recently announced its intent to merge with Menlo Park-based Unidym Inc., opted not to pursue its original goal, announced a couple of years ago, of being able to produce 1000 pounds of nanotubes per day. The decision had nothing to do with test and measurement, but rather resulted from a change in strategic direction.
The company sells research-quality, SWCNTs for $375 to $2,000 per gram on its Website, depending upon purity and grade. Its commercial products, available at lower prices in bulk, are used in inks, composites under development, and commercial plastic parts for chip fab plants.
Carbon nanotube grades are defined at three structural levels. At the primary level, spectroscopy allows developers to see different chemical and structural characteristics. At the secondary level, electron microscopy detects how the primary elements pack together. That entails the self-association of nanotubes together into ropes or bundles. The tertiary structure, visible to the unaided eye, can be fluffy like cotton candy, or solid like a hard candy, or powdered.
“We have the analytical equipment we need” to look at all of these levels, says Ken McElrath, vice president for product development.
MEMS pioneer and entrepreneur Henry J. Klim, managing director at MST Technology Systems in Boston, is a MEMS test equipment expert who is doing testing for start-up LVSI Sensors. Klim started his own company after splitting off from ETEC (which he headed) in Massachusetts, a firm that did pioneering work in MEMS testing.
Standard test-equipment manufacturers offer too many features and options for what MEMS manufacturers require, he says. In his experience, “the magic is not in the test equipment-it’s in the fixturing.”
Klim says that testing-of inertial devices, sensors, and optical devices, for instance-is slow primarily because changes in physical phenomena are often slow (consider, for instance, how long it takes for your oven to heat up). To get around that, testing must be done in parallel, he explains.
There are two common production methods to test MEMS systems: batch and inline flow. In a batch flow, the system is set to one temperature, tested, and then rerun under the next temperature set point. This process is run through many times. With an inline flow product, the MEMS devices enter at one end of the test flow and move through various stations as needed.
Klim is a proponent of batch flow as opposed to inline flow, since inline systems are most often custom systems and aren’t as reliable as off-the-shelf systems.
Howard Wisniowski is marketing program manager for Norwood, Mass.-based Analog Devices’ MEMS accelerometers and gyroscopes in automotive and consumer products. He is happy with the tools available, he says.
“ADI has settled on Labview PXI for bench and Teradyne for test,” he says. “We are, in fact, getting the quantity of test-and-measurement instruments we need, and our future looks bright with our selections.”
In addition to looking at MEMS devices via microscopy, testing also includes handling processes. “In light of the many emerging high-volume consumer markets using MEMS sensors, ADI’s approach is to maximize the usage of standard handler/tester configurations,” says Wisniowski. “These configurations allow for predictability, increased capacity, and re-use of assets.”
Kurt Petersen, CEO and chairman at SiTime in Sunnyvale, Calif., says, “We’re using off-the-shelf test systems. Usually, you have to customize the hardware a bit, so we do that or have it done.”
SiTime uses a Bosch-licensed process to make ultra-stable mechanical oscillators that are integrated into standard silicon chips. According to Petersen, the oscillator tests are routine electrical tests, so there’s no problem about production slowdowns. As with any other integrated circuit parts, they have to take a lot of data on a lot of parts and make sure the test system is working.
Petersen notes that MEMS test and measurement involves the wafer probe test and the final test of packaging before oscillators are shipped to the customer. Both use standard equipment.
Packaging testing requires a handler-a bowl feeder that moves packages onto a system that then delivers them to where the electrical contacts are made. It’s a routine handler, but SiTime had to customize the system that takes electrical measurements.
“The wafer probe is different,” he says. “We do a complete wafer probe to make sure the MEMS are operating properly.”
Petersen is a veteran familiar with a range of MEMS products. He says with optical chips, it’s necessary to make sure all the mirrors are working properly. Pressure-sensitive products have to pressure-test sensors on the chip. With sensors it’s hard to guarantee performance over a range of temperatures. And accelerometers have more-severe testing problems than oscillators.
An advantage of single-crystal silicon systems is that they exhibit very predictable, reproducible, and repeatable mechanics over various temperatures. “That’s absolutely critical for a commodity product like our oscillator,” Petersen says. “The quartz industry, which we compete against, has standardized on testing machines that measure frequency over temperature very rapidly.”
Saunders and Associates is a Phoenix, Ariz.-based company that makes quartz-crystal test-and-production machines to do that testing for SiTime.
Developers’ wish lists
The wish list for improvements in test-and-measurement equipment is ambitious.
“We would like to see software that manages and collects data from different tests and measurements, a kind of database,” say Raymor’s Robert and Larouche. Such software would facilitate data collection by automatically importing data files from specified folders. Currently, different instruments support different types of files or presentation formats. A standardized database would be helpful for data interpretation and would provide uniform presentation of plots and tables for records, they say.
Other improvements the Raymor executives would like to see include monitoring tools for nanotube production. Such tools would enable detection of any variation in production quality and prevent important time losses.
On the wish list for SiTime’s Petersen is more-routine use of equipment that probes several chips at the same time. Right now, it takes a long time to probe a single wafer. He’d also like to see faster steppers. Current steppers have to pull up their wires, move to a new location, line up, and measure again for every iteration. “In general, the closer you can have your MEMS part match the standard integrated circuit, the better the testing infrastructure will be. And the lower your costs will be,” says Petersen.
Carbon Nanotechnologies’ McElrath sees a need for new kinds of analytical-and-detection equipment that could speed the process of making very high value grades of nanomaterials. This equipment could separate and identify nanotubes by type, whether metallic or semiconducting.
McElrath says he’s not alone in envisioning his dream instrument: a total analyzer robot. To use it, you’d just insert a sample-and it would report back with details on type, distribution, impurity level, etc. That’s the general wish, he jokes, “but most of us have our feet planted in reality.”