INDUSTRY NEWS

Metrology start-ups, spin-offs turned up business in downturn

GOOD AS GOLD: Qcept's contact potential difference imaging technology can detect less than a monolayer of material. The raised areas in the wafer map (inset) represent areas of deposited gold on a 100-mm silicon wafer. Graph shows gold surface concentration and sensor output voltage data.

So, here's the big picture: a flabby national economy, an industry barely out of a prolonged slump, a transition to a larger wafer size, transistor shrinks, the introduction of new process materials. In other words, it's a great time to be a semiconductor equipment supplier, right?

Some vendors plying their trade in metrology and advanced process control (APC) certainly hope so. Although it seems counterintuitive, it's a truism that when the industry's notorious roller-coaster business cycle heads for the dips, chipmakers actually have the time and, more important, the inclination to return phone calls.

" In this industry it¡¦s always the case that during downturns the fabs redeploy people from production to developing the processes, optimizing them, and getting them ready for the next upturn," says Lee Smith, CEO of Oraxion, a Southern California–based start-up. "It's easier to get customers' attention during a downturn."

Allen Vance, senior vice president at Qcept Technologies in Atlanta, says the fledgling firm has had success setting up meetings to demonstrate its new nondestructive metrology method. "We've gotten the meetings we wanted to have with yield management and process people." Vance notes that Qcept has managed to meet 13 of the 15 potential clients the company contacted.

A developer of a wafer-temperature measurement system has also benefited from the recent downturn. "It gave us the opportunity to work more closely with our customers to develop new applications," says Rod Browning, president and CEO of three-year-old OnWafer Technologies.

Of course, you can have your foot in the door, but you still must have something worth buying. With yields always uppermost in the minds of chipmaking executives, metrology and APC firms are likely to find the welcome mat out more often than not. In addition, a number of metrology start-ups and new spin-offs may benefit from the "convergence" of factors cited above, as Oraxion's Smith puts it. In interviews, they say that their tools are attracting the attention of device manufacturers and OEMs alike.

In OnWafer's case, Browning says the company¡¦s wireless SensorWafers measure temperatures nonintrusively and give a clear picture of what¡¦s actually happening on the surface of product wafers during processing. Accompanying OnView software is designed to collect data from the wafers' in situ sensors. The SensorWafers collect data in real time and download the information using an infrared link, the company says.

In late 2002 Browning left KLA-Tencor to join OnWafer, which had been cofounded two years earlier by Costas Spanos, a professor of electrical engineering and computer science at the University of California, Berkeley. "Basically, at that point they had developed these wafer-based wireless sensors for photolithography and plasma etch applications, and were using them to get never-before-seen process data."

At the time, he notes, "people were struggling with critical dimensions as they were trying to advance through the 100-nm node, and even having difficulty yielding at 130 nm. Having applications in the pattern-transfer area (litho and etch) seemed the perfect place to be. The key question was, could we use this novel metrology approach, which 'sees what's happening on the wafer surface' to help optimize the process?"

As effective as sensors are for gathering data, though, "if you can't do something with the data, you can¡¦t really optimize your process," Browning points out. As they developed new software applications and he saw that the technology could actually accomplish that goal, he was sold. "I think that's really when we finally said this is going to work."

OnWafer essentially doubled its revenues last year and expects "to more than double revenues this year," Browning says. "Our primary focus is in pattern transfer to improve critical dimension and CD uniformity." Business has been so good that the firm counts as customers "nearly all the top 25" major chipmakers. The company recently moved to larger quarters in the east San Francisco Bay Area suburb of Dublin.

Oraxion is another metrology launch with roots in academia. The company¡¦s CGS series of inspection tools is based on technologies developed in the 1990s by a California Institute of Technology professor working on U.S. Department of Defense research in the area of "ballistics impact in materials," explains Smith, the start-up's chief executive. The technology—coherent gradient sensing (CGS)—is "a type of interferometry that is new in the semiconductor industry."

The technology remained lab bound until 1999, when the Massachusetts Institute of Technology became involved in the research effort, and the idea that the technology could apply to semiconductor manufacturing began to take hold. Caltech contacted Idealab, a high-tech incubator based in Pasadena, CA, for venture funds, says Smith. The company was launched early in 2002.

Was it a hard sell or did the technology sell itself? "Largely, the latter," Smith replies. "The company was formed in March 2002. In December of that year it shipped a prototype to International Sematech, and this coming summer it¡¦s shipping its first product." The fully automated tool, the CGS300 wafer-inspection and defect-prediction metrology system, will be sent to "a flagship 300-mm fab in the United States. That makes it two and a half years from inception to installation."

"There is a convergence between this technique and a need that¡¦s been developing in the industry, accelerated by the 200- to 300-mm conversion," Smith says. "The wafers got bigger in diameter, but the thickness changed little." Mechanical instabilities processed into this "large platter are sitting there waiting to spring forward. We think of a silicon wafer now as a spring. All it takes is something to trigger it to assume a distorted shape of lower energy . . . in order to relieve some of the stresses that are put on it. And processing delivers those triggers big time in the form of nonuniformities in heating rates, cooling rates, and lack of axial symmetry in all kinds of processes, like chemical-mechanical polishing.

"Because features are getting so small, they are more fragile and easily damaged by these wafer distortions," Smith adds. Large process lines "may have 400 steps, and 100 of them may be thermal cycles." The "spring effects" cause nonuniformities that deform the wafer "into a potato chip," Smith jokes.

The hundreds of microns of "nonflatness" that can develop are certainly no laughing matter. Smith lists the potential problems: vias pull out, interconnect lines crack, low-k dielectrics crack, films delaminate, STI leaks, strained Si has control issues, etc. The industry trends point only to smaller features with less and less tolerance for stress-driven wafer distortion, a trend that demands better process control," the executive says. Oraxion claims that its in-line production monitoring tool enables fabs to understand and monitor the buildup of deleterious stress in product wafers for the first time ever.

Nonuniformity issues across the wafer also are the focus of Qcept Technologies' new offering. Based on technology licensed from the Georgia Institute of Technology, the company has developed a tool to measure chemical nonuniformities on semiconductor wafers. The technology, called contact potential difference imaging, measures spatial variations in work function that correlate directly to variations in the wafer¡¦s surface chemistry, the company says.

Integrated into the tabletop Chemetriq tool, the scanning metrology system is capable of scanning a 200-mm wafer, either patterned or bare, in less than 5 minutes. Qcept expects to sell its first commercial version of the Chemetriq tool in April, says Vance. The tool works with copper processes, CMP, and different cleaning chemistries, because the start-up is trying to embrace as many applications as possible, the company's senior vice president explains.

"We're initially looking to deliver this system in an R&D environment as a tabletop tool, but we believe this technology has a place in advanced process control in the fab," he emphasizes. He adds that the operating speed that enables the Chemetriq to rapidly assess chemistry uniformity is a big selling point. "Obviously, throughput is key."

HyperNex in State College, PA, is a metrology spin-off also looking to position itself during the chip industry's long-awaited upswing. Originally a research arm of ATMI, the venture specializes in x-ray diffraction analysis. In February 2001, it signed an agreement with IBM to perfect its wide-angle x-ray diffraction technology for copper processing.

The start-up has developed an automated 300-mm system, the AutoTex, incorporating crystallographic microstructure analysis for advanced fabs. Damascene copper processing is particularly prone to problems associated with stress migration, electromigration, defects, and resistivity. Polycrystalline device layers such as barriers and metal lines can hammer both device yield and performance, the company emphasizes, adding that yield ramps in copper production lines, particularly at the 130-nm node and below, are more punishing than they¡¦ve ever been. The company touts its new system as a fast-working alternative to off-line tools used for examining crystallographic microstructure performance during processing.

Dave Kurtz, the company president who has a background in material science, notes that raising money to get a system into the marketplace, especially "in this demanding environment," is tough. It's particularly difficult when you¡¦re trying to accomplish this "in a cost-effective way and get to a level of profitability without spending a fortune."

The metrology segment of the market faces more of a challenge today than it did 10 years ago, "and certainly more than it did 20 years ago," Kurtz asserts. A tabletop analytical tool was more likely to find a spot on a fab manager's order list back in the day. "Today, it's a significant challenge. I won¡¦t give you numbers, but you can imagine the amount of venture capital investment it takes to fully develop and put something in a fab."

Why even try then? Kurtz chuckles and answers only half in jest, "Sometimes you're naive enough to not know what you're up against." A development as "disruptive" as the introduction of copper damascene, however, creates opportunities for spin-offs such as HyperNex, because "it presents new issues for both the process and metrology."

Kurtz says the amount of cash needed to develop instrumentation or process gear is "very substantial. A number like $15 million or $25 million is not uncommon in this industry, and look at the amount of business you need to earn that back. I think we're unique in that we¡¦ve done quite a bit with quite a little."

HyperNex has been able to parlay U.S. Small Business Innovation Research program support and Defense Department–based research into a government-backed success story. The help "enabled us to create expertise and an infrastructure that we could then translate into a commercial effort. It did not directly pay for our product, but it gave us that base expertise and infrastructure."

Kurtz factors into this developmental calculus the belief that the process equipment segment has a higher reward potential than its metrology counterpart. Process OEMs compete, and one technology will win out and the other will lose out, he argues. "But the one that wins, wins big. Metrology doesn¡¦t quite work that way." One metrology tool might be accepted in one part of the chip business and another measurement and monitoring solution in another segment, he says.

With that in mind, Kurtz, like the other executives here, agrees that the recent downturn gave companies such as his the opportunity to lay the groundwork for bigger payoffs down the road. As Browning of OnWafer says, "The slowness of the industry enabled us to flesh out early."

MicroHome | Search | Current Issue | MicroArchives
Buyers Guide | Media Kit