Irish start-up senses plasma turning point down the road

Build a better sensor and the world will beat a path to your door. Or at least that part of the world that makes semiconductors. That's the hope, anyway, of Scientific Systems. The principals behind the small Dublin-based company, which specializes in plasma process components such as impedance monitors and ion flux probes, believe they're getting closer to helping chipmakers overcome one of the major challenges identified in the 1999 International Technology Roadmap for Semiconductors (ITRS).

LOOKING AHEAD: Michael Hopkins, left, and Ciarán Ó Móráin examine a current voltage probe, one of their firm's products, at Dublin headquarters.

Improved sensors are one of the sine qua nons for metrology in the next five years, the ITRS makes clear. Published in November, the updated document targets them as one of metrology's so-called difficult challenges: "Better sensors must be developed for trench etch endpoint, ion species/energy/dosage (current), and wafer temperature during RTA." Ultimately, better sensors will enable more control of the ionized gases, resulting in yield improvements.

Scientific Systems believes it has a solution close at hand. The five-year-old firm has been working with chipmakers in Europe, Japan, and the United States as well as OEMs such as Lam Research and Applied Materials to perfect an automatic sensor for measuring and controlling plasma etch and deposition. Michael Hopkins, president of the company, thinks the technology, now in beta site tests, represents a breakthrough.

"We would think of it like that, yes," he replies. "It's basically a paradigm shift. We feel that the plasma process has been a little bit of a black art in the industry. And as the industry moves to a more mature manufacturing level, then you're going to have to require a more robust process. We use the analogy of 'engineering the plasma.' "

Hopkins has a background in demystifying the "black art." He headed the Plasma Research Laboratory at Dublin City University (DCU), where he began developing the technology while developing the lab into a world-class facility. A former DCU student, Ciarán Ó Móráin, contacted Hopkins from Japan, where he was working as a senior product development engineer with Irie Koken, to alert him to the commercial possibilities in plasma diagnostics. Ó Móráin eventually returned and helped to launch Scientific Systems in 1995 as a campus-based company. Hopkins left DCU in November 1998 to concentrate on Scientific's fortunes.

As vice president of the company, Ó Móráin oversees Scientific Systems's U.S. operations from its office in San Jose. He credits Hopkins with forging a top-drawer plasma research facility before leaving DCU. "At DCU the basic research Mike's group did in terms of plasma modeling and diagnostics would definitely make it one of the top ones in the world at a fundamental level."

That expertise proved effective in raising operating funds. Scientific Systems received a total of approximately $2 million from venture capital firms and Enterprise Ireland, a government agency. If its new technology is adopted by the industry, the company expects to generate up to $15 million in revenue within the next couple of years.

Some 34 positions will be added to its staff, composed primarily of process and electronic design engineers. Scientific Systems' initial goal "is to get a hardware base in the fab," Hopkins says. He adds that the company also needs to have both the proper software available for process control support and the requisite service infrastructure.

A major U.S. chipmaker is evaluating the technology. "We would hope we'd be moving from beta-type products to...turnkey solutions by the end of the year," Hopkins says. The company expects to complete the evaluation before the end of this spring.

Hopkins and Ó Móráin both note that companies with similar plasma diagnostic or sensor products--firms such as Advanced Energy Industries in Colorado and Hiden Analytical in the UK--do not necessarily compete directly with Scientific Systems' technology. "We would have a wide range of competitors on the periphery, if you like," Hopkins says.

"Others make similar products," Ó Móráin says. "We're different. We're tailoring our product toward end-users, although we think we're probably focusing more on OEMs at the moment. I'm not saying that we're ignoring the OEMs, because they're part of the food chain."

Terry Long, vice president of instrumentation for Advanced Energy Industries, agrees that in situ process monitoring is becoming a reality. Addressing the topic in the "MICROutlook 2000" article appearing in the November/December 1999 issue, the executive says, "After 10 years of pushing a rope here, I am finally able to see the adoption of in situ monitors. The move to 300-mm wafers really is pushing that change through. We can't afford to lose those things."

Long notes that "the defect-related development" he sees causing--"pardon a worn-out expression here--a real true paradigm shift in the way we're running manufacturing environments is the integration of in situ monitor capability on the process tools coming from the OEMs." A few "major plasma tool suppliers," particularly manufacturers of etch equipment, are offering integrated advanced process control instrumentation on next-generation systems, Long says.

Scientific Systems' purported advantage is its ability to transfer its "strong background in plasma expertise," as Hopkins puts it, from the lab to the fab. "First of all, you have to have a basic understanding of what plasma involves in order for you to measure it. Then you have to have the ability to control it.

"There are two ways of trying to control the process," he continues. "One is to use metrology to measure the wafer state and then try to modify the process in order to control it. The other way is to keep the process locked and under very tight control, and that might alleviate much of the need for metrology to measure things."

Both company executives believe the technology has great potential for defect reduction. "There's lots happening with plasma arcing that will clearly lead to defects. There are many subtle things that happen," Hopkins points out. "A lot of [chipmakers] measure particle density or production and correlate that to defects. They lack an understanding of plasma and how plasma drops particles on the wafer."

Ó Móráin sees "a huge opening for us on yields," a belief borne out so far by the feedback he's getting from developmental sites. "Where we've been very successful to date is improving the quality of the plasma manufacturing steps. Improving that and relating that back to cost of ownership or overall equipment effectiveness is difficult. We've improved the flow of the manufacturing step. . . and have been able to implement methodologies with our customers that have allowed us [to eliminate] bottlenecks. That's a big impact."

He claims other savings are possible on materials "if you take the consumables like gas so endpoint chamber cleans can be done sooner." Tapping into its large base of manufacturers, Scientific Systems "believes we can work toward developing end point for very low open areas, via contacts, and things like that."

The company has been able to team up "with the end-users to generate solutions for their particular plasmas. We do chamber matching primarily and baselining the process, simple statistical process control." If a process engineer has "stripped down the tool for no reason at all, thinking [the problem] is the plasma, or thinking it's the RF, we can give him a fingerprint of what the plasma SPC limit should be for a given process.

"We've been very successful in selling at the end-user level," Ó Móráin continues. "We basically have a sensor that can integrate into existing tools." Making "a sensor head for a given tool set, gives it almost a plug-and-play capability from a mechanical point of view." The executive stresses, "We don't shift the process," adding, "our components are essentially transparent to the power delivery path."

Niall MacGearailt, a process engineer for Lam Research Ireland, has been working with the technology at a leading European fab. Lam has one probe in a production tool at the facility as part of a beta site evaluation. He calls the component "a nonintrusive, in-line, real-time sensor that yields the fundamental process parameters. They complement this technology with the ability to build very small sensors that can be installed in-line on any tool."

Asked whether he considers the technology a breakthrough, MacGearailt replies: "These fundamental plasma process parameters have been measured before in the academic arena. However, Scientific Systems is one of the first companies offering a sensor suitable for a large semiconductor fab. The data from these sensors have proven invaluable for understanding both process tools and their interaction with production material. Next-generation tools will have sensor technology like this built into the tool and will use the data for tool control."

The technology has shown three yield-related advantages so far, according to the engineer. "One, the data from these sensors have helped us identify the tool-to-tool variations that affect yield. Two, the sensor can be used for excursion prevention, allowing us to shut down tools when faults occur or when tools drift out of specification. Three, the data from the probe will help us ramp technology faster by decreasing the time to qualify new tools."

As for Hopkins's claim that companies don't understand the nature of plasma or how plasma drops particles on wafers, MacGearailt says: "Plasma processing is very well understood. However, as new problems occur, knowing the fundamental process parameters for affected material may lead to faster conclusions."

"The approach we took from a technical standpoint definitely made our technology different," insists Ó Móráin. "The sensor was designed so the head was not invasive. You put it into the power delivery path, and it didn't shift the process. There's a postmatch power delivery between the match network and the plasma load. Competing technologies would shift the process significantly so you'd have to requalify your process. We wanted to make a sensor we could give to end-users and let them evaluate the technology without having to requalify the process. It was one of the fundamental design considerations from the start. On the data acquisition side...we developed electronics that give much higher resolution than is available with current sampling techniques."

The high-resolution "in both current voltage and phase...allows us to see very small changes in the plasma load induced by the process," the company executive asserts. In addition to, say, Lam etchers, the technology can be used in PECVD tools. "If you want to see small changes in plasma load as the process steps carry through, we believe the resolution that we have allows you to identify events that either change or adversely affect the process yield."

"When we started looking at the roadmap a couple of years ago it said that pilot lines around this time frame would need these types of sensors," Ó Móráin concludes, as he and Hopkins wait expectantly near the front door.