The 27th annual event will feature special joint sessions on contamination, wafer inspection, defect reduction, and yield improvement during four major conferences. Among the issues covered in the special sessions are hydrocarbon impurities on the optical surfaces at 157 nm, direct-to-digital holography for wafer defect detection, the effect of a scanned electron beam on an advanced logic device, and SEM inspection of optically invisible defects.

The broad-ranging microlithography symposium is set for March 3–8 at the Santa Clara Convention Center and the adjoining Westin Hotel (Santa Clara, CA). The four conference topics are emerging lithographic technologies; metrology, inspection, and process control; resist technology and processing; and optical microlithography. The symposium opens with a keynote address by Bob Helms, president and CEO of International Sematech. The title of Helms's speech is "Silicon Technology Roadmap: Bigger Challenges—New R&D Business Models?"

A high number of submissions have convinced at least one conference chairman that metrology has assumed even greater importance in the field than it has held previously. "We're seeing an emergence of a [greater] need for metrology, which I think is a big reason for the increase in submissions," notes Daniel Herr, chairman of the conference on metrology, inspection, and process control.

In particular, Herr, the director of materials and process science research for SRC, says equipment suppliers are stepping up efforts to address chipmakers' needs. "We're seeing the emergence of tools for enabling nanoscale metrology. What's interesting is that scatterometry has been around for quite a while, [and] I think it is finally coming into its own and emerging as [an applicable method]."

Herr says over the last five to seven years the metrology committee "has settled in on a few criteria" to rate the quality of submissions. "The first is; is it good science? Second, what is the level of interest to members, in terms of relevance? Third is a lack of commercialization.

"We had a tremendous response this year," continues Herr. He notes that the selection committee received "140-some-odd abstracts." The committee "set the bar pretty high" for submission acceptance. "I think the acceptance rate for the [oral presentations] was less than 50%," he points out.

Several sessions will address 157-nm issues for an industry making a run at the sub-100-nm process nodes. Lithography at 157 nm is much more sensitive to contaminants on optical surfaces, and the process requires more stringent defect-reduction methods than at 193 nm. These methods may even include form of in situ cleaning, according to some experts.

International Sematech addressed the matter of challenges posed by the adoption of 157-nm lithography during a technical data review held in Orlando, FL in December. The participants identified the calcium fluoride birefringence issue and pellicles as two of the more profound difficulties. However, technical experts at the review were encouraged by the progress made in developing the infrastructure for 157-nm equipment, according to Sematech.

In May 2001 a NIST researcher working on a Sematech project discovered the double-refraction phenomenon bedeviling the calcium fluoride–based lenses used in 157-nm applications, says Tony Yen, International Sematech's director of lithography. In a July 2001 symposium hosted by Sematech, "toolmakers got the message" about the problem, he asserts. "They went back and started to modify their optical designs."

Yen says major stepper manufacturers such as ASML, Nikon, and Canon are confident they can lick the birefringence problem. To fully compensate for birefringence requires rearranging the calcium fluoride pieces in orientations other than the common <111> direction. Yen notes that crystals oriented in the <100> direction can be grown. In addition, lens blanks in the <100> direction can also be obtained from <111> crystals by slicing the <111> ingot in a direction other than horizontal.

Recent developments at Hinds Instruments and Corning also bode well for the introduction of 157-nm processes. Hinds Instruments has received a contract from Sematech to further develop a DUV birefringence measurement system. The Hillsboro, OR–based supplier is extending the capabilities of its Exicor measurement tool to characterize birefringence in any optical material transparent at both 193 and 157 nm. The company expects to have an engineering model available for calcium fluoride manufacturers quite soon and a production-worthy system ready by the second half of this year. That system will be capable of measuring samples ≤400 mm diam x 270 mm thick.

Adapting the Exicor tool to 157-nm wavelength "is a nontrivial task," notes Mike Ward, marketing manager for Hinds. "The difference is that in the visible [spectrum] we use laser sources and silicon detectors, and we operate in air environments. [In the case of 157 nm] we don't have any of those luxuries, so we're having to use noncoherent sources...because of the light levels and photomultiplier detectors, as opposed to silicon solid-state detectors. We also have to operate in a nitrogen-purged environment.

"One of our experimenters...is doing prototyping in a glovebox in the back," Ward continues. "He said he's certainly glad he's been practicing his tai chi, because the flexibility is needed in order to align the system."

Ward is confident that the DUV capability exists. The nitrogen-purged glovebox poses problems, though. "It's pretty commonly known that [working] in a nitrogen-purged environment is by far the most difficult thing to do. You have to maintain a parts-per-million level of oxygen. The problem is that the oxygen absorbs 157-nm light. That degrades your signal—that's why it's called deep UV, actually. Around here we called it vacuum UV, before deep UV came out. But that's what's difficult: to build a system that makes consistent measurements...in a vacuum environment."

Corning has completed the expansion of its calcium fluoride plant in North Brookfield, MA, to address a severe shortage of the product. The company is also in the process of expanding another plant in Canton, NY, says Bob Sell, manager of marketing and strategy development for Corning semiconductor materials.

He says the supplier is interested in evaluating the Hinds' system. "We certainly will be talking to them about...where it fits in the process." Sells says all crystal manufacturers need to improve their yields, which are "probably in the 10% range, plus or minus a point. We all need to double or triple the yield levels."

Time, not money, is the most important issue, Sell points out. To avoid the intrinsic birefringence of the <111> stacked-cube direction manufacturers can try to rotate the cubes or place them on their points in either <110> or <100> planes. Since it takes approximately 30 to 45 days to grow a crystal, however, a manufacturer such as Corning can conduct such experimental work only "about six times a year," he notes.

"We sell a lot of fused silica for 248- and 193-nm systems," the marketing manager says. "We can make a ton of glass in 7 to 10 days. In this instance, you're making 100 to 200 pounds, depending on the furnace size, in a month. That's just nature. It's science. We'll get the science. It just takes time."

Despite a "slight delay" in 157-nm–capable steppers, Sematech's Yen says the "first batch" of R&D tools will arrive in 2003. A limited number of production-worthy systems will be ready in the second half of 2004. Chipmakers can expect full-scale ramp from "all three major vendors" in 2005–2006, Yen asserts, adding that he's optimistic a single-layer 157-nm photoresist will be ready for the first batch of steppers.

As for photomasks, Yen expressed optimism that solutions for NGL processes, particularly for extreme-UV lithography, will be ready, despite "a lot of hurdles."

Other important developments on the eve of the SPIE symposium include:

Additional symposium highlights include:

The exhibitor list for SPIE Microlithography is found here. More information is available at www.spie.org/info/ml.