What you can't see could cost you millions
An effective fight against airborne contaminants requires a combination of filters, purifiers and monitors.
by Sarah Fister Gale
Wafer technology may be getting better faster and smaller, but with each technology jump forward, airborne molecular contamination (AMC) seems to set the cleanroom battle a little further back. Unlike particular contamination, AMC is the invisible enemy, wrecking havoc on yields—even in the tiniest quantities. Thinner technologies are even more susceptible to AMC deposits than they were just two years ago.
Every AMC event has the potential to impact productivity, and it may not be discovered for hours or even days, depending on the type and duration of the exposure. Problems can stem from something as simple as excessive ammonia during a shift change or as complicated as the recirculated contaminants from a chemical leak that happened days earlier somewhere else in the facility.
Weapons in the fight
Cluster tools, purge gasses and vacuums can help reduce AMC's effects by limiting the amount of air with which wafers come in contact. But in large-scale environments, this isn't an economically feasible solution, says Mark Camenzind, a senior technical advisor with AMC analysis experts Air Liquide-Balazs Analytical Services (Fremont, Calif.; www.us.airliquide.com).
"There's no one magic bullet, and each company has its own proprietary solution," sums Camenzind. An effective fight against airborne contaminants, he says, requires a combination of filters, purifiers and monitors.
As semiconductor device geometry and size continue to decrease into the deep sub-micron level, chemical contamination has become as important as, if not more critical than, particle contamination, says Peter Maguire, Taiwan-based sales manager for Lighthouse Worldwide Solutions (Milpitas, Calif.; www.golighthouse.com). "The level of basic impurities in the air can be critical at very low concentrations down to parts-per-billion (ppb) levels."
The target control levels for chemicals, such as ammonia, solvents, and molecular acids, inside the process tool environment are now under 1 ppb. And for others, it's even lower. This is further complicated by the changing nature of the environment and its impact on processes, allowing for an infinite number of possible hazardous combinations. Every year, acceptable levels of contamination drop just to maintain the yields of previous years; yet there are no definitive solutions for achieving those levels or for determining what chemicals in what combinations will have a detrimental effect on a particular product or process.
"How do you hold every variable constant throughout the process flow and correlate variations to possible AMC?" asks Fred Lakhani, senior member of the technology staff at International Sematech (Austin, Tex. www.sematech.org). "When you have 400 process steps, and each step has its own set of variables, on top of environmental concerns it becomes too much to handle."
Lakhani thinks that testing ultimately has to tie into the product itself. "The only way to maintain control over AMC is to understand what the yield detractors are," says Lakhani. "For example, with the right kind of test structures and methods, semiconductor manufacturers will be able to start tying trace elements in the air and consumables to their impact on yield."
Others believe a more proactive approach is necessary. Traditional methods of AMC detection-and-response require contaminations to occur and be recognized before they can be addressed; in essence, waiting for decreases in yields to show there is, or was, an AMC event.
"We love to do grab-sampling," Camenzind says, but he adds that discovering a problem two days after it happened can cost millions of dollars in lost product. "The industry needs better early warning systems."
Along with chemical filters and purifiers, Camenzind sees online monitoring as a growing trend that will help companies protect themselves from contaminations in real time and reduce the effect that AMC has on productivity. "Whenever it's cost-effective, it's ideal to have constant monitoring," he says.
Chris Muller, technical services manager at AMC filter maker Purafil, Inc. (Doraville, Ga.; www.purafil.com), agrees: "Online monitoring allows you to respond to events as they happen." But it can be expensive, he says, if you are using real-time direct gas monitors, which measure specific gases in the atmosphere.
Copper and silver sensors showing varying levels of corrosion due to AMC contamination.
Muller suggests that online reactivity monitoring may offer a more cost-effective, real-time alternative. Reactivity monitors detect the interaction of contaminant gases on copper, silver, and other metals by measuring the resulting corrosion film thickness. They gauge the overall effects of different contaminants, such as acids or bases, without attempting to pin each down to a specific type or source.
Unlike direct gas monitors, reactivity monitors don't give you specific details about the types of gases. But they let you see the total impact the atmosphere has on processes in real-time, and at 10 times less cost. They can be used in real-time by tapping into the facility's monitoring network, or passively through the use of wafers or coupons left in the environment and periodically tested for corrosion.
"Reactivity monitors have been used for years in process industries, and many of the same contaminants of concern to these industries are also of concern to the semiconductor industry," Muller says. An air-quality classification scheme based on reactivity monitoring has gained wide acceptance throughout the semiconductor industry. It also directly indicates the potential for AMC-related process effects.
Muller believes that if cleanroom managers establish an air-quality baseline, reactivity monitors can then identify and investigate when specific contaminants reach warning levels—before they become a problem on the process floor.
"The million-dollar question is, 'what's the impact?'" Muller says. When you collect data and tie negative results to specific incidents, you can begin to quantify the cause-and-effect relationship.
Extraction Systems (Franklin, Mass.; www.extractionsystemsinc.com) and Molecular Analytics (Sparks, Md.; www.ionpro.com) also make measurement systems. Those from Extraction Systems are aimed at monitoring lithographic areas, while Molecular Analytics' AirSentry monitors entire facilities for multiple trace contaminants.
Facilities can incorporate all of these technologies into their overall preventive maintenance program, and they can help reduce the number of AMC-related incidents by allowing proactive investigation of potential problems.
While there's no single solution for AMC, Dan Rodier, applications engineer for Particle Measuring Systems (Boulder, Colo.; www.pmeasuring.com), says the best defense is to know what chemicals have negative effects on specific process steps by setting an air-quality baseline, then monitoring changes. "You can't control something that you don't measure," Rodier adds.
With the right monitoring, he says, companies can begin to identify when chemicals are increasing and what effects that has on yield. With that information, companies can choose the best filters for their needs, better schedule their filter changes, and react to accidental contaminations more aggressively.
Standards on the horizon
Recognizing the need for more specific knowledge about AMC's impact, Working Group 8 of the Geneva-based International Organization for Standardization (ISO; www.iso.ch) 209 technical committee is working on its ISO TC 209 document to define new standards for equipment, facilities and operational methods associated with control of the cleanroom environment. The goal of the document, which the group has been developing since 1999, is to create a set of standards applicable to all precision products made in cleanrooms that can be classified in terms of AMC concentrations in various contamination categories.
Allyson Hartzell, senior staff scientist in CAD and MEMS design engineering at Analog Devices Inc. (Norwood, Mass.; www.analog.com), and the United States' delegate to the group, explains, "We tried to make our classification scheme as broad as possible so that not only can it include semiconductors but it can also include other industries that are affected—such as thin-film head manufacturers, biotech, or anybody that has a cleanroom and is making a product."
Hartzell admits that this scheme is tougher than classifying particular contamination because you aren't just measuring the size of the contaminant. Every type of airborne contaminant can have different effects on different technologies as a function of chemistry and surface-adsorption kinetics. "There's no one definitive analytical technique for monitoring AMC," she says.
Reactivity monitors detect the interaction of contaminant gases on copper, silver, and other metals by measuring corrosion film thickness.
The final TC 209 document, which could be available as soon as early next year, will include an informational annex with examples of analytical techniques to test airborne chemicals. These sections won't be requirements but rather guidelines to teach users about the relationships between technology and AMC. The informational references will suggest ways to test chemicals and species, and discuss the impact those chemicals may have on certain technologies, products and processes.
"Users have to understand what is deleterious to their products," Hartzell says. The document will give cleanroom personnel a way to standardize and regulate the AMC that could most hurt their products. "It's a method in which we can help the cleanrooms of the world define how clean they are in respect to AMC, and define it quantitatively," she says.
The ISO 209 Group hopes that its informational classification system can help educate users about the risks in their own environments as well as begin to create a classification system that could be used around the world. As Hartzell notes, "The industry has never had that in the past."
Dave Ruede, filter group manager for Extraction Systems, believes the industry needs such information to gain an edge against the ravages of AMC. In addition, he hopes companies will begin sharing their research more openly to help the industry develop better tools, filters and response systems. "There's a big payback if we all share data about AMC," he says. "When we pool our research, we can work together to find better solutions."
Camenzind also points out that the 2003 edition of the International Technology Roadmap for Semiconductors has new and valuable information for the fight against AMC. Table 114a in the Roadmap sets proposed limits and tests for surface molecular contamination (SMC) due to AMC. "It's a huge change," he says.
SMC has been identified as one of the greatest potential risks to yield as manufacturers reach and surpass 300-mm, 0.13-µm semiconductor processing. Camenzind believes that tracking SMCs will deliver better return on investment because it will help target the biggest issue—what's landing on the wafer—first.
Real-time SMC monitoring of process tools, chemical filters, mini-environments and pressurized gas lines is gaining attention because it provides the ability to detect events that can cause irreversible damage to critical product or process surfaces.
"You could test for everything but you'd go bankrupt," Camenzind says. "So, you have to ask yourself, 'where will I gain the most return on investment in testing?'"
SARAH FISTER GALE is a freelance business writer based in Minneapolis, Minn. She can be reached at: firstname.lastname@example.org