Hazards of molecular contamination grow over space and time
Wafers are increasingly exposed to the outgassing of molecular contaminants, the effects of which are difficult to predict and can take time to manifest
By Carolyn Mathas
More than 40 years ago, in the early 1960s, the space program began to experience firsthand the effects of molecular contamination. Problems were apparent within the optics-rich vacuum environments well before the first spacecraft was launched. Although NASA rapidly established a criterion for the outgassing of material to reduce the obvious contamination, it wasn’t until the life expectancy of spacecraft systems dramatically increased that the magnitude of problems were known; the longer they were in use the more deposition would form. Today, a similar model is playing out in wafer manufacturing. Wafers are increasingly exposed to, and masks are stored for years in, an atmosphere of outgassing molecular contamination reminiscent of the early space program.
Trends in wafer manufacturing
“Several global conditions influence and impact the wafer manufacturing industry, including globalization, the reduced number of tool makers and chip makers, and the growth of the market in Asia,” says John Higley, worldwide GPS director for the Gas Microcontamination Control Group at Mykrolis Corp. “The technology roadmap continues.”
“I like to describe the environment in terms of concentric circles of cleanliness,” says Higley. “There’s the outdoors, where all kinds of molecular dogs or cats live together; then there’s the cleanroom and its level of cleanliness. Inside the tool is another level, as is the wafer surface and the point of use at the lens. Yet another level is actually inside at the lens surfaces, where there’s high-purity gas or purged gas. So there are multiple layers of cleanliness and all have molecular requirements that require filtration. Measurement and control are required at each level, and achieving that becomes increasingly critical and more difficult,” adds Higley.
Molecular contamination expanded from being a problem that was solved by simply adding a chemical filter on a tool to a whole new complex level that requires the industry to measure for acids, bases and organics. Companies are finding that filters can fail-or last two to three times longer-depending on whether or not they are monitored and contamination is adequately measured. And, attempts to lower the cost of ownership are always at the forefront.
According to Jürgen Lobert, metrology group manager with Mykrolis, and Dr. Robert Andersen, principal scientist with Analytical Services, “Fabs are cleaning up over time. Just five years ago, concentrations of ten to fifteen parts per million of ammonia existed. What we measure today, especially in 193-nm fabs, is a concentration of around 5 ppm in the ambient air. Some of our customers are able to reduce concentrations even further and it’s no longer unusual to see ammonia concentrations of 3 ppb in cleanroom ambients. In general we see a steady improvement in air quality with respect to gas contamination.”
Not everyone is as optimistic. “Molecular contamination measurement and control requirements are part of a roadmap that is increasingly complex, particularly in lithography with 193-nm wavelengths. Today, 248-nm lithography is mature. In 193-nm [processes], smaller wavelengths react more readily with contaminants, causing problems. The lens in the 193-nm exposure tool is much more expensive so contamination becomes more costly. The chemistry is more complex, and the equipment is far more expensive. There is a demand for more complex measurement and more complex filtration,” adds Higley (see Fig. 1).
Over the past several years, lithography has moved from 248-nm, to an attempt at 157-nm, and into the adoption phase at 193-nm processes. “We knew it was going to be tough. People were used to operating mature 248-nm systems; they knew they could change the filters every so often-two years, six years-no problems. Now, with a 193-nm tool, the chemistry is a lot more complex. They’re still filtering out bases, but now they’re also filtering acids and organics, and some organics are more difficult to measure. All of a sudden, they’re getting lens contamination; and the rooms where the tools are don’t seem to be clean enough. These are the same types of issues that came up when 248 nm was new, except now it’s not just bases we’re concerned about-it’s a range of organics and acids as well,” adds Higley.
For a brief time, the industry attempted to migrate to 157 nm-where acids, bases and organics that today threaten 193 nm were a larger threat still. The shorter the wavelength, the easier it is to literally break molecules apart and create deposition on lenses. The 193-nm lithography process is conducted in ambient cleanroom air without being purged with an inert gas such as nitrogen. It’s more forgiving than 157 nm, but more sensitive than 248-nm lithography.
According to Lobert, the challenges became too great with 157 nm, especially with respect to contamination control. “The 248-nm process is able to tolerate a few ppb of ammonia, the 193-nm process cannot tolerate anything more than .8 to 1.0 ppb of ammonia, while the sensitivity of 157 nm was just too high to accommodate any contamination. That’s one of the many reasons the program died-it was just too expensive,” says Lobert. The 193-nm immersion alternative is much more promising, using existing technology and adding a small amount of water at the very last process step. “It’s easier to handle and less expensive than 157 nm.”
“When you’re looking at a lithography tool with a $5 million lens inside operating a molecular control system with multiple complex filters, different process gases, and room air, it’s akin to flying an airplane without a fuel gauge. The consequences are much higher than just losing a few wafers from ammonia contamination. Now we’re looking at losing the usefulness of a tool, taking it out of operation for days or weeks while trying to clean the lens-that’s a huge deal,” adds Higley.
Figure 1. This mock-up shows the complex filtration and purification processes used throughout Mykrolis???s photolithography environment. Photo courtesy of Mykrolis Corp.
The real ramp-up of 193-nm lithography is just occurring, resulting in an increased demand for complex measurement-the measurement of acids, bases and organics that are fugitive in the ambient air, purged gas, and high-purity gas that serve 193-nm lithography. The rate of evolution of 193 nm is related to measurement. “With 193 nm, you really have to measure for acids, bases and organics. There’s a push for real-time monitoring. You’re looking for organics that are particularly damaging and monitoring is not available for everything,” explains Higley.
Several standards bodies actively address airborne molecular contamination (AMC). The IEST, as Secretariat for ISO TC209: Cleanrooms and associated controlled environments, completed a draft international standard (DIS) affecting classification of AMC in October 2004. Currently, technical comments are being incorporated in a final draft that is expected to go out for vote in the third quarter of this year. It is anticipated that there will be an ISO standard for airborne molecular contamination by the end of 2005.
The standard will not be industry specific. Instead, it specifies a classification system for identifying and quantifying airborne molecular contamination, offering nine sampling methods and twenty-seven methods of analysis as a starting point. In comparison, IEST Working Group CC035 is in the earlier stages of working on AMC guidelines that are expected to be industry specific. Ultimately, however, it is the user that must make the final determination as to whether or not the contamination levels that exist are detrimental to a company’s process.
“In general, the challenge is that airborne molecular contamination may exist for years in an environment without causing an issue. Once a problem is identified, such as corrosion on a chip, however, it’s detective work to understand its cause,” states IEST’s Dick Matthews. “How much contaminant is there? Where is it coming from? Is it generated inside the room or transferred through the air-handling system? Molecules go through air filters. Companies may do a wonderful job at removing particulates, but there may be gases that require another stage of filtration to prevent them from entering a controlled environment. As we get further into nanotechnology, for example, it will be an issue of how many different stages of a gaseous filtration system might be required to prevent contamination to the product. A particle is easy-it’s one thing. A gas is multiple things; there’s a big difference,” adds Matthews.
A majority of the standards activity is being driven by the microelectronics industry as it encounters specific problems. To date, molecular contamination is not as significant a concern within the pharmaceutical and medical device industries.
The IEST’s responsibility and mission is to take the generic criteria created by ISO and create industry-specific recommended practices-that’s what Working Group CC035 is doing. ISO also is considering creating a standard for surface molecular contamination (SMC) as a companion to its AMC document. “The result of AMC is surface contamination. A molecule doesn’t necessarily cause problems until it parks itself somewhere,” says Matthews.
Simultaneously, the International Technology Roadmap for Semiconductors (ITRS) wafer environment contamination control (WECC) is attempting to create guidelines for airborne molecular contaminants regarding the level to which air quality should be controlled. “We’ve controlled air quality for particles for many years,” says ITRS WECC Chairman John DeGenova. The WECC is attempting to set guidelines for air quality in response to the numerous defects that appear at various process steps, depending on the contaminants in the air stream and the length of time wafers are exposed. Air is the most abundant chemical a wafer encounters, and at every step of the process. “Yet we’ve never considered it a process step in the past,” DeGenova admits. “Now we have to realize that wafers are being ‘treated’ with air in between process steps.”
“We probably should have pushed standards years ago. Now, with smaller feature sizes, it’s becoming more critical. We seem to run so fast, advancing technology, [that] at times we get ahead of ourselves. We spend too much time getting to the next technology node rather than really understanding where we are with the present one,” adds DeGenova. “Yet it’s a hard sell to management to convince them that this is serious. It’s a difficult concept that there is a whole chemical factory inside the air we breathe. What makes them a believer is loss of yield.”
Chris Muller, technical services manager at Purafil, Inc., serves on the Yield Enhancement Working Group of the WECC. He explains, “The WECC is setting levels for AMC and SMC. The roadmap is updated every odd year, so a vote is coming to approve the 2005 roadmap. In even years, there are interim releases in case anything needs to be changed that can’t wait until an odd year. At this point we’re actually raising some levels because we could not justify keeping them as low as they were.”
Muller clarifies the group’s position: “In order to keep it realistic in terms of what’s achievable both in terms of measurement and control, we can change the levels to make it easier to monitor and control. We are breaking things down a little differently since the wafer environment has changed from a few years ago when the wafer environment was the cleanroom. As we move forward, the environment is actually shrinking to the FOUPs [front-opening unified pods]. Even though the environment is smaller, the concerns are the same.”
The potential problems are many. For example, outgassing within the FOUPs causes potential adhesion failures. If anything adheres to a wafer’s surface, it may affect the lithography pattern, and acids can chemically react with metals to form corrosion or a chemical reaction. Today, width lines are so close together that a gas molecule or chemical molecule may bridge the lines, causing short-circuiting. A number of microfailures may exist without causing problems; however, a number of microfailures can collect throughout the processing, leading to a macrofailure, which causes device failure.
The WECC consists of diverse organizations-semiconductor companies, suppliers, and laboratories-that are coming together to establish a realistic guideline based on what’s currently known and the anticipated needs of the future. “There are so many variables in our processes, it’s really hard to pinpoint which particular variable is critical. How clean is it? How clean does it need to be? How do you know how clean it needs to be without a lot of experiments?” DeGenova asks. “For instance, there may be 100 parts per billion of ammonia in the air. The level may be problematic at one step, but not in other steps. There are hundreds of different contaminants-acids, bases and organics. Which ones are critical, and at what levels?”
Additional real-time instrumentation has been identified as a critical need. “No one instrument tells us everything we need to know,” DeGenova adds. Within the standard mechanical interface (SMIF) environment things are changing as well. “The minienvironments are, on one hand, protecting wafers from some contamination, and on the other, given that they’re plastic and outgas, they are a source of AMC. They may protect wafers from an acid in the air, but then contaminate wafers with an organic that comes directly from outgassing. We need to understand the materials we use, the outgassing characteristics of our tools, building materials, and everything that goes into a cleanroom,” says DeGenova.
Mark Camenzind, senior technical advisor at Air Liquide America Electronics, Balazs Analytical Services, and SMC specialist, says that the same problems have been happening for twenty to thirty years. “The good news is that the ITRS covered AMC already, addressing most issues except AMC on steppers to avoid optics hazing. The roadmap two years ago added surface molecular contamination. We are looking at witness wafers exposed to cleanrooms for 24 hours, or to FOUPs or boxes, and setting specs for the amount of organics that land on the wafers. We also set specs for dopants and specs for metals landing on wafers within 24 hours. Table 114a of the 2004 Update provides AMC and SMC levels for witness wafers. What must be added are specs for reticle storage-and that’s being worked on right now.”
Wafers are exposed to air for a few hours at a time. Masks will be used for a year and cost approximately $3 million. A question becomes: How clean an environment is required for mask storage boxes and reticle stockers where masks are stored in a kind of “library” waiting to be used? They spend years in this storage and could be affected by the outgassing of plastics or contaminants in the air. The main concern is that organic compounds that stick to the reticle-sulfur, silicon, and phosphorous-leave nonvolatile residues on the lens that will build up over time, requiring the masks to be cleaned.
Although the ITRS provides a guideline, issuing one value for a particular area, Camenzind notes that clients, depending on their actual processes and products, may have completely different sensitivities. “What is new is that we have methods to analyze contaminants at the levels needed to find them when they cause problems. Now we can look at organics on wafers using a standard test method (SEMI MF1982-1103) for organics on wafers to one-thousandth of a monolayer. We can detect dopants down to below one-thousandth of a monolayer, which can affect some processes, and metals to a millionth of a monolayer.”
Are standards enough?
Though standards activity will certainly bring attention to the issues, questions remain whether it will be enough. “Even today, companies don’t understand airborne contamination,” says Latif Ahmed, technical support manager for Chemtrace. “Every day we see problems in the lab and people are requesting tests, yet they don’t understand the significance of the tests. They often will not follow recommendations; and there’s a basic lack of understanding of how important AMC and SMC are. It often takes being in crisis mode to finally get it.”
Ahmed admits that the industry also hasn’t provided the necessary guidelines in some cases. “There isn’t any standard, for example, as to the organic contamination present due to the manufacturing processes. They go through a lot of cleaning and through a certain process where there’s a possibility that the organics can be left behind on the chamber body or the chamber components. It is not known at this point exactly to what level those organics are harmful to the manufacturing processes.”
Organic contamination can be easily attributed to other problems as it sometimes manifests in odd ways such as hazing during a drying process. Finding the source is difficult. “It’s not only the yield loss from the wafer point of view that accounts for monetary loss. There is also tool performance and time between cleanings, the time it takes to bring the tool up-all of these are problematic,” explains Ahmed.
Another problem is that many companies are operating without a baseline understanding of their existing AMC and SMC levels, which is critical so that testing means something. A baseline can be established, even after a facility is brought on-line, by routinely monitoring on a monthly basis for a few months and charting out contamination levels. Once an understanding of contamination levels is achieved, the facility can monitor quarterly to see if there is any incursion.
There are many potential culprits of molecular contamination and all materials coming into the fab should be qualified. For example, gloves and other consumables can be contributors, but are often ignored. Air-handling systems should be designed so that they do not cross-contaminate each other. HEPA filters, floor tiles, paints and humans are also to blame.
Each fab has its own standard levels and sets its own criteria. Suppliers of AMC and SMC controls attempt to look at the big picture, balancing the total scope of what can be accomplished against what companies are willing to do. From one company to the next, even though they make the same devices, diverse materials are used. As a result, there are few “cookie cutter” solutions.
Although concern for particles far overshadows molecular contamination control, tests are continually improving. Multiple standard methods are now available with adequate sensitivity to find contaminants on the wafer and in the air at levels that affect processes. The next challenge for companies is to understand the levels that affect them and their processes, set specifications, and utilize protective measures such as purifiers or carbon filters to remove contaminants to a level where they no longer are likely to affect yield.
“The future will be challenging in the areas of immersion lithography. How clean will the water or other fluids that may be used need to be? There are thinner barrier layers, thinner gates and copper layers, and the potential contamination of tools. All these issues are a huge focus for semiconductors and reticles, but they also apply to flat panel displays, disk drives, lasers and other optics, across multiple industries,” says Camenzind.
Sameer Abu-Zaid, senior manager of contamination control for Asyst Technologies, Inc., believes that the industry will become more aggressive in controlling AMC in both front-end and back-end processes, as well as integrating AMC measures in manufacturing facilities that incorporate early detection control and active protection of the wafer environment (see Fig. 2). “In situ monitoring will facilitate early detection of AMC and off-line measurement will be used to limit AMC damage to wafers.” III
SMC in space
When deposition first clouded the optic lenses of the spacecraft instrumentation, NASA rapidly began to solve the dilemma. Initial studies performed by the Jet Propulsion Laboratory (JPL), under contract to Stanford Research Institute, culminated in the publication of a two- volume report. The spec that resulted evolved into the American Society for Testing and Materials (ASTM) Standard E-595, a quick-screening test for outgassing. Under the test, materials must meet requirements for total mass lost in 24 hours and for collected volatile condensable material in the same 24-hour period. The test, used internationally and modified over the years, is referenced in an ISO standard for space system contamination and cleanliness control. A more sophisticated test, ASTM 1559, focuses on outgassing but goes further to validate materials and model what happens in a space environment.
Of major consequence is the effect of radiation on the deposition of molecular contaminants. “Radiation enhances the deposition process,” according to Gene Borson, contamination control engineer at Swales Aerospace (Beltsville, Md.). “In space, ultraviolet radiation from the sun causes materials that would normally not condense, even if they’re low molecular weight, to form on the surface and degrade. In this environment, for example, an optical window darkens in time.”
When contamination occurs in launched space aircraft, with the exception of a space station or the Hubble telescope, literally little can be done to clean or replace parts. Within the extreme ultraviolet (EUV) range, where wavelengths are very short, it takes less contaminant deposition to affect instrument performance. “The requirements within the space program become so stringent there’s a point where it’s impossible to measure or verify them directly,” adds Borson. “So we control our procedures to prevent deposition of contaminants in the cleanroom, [are] careful in the assembly process, and do everything possible to control the environment to keep deposition to a minimum. It’s all about monitoring the environment and following extremely stringent procedures.”
Two methods used to monitor for deposition include quartz crystal microbalance and surface acoustic wave (SAW) microbalance. In the first case, a quartz crystal oscillates and a slight change in mass of an exposed quartz crystal affects the frequency of the crystal, which can be measured and related to the mass that deposits on the surface. The SAW microbalance has a slightly different oscillation mode and it is used to monitor an environment to indicate any increase in deposition.
“There’s a lot of work being done with SAW devices today. As electronic circuits get smaller, molecular contaminants have a greater effect. Today’s manufacturers are going down into the ultraviolet wavelengths to generate circuits. And, the UV itself may also have an affect on the deposition process if airborne molecular contamination is present,” said Borson. “We were aware of this in the space program long ago. Now they’re looking for ways to control the environment and keep contamination low and yields up in the semiconductor industry.”