Squaring off against new contamination challenges
The semiconductor industry continues to present challenges to contamination control scientists and solution providers, and the expectation is that with each new generation of semiconductor products will also come a parallel set of new contamination concerns.
BY JOHN WILLIAMSON
Technology constantly advances, with the impossible of yesterday frequently found to be commonplace today. Given this truth, some might wager that the semiconductor industry will likewise someday completely conquer the challenges of contamination in its manufacturing processes. But, not all industry leaders see it that way, suggesting the pursuit of cleanliness would be more accurately regarded as a lifetime occupation.
In either case, however, the semiconductor industry's assault on contamination is continuing on many fronts—from product isolation, to improved process equipment and filtration systems, to the sourcing and control of airborne molecular contamination (AMC).
Ken Goldstein, principal of Cleanroom Consultants, Inc. (Phoenix, Ariz; www.cleanroom-consultants.com), doesn't see any "magic" in today's cleanroom environments: "The basic equipment is approaching commodity status, and there's heavy price competition." Instead, Goldstein sees future technology investment being made in materials handling and control, AMC, and minienvironments. "These are the tools, and these are the areas where major investments will be made," he suggests.
Over the long term, Goldstein says minienvironments will rule as products are isolated from the cleanroom itself. "We probably won't see a lights-out fab," he comments, "but we will see increasing investment in minienvironments. There'll be a trade-off between the costs involved in the tool sets and the need to recoup these investments as quickly as possible—ideally, after one year of production."
Phil Naughton, Freescale Semiconductor (www.freescale.com), assignee to International Sematech (Austin, Tex.; www.sematech.org), agrees: "Today, the mini-environment is the factory of standard design and we must take advantage of that and look at the impact of these mini-factories on the main factory."
Naughton points out that some firms respond to this trend by allowing higher levels of contamination in the main plant, relying on their minienvironments to protect the product, but notes this raises another question: 'What is the minimum level of cleanliness that can be tolerated without risking product quality?' "The industry is becoming more aware of contamination issues that can occur within minienvironments, but may not have the answers to this question," he says.
Arizona State University's Dr. Alan Chasey, who serves as director of CREATE (Construction, Research and Education for Advanced Technology Environments; www.create.asu.edu), warns that minienvironments don't obviate the need for caution. "With the trend toward minienvironments and pods, fab operators may elect to reduce the level of contamination control elsewhere, say from a Class 10 to 100 to 1000," Chasey says. "But as this is done, concern should mount as to what happens when equipment is opened for cleaning." Chasey suggests one approach may be the use of "mini cleanrooms," into which pods can be placed for cleaning and maintenance.
Today, the product is the greatest challenge of contamination control, says Chasey. "In the old days, people were the dirtiest items in the fab, but with mini-environments, the question to be answered is 'what's going on in the machine?' Everything that goes into the mini-environment must be compatible with the process. For example, FOUPs (Front Opening Unified Pods) may give off emissions that contaminate the wafers. Robotic operations also wear down, and where do those particles go?"
Theron Colvin, industrial engineer with Microchip Technologies, Inc. (Tempe, Ariz.; www.microchip.com), makes a similar observation: "Most contamination is caused by the equipment itself. This means we must work on cleaning the processing tools so they don't create damaging particles between process steps and chambers."
One way to approach this, Colvin says, "is to first get the process to work, and then clean it up."
Materials and material handling
Both process and material-handling equipment face increasingly stringent requirements as line diameters and wafer thicknesses decrease. Overall factory design can also create impediments to optimum contamination control. For example, says Colvin, "when material-handling equipment migrated from tracks on the floor to overhead conveyors, this immediately also constrained the process managers' ability to facilitate changeout. Factories must be built with the flexibility to do this as equipment changes in size and shape."
Colvin also isn't convinced that the current generation of FOUPs for 300-mm wavers will take off: "The industry must be concerned about the environment between the FOUP and the process tools. It would be productive if the industry standardized on a load port."
"As technology evolves within the fab, all process changes have to be taken into account," says Patrick McKinney, president of the semiconductors group at Entegris, Inc. (Chaska, Minn.; www.entegris.com). "The cry is 'please don't change things within my systems —it will change things elsewhere in the production line.'"
"In short," says McKinney, "to address contamination control efficiently, we must look up- and downstream of a proposed change to see how it impacts product quality." He notes these concerns extend to all components used in the production cycle—the plastics, their potential for outgassing and how to address it, how to handle electrostatic discharge, and what the impact of that solution has on the production process. Among the list of concerns:
- Tight control of materials within wafer baths;
- Measuring and controlling the flow of process chemicals (for example, rate of flow can damage wafers);
- Specifying very pure materials in manufacturing and engineering, then controlling the chemical usage at the point of use;
- The effect of a change in processing materials on the bulk shipping containers used to transport them from the manufacturer to the fab.
Wiley Wilkinson, Entegris marketing director, points out that while "most fluid-handling products are OK, as chip lines shrink, contamination becomes more of an issue. In other words, fewer contaminants cause more problems."
To help address this challenge, SEMI (http://www.semi.org) has established a set of guidelines for specifying fluid-handling systems. For example, SEMI F57-0301 specifies polymer and components for ultra-pure water and chemical distribution systems. (See CleanRooms, January 2004: "Using SEMI F57 compliance to improve semiconductor fluid-handling systems.") Others are SEMI E49.2 and .3, which provide guidelines for polymer assemblies used in ultra-pure water and chemical distribution systems. Other standards apply to gases, wafer handling, and other products and components associated with the production of microelectronic devices.
Wilkinson observes that these standards become more important in single-wafer processing where different chemicals and concentrations may be used when switching from one product to another. "Non-compliant piping and fittings may provide places for entrapment despite flushing the lines between batches," he explains. "These chemicals can precipitate and impact the processing of the next batch. OEMs and end users should be assured that the product meets specifications, either through their own testing or, more ideally, through an independent testing laboratory."
Barry Gotlinksy, Ph.D., vice president of scientific laboratory services at Pall Corp. (Glen Cove, N.Y.; www.pall.com), says the drive toward greater production yields and greater product performance in the face of shrinking device geometries has also intensified the need for filtration at every stage of semiconductor fabrication. "As a general observation, we're solving contamination problems today that in the past seemed insurmountable," says Gotlinsky. He points to process lines where classic filtration is now down to nanometer-scale sizes, noting that purification is increasingly important as line sizes and thicknesses decrease, and with changes in the types of chemicals being used.
"We've found that while UHP [ultra-high-purity] water may meet spec, batch variations within the spec affect product performance," Gotlinsky says. "Therefore, steps must be taken to ensure consistency, and this is accomplished with advances in filtration equipment, and installing the equipment at the beginning of the line."
The performance of air management devices is another area requiring renewed attention within the changing shape of semiconductor production. Ron Sumner, director of marketing at SemiNet, Inc. (Valencia, Calif.; www.seminet.com), points out that the introduction of closed pods has had a significant impact on air-handling system requirements: "As long as the transit areas are protected and maintained at reasonable cleanliness levels, the contamination pulled into the process areas by piggybacking on the pods is reduced; but airflow velocities can impact this."
Sumner notes that, historically, airflow velocities have been specified higher than they needed to be. He recommends that when minienvironments are used, air velocities should be lowered where possible to minimize turbulence that could potentially let particles migrate into process areas.
Another common problem can be sourced to the design of the enclosure itself. "Horizontal surfaces that can deflect or trap the laminar airflow are symptoms of inadequate design, but cannot always be avoided," says Sumner. "Airflow management techniques can sometimes overcome the limitations that can't be designed out."
When designing any minienvironment system, Sumner says one should first determine whether the system is going to exhaust the airflow directly into the ambient fab air or into a dedicated exhaust system, or whether there will be a need for a recirculation system to control particulate or AMC contamination.
Noting that cleanroom airflow is also used to control room temperature, humidity, and pressure, Sumner says it becomes even more critical when a recirculation system is being used. Temperature and humidity are not flushed out as in a single pass system, and any temperature increases generated by the process will also be recirculated and build up over time.
"For example, if electronic equipment is generating heat, the need for cooling may be increased dramatically," Sumner explains. "To achieve this, a certain amount of 'cold' air must be injected into the room." Sumner suggests a make-up air environmental control unit (ECU) that introduces conditioned air into the recirculated air stream as one way to control these parameters. "We also see a growing need for recirculation environments using high-purity nitrogen to protect the process," he adds.
AMC remains a major challenge for all advanced semiconductor manufacturers. As described by Arizona State's Chasey, "Controlling AMC requires an understanding of the physics tied to contamination problems, and every generation of product presents new chemistries and new reactions. Therefore, our approach to contamination control must move with each technological advancement. It demands flexible thinking and perhaps a willingness to design for one process, build for a second, and produce under a third."
John Plata of Texas Instruments Worldwide Construction Strategic Planning sees AMC as the issue of today and tomorrow. "What we have to solve is where contaminants are coming from, how serious they are, and how we can address them," he says.
Plata adds, however, that the industry also has to learn how and where to monitor processes to accurately determine this information: "AMC is not wafer-diameter specific; it extends across the industry, and each attempt at minimizing AMC can introduce another issue."
For example, Plata notes, if the coated concrete walls in a minienvironment service bay are epoxy-based, then there are new concerns, such as whether the epoxy will outgas and become another source of contamination. "It's a never-ending learning process. We've built and equipped generations of fabs, and yet each time we learn more about AMC."
Standards and standards development efforts remain a major concern of both the semiconductor and contamination-control industries. As described by Cleanroom Consultants' Ken Goldstein, "For the most part, there are no 'real' rules and regulations as there are in the biotech and pharmaceutical industries, where FDA standards apply and are down on paper." Instead, Goldstein sees the semiconductor industry relying largely on self-policing efforts, "with some 'standards' being good and some not so good," and most more accurately termed recommended practices. In terms of airborne cleanliness, however, Goldstein says, "the microelectronics industry is far ahead of both the pharma and biotech industries."
Though the International Standards Organization (ISO) has released its 14644 series of standards and draft standards for semiconductor manufacturing, Goldstein says there is significant inconsistency in their value to the industry. For example, he says, "14644-1 is good in updating, extending, and supplanting Federal Standard 209E, but the others are not as clear or as realistic. [14644-3] is acceptable but not perfect in monitoring ongoing operations, and presents a target to achieve; [14644-3] and beyond fall short of telling what to do and how to do it regarding designing, planning, and working in cleanrooms. There are no 'thou shalts' or ways presented to quantify performance."
Goldstein's concern is that although the standards may state the benefits of a particular approach (such as ISO 14644-4, which says, for example, that there must be a clearly agreed-on tool list and utility matrix prior to design), "they are unrealistic in that what we're really doing is trying to plan for future manufacturing while the tools and the processes are undergoing continuous revision." Instead, he says the industry must design for flexibility and be able to adapt to change.
Goldstein's biggest concern, however, is that ISO 14644-6 has yet to be released. This standards document deals with the vocabulary of semiconductor manufacturing and, as noted by Goldstein, "Normally, you want an agreed-upon set of definitions when discussing any contentious subject in order to avoid the 'tower of Babel' effect." (The ISO 14644 series was prepared by ISO/TC 209, and can be found on the Web site of the Institute of Environmental Sciences and Technology; www.iest.org.)
Risk assessment is part and parcel of any semiconductor industry trend analysis, and this is certainly true wherever contamination control is concerned. As pointed out by Arizona State's Chasey, "it's an up-front investment situation. A 300-mm fab can cost $3 billion, with 80 percent of that in tools, the balance in the facility itself, planned over a five-year stretch with payback starting in a minimum of 15 months."
Risk analysis and perception
The implementation of minienvironments is certainly one major consideration driven by risk assessment and analysis. "Manufacturers must be comfortable with the minienvironment approach," says Chasey. "Some will be and some won't be, but all will make the choice based on their assessment of risk—of losing product."
Though Freescale Semiconductor's Phil Naughton believes that, in the drive toward cost reduction, the opportunities lie outside the minienvironment, he agrees it ultimately comes down to risk management. "How much can we 'degrade' the overall environment? What is the minimum level of cleanliness that can be tolerated without risking quality of production?" he asks. Naughton observes that while some companies are permitting a relatively high degree of degredation outside their mini-environments, they're balancing their risks with measures that ensure products are not exposed to this environment.
Citing the extensive investment being made in automating the production of flat-panel displays, Theron Colvin at Microchip Technologies observes that Asian firms are much less risk-averse than their U.S. counterparts. "U.S. companies analyze things to death, whereas Asian firms are much more willing to take risks, learn from their mistakes, and move on," Colvin says. "They see risk as an opportunity to succeed."
Given the price tag associated with the processing of a single 300-mm wafer, and the multibillion dollar up-front investment in 300-mm fabs, contamination control will clearly remain a critical concern for semiconductor manufacturers, whether within or outside of minienvironments. As the future brings a continuous evolution in semiconductor technology, it will also demand an evolution in how semiconductors are manufactured and in the contamination-control technology employed to protect both product and personnel.
JOHN WILLIAMSON is a freelance writer living in Lebanon, N.J. He can be reached at: firstname.lastname@example.org
Focus on green, efficient and safe
As semiconductor production is increasingly handled in minienvironments, the quality of both the workplace and natural environments—as well as that of the product—may be impacted. Says Cleanroom Consultants' Ken Goldstein, "We can envision manufacturing in an office environment, a shirtsleeve and plant-on-the-desk environment, with the actual fabrication being done in pods. In this type of setting, there'll be less need to worry about temperature, air particles, and humidity. And it's in this scenario where opportunities for improvement lie."
Robert Morris, president and CEO of Flow Safe Inc. (Denville, N.J.; www.flowsafe.net), agrees; in fact, he sees a growing trend toward "green" cleanroom design, one that is both worker- and environment-friendly. "The semiconductor industry, as it undertakes new construction, will implement experiences from the chemical laboratory, installing systems that make better use of water and air while protecting the worker and the environment at a higher level," says Morris. An additional benefit, he claims, will be reduced operating and capital costs from more energy-efficient equipment.
To get there, Morris says the industry will need to think beyond just the use and impact of minienvironments but rather of the entire system—and of the long-term costs of equipment rather than the initial price tag. He suggests, "Think downstream and consider the statement, 'We have a wonderful product but the building doesn't work.'" Among possible useful resources, Morris points to the U.S. Green Building Council's LEED program (Leadership in Energy and Environmental Design www.usgbc.org/leed/leed_ main.asp), which works to boost sustainable design and construction in the U.S. Morris predicts,"as the semiconductor industry begins to grow and reinvest, green applications will take greater root. It's a necessity."
Improvements to processes can also benefit the environment. As an example, Pall Corp.'s Barry Gotlinksy points to the use of ultra-high-purity (UHP) water: "Many of today's filter products are constructed of all-fluorocarbon components, such as polytetrafluoroethene (PTFE), but the difficulty in using PTFE for membranes is that they are hydrophobic and must be pre-wet." Traditionally, this has been accomplished using a low-surface-tension fluid, such as isopropanol (IPA). But as noted by Gotlinsky, "with increasing focus on making today's semiconductor fabrication processes more 'green', such methods are losing popularity."
Today, observes Gotlinsky, these filter membranes can be delivered and stored pre-wet, eliminating the need to use low-surface-tension organic wetting solutions in the fab. "This minimizes environmental exposure and provides a significant advantage to semiconductor manufacturing plants," Gotlinksy says. "It also increases the throughput of a typical point-of-use chemical wet bench as it reduces the amount of time required to pre-wet and 'rinse up' filter cartridges during a change-out period."
Bernie Frist, managing principal of Environmental and Occupational Risk Management, Inc. (Sunnyvale, Calif.; www.eorm.com), points to another opportunity for improving the work environment and to further reduce risk to employees during equipment maintenance (such as when purging chambers or opening lines). Referring to exhaust ventilation and to Semi S2-0703 (covering the design of semiconductor manufacturing tools and the control of potential chemical exposures: Environmental, Health, and Safety Guideline for Semiconductor Manufac-turing Equipment), Frist notes that "section 22 suggests that equipment exhaust ventilation be designed to prevent potentially hazardous chemical exposures to employees under three types of operating conditions:
- As a primary control when normal operations present potentially hazardous chemical exposures by diffusive emissions that cannot be otherwise controlled, such as with wet stations;
- As supplemental control when intermittent activities present potentially hazardous chemical exposures that cannot reasonably be controlled by other means, such as during maintenance;
- As a secondary control when a single-point failure presents the potential for employee exposure to hazardous materials, and this exposure cannot be controlled by other means."
In addition, section 23 of the guideline pertains to chemicals and suggests that there should be no chemical emissions to the workplace environment during normal equipment operation, and that chemical emission during maintenance activities or equipment failures should be minimized.
Says Frist, "The guideline provides airborne chemical control criteria to demonstrate conformance. Ambient air concentrations must be less than one percent of the Occupational Exposure Limit (OEL) in the worst-case breathing zone during normal equipment operation, and less than 25 percent of the OEL in the anticipated worst-case personnel breathing zone during maintenance activities or realistic worst-case system failure."
Other standards and guidelines apply to equipment and materials used in the cleanroom, and can be found on the Semi Web site: www.semi.org—JW