Cleanroom facelifts more than skin deep
Well-planned cleanroom-retrofit investments can simultaneously provide advanced contamination control technology while ultimately adding to the bottom line
By John Haystead
Industries from semiconductor to disk-drive to pharmaceutical and food processing are all looking for ways to increase production and lower operating costs while minimizing capital outlays. The solution is often found in retrofitting and upgrading existing cleanroom facilities. While the scope of a cleanroom retrofit program can range from a relatively small upgrade of individual process stations to a facility-wide overhaul, one of the key benefits of a retrofit over new construction is that it allows companies to make incremental, low-risk capital investments.
According to research concern, The McIlvaine Co. (Northbrook, IL), the cleanroom retrofit business accounted for $1 billion of the total $2.8 billion worldwide cleanroom construction market in 1998. And, while the semiconductor industry continues in a period of reduced profits and tight budgets, it, in particular, is turning to cleanroom retrofit programs to keep pace with advancing submicron device technology. As linewidths shrink to 0.15 micron and below, the control of particulate and molecular contamination, directionality of airflow, temperature and humidity become ever more crucial.
As one of the earliest and largest implementers of cleanroom technology, the semiconductor industry also has a pressing need to upgrade many of its facilities. As Bob Esser, general manager, Midwest Cleanroom Associates (Grand Rapids, MI) notes, “many first- and second-generation cleanrooms built in the 1970s and 1980s can benefit from retrofit. The majority of these are in the microelectronics industry, which has generally had facilities the longest.”
Moving up in class
According to Esser, the most common goal of cleanroom retrofit programs is to increase the classification level, such as from Class 10,000 to 1,000 or from 1,000 to 100 or 10. As a result, most typical retrofit programs start with the removal and replacement of existing ceiling filtration systems with more air filtration units, but this requirement then dominoes into other areas such as the installation of more efficient heating, ventilation and air conditioning (HVAC) and greater capacity chilled-water systems, as well as architectural changes to handle the removal of air from the room.
For example, MCA recently completed a project with Dupont Photomask (Kokomo, IN) which began using cleanrooms in 1983, but by 1995, had started moving to much higher classification levels to keep pace with the semiconductor industry`s increasingly smaller geometry requirements for photomasks. Dupont has now upgraded several production areas from Class 100 to Class 1 to accommodate advanced process lines, including suites to hold electron beam lithography and inspection tools. Most of the company`s work is currently 0.35 micron technology, but it now also has 0.25 micron technology in production and is planning for 0.18 micron technology. Says Mike Ward, plant manager at DuPont Photomask, “Although we began looking at our upgrade program from scratch, we`re now planning for future growth such as the addition of suites to our advanced process line and making sure we have adequate chiller capacity to accommodate it.”
Paul Enderle, sales manager at Clestra Cleanroom (N. Syracuse, NY), agrees: “there are a lot of ways to attack a retrofit program, largely dependent on the end-user industry requirement.” For example, Enderle observes that while the classification level of an electronics assembly facility may be best upgraded through the installation of new fan-filter units into the existing ceiling system, a semiconductor manufacturing environment may be better served by minienvironment enclosures that provide localized ultraclean spaces in lieu of upgrading the entire facility.
Scott Mackler, Clestra marketing manager, notes, even if a decision is made to go with a minienvironment approach, there are still different levels of installation possible. “One approach is to hang fairly simple enclosures off a new or existing ceiling grid system with or without upgrading the existing HEPA coverage; for more critical processes, such as photomask or CD-ROM production, an entire standalone minienvironment system may be installed with its own dedicated environmental control unit — either sub-fab, above the ceiling or in some cases in-fab.”
Old fabs work like new
Asyst Technologies (Fremont, CA) recently announced an alliance with architectural/engineering firm ADP Marshall (Rumford, RI) aimed at pursuing the semiconductor retrofit market from a turnkey perspective. According to Sal Abuzeid, Asyst vice president for Semiconductor Fab Business Development, the ongoing push to 300-mm wafers, together with continually shrinking die sizes, is combining with the industry`s current downturn to make retrofit an increasingly appealing alternative to new construction. “The cost of a total facility upgrade to accommodate 0.18 or 0.15 micron technology is only a fraction of that of building a new fab — on the order of $50 million vs. $300 million.”
The Asyst/ADP Marshall approach to retrofit starts with an extensive operational analysis study aimed at improving the fab`s performance and reducing costs. Says Abuzeid, “many of the fabs built into the late 1980s are now reaching the end of their useable life. Upgrades can provide a substantial mid-life kicker.” Abuzeid points out optimizing the infrastructure of these fabs not only allows state-of-the-art devices to be manufactured in older facilities but provides cost-effective productivity enhancement as well. “If you`re making 0.5-micron devices, you`re not going to reap the same premium on the product as someone doing 0.25 micron.”
Although most older 8-inch fabs are largely distributed evenly around the world, Abuzeid notes that in the case of 6-inch fabs, Japan offers the largest potential opportunity for retrofit, since both more and larger 6-inch facilities are located there. In fact, particularly for Japanese semiconductor companies, tight budgets and time-to-market constraints may have more impact on any decision to swing away from ballroom facilities toward minienvironments and closed-container automated handling systems than an overall change in manufacturing philosophy. Though Abuzeid says it`s still too early to tell how the Japanese companies will eventually respond, he observes that they appear to have the most to gain from such a retrofit strategy and are “certainly listening more closely now.”
The Asyst study encompasses all aspects of a facility`s capabilities including process tools, gas and chemical delivery systems, deionized water, HVAC, safety systems, as well as its automated handling systems. Abuzeid notes that all of these systems have to be evaluated in terms of both the required process specifications as well as current and future production capacity. The incorporation of new semiconductor process technologies such as CMP and copper are particularly well suited to targeted retrofit programs as manufacturers look for ways to isolate these processes from the rest of their production lines and ensure they can be implemented without risking yields.
In addition, Abuzeid notes that the ability to continually operate the fab during the upgrade process is another important consideration for many users. “This has a direct impact on time-to-market, since as soon as a company gets its new process tool set, it can very quickly begin producing next-generation products.” This observation is echoed by Tim Loughran, manager of business development at Performance Contracting (Lenexa, KS), who points to one semiconductor customer site being upgraded from Class 100 to Class 1, retrofitting one process tool (16 to 24 square feet) at a time, putting in new grid ceilings and filters.
The cost of a semiconductor fab retrofit is most directly related to the age of the original facility. According to Abuzeid, whereas the basic infrastructure of fabs first built in the early 1980s is compatible with the requirements of modern systems incorporating, for example, full basements, raised floors, and high ceilings, other older facilities will require much more extensive infrastructure work. For example, although many fabs received ex tensive process-system upgrades in the 1980s, the facilities themselves may have been built as early as the 1960s with small basements and low ceilings which can pose difficult obstacles to modern transport systems and stockers. Still, Abuzeid says there are usually ways to get around these issues.
Retrofit isn`t absolutely always more cost-effective than new construction, however. As Loughran points out, “while a large part of our business is retrofit vs. new construction, there are cases where new construction is more cost-effective or quicker in the long run.” Issues such as existing contamination levels, cost of demolition and building code issues, can play a role. For example, Loughran points to one case where the cost to bring an existing facility up to seizmic code compliance far outweighed that of building a new structure from scratch. “Such cases are isolated, but each needs to be examined on a case-by-case basis to determine whether retrofit really saves time and money over new construction.”
In contrast to the semiconductor industry, yield considerations and profit margins rarely drive the pharmaceutical industry to upgrade the classification levels of its facilities. Instead, as Clestra notes, pharmaceutical facilities are generally designed to meet specific regulatory requirements on the preparation of products and will generally continue to operate at those same cleanliness levels, whether Class 100,000, Class 10,000 or sterile. On the other hand, there are also instances where regulatory and validation levels are changed, automatically requiring that facilities be upgraded. For example, as Mackler of Clestra describes, the FDA recently mandated that nasal sprays be processed in a manner akin to sterile processing. “This led to a major retrofit requirement for an entire cleanroom facility in an area that was previously a controlled environment but not classified.”
The requirements of a pharmaceutical cleanroom retrofit may also be considerably different and more expensive than those of a semiconductor facility. For example, Mackler points out that compared to the basic negative pressure plenum of electronics applications where customers are largely considering only particle counts, pharmaceutical cleanrooms involve very high-integrity envelopes, hard-ducted air supplies and returns and high levels of test and certification. “Costs can in some cases be incremental but they can double or triple in others.”
In particular, an upgrade to Class 100 pharmaceutical requirements represents much more than an incremental reduction in particle counts. As Mackler explains, as opposed to the doubling of total airflow when going from Class 100,000 to 10,000 or from 10,000 to 1,000, the move from 1,000 to 100 requires a four-fold increase with 90 to 100 fpm flow to the room vs. the 25 to 50 fpm of Class 1,000. In addition, pharmaceutical Class 100 applications require 100 percent ceiling coverage requiring architectural changes to accommodate the HVAC. “By and large,” says Mackler, “if you have a Class 10,000 room and want to get to a Class 100 room, you will double or triple the cost per square foot of that room, primarily due to accommodating the HVAC.” These estimates also assume that the customer will maintain the existing pressure, temperature and humidity conditions, and, as Enderle notes, don`t include the cost of any demolition required to the existing facility.
Ultimately, says Mackler, the key to keeping down the cost of a future retrofit program, particularly for pharmaceutical applications, is to plan for it. “Decisions should be made as part of the value-added engineering and energy analysis when designing a cleanroom or when implementing a major retrofit.” For example, Mackler points out that while the upfront cost of placing an air handler on a platform as opposed to floor level may be somewhat more, system upgrades down the road will be far less of a project. Similarly, the installation of more capacity ductwork and equipment housings than are initially needed will allow the cost of future upgrade to be incremental rather than major.
The continuing cost of clean
Although the principal driver behind cleanroom retrofit programs is increased yield or regulatory considerations, there can often be ancillary cost-saving advantages as well. For example, reduced energy usage and lower operating costs through modernization of HVAC systems can frequently deliver sizeable facility-wide cost savings.
Still, the desire to upgrade cleanliness classification levels must in general be regarded as in direct conflict with the goal of reducing energy consumption and improving operating efficiency. As Enderle of Clestra notes, “moving to a higher classification level doesn`t equate to energy efficiency. Everything else being equal, if you have Class 10,000 and want to upgrade to Class 1,000, it`s going to cost you more because the cleaner the cleanroom, the more energy it will take.”
Still, Enderle agrees that many existing facilities may be operating at far from the optimum efficiency levels capable with today`s technology. For example, he suggests that even when class level is not of concern, users of older cleanrooms may still want to investigate replacing motors and fans with higher-efficiency systems. Filter media replacement may also be justified because many newer media usually produce lower pressure drops, which can provide some cost savings.
To better understand the relationship between cleanroom upgrade and energy efficiency, the Lawrence Berkeley National Laboratory (LBNL) has been working with Supersymmetry USA Inc. (Oakland, CA) to develop case studies of energy usage and energy efficiency in cleanroom environments (“Study illuminates energy savings in cleanrooms,” CleanRooms, January 1999, p. 4). Supersymmetry is a design and consulting concern based in Singapore, with offices in the U.S. that specialize in innovative and efficient cleanroom HVAC designs.
As Supersymmetry`s project engineer Eric Concannon describes, the team is now evaluating the results of three HVAC retrofit case studies conducted at Applied Materials (Santa Clara, CA), STMicrolectronics (Singapore) and Hine Design (Sunnyvale, CA). According to Concannon, although results are not yet tabulated, all of the projects demonstrated clear cost effectiveness. For example, as part of the Applied Materials retrofit program, a chilled-water system was installed that achieved close to a 25 percent operating cost savings on the system itself and on the order of 5 percent to 10 percent across the facility. At STMicroelectronics, the company was able to reduce its energy usage by roughly 30 percent, while increasing chip output by 20 percent over the same period. Similarly, Hine Design, which designs and manufactures robotic arms used to handle silicon wafers in semiconductor processes such as CMP, was able to double its cleanroom space while maintaining a constant energy bill. Refurbished by Flowstar Corp (Gilroy, CA) the company`s cleanroom facility is segmented into six Class 100 cells within a Class 1,000 room, all controlled by a central airflow management system. At an installation cost of $49,000, facilities manager Dominic Credi says the system is saving him over $100,000 per year in energy costs.
Many companies tend to look at cleanroom efficiency and the benefits of cleanroom retrofit only in terms of their processes, Concannon observes. Instead, he says, “they need to consider their central plants, including recirculating fans, etc., in the equation.” The Applied Materials case study, for example, included an analysis of automated fan-velocity control based on particle count measurement and occupancy sensors. Although particle count levels were the primary control mechanism, the system was also set up to reduce fan velocity (from 70 to 80fpm to 50fpm) whenever the facility was vacant. “Since individual bays are controlled separately and there aren`t people in all areas 24 hours per day, this can save a lot of energy,” Concannon says.
Planning for savings
While no one disputes the merits of an energy-efficient cleanroom program, the business reality is that such considerations alone rarely convince companies to invest in a significant retrofit program. From his point of view, Esser of MCA notes that while “operational cost savings are usually discussed with customers as part of a retrofit proposal, they are never a primary motivating factor.” Clestra`s Mackler agrees, explaining that while most energy savings projects will offer some payback, the time frame is usually around 3 to 5 years, which is well beyond that of the 18 months or less of most new chip technologies. “The truth,” says Mackler, “is unless the retrofit directly contributes to increased yield, it`s not going to make a compelling financial case.”
On the other hand, experts are quick to point out that there`s no reason why energy-efficiency benefits, lower-cost retrofit and class-level upgrades can`t all be achieved through carefully considered long-range planning. According to Enderle, planning for future upgrades and retrofits should be a central part of any cleanroom facility construction as well as a major retrofit program. The Asyst/ADP Marshall approach to semiconductor fab retrofit also includes provisions for future upgrade possibilities. Says Abuzeid, “although we don`t include the cost of these future upgrades in our cost analyses, we do try to incorporate provisions for implementing future technologies and point out the long-term cost benefits.”
As a final word of advice to anyone considering a retrofit program, Mackler of Clestra recommends that customers be up front with their contractor(s) on the budget available for the project. “In the end, this will save them a lot of time and money and allow the vendor to propose a realistic solution that will best meet requirements,” he says.
John Haystead is a freelance writer in Surry, ME, and was formerly the editor of CleanRooms.
Top: An example of a U.S. fab upgraded to include minienvironments and 200-mm load port transfer systems. (Photo courtesy of Asyst Technologies Inc.)
Top middle: An example of a retrofit that employs Clestra`s 55-mm grid. At left, during construction. Bottom middle: after. (Photos courtesy of Clestra Cleanroom.)
Bottom: A view of the construction of the retrofit of the microelectronic filter cleanroom at Pall Corp. (Photos courtesy of Clestra Cleanroom.)
At left: Like many semiconductor facilities built in the 1970s and 1980s, DuPont Photomask`s facility in Kokomo, IN, has used the retrofit option to keep pace with advancing process technology requirements, most recently upgrading several production areas including gowning rooms from Class 100 to Class 1. (Photos courtesy MCA.)
Cleanroom bays 1 and 2 at Hine Design`s Sunnyvale, CA, facility are joined at the center to form one large bay. In total the facility has 16,000 square feet of cleanroom space divided into six cells. The company`s airflow management system dramatically reduces energy costs by allowing the airflow in each cell to be reduced to 15 to 20 percent of capacity during off hours and 65 percent of capacity during operation.