New ESH report details ways to cut fab water, energy costs

Afab can save 5 million kilowatt-hours per year by reducing ultrapure water (UPW) use by 10%. Assuming a cost of 41/2 cents per kilowatt-hour, that reduction translates into an annual savings of approximately $225,000. An efficient cleanroom can save more than 5.4 million kilowatt-hours of energy annually, while a cleanroom running at optimum efficiency can save an additional 3.6 million kilowatt-hours.

Those energy- and money-saving tidbits, or reminders as the case may be, are among a plethora of suggestions contained in a new 31-page report issued by the Pacific Northwest Pollution Prevention Resource Center. Titled "Energy and Water Efficiency for Semiconductor Manufacturing," the publication includes advice on reducing energy costs by improving HVAC performance, lowering water costs through recycling and reuse, and reducing UPW system maintenance costs. It also makes note of the many reasons why the semiconductor industry can't or won't take the efficiency steps. A list of relevant Web sites and resources is attached to the end of the document. The PPRC, as it's known, is a nonprofit organization that collaborates with government agencies and private industry in the region to "promote environmental protection through pollution prevention."

The report is written primarily for technical assistance providers, or TAPs, working for the State of Washington's Department of Ecology, says Catherine Dickerson, a technical lead at the PPRC and the author of the report. As their title suggests, TAPs visit industrial sites on behalf of their department to offer advice and guidance about ecologically and economically sound manufacturing practices.

This publication and other so-called "topical reports" provide an overview of the specific industry in question so that TAPs at both the state and local level "can go in and talk to these people and not be totally behind the eight ball . . . or 20 steps behind" everyone else in their level of knowledge, says Dickerson.

The new efficiency report can also prove helpful for fab managers stalled at the crossroads where environmental regulations, production costs, and yields come together. Dickerson agrees that the bulk of the report and its recommendations are probably familiar to the semiconductor industry. Still, she insists, "I'm not convinced it has ever been in one place before." PPRC's intention was to "put the information in one spot, highlight the low-hanging fruit and the hard-to-get-at stuff. Some of it is new for a lot of people."

In an attempt "to provide objective information that everyone can use," the PPRC sought feedback from industry sources. Dickerson says the input helped her avoid obvious overstatements. One such "off-base" claim regarded "a statement on how much of an efficiency gain you could make," Dickerson recalls. Her research showed savings of up to 90% in certain production areas, which, her source informed her, was "really pie in the sky. . . and wasn't feasible."

The report suggests that the industry's rapid rate of product innovation, codified by Moore's Law, gives chipmakers and suppliers the unique ability to introduce "energy efficiency options with each new chip generation." Dickerson says she put this argument to "one gentleman in particular," a source with a nonprofit organization "who works with industry on energy efficiency issues throughout the Northwest, mostly the electronics industry.

"It's a difficult topic to push, I guess," she recalls. "Although it's certainly true that some of the processes change very quickly, the physical structure of the building does not. I tried to make that apparent in the report. He agreed that there was the opportunity there to institute some of these efficiencies, more often than in the auto industry, for example."

But, as the report acknowledges, the relentless drive to the next generation of chips means "the product output schedule dominates." It's one of the myriad examples cited as possible obstacles to more-efficient operations, according to Dickerson. Calling it a classic "double-edged sword," the author notes that the obvious "pressure for profits . . . sometimes makes it more difficult to institute change."

The report makes a strong case for the benefits of efficiency, noting that the semiconductor industry is estimated to waste "at least half" of the energy it uses. "What does saving energy return to a facility? There are many benefits, beyond saving money on electric bills. In many cases, saving electricity will increase product yield, through time savings and improved productivity. Energy efficiency can also provide a fab with more flexibility than competitors who do not address efficiency issues, since [the fab is] less susceptible to price changes in resources. The bottom line is healthier, making the plant more stable in times when industry has downturned [sic]; overhead is lower; and there is a lower overall demand for natural resources, making [the manufacturer] more desired in communities." Wasted energy could be converted into a profit "with returns exceeding 30% ROI," the report goes on.

Christa Hernandez is one of the technical assistance providers for whom the report is written. A chemical engineer, she works for Washington's Department of Ecology in the toxics reduction unit, which is part of the state's hazardous waste program. "A lot of what I do is look at a facility's process . . . and I use my knowledge of regulations and my experience in chemical engineering to make suggestions."

Hernandez has no enforcement authority. Companies may decide to take her suggestions, or they may not. "It's all up to them," she says, adding, "a lot of the time I'll do varying levels of documentation to show them why my idea is a good one. If it's a question of law, though, usually they just take my word for it."

Occasionally, Hernandez has had to tell a company, "You can't put this in the garbage. It could be, for example, a plating line where I'd say, 'If you add an extra rinse tank here and here, then you can cut your water use to 25% of what it is now.'" She has a rinse-water modeling program on her computer at work.

By early March the pollution prevention expert had not yet had the opportunity to read the report, but in her experience the microchip industry knows quite well what it can't put in the garbage, she indicates. "Usually, those companies are really well educated about what they need to do. That industry changes so rapidly. They're always coming up with innovations and new ways to do things . . . as a part of their business. If they don't make those changes they're not in business very long. The companies I end up doing the most good for are companies who have been doing business the same way for 20 years."

 

Dave Stangis is certainly familiar with the suggestions presented in the report. The ESH external affairs manager with Intel agrees with Hernandez's comment about the bleeding-edge nature of chipmaking. Many observers look at what he calls "the overall churn of the industry" as a downside, but Stangis says Intel tries to view this characteristic as an opportunity for improvement.

"Some of the long-term products of this industry might be two years away. I was talking to another group last night [about] design for environmental, health, and safety for a process that won't be out for three or four years." As an example, Stangis cites the industry's two-prong approach to PFCs. First, the industry is searching for PFC substitutes because the gases are now known to have high global-warming potential. Second, he notes the drive to find "other options in terms of abatement, collection, or recycling."

Stangis acknowledges that the industry is generally risk averse and that this conservative approach is in great measure caused by the relentless production schedules. Both the PPRC and Stangis note the double-edged nature of this line of argument. From Stangis's point of view, on the one hand the cycles raise the level of innovation; on the other hand, Intel's copy-exact philosophy means that in trying to ramp up quickly it can replicate the design of a New Mexico fab for its new plant in Arizona. "However, if there's a mistake, the potential to copy a mistake is in this process as well." One industry expert cited in the report's footnotes says industry wags calls this possibility "infectious repetitis."

Intel employees' performances, however, are measured in part on their willingness to take risks, says Stangis, who notes that this practice is "more Intel specific" than the industry norm. The chipmaker's natural resources group is charged with "looking at the process four or five years down the pike to find technical solutions that don't exist and making dollar decisions on the right thing to do in terms of the environment."

Risk aversion acts as a strong counterweight to open-arms acceptance of energy-efficiency strategies, Stangis indicates. The mentality "is translated through the process. The people who make the pumps that then work the tools that connect to the vacuum system. . . oversize to be conservative. They overstate in terms of the energy demand so that there's no problem with the equipment." This anxiety over maintaining high uptimes leads to what Stangis calls "one of the biggest external arguments, that the industry is not paying attention to this kind of design excess."

Unlike the steel industry, for example, the raw materials used by the semiconductor industry "are really R&D and depreciation," argues Stangis. "We're talking billions of dollars." This is among the issues that the whole industry needs to examine, he maintains: "What are the real costs in terms of the products?" and is the fundamental argument that "if you save water and energy it pays for itself" valid?

Even as the report notes the many reasons why the semiconductor industry may not rush to embrace energy-efficient solutions, it tries to counter these reasons with an array of statistical arguments and comparisons. The largest savings can be found in modifying HVAC systems, it notes. Their lifetime of 10 to 20 years contrasts with that of process tools, which have a span of 3 to 5 years and whose suppliers look first to keeping price down before they will consider ways to increase tool efficiency. Designing or retrofitting a more efficient HVAC system "can have a lasting impact on energy use," the PPRC document states.

An additional benefit is that energy-efficient fabs and tools make the facility more reliable. "With less wear on filters, pumps, and motors, maintenance and operating costs are reduced. Perhaps most important, an energy-efficient design has lower pressure and slower airflow, which improve filtration efficiency from these systems. The end result is reduced particle contamination on the semiconductor, generating improved yields and product quality." Targets are vacuum systems, exhaust, compressed air, and DI water.

Furthermore, with 300-mm processing on the verge of becoming the state of the art, IC manufacturers have even more incentive to reduce their UPW use. The report says that chip production on 300-mm wafers will require "at least 11/2 times more water than 200-mm fabs [require], and as much water as is needed annually by a city of 60,000 people."

The electricity use per square foot in a fab can be up to 100 times the energy demands of a modern office, the report comments. "One industry representative indicated that the cost savings achieved for one year's worth of energy savings at a fab is equivalent to the total cost of one day of production," the report comments. "Put another way: energy accounts for about 1­2% of the cost of semiconductor production. Considering the amount of money at stake, it is easy to understand industry reluctance to invest in unfamiliar efficiency technologies, which could require costly production downtime."

"Energy and Water Efficiency for Semiconductor Manufacturing" is available from PPRC's Web site. The address is http://www.pprc.org/pprc/pubs/topics/semicond/semicond.html.