For some time now we've been concerned about energy conservation in semiconductor manufacturing cleanrooms. Papers have been written, seminars have been conducted, consultants have proliferated and engineers have gone back to drawing boards where they've produced solutions-some quite innovative and others rather exotic.
by Mike Fitzpatrick
We have even reduced the airflow in our cleanrooms, taking our most widely used process fluid-clean air-and diminished its capabilities all in the name of energy conservation.
We've turned off our lights, adjusted our temperatures, tweaked our exhaust, reduced our area under filter and replaced our equipment with more energy-efficient models. We've even reduced the airflow in our cleanrooms, taking our most widely used process fluid-clean air-and diminished its capabilities all in the name of energy conservation.
Dr. Ken Gold stein's article in last month's issue (see Clean Rooms, March, 2002, p. 62) raises some interesting questions. Why do we care about energy conservation in cleanrooms? And what brought about our current crusade to wring every excess watt from our facilities?
We profess to care about energy con servation for two reasons: to reduce manufacturing costs and to protect the environment. Goldstein's article demonstrates that although the cost of power in the typical wafer fab may seem high, it represents a minor portion of the overall cost of producing the product. Any cost savings that might be obtained by reducing facility energy demand are not large enough to motivate most plant managers.
If cost is not a major concern, then protecting the environment certainly is. Here again energy conservation represents only a slice of the pie.
Industry has responded
The semiconductor industry has been responsive to environmental issues over the last decade with many manufacturers routinely exceeding the requirements set by law. Use of solvents has been reduced significantly. Exhaust streams that once were discharged into the atmosphere are now neutralized and abated. Water recycling and the use of gray water are common. It has been an evolving process, and it continues.
However, the environmental zeal of manufacturers has not caused them to focus on energy conservation as their major contribution to the environment.
Because costs are not the real issue, and because we are not considerably greener than we were a year ago, what prompted our recent fixation on energy conservation?
Our present preoccupation with energy conservation has its roots in the energy crisis of early last year. The crisis was caused by the normal cycle of the supply of a commodity (natural gas prices rose) coupled with a scarcity of production capacity (power plants).
Nowhere was the situation as grave as it was in California. Prices of electricity at both wholesale and retail levels skyrocketed. Power shortages and rolling blackouts were common. Factories shut down and gray-haired grandmothers from Marin County were forced to choose between hugging their trees and burning them for firewood.
This energy crisis may have been limited to California, but from a national perspective we had been setting ourselves up for an energy crunch for some time. Demand had been increasing and capacity had been decreasing for nearly a decade. The National Energy Information Center estimated that between 1990 and 1999 our overall demand had increased by 11.3 percent, while our production capacity had decreased by 1.7 percent. If we did not bring more capacity on-line we soon would start turning out lights nationwide.
According to a recent article in Engineering News Record, more than 83,000 megawatts of new capacity came on-line between 1999 and 2001, increasing overall capacity by 10 percent. New power-plant construction ran at such a feverish pace that we're seeing plans for additional plants shelved in anticipation of a power glut that will last until about 2005.
What does all this mean to the operators of large semiconductor facilities? We most likely got excited over an issue that had little impact on our lives-one that is readily cured by the application of free-market economics.
It means that any payback calculations we used to justify energy-conservation expenditures were probably based on energy costs at their peak, rather than their long-term averages. Paybacks calculated on that basis may prove to be of longer duration than we anticipated and may even yield negative returns. It also means that some of us fiddled with our process parameters-like cleanroom air velocity-in the name of energy conservation when we should have let our process requirements make that determination.
The semiconductor industry has made significant progress in reducing energy consumption. Long before the recent crisis, it had been taking steps to conserve energy.
The concept is hardly new. Energy-efficient motors and compressors are now the norm rather than the exception. Airflow velocities have been reduced as we have moved away from the “more is better” frame of mind. Owners and designers continue to work to minimize the area under filter and engineers continue to develop energy-effective designs.
Most importantly, the 2001 Edition of the International Technology Roadmap for Semiconductors (ITRS) established aggressive goals to “reduce the energy use of process equipment” and to “reduce the energy use of the manufacturing facility.” Even modest progress in decreasing the energy consumption of process tools will result in significant energy savings.
Does this mean that we can declare victory and do nothing? Not quite. We need to be aware that energy has a cost, both monetary and environmental. With that in mind we must continue to look for opportunities to reduce those costs where prudent.
The suppy side considered
While we have focused our attention on reducing consumption, we have failed to examine the supply side of the equation. Here, several opportunities to cut costs present themselves. Power distributors bill their customers based on two rates, a Time-Of-Use (TOU) tariff and an Actual Demand (AD) tariff.
Under the TOU tariff, customers are billed at various rates depending on the time of day, the day of the week and the time of year. Higher rates apply to those times when overall electrical usage is high-normal working hours, weekdays and summer.
The AD tariff bills the consumer based on the highest kilowatt demand in any 15-minute period during the month. Under this tariff, if a plant is running at the maximum demand stipulated and the RO regeneration system starts up, the facility will be billed at a higher demand tariff for the entire month.
Manufacturers can significantly reduce their energy costs by reducing their consumption during peak load hours, and by maintaining a low demand profile-a practice called “peak shaving.” This can be accomplished through energy management or by distributed generation.
Energy-management systems enable a manufacturer to identify and turn off non-essential services during peak hours, or to develop a strategy of energy consumption that reduces peak loads. They consist of real-time-demand monitoring systems and software that enables the viewing of electrical consumption fluctuations over a 24-hour period, often in 15-minute increments. The electrical usage of a specific facility's systems can be identified and strategies can be developed to reduce and schedule overall plant demand.
Distributed generation provides an onsite power source that can reduce the power required from the utility company during peak hours. The source of the onsite power may be a gas or diesel-driven generator, but most frequently it is a gas-turbine generator. When plant energy-monitoring systems indicate demand starting to exceed tariff limits, the generators start up and help to carry the load of the facility, thus keeping the load to the utility below tariff limits. Distributed generation systems also provide a source of standby power in case of power outages or brownouts.
Changing the methods we use to buy electricity also may reduce our energy costs. Deregulation now enables consumers to take competitive bids for power supplies.
Power aggregation enables end users to combine their loads to obtain electricity at reduced rates, and energy-procurement services shop for power nationwide on behalf of their clients.
One thing is certain, power quality and reliability will decrease in the short term due to deregulation, while costs will fluctuate. Many available options enable us to minimize our electrical costs. Reducing energy costs by cutting our cleanroom airflow is probably not the best direction to go.
Michael A. Fitzpatrick is program director of microelectronics for Lockwood Greene Engineers. He has participated in the design and construction of semiconductor facilities for more than 23 years and has been instrumental in the development of methods to decrease time-to-market in the construction of manufacturing facilities. A senior member of the Institute of Environmental Sciences and Technology (IEST), he is Chairman for WG012 (Considerations in Cleanroom Design) and WG028 (Minienvironments).