Reusing/Reclaiming High Purity Deionized Water
With the transition from 200 to 300 mm wafers, never before have replacement, reduction, recycling and reclamation been more critical. Technological advancements in the life sciences and regulatory changes in USP 23 and WFI are making the production of high purity water free of TOCs (total organic carbons) a top priority.
By SUSAN ENGLISH
Today, companies large and small are utilizing some form of water conservation or reuse. The need to recycle, reclaim or in some way reuse, ultrapure deionized water (UPW/DI) water in fabs is apparent. It is the most heavily used–and no longer the cheapest–chemical that removes other chemicals from wafer surfaces. Indeed, UPDI can have as much effect on wafers as any other chemical. The concentration of impurities in water must be much lower than even the most sensitive chemicals, such as hydrofluoric acid, because water is either dried on the surface of the wafer in spin dryers or displaced from the surface with isopropyl alcohol. Either method can leave contaminants on the wafer`s surface.
However, manufacturers of wet benches, water purification systems and other suppliers are not jumping on the collective recycling bandwagon just yet. The industry was badly burnt by trying to recycle back in the `80s, when attempts at ultrapure water recycling caused gigantic production “upsets” and fab shutdowns. Contamination from organics was picked up during the cleaning process–primarily from photoresist materials. Unable to handle the contaminants, which were then adsorbed onto wafers, these older water purification systems broke down, causing a total yield “bust.” Until the ability to measure TOCs in real time is possible, the industry will continue to use alternatives: optimizing processes to reduce consumption and reusing water/chemicals.
Some experts insist recycling`s time has come. Marjorie K. Balazs, president and founder of Balazs Analytical Laboratory (Sunnyvale, CA), sets pure water standards for the industry. Her recommendations are adopted by SEMI/Sematech as “acceptable criteria” and are widely considered the highest in the industry. She says: “It is past time for the United States to incorporate recycling into their processes. The differential in cost between those who do and those who don`t will affect the cost of products and our competitiveness on the world market.” Balazs points out that because all UPW used today contains traces of low, “usually unmeasureable” amounts of organic material that does no harm to processes, any added contamination would have to come from the fab. Therefore, she recommends studying these materials to determine how to build a system to handle them. One approach she proposes is to heavily ozonate all recycled water to completely eliminate organic material, decompose the excess ozone, and pass the water through the prefiltering system at the beginning of the makeup loop “as an insurance policy,” she says.
Other approaches are being developed. While many companies are already recycling and conserving waste streams in more non-critical areas–reclaiming UPW for cooling towers or reducing the flows on sinks, for example–such efforts are somewhat ad hoc and do not involve system redesign or new tool technology, which experts like Balazs say are essential to arriving at a reliable UPW/DI water system which can withstand upsets. On the contrary, most fabs are not designed to recycle water in any appreciable quantity. Although most have built some kind of system, with room for future recycling possibilities in mind, Balazs says they have no clear idea about what such a system would involve, how it should be configured, or have not left enough space to accommodate it.
Recycling water for reuse on wafers would be a great money-saver because it takes much fewer chemicals and supplies to produce UPW from recycled pure water than from source water. However, Balazs says the reason more users don`t take advantage of recycling is that they don`t realize it would mean a great reduction in the size of the makeup section, which would reduce RO membrane, DI resin, chemical and electrical costs. Surprisingly, even the transition to 300 mm wafers would cause no significant increase in the size of makeup units if the system were properly redesigned.
Meeting the recycle/reclaim challenge
Charged with the task of creating a multidisciplinary culture “to educate a new breed of engineering leaders and to produce critical precompetitive science and technology for environmentally benign semiconductor manufacturing,” the NSF/SRC Engineering Research Center for Environmentally Benign Semiconductor Manufacturing (CEBSM) has two primary objectives: “the modification and optimization of existing processes and tools to minimize waste generation and maximize resource utilization,” and “the development of totally new approaches and processes which are environmentally more desirable, with emphasis on the recovery and recycle of chemicals.” A consortium of the University of Arizona (Tucson), M.I.T., Stanford University, and the University of California at Berkeley, the CEBSM will use SIA Roadmap goals as guidelines while undertaking projects involving university/industry collaboration and joint research with other institutions.
Projects are organized within six “thrust areas,” which include Water Purification, Distribution and Use (Recycle and Reclaim) and Metrology and Sensors for Environmental Application. CEBSM Director, Dr. Farhang Shadman, Professor of Chemical and Environmental Engineering at the University of Arizona, says projects have been carefully selected with an eye to improving yield and reducing costs. “The way the industry is, you can`t just push something for environmental reasons only. `Design for environment` also has to affect the bottom line, and it has to make sense.” Most of the impurities in recycled water coming from the fab are process generated and are, therefore, a “known quantity,” says Shadman, while feedwater is unknown, due to its variety of contaminants. He says research shows that with proper system design, the use of recycled water has actually improved the efficiency of some units. The only catch, he warns, is that as fabs start recycling, new impurities will be introduced into the system. These impurities are trace metals and organic chemicals, or surfactants, which Shadman terms “recalcitrant compounds.” Therefore, a major thrust area will be the development of sensors and metrology for measuring resistivity and detecting upsets. “Right now,” he says, “measurement tools are sensitive in measuring low concentrations, but not fast enough to detect upsets and take corrective measures in a timely fashion. Maybe in about a year or two, we`re going to have a new group of analyzers with emphasis on response time.”
Conservation vs. recycling
Texas Instruments (Austin, TX) is one of the participating companies in the CEBSM`s efforts to recycle and reclaim DI water. Paul Gowen, TI`s corporate environmental specialist, says much of the U.S. semiconductor industry is “very apprehensive” about recycling UPW because of the risks of organic contaminants and the lack of real-time TOC analyzers. In fact, some companies, like Intel and Motorola, have taken a public stance against water recycling until TOC analyzers can speciate and give results almost immediately. “Right now,” Gowen says, “3-6 minutes is the range.” TI has already spent a lot of money recycling wastewater with on-line analytical capability. The company also takes wastewater from some of its DI water plants and uses it for other less critical operations like cooling tower makeup. In addition, it has corporate goals of “zero waste generation/zero injuries/zero preventable illnesses.” Says Gowen, “The concept is that, eventually, we will reduce down to zero wasted resources.”
In spite of its initial reservations about the lack of rapid sensor technology, Intel Corp. piloted its own recycling effort at one of its development facilities in Oregon in order to evaluate the technical and economic viability of recycling UPW rinses. Selected wet benches were used to minimize the risk of organic contamination. Resistivity, oxygen reduction potential (ORP) and TOC monitoring were employed across an array of different purification technologies, such as organic scavenging resins, vacuum degasification, ozonation, RO, ion exchange and UV sterilizers in various combinations.
According to Rich Poliak, Intel`s manager of chemical strategies, it was discovered that because process engineers were already working to reduce the total amount of water by optimizing their processes, flow reductions actually increased the concentration of impurities coming from the individual wet benches, considerably diminishing the total quantities of water available for recycling. In fact, the higher impurity concentrations and lower volumes of wastewater decreased the benefits of UPW recycling to the point where the team decided that focusing on reuse and reductions provided greater gains and acceptable process risks.
Says Poliak, “With recycling, the risk is the ability to detect upsets to the system in terms of organic chemicals like isopropyl alcohol and some of the organic bases we use in our processes. When you rinse wafers, you`re trying to rinse the chemicals and particles off, so the resultant wastewater has very low levels of different kinds of chemical species in it, not the same as the ones in the incoming water stream. Although they`re extremely low levels, some of those chemicals have a tendency to concentrate in separation processes like RO membranes and ion exchange resins.”
After thoroughly investigating the “pollution prevention hierarchy” of “replace/reduce/reuse and recycle,” Intel decided not to recycle ultra pure water for ultra pure water use but instead to reuse significant amounts of wastewater. Poliak sums it up: “If we weren`t doing water conservation in other areas through reduction, replacement, and reusing it elsewhere, we probably would have gone to recycling and probably over-engineered it. But because we`ve been able to conserve water very significantly through other means, we think the payback and risk/reward ratio of ultra pure water recycling isn`t there yet until the technology to detect these specific chemical species is.”
Intel has also been working with its suppliers to find more efficient ways of producing ultrapure water. Poliak says, “For every gallon of raw water, we produce as much ultra pure water as possible. The difficulty is that the ability to do that changes, depending on the water source because water quality is not the same all over the place. In some locations, it`s very easy to do because the feedwater is very clean; and in others, it`s not.”
The approaching transition from 200 mm to 300 mm wafers has some experts predicting a seven-fold increase in UPW usage. Others are determined to hold the line and are demanding better process and design technology, as well as a aller footprint, for the new generation of wet processing tools. Interlab Inc. (Danbury, CT) is chiefly concerned with DI water for semiconductor applications and it also builds cleaning systems for the precision optics field.
“The 300 mm wafer set will use approximately seven times as much water as the 200 mm. So we`re jumping from something like 1,000 gallons to 7,000 gallons,” says Marj Balazs. She warns that merely cutting back on the use of water to keep the system clean is not the answer, but could reintroduce the old microbial contamination problems. “It takes a system redesign, but people want to do it as an add-on. They just want to take it and repolish it. In my opinion, that`s a dangerous approach.”
CEBSM`s Dr. Farhang Shadman describes the transition as a major challenge requiring major changes in tool design. “The whole idea here is that if you continue using the old design and just make it larger–just scale it–you`re going to pay a big price.” He observes some fabs changing some of their processes and switching to single-wafer processing, a step that will, in turn, require a different kind of cleaning.
Intel is actively and aggressively keeping up the pressure on equipment suppliers to hold chemical and water usage on a per wafer basis within bounds. Says Rich Poliak, “We`ve already been working with our equipment suppliers on the next-generation tools so that the equipment requires less water to rinse wafers. Our new technologies–quarter-micron-generation and beyond–are using a newly designed piece of equipment that requires significantly less amounts of water and chemicals.”
It was not a simple model to arrive at. Reducing water in one area doesn`t necessarily mean you`ve reduced the amount of cooling water, etc., he warns. “Our estimate for a typical factory is that the new wet stations will save us about 300,000 gallons a day. For a nominal-size factory, it might be a bit more. For a very big factory, it might be a bit less.”
One supplier who has already redesigned its equipment to conserve both chemical and water resources is SCP Global Technologies (Boise, ID). The company is designing a new 300-mm wafer capacity wet bench that will have the same specs as their 200 mm tool. “It`s not easy to do,” says Joe Hartman, vice president and SCP`s director of engineering, “but we`ve got a lot of advances.” These include informal tanks built to conform to the shape of the wafer. “Instead of having round wafers in a square tank, we have round wafers in a roundish or half-round tank,” he says. “If you do the math, it`s 10-20 percent just off the top.” The company is also researching spray techniques–how water gets introduced to the tank–in an effort to get the same amount of cleaning with less flow time, an innovation that will save even more water than the informal tanks.
The microorganisms of concern to laboratory water purification systems are bacteria. A typical bacterial level for a potable laboratory water supply is one colony-forming unit per milliliter (cfu/ml). Bacteria will enter an unprotected water purification system from the feedwater, any breaks in the system, or through the dispenser. Once in the system, it secretes a slimy polymeric substance that adheres bacteria to the surfaces of storage tanks, deionization cartridges, plumbing and hard-to-clean complicated surfaces. Although bacteria can be killed with disinfectants like hydrogen peroxide, hypochlorite and formaldehyde, their polymeric secretions and lipopolysaccharide cellular fragments remain and may be a source of contamination if not removed. All these bacteria must be removed to produce ultrapure reagent quality water. To produce ultrapure water, deionization via a mixed bed system is the most economical way to go.
The absence of dissolved organics is very important when performing analyses of organic substances, such as High Performance Liquid Chromatography, electrophoresis and fluoroscopy, or tissue culture research. Pharmaceutical grade water must be pyrogen-free.
The National Committee for Clinical Laboratory Standards specifies five types of water: I, IIA, IIB, III and Special Purpose water. Ultrapure, or Type 1, water must be clean enough to prevent interference with atomic absorption, flame emission spectrometry, and various other analytical techniques. Considered a specialty grade of Type 1 water, at 18.3 megohms, semiconductor water is actually a little better than Type 1 water. Further treatments beyond pre-treatment and deionization are needed to produce specialty grade Type 1 water. In response to these recommendations, the water purification industry is scrambling to develop more durable RO membranes capable of reducing water waste and ultrafilters with smaller pore sizes.
To avoid resin-regeneration costs and downtime, Millipore Corp. (Bedford, MA) has developed electrodeionization systems that produce analytical-grade water directly from tap water. By combining reverse osmosis and the company`s patented continuous electrodeionization technology, which uses electrical current instead of chemicals to regenerate resins, electrodeionization technology ensures a continuous supply of consistent quality water, eliminating the fluctuating water quality experienced with both distillation or traditional deionization, according to the company.
A new generation of wafer geometries and highly sensitive analytical instruments demand ultra pure DI water. Interestingly, industry observers predict it won`t be long before pharmaceutical water systems will closely resemble those found in microelectronics facilities. The pharmaceutical/biotech industry is more concerned than ever before about final water purity for WFI and for U.S. Pharmacopeia (USP)-quality water.n
An example of a typical factory water flow diagram with reuse loops. Note how the incoming city water proceeds through the system for purifying and how it is also used for domestic uses.
For 18 megohm DI water at point-of-use, Interlab, Inc. (Danbury, CT) manufactures its Micro-Rinse water polishing system, which provides an independent source of pressurized DI water from tap water for low consumption applications.
Anions and cations in feedwater (a) pass through ion exchanger resins (b) and replace the attached hydrogen and hydroxyl ions (c). The hydrogen and hydroxyl ions then combine to form pure water molecules (d). Reproduced with permission of Elga Ltd., High Wycombe, Bucks, England.
Millipore`s Milli-Q and Elix Systems supply Type 2 (15 megohm) purfied water for media pre- paration, instruments or feedwater to a polisher via EDI technology. EDI uses an ion exchange resin, ion exchange membrane and dc voltage (electricity) to remove ions from water.