Ultrapure Water: Rewards of Recycling
SEMATECH'S programs in ultrapure water (UPW) target conversation, recycling and optimization of UPW plants.
Laura Peters, Senior Editor -- Semiconductor International, 2/1/1998
Because of the low cost and abundance of quality water in the United States, recycling of ultrapure water (UPW) is not widely practiced. Companies active in recycling include Texas Instruments, Hewlett Packard, Digital, and other firms at roughly a dozen sites in the United States. So while North American recycling practices have seriously lagged conservation protocol in Japan, Singapore, Taiwan and Korea, SEMATECH aims to change this condition. The consortium is identifying the most successful strategies for water conservation, setting up a state-of-the-art re-claiming and recycling facility in Austin, Texas (lead photo), and providing a software package used to optimize new and existing UPW plants.
| A typical 200 mm wafer fab processing 40,000 wafers per month uses between 2 and 3 million gallons of water per day. Conservation strategies are driven by the increasing demand on local municipalities to supply adequate water to existing and new wafer fabs, which could double the industry's water requirement over the next five or six years. However, John DeGenova, project man-ager of SEMATECH's Environmental Safety and Health (ESH) Group, claimed that benefits to UPW recycling reach far beyond conservation goals. "It's been proven that we can actually get better UPW quality at the point of use with recycling." |
Recycling provides further cost savings as the amount of chemical required for regeneration is lower and the fab produces less wastewater, thereby lowering discharge costs. DeGenova explained that by saving water, recycling systems can pay for themselves within 18 to 24 months, depending on local water costs. Walter Worth, SEMATECH's program manager for ESH, said that a typical payback period for the industry is 30 months.
Other industry efforts that have successfully reduced water consumption in fabs include the optimization of bath designs, use of reduced cassettes and optimization of wafer cleaning processes. New processes are making increased use of dilute chemistries, hot deionized (DI) water rinses and megasonic cleaning, as applicable, to conserve UPW water and chemicals.1 SEMATECH members believe that these changes could account for a relatively stable level of water use in fabs (Fig. 1). In other words, there is little correlation between mask level increases and UPW usage, although fairly significant variation existed among the 16 U.S. companies surveyed by SEMATECH. Given the values in Figure 1, total demand on the UPW plant is ~25% higher than the values shown, because of losses in the purification process.
A chart from the 1997 SIA National Technology Roadmap for Semiconductors (Table 1) summarizes strategies being developed to reduce feed-water and UPW use and lower water purification costs. One of the most elusive, yet important goals is correlating water purity levels with purity needs. "There isn't much data to suggest that if you want a yield of 90% you need this quality of water," DeGenova said. However, with each new device generation, impurity specifications for low ionic levels, organic levels, quantity of dissolved oxygen and particle levels become tighter. Impurity levels now rival their detectable limits, using the best analytical tools available today. Interestingly, fab managers now emphasize that it is the reliability of the UPW system itself that is in need of improvement (i.e., >90% uptime) not the quality of the water.
Recovery, reuse, reclaim, recycle
In wafer fabs, UPW is either reclaimed, recycled or discharged. Reclaiming, practiced in the majority of fabs worldwide, refers to the reuse of UPW in non-process applications, such as in cooling towers and scrubbers. Many types of water can be reclaimed including water used in analytical applications as well as the reject water from the RO process. A typical reclaim process is illustrated in Figure 2.
Recycling refers to the recovery of spent rinsewaters from the wafer cleaning system, which is monitored and, based on the results, either sent back to the beginning of the UPW process (feed-water inlet) or to any of several intermediary tanks throughout the UPW system. Holding tanks can precede the RO system, the ion exchange system or the final polishing step. Figure 3 quantifies water quality levels through various points in an UPW facility, with and without recycling.
Typically 60-70% of the UPW used in fabs can be recycled cost-effectively. UPW rinses used in pre-furnace cleans are ideally suited for recycling, perhaps only requiring the final polish. Rinsewater containing acids or bases are neutralized first and then recycled to the feed-water inlet or some intermediary tanks. Rinsewaters from resist stripping hoods are not readily recyclable and usually require additional treatment. Rinsewaters containing organic solvents are normally sent to industrial waste.
Risk management
The main risks in implementing a water recycling process include the introduction of impurity spikes to the system, the buildup of recalcitrant compounds, the possibility of new chemical interactions and the possible introduction of process-generated compounds that will not be removed using current purification methods.1 Any serious diversion from specifications is likely to shut down the UPW system and wafer fab. The downtime associated with such catastrophic failures is unacceptable in semiconductor manufacturing. As more new materials and chemicals are introduced to the wafer fab, some of these risks increase. 
1. A survey of 16 U.S. chip anufacturing facilities shows that despite increases in the number of mask levels, water usge is not trending upward because of effective conservation, reclaiming and recycling strategies.
Managing the risks associated with the recycling of water in a wafer fab depends on the configuration of a reliable UPW plant. The efficiency of the system relies on the use of quick-response sensors to monitor water quality, giving engineers enough time to decide whether the water is acceptable for recycling and to which holding tank it should be directed. For instance, recycling spent rinsewater to the final polishing loop provides the greatest benefit but also carries the highest risk. In general, the system must be designed to respond to inevitable changes in water quality.
The most serious offenders in UPW systems today are organic compounds. In fact, there is very little data available on the removal of organic compounds typically used in wafer processing.1 Detecting organic compounds requires a fast response total organic carbon (TOC) sensor to provide in-line testing of water going to a holding tank, generally located in the sub-fab or mechanical building next to the fab. The sensor must provide a response rate that is fast enough to get an analytical reading and be able to divert a contaminated stream to industrial waste before it can affect product. 
2. A typical water reclaiming system reuses water in non-process realated applications, such as in cooling towers and scrubbers.
Today's TOC testers provide information in ~2-3 min, and SEMATECH's goal is to eventually reduce this to 30 sec. "On these sensors, fast response is actually more important than accuracy, because the organic compounds are either there or they're not, so ±30% accuracy is fine," DeGenova said. The faster the response time, the smaller the necessary holding tank. 
3. Improved quality levels are obtained when UPW water is recycled.
Early attempts at recycling often focused on treating the industrial waste effluent. DeGenova explained, "I think a lot of failed attempts to recycle in the past were due to the fact that the easiest place to get the water was at the end of the process. Unfortunately, these concentrated streams were very unpredictable, they put a heavy load on the ion exchange media and RO membranes, and often could not be brought to acceptable purity levels." Today's procedure of selecting specific rinsewaters, the purity of which is considerably better than the purity of the feedwater supply (Fig. 3), is much more reliable.
"Many people believe that implementing recycling in an existing fab is nearly impossible, but essentially, that is not true, and we're going to demonstrate that in our facility," DeGenova said. The cost of retrofitting a recycling system typically adds $250,000 (if a rinsewater collection system exists) to $1 million worth of additional plumbing. The challenge, as explained by Bob Duffin, director of ESH at SEMATECH, is finding space for the additional plumbing, while also accommodating the person performing the installation.
The rewards
Obviously, cost savings and improved water purity at the point of use are the most intriguing benefits to recycling. With a higher quality of feed-water at the source, the unit processes in the UPW facility operate with improved efficiency, and the load on the UPW plant is significantly reduced. The improved processing reduces UPW system downtime, resulting in a safer, more reliable facility. The benefits of conservation of water, chemicals, and energy used in the fab are attractive and will become even more inspiring in the United States as more fabs are built and the demand for water increases .
With recycling, there is also a reduction in quantity of wastewater requiring treatment. Some skeptics argue that if you recycle 50% of your water, you will double the level of contaminants in the wastewater, and you may have a problem meeting regulated levels for discharge. As DeGenova explained, contaminant levels will not double as you will have much less waste generated in the UPW plant as compared to a system treating municipal water exclusively.
Optimization of UPW plants
SEMATECH's UPW simulator is unique in that it allows users to compare water quality anywhere in the UPW system. The software simulates the thermodynamic and kinetic reactions in each of the UPW plant's unit processes, generating impurity outputs from each of the processes, including RO, ion exchange, degassification, polishing, etc. The simulator allows engineers to test many "what-if" scenarios to see the benefits of recycling, to compare recycling schemes, to move equipment around and to compare UPW systems from different vendors. It allows companies to determine how recycling would affect water quality in existing facilities. Perhaps most importantly, the software could be used to monitor UPW systems for optimum efficiency while notifying engineers when recovery to specifications returns.
References
1. J. DeGenova and F. Shadman, "Recovery, Reuse, and Recycle of Water in Semiconductor Wafer Fabrication Facilities," Environmental Progress, Vol. 16, No. 4, AIChE, Winter 1997, p. 263.
