Ultrapure Water Recycle in the Semiconductor Industry
John DeGenova
Department of Chemical and Environmental Engineering
University of Arizona
ã 1999 Arizona Board of Regents for The University of Arizona
This paper summarizes the benefits and risks associated with water recycling in the semiconductor industry. The specific items that will be covered in this paper include characterization of the spent rinsewater types, the composition range of key impurities, and different recycling strategies. The discussion will include the development of new purification methods for the removal of organic impurities, and the development of a computer model for simulating the effects of recycle.
INTRODUCTION
The semiconductor industry has recently experienced rapid growth at an unprecedented rate. This expansion is causing concern in some communities due to the large quantities of water presently required for semiconductor wafer manufacturing. Along with this expansion comes the construction and installation of new wafer fabrication facilities (fabs). Each new fab will use 1 to 3 million gallons per day for just wafer processing. ;Tthe water use in some locations may approach twice that range. The only cost effective, long term solution is proper segregation and collection of waste rinsewaters as part of the implementation of a true recycling strategy.
Recycling of water in the semiconductor industry is the reuse of water, that had been previously purified to an extremely high quality, used in wafer processing, collected, retreated, and used again to process wafers. The practical range of recoveries for recycled water is from a low of 10-20% to over 95%. Fab rinses can be relatively easily recycled because they consist of water, already extremely contamination-free, with low levels of easily removed ions from wafer processing chemicals. Much of the collected rinsewater is already cleaner than the municipal supply water it replaces. This water can be directly re-used as a replacement for some of the municipal water supply, without additional treatment. Various amounts of recovered water can be recycled. Some of the water may require treatment prior to reintroduction to the UPW system.
Figure I indicates the wide range of water usage rate reported by semiconductor manufacturing facilities. This chart is the result of a survey performed by Sematech, along with data obtained from its member companies. The amount of UPW used per wafer produced varies with wafer size and from site to site between the different manufacturers. Each wafer fabrication facility typically uses 1-3 MGPD (million gallons per day). Some companies have 3-4 fabs per site, resulting in consumption of UPW exceeding 10 MGPD. This amount does not refer to the municipal water demand but rather only the demand for UPW. The actual demand on the municipal water supply is approximately 25%-40% greater than the quantity of UPW due to losses in the purification processes. These demands can be quite significant on the municipal water supply and have been a problem for some communities, especially those located in arid regions.
Most of the UPW consumed is used for wafer rinsing purposes. Figure II indicates some typical semiconductor process tool setups and a general indication of the various types of wastewater generated1.
The first and primary strategy to conserve UPW is simply the reduction of water used in the wafer rinsing process. Some of the techniques, indicated in Table I are quite simple but can make significant impacts to the overall water consumption. Spray type rinsing has been shown to use much less water than the typical overflow or dump-rinse methods. It can reduce water consumption by 75-95%, and dramatically lower processing times.
· Spray rinsing vs. overflow/quick dump· Rinse tank geometry improvements
· Hot UPW vs. cold UPW
· Megasonic rinsing
· Idle flow rate reduction
· Analytical monitoring of rinse water
· Computer modeling, convective/diffusive
Table I: Rinse Water Reduction Techniques
Megasonic excitation during rinsing is a typical method of indirectly improving the rinse process. By adding megasonic action, the cleaning process in some cases can be optimized, and possibly allowing for a reduction of necessary rinse water.
Until recently, rinse tanks were typically designed with relatively large volumes, significantly larger than the wafer carriers (boats) which hold the wafers to be rinsed. As the wafers are typically loaded into the boats tightly spaced relative to one another, the path of least resistance for the water flow is actually around the wafer boat, rather than in between the wafer product spacing.
As can be seen in Figure III, in an overflow rinse process, water flows at a constant rate from the bottom of the tank overflowing out of the top of the tank, for usually 5-10 minutes. The initial turbulent mixing mechanism while wafers enter the tank changes quickly to one of non-turbulent, laminar flow. In this case, most of the water flows to the outside of the wafer boat, bypassing the wafers entirely. It has been shown that nearly 80% of the rinse water actually bypasses the product4. A smaller rinse tank, or one that confines the flow to the wafer volume, actually provides for a more water-efficient rinse. New semiconductor process tools are being designed with smaller rinse tanks and with directional flow patterns to force the water in between the wafer spaces, producing faster and more effective rinsing. Less process time is required for the rinse, and better process control is achieved.
Studies are under way to evaluate sensors for monitoring the quality of water in each rinse tank to determine when adequate rinsing is achieved. Other investigations have focused on the development of rinse models using both diffusion and convection equations to help optimize the rinsing process5. These models incorporate the processes that occur in between the tightly spaced wafers, including the desorption of the chemicals from the wafer surface and the diffusion of the chemicals through the boundary layer into the bulk fluid, where impurities are carried away by convection. Further work needs to be done in the area of turbulent mixing, dump rinse efficiencies, and elucidating mechanisms for spray rinsing.
Reduction of UPW flow rates during idle periods can also make significant differences in water reduction. A minimum flow is required to prevent bacteria colonies from forming colonies on to the pipe walls.
A second water reduction strategy that is currently underway is in the re-use of water in other areas, in which the quality of water is not a primary concern. Some of the more common applications of water re-use are listed in Table II. The reject water from the reverse osmosis process is a good candidate for this type of re-use. Spent rinse water can also be used for some of these applications. However, since it may contain corrosive ions, this water may be quite aggressive and could corrode equipment components. Additional treatment may be necessary prior to re-use of this water, such as alkalinity adjustments.
· RO Reject; C/T, Air Scrubbers, Irrigation· Ultrafilter Reject; UPW system rinsing
· Analytical instrument discharge; various use
· Spent rinse water; C/T Makeup
Table II: Water Reclamation /Reuse
Another water use optimization and reduction strategy is the recycling of spent rinse water back into the UPW treatment system. Because of the relatively high purity level of the rinse water, it can be added back into the UPW purification process at various points within the system. It can be combined with the feed water at the treatment input or be combined at other points such as with the reverse osmosis purified water. Depending on the possible contaminants in the spent rinse water, it can also be added back into the UPW process at a purer stage, such as in the ultrapure water storage tank where it will be re-polished prior to use. Table III lists these options, which are also illustrated in Figure IV. Figure IV indicates a generic UPW treatment process, replete with primary treatment, ion exchange or secondary treatment, and a final polishing treatment step with holding tanks. The spent rinse water can be brought back to the holding tanks at any of these three major stages.
· Feed Water storage tank· Semi-Pure storage tank
· Ultrapure Water storage tank
Table III: Recycle Flow Options
There are significant benefits, related to both cost and processing, associated with recycling. For example, one is the improvement in final water quality that is achieved. With certain ions and most organic compounds, recycling is able to lower contamination levels proportional to the amount recycled; an 80% recycle amount cuts contamination by a similar amount.
Some of the key benefits to recycling are included in Table IV. Note that the improvements will vary with the amount recycled and the technologies used to treat recovered water. This is not an exhaustive list.
1. Improved final UPW quality /2. Rinse recipes can be optimized for cleanliness and throughput, instead of minimized water consumption /
3. Improved reliability of UPW facility, less downtime: /
Reduced frequency of RO membrane cleaning processes /
Reduced frequency of ion exchange regenerations /
Reduced frequency of filter backwashes/rinses /
Improved efficiencies in UPW treatment processing /
Improved control of feed water problems /
4. Reduced chemical usage for ion exchange regenerations and waste treatment /
5. Significant cost savings: /
Less feed and discharge water costs /
Less regeneration chemical and backwash water costs /
Less industrial waste treatment cost /
Less maintenance and cleaning chemical cost /
6. Possibly Improved RO reject quality for other reclamation purposes /
7. Less demand on the municipal water supply and waste water treatment systems /
Table IV: Benefits Associated with Recycling Water
There are also risks, however, associated with recycling spent process rinsewater. These risks include the items listed in Table V. Notice, though, that some of the benefits of recycling directly answer risks associated with municipal feed water supply and quality; and that some of the risks are similar for either water source. The key is to manage the risks with the best engineering practices wherever possible. In some cases, this can only be done with great difficulty, cost, and/or complexity on municipal feed streams, but can easily be achieved with a comprehensive recycle strategy.
Recycled water /Municipal Supply Water
1. Introduction of impurity spikes into the system. / Same2. The potential buildup of recalcitrant compounds. / none
3. Qualification of the present purification methods in removing process generated contaminants / none
4. Risk of new chemical interactions caused by recycle / none
5. Contamination due to biofouling / Same
Table V: Risks Associated with UPW Treatment
The risks of recycling, listed in Table V, include the possible introduction of new and unknown process contaminants into the UPW purification system. Compounds that are not properly removed could remain and possibly even build up in the system, and the product yield could be negatively impacted. There is very little data available on the removal of organic compounds typically used in wafer processing. The presence of any of these compounds in the UPW product water at the point of use could have devastating consequences. Hence, the proper treatment or segregation of the spent rinse water is imperative in order to keep these compounds out of the polish loop. The characterization of the behavior of specific compounds subjected to standard units processes, such as activated carbon, Reverse Osmosis, and UV oxidation in a UPW system is an ongoing task.
DISCUSSION OF WATER CONSERVATION STRATEGIES
Spent ultrapure rinse water is contaminated with residual chemicals carried over from the process flow. Rinses that typically follow acid/base chemistries can usually be readily recycled; the contaminants can be removed from the water using standard separation techniques, such as ion exchange. One must only segregate these rinse water drains away from the industrial waste drain to collect the water for recycle purposes.
Typical compounds found in spent rinsewaters are listed in Table VI. These compounds are typically quite easy to detect with available analytical instruments, and are quite easy to remove with standard unit operations found in most UPW treatment facilities. Through proper segregation, detection, and diversion techniques, most of the spent rinsewaters generated in wafer fab facilities is readily recyclable.
Due to the requirements of some of the semiconductor processing chemistries, some rinse water may contain compounds that are not readily removed with standard separation techniques. For example, organic surfactants and co-solvents may be present. The rinse water after photoresist strip chemistry may contain organics if the resist oxidation process is somehow only partially complete. Certainly, the rinsewater that follows organic chemical baths will contain compounds that may not be readily removable. Characterization and correct treatment, or even segregation of these waste streams, are key to achieve high levels of water recovery concomitant with guaranteed quality.
The contaminants in rinse water that generate most of the concern for recycling in the industry today are organics. The method of choice for the removal of these organic compounds in a recycle system is diversion. Metrology is available to rapidly detect and divert organic-containing water.
As recycle systems become more robust and take on the role as the primary source of water, organic oxidation will be the mainstay. The concept of photocatalytic oxidation reactors as applied to UPW recycle systems has been developed at the University of Arizona in Tucson 2. The photocatalyzed oxidation of process-generated impurities like organic solvents, surfactants, and trace-chlorinated hydrocarbons in ultrapure water is promising.