Recycle Spent Rinse Waters with EDI

Robert P. Donovan and, Dennis J. Morrison,
Sandia National Laboratories, Albuquerque, N.M. -- Semiconductor International, 7/1/1999

Recycyling of the spent rinse water discharged from wet benches commonly used in semiconductor processing is one tactic for responding to the targets for water usage published in the 1997 National Technology Roadmap for Semiconductors (NTRS).¹ Not only does the NTRS list a target that dramatically reduces total water usage/unit area of silicon manufactured by the industry in the future, but for the years 2003 and beyond, the NTRS actually touts goals that would have semiconductor manufactureres drawing less water from a regional water supply per unit area of silicon manufactured than the quantity of ultrapure water (UPW) used in the production of that same silicon. Achieving this latter NTRS target strongly implies more widespread recycling of spent rinse waters at semiconductor manufacturing sites.

Although, by most metrics, spent rinse waters are of much higher purity than incoming municipal waters (Table 1), recycling of these spent rinse waters back into the UPW production plant is not a simple, straightforward task. The rub is that certain chemicals used in semiconductor manufacturing, and thus, potentially present in trace concentrations (or more) in spent rinse waters, are not found in municipal water supplies and are not necessarily removed by the conventional UPW production sequence used by semiconductor manufacturers. Some of these contaminants, unique to spent rinse waters, may actually foul the resins and membranes of the UPW system, posing a threat to UPW production and potentially even causing a shutdown.

Because of its capability to remove many of these potentially harmful contaminants, electrodeionization (EDI) -- a technology growing in importance as an alternative, primary ion exchange process -- is also a candidate treatment module for insertion into a loop for recycling spent rinse waters. In this application, the role of the EDI unit is to protect the UPW system from contaminant spikes or excursions in the recycled rinse water and to upgrade the overall quality of the spent rinse waters, making a larger percentage of these wet bench discharges suitable for recycling. This article reports the performance of a small (0.7 - 0.8 gpm), separate bed EDI cell (the El Ion by SG Water Systems) when fed by representative spent rinse waters from rinsing operations following: 1) an SC-1 clean (RW #1) and 2) a dilute sulfuric acid clean (RW #2). Tables 2 and 3 (RW #1) and Tables 4 and 5 (RW #2) list the properties of grab samples of the feed water and product water collected during this evaluation.

Note that the concentrations of calcium and magnesium in both feed waters (Tables 2 and 4) are orders of magnitude less than the concentrations of calcium and magnesium listed in Table 1 as being typical of municipal water (the ionic concentrations of the feed waters in Tables 2 and 4 are given in ppm; the EDI product water ionic concentrations are in ppb as are all the concentrations listed in Table 1). Calcium and magnesium typically dominate the conductivity of municipal feed water and are the species that determine the hardness of the water. Their absence in spent rinse waters implies that the limits on the conductivity of EDI feed water, as specified by most EDI manufacturers and based on experience with municipal, hard waters, are not necessarily valid when the EDI feed water is spent rinse water from semiconductor wet benches.

In spent rinse waters, the dominant conductive species differ from those of municipal waters and, in particular, are not calcium or magnesium, the primary foulants of an EDI cell. Most manufacturers of EDI equipment recommend: 1) the conductivity of the feed water be no higher than (~50 µS/cm and 2) a reverse osmosis (R/O) module be used, if necessary, to meet this requirement. However, the conductivity of RW #1 was 62 µS/cm and that of RW #2 was 102 µS/cm. No upstream R/O unit was used. In spite of these departures from manufacturers' recommendations, the data in Tables 2 - 5 show that the EDI product water from RW #1 and RW # 2 is highly suitable for recycling and is, in fact, cleaner than the product water from a typical R/O stage of a conventional UPW system (Table 1). The usual conductivity limitation on EDI feed water seems to not apply when the EDI feed water is spent rinse water and neither does the need to include an upstream R/O unit as part of the complete EDI installation.

While the results reported here are based on relatively short-term operation, Parker² has reported similar EDI performance in a recycle role over four months of around-the-clock operation. Parker operated a mixed bed EDI design (by Ionics Inc.) fed by spent rinse waters of properties generally similar to those of RW #1 and RW #2 and, for one short period, of conductivity exceeding 1000 µS/cm. EDI product water under all such challenges and over the entire test period remained compatible with recycling to the R/O product tank.

Acknowledgements

This project has been supported jointly under a SEMATECH/DOE CRADA and an EPA contract. SEMATECH's S116 Project Technical Advisory Board, led by John DeGenova and Tracy Boswell, has provided mentoring and valuable guidance throughout the project life. Project work was conducted in the fabrication facility of Sandia's Microelectronics Development Laboratory, Alton Romig, Director. Sandia is a multiprogram laboratory operated by Sandia Corporation, a Lockheed Martin Company, for the United States Department of Energy under Contract DE-AC04-94AL85000.

It is a pleasure to thank Dietmar Steudten and Phil Sampson of SG Water Systems for making the EDI test unit available for this evaluation and for their continued support during the startup. John DeGenova, Texas Instruments Inc., prepared Table 1; Chemtrace, Hayward, CA, performed the off-line analyses of the grab samples summarized in Tables 2 - 5.

References

1. The National Technology Roadmap for Semiconductors, 1997 Edition (Semiconductor Industry Association, 181 Metro Drive, Suite 450, San Jose, CA 95110).
2. Parker, R., 'Electrodeionization in a Semiconductor Fab Recycle System,' pp. 219 - 238 in 1999 Proceedings of the 18th Annual Semiconductor Pure Water and Chemicals Conference, March 1 - 4, 1999, Santa Clara, CA (Balazs Analytical Laboratory, 252 Humboldt Court, Sunnyvale, CA 94089).