The many options for managing CMP wastewater
07/01/2004
Waste streams from CMP can vary greatly depending upon a wide variety of factors in wafer processing, and one "right" solution for treatment of CMP wastewater may never exist. This article provides an overview of treatment technologies to supplement the experiences of fab engineers and the support of suppliers.
As the geometries of critical features shrink, chemical mechanical planarization (CMP) process requirements become increasingly more stringent. Deviations in planarity and uniformity, microscratching, and defectivity become less and less acceptable and contribute more directly to overall yield loss. In order to satisfy the changing demands of IC manufacturers, CMP slurry providers must continuously improve products and reduce overall product costs while providing better performance, and more recently, look to new chemical components and abrasives.
Accompanying all of the performance changes in wafer processing, environmental protection agencies worldwide are imposing tougher restrictions on the types and amounts of fab waste they allow to be discharged into public water systems. To be in compliance with discharge limits, semiconductor companies must treat waste streams to remove harmful contaminants. In CMP waste streams, this means abrasive particles need to be removed; hazardous chemicals need to be removed or reacted; and the pH must be brought to near-neutral levels.
From the outset, it is important to consider many factors to determine the "best" CMP waste-stream treatment for a particular facility. These factors include discharge permit requirements, corporate environmental goals, CMP waste-stream dilution by other fab effluents, treatment process risks and reliability, flexibility for system expansion/contraction in response to fab operations, and the capabilities of existing waste-treatment equipment and processes. Other considerations include fab-drain system design for solids handling and for wastewater segregation, space available for additional treatment equipment, driving forces for water reclaim, drain system segregation on the CMP tool, and waste-treatment system capital and operating costs.
CMP wastewater composition
Slurry waste streams from CMP equipment are generally composed of dissolved, suspended, and settled fine particles; oxidizing agents; organic complexing agents; inorganic ions; and dissolved metals. The waste-stream compositions can vary widely depending on the original slurry composition, the type of dielectric or conducting layer being removed, CMP tool design, and operating parameters. Polishing slurries for oxide and metals are usually aqueous formulations with a large pH range (pH of 1–11). Among other materials, slurries generally contain the following:
- particulate abrasive compounds at 1–10 wt% (alumina, silica, or cerium oxide);
- chemical oxidants at 1–10 wt% (organics, such as peroxides and peracetic acid, and/or inorganics such as ferric nitrate, salts of iodates, persulfates, etc.);
- buffers;
- KOH, NH4OH, or monoethanolamine, for oxide CMP slurries;
- complexing agents at <1 wt% (organic acids and potassium salts thereof, as well as a wide variety of other organics); and
- surfactants at <0.1 wt% (sodium dodecylsulfate, and others).
During polishing, the slurry is diluted by a factor up to 60 by ultrapure water, and it picks up material from the surfaces being polished.
Waste treatment challenges
For discharging into public water systems, the waste treatment challenges posed by these streams include pH adjustment, residual oxidation-reduction potential, total organic carbon (TOC), chemical oxygen demand (COD), biological oxygen demand (BOD), suspended solids, dissolved solids, and dissolved metals.
Additional challenges with respect to reclaim for less critical uses — such as in cooling towers, scrubbers, and irrigation applications — include residual organics from chelants and organic oxidants, ammonia, and surfactant compounds. For reclaim to high-purity water applications, extensive treatment would be required for complete removal of all suspended and dissolved solids, organics, ammonia, and ionic species.
Treatment technologies
Working with the challenges for discharge only, a short list of well-established technologies can be constructed for options to meet the various needs, based on the known properties of CMP waste solutions (see table). These options are predicated on a facility needing a solution other than simple dilution by the final fab effluent in order to achieve its process and environmental goals.
|
Additional process requirements for suspended solids removal and copper precipitation include subsequent sludge dewatering and sludge disposal. Disposing of concentrated copper-containing wastes is an important consideration for ion exchange treatment.
The two greatest challenges from a treatment-for-discharge perspective are the suspended solids and the copper. For some fabs, both the suspended solids and copper may be serious issues, in which case they could be removed at the same time in a single-unit operation combining chemical precipitation and solids removal, or they could be removed in sequential fashion. For other fabs, only the solids or copper may be considered problematic.
Reverse osmosis is technically feasible to apply, but is not discussed in detail here because it tends to produce a significantly less concentrated waste (and thus leads to more costly disposal). This approach also has greater pretreatment requirements than ion exchange. The one big advantage of reverse osmosis is that it maximizes the potential for water reuse. Treatment systems for reclaim are not discussed here as they can vary enormously depending on the water quality requirements of the process receiving the treated wastewater.
Membrane filtration for solids removal
Filtration systems include both open-end and dead-end filtration devices. In dead-end filtration, all solids are collected within the filtration element, rather than staying in suspension. The filtration element must be backpulsed or backwashed periodically to remove solids from the system. Dead-end microfiltration devices include both rigid polymeric membranes and coated sock-type filters. Open cross-flow filtration membranes can be rigid monolithic (ceramic), tubular (polymeric), or spiral wound (polymeric). The choice depends on issues such as the type and size of solids to be filtered, solids loading, available footprint area, cleaning procedures, and so on. Tubular systems provide a wide tubular flow channel for the working fluid and are less prone to fouling by large particles.
Membrane filtration has been investigated as an option for solids removal from CMP waste streams by treatment equipment suppliers as well as CMP process users and equipment designers. In membrane setups, both ultrafiltration (particle filtration range = 0.002–0.2µm) and microfiltration (particle filtration range = 0.05–2.0µm) have been applied to CMP wastewaters.
A schematic of a membrane filtration treatment system. |
Membrane filtration is capable of removing copper as well by adding suitable precipitation aids to the mixing tank in advance. In both pilot and commercial applications, the setup shown in the figure on p. 61 has demonstrated suitable treatment of CMP wastewaters for eventual discharge. These membrane systems are operated in a recirculating configuration rather than single-pass mode to concentrate the solids before discharge to the filter press. A sock-type filter element produces a discontinuous backwash waste stream with concentrated solids.
Coagulant chemistry is often used to enhance membrane flux (usually by a factor of two or more) and run time by agglomerating the fine solids that can most easily clog the membrane pores. The term "flux" is used to define flow/unit area of membrane/unit time. In pilot testing, fluxes of 40–150 gallons/sq. ft of membrane area/day (gfd) and even higher have been observed. Periodic backflushing and chemical cleaning of the membrane are used to restore the membrane flux to the design value once it declines to an unacceptable value. Fouled membranes must be replaced periodically when the flux no longer returns to an acceptable value. Sock-type filters are simply replaced when performance degrades.
Advantages to membrane filtration for solids removal include:
- consistent effluent water quality — reduced sensitivity to upstream process upsets;
- solids dewatering up to 10 wt% in membrane filter unit;
- modularity (meaning additional treatment capacity is easily added as CMP processes ramp up);
- plant personnel already familiar and comfortable with membrane treatment operations; and
- small equipment footprint.
Membrane filtration is required for pretreatment if reverse osmosis unit is to be placed downstream.
Some disadvantages to membrane filtration include:
- Long-term membrane flux, stability, and cleaning protocol data are needed.
- Periodic membrane cleaning and eventual replacement is required.
Ion exchange for copper removal
One option for copper removal is the use of cation exchange. It may be desirable to break the removal of suspended solids and dissolved copper into two steps for a number of reasons:
- disposal costs for metal-containing sludges, depending on local regulatory environment;
- reclaim metals for reuse within the facility to reduce environmental liability;
- removal of suspended solids is not seen as a critical issue for wastewater operations;
- potential economic advantage exists by using alternatives to chemical precipitation for metals removal; and
- part of the system for total CMP stream clean-up can be reused in high-purity applications.
The first three factors listed are all site-specific fab conditions. The fourth is a straightforward comparison of capital and operating costs. Removal of problematic organic compounds is the biggest impediment for reuse in high-purity applications.
Pretreatment considerations for application of ion exchange for metals removal include the removal of oxidation-reduction potential to prevent resin degradation and solids removal to the extent needed to prevent resin fouling.
Ion exchange is a well-established technology with a long history in the metals removal field. Since copper is cationic, only this type of ion exchange is discussed here. The basic process is to pass the wastewater bearing dissolved metals through a bed of ion exchange resin beads that adsorb the cations from solution and replace them with either H+ or Na+. Once the adsorption potential of the bed is exhausted, it is regenerated by applying a concentrated (1–10 wt%) solution of acid (HCl, H2SO4) or brine (NaCl) to displace the adsorbed cations and replace them with cations from the regenerant. The waste regenerant solution contains the desorbed cations from the original wastewater and leftover regenerant. The total water flow needed for regeneration may amount to 5%–25% of the original waste-stream flow depending on the system configuration and wastewater chemistry. Thus, the concentration factor compared to the original wastewater can vary from 4 to 20.
One of three cation exchange resins may be selected for use in the process: strong acid resins, which exchange virtually all cations from the wastewater; weak acid resins, which have an even stronger affinity for hardness than do strong acid resins, and are able to produce softer water, as long as the alkalinity is in the right range; and chelating resins, which are well-established specialty resins incorporating iminodiacetate adsorption sites to break complexes of chelated heavy metals and adsorb the metals onto the resin.
Because ammonium and potassium salts are favored for many of the additives in the slurry formulation, using the strong acid cation exchange approach will likely result in very high resin loading with these compounds, taking away resin capacity for the heavy metals and resulting in excessive demand for regeneration. Weak acid cation exchange would be preferred for more efficient operation; however, the slurry waste will also contain residual chelating compounds that likely necessitate the use of chelating resins to achieve acceptable metal removal rates. Other metal compounds in the waste stream, such as iron, can adsorb as well, competing with the compound (e.g., copper) that the system was designed to remove. Careful waste-stream analysis and pilot testing are recommended before selecting a specific option.
Concentrated waste disposal
The concentrated ion exchange regenerant waste must then be disposed of in an acceptable fashion. Two commonly used techniques are electrowinning to recover the metals on-site (plate them out on an electrode), and evaporation to recover the water and produce a relatively small amount of hazardous waste solids.
The electrowinning process requires a reasonably concentrated waste stream to be cost-effective, so it is not a good fit for the waste coming directly off the CMP tool. In addition, wastewater treated by electrowinning should be free of chlorides because chlorine gas would be evolved at the counter electrode in the electrochemical processing. High levels of dissolved iron can also be detrimental to electrochemical recovery processes.
Evaporation is very costly, and so is generally used only on very concentrated problematic waste streams, but it does provide recovery of good-purity water. For copper CMP waste, evaporation suffers from the fact that it generates a sludge containing unstabilized copper and virtually all the dissolved solids from the original slurry. For 1000ppm of dissolved solids in a 20gpm CMP waste stream, this corresponds to disposal of a minimum of 3.7 tons/month (dry solids basis) of hazardous waste.
A third option, provided by environmental contracting firms, is services to take charge of the exhausted resins to reprocess them at waste recovery facilities, while supplying fresh resin in canisters.
Electrowinning the concentrated waste may ultimately be the preferred choice for two other reasons. First, the equipment may already exist on-site for certain plating or etch bath waste streams, and second, it allows recovery of what would otherwise be a potentially toxic waste on-site. Note that the actual resource value of the recovered copper is negligible: at 10ppm Cu in 20gpm stream at $0.75/lb for high-grade copper, the value is only $660/yr.
Some advantages to ion exchange are as follows:
- excellent copper-removal capability with appropriate resin choice and operating conditions;
- compact unit operation (bed cross-section typically only 0.1–0.5 sq. ft/gpm) with system size mainly affected by pre- and post-treatment unit operations;
- common and familiar unit operation in the fab environment; and
- toleration to fluctuations in influent water chemistry without impact on effluent water quality.
Some disadvantages to ion exchange are:
- Pretreatment systems are required to prevent resin fouling and degradation.
- Relatively expensive regeneration wastewater-disposal systems are needed.
- Potential difficulties exist in adding design flexibility needed to handle widely varying waste-stream compositions and respond to changes in upstream operations.
Summary
There are a variety of well-established technologies for cleaning up aqueous wastes contaminated with metals, suspended solids, and oxidizers, such as the typical CMP waste stream from metal-layer processing. The technology choice for a particular fab depends on a variety of fab-specific and corporation-specific goals and constraints. In any case, the keys to success are a careful pre-screening procedure to identify technologies best suited to the local level and then a rigorous program of pilot testing to determine the ability of various alternatives to handle current and future needs.
Acknowledgment
This article was prepared with major contributions from Mary Reker of SpeedFam Corp., a unit of Novellus Systems Inc.
For further reading
M. Reker, M. Lenart, S. Harnberger, "Treatment and Water Recycling of Copper CMP Slurry Waste Streams to Achieve Environmental Compliance for Copper and Suspended Solids," Semiconductor Fabtech, 8th Ed., pp. 141–150, 1998.
L. Mendicino, P.T. Brown, "The Environment, Health and Safety Side of Copper Metalization," Semiconductor International, pp. 105–110, June 1998.
Water Environment Federation, Pretreatment of Industrial Wastes, Manual of Practice No. FD-3, Water Environment Federation, Alexandria, VA, 1994.
J. Mallevialle, P.E. Odendaal, M.R. Wiesner, eds., Water Treatment Membrane Processes, American Water Works Res. Foundation, McGraw-Hill, New York, 1996.
A. Rushton, A.S. Ward, R.G. Holdich, eds., Solid Liquid Filtration and Separation Technology, Weinheim Publishing, New York, 1996.
B.A. Bolto, L. Pawlowski, Wastewater Treatment by Ion Exchange, E&FN Spoon, New York, 1987.
Brian V. Jenkins received his BS in chemical engineering from Northwestern U.'s Technological Institute. He works in the Industry Technology for Microelectronics unit at Nalco Co., 1601 W. Diehl Rd., Naperville, IL 60563-1198; ph 630/305-1031, e-mail [email protected].
Craig W. Myers received his PhD in chemical engineering from the U. of Wisconsin and his BS in chemical engineering from the U. of Illinois. He is a staff scientist in the Global Water Research unit at Nalco Co.
Stan Lesiak received his BA in business management, and is a business development specialist at Cabot Microelectronics.
Richard Viscomi received his BA in biology from Lafayette College, and is a product support engineer at Cabot Microelectronics Corp., 870 N. Commons Dr., Aurora, IL 60504; ph 480/634-1042, e-mail [email protected].