Removing Esoteric Contaminants from Drinking Waters: Impacts of Treatment Implementation

At first blush, the production and distribution of drinking water seems to be a very straightforward process. There is a need to remove microbial agents and any anthropogenic or autochthonous contaminants that may be a health concern. A disinfectant is often added to maintain a microbiologically-safe water throughout the distribution system. However, for those that design or operate drinking water facilities or conduct drinking water research, there are many subtle issues that arise in both individual processes and in the combined treatment train. To elucidate these issues, evaluation of the removal of a specific contaminant in a controlled setting can shed a great deal of light on treatment and what the implementation of treatment entails for residual and distribution system plans.

A valuable contaminant to investigate in this light would be one that has chemical properties that differ from those of other contaminants for which treatments are already in place. In recent years, due to more sensitive analytics, the contaminant perchlorate has been found in a number of natural waters (USEPA 2004a,b). Regardless of how perchlorate ended up in the environment or its regulatory status, the unique chemistry of perchlorate offers an opportunity to look at the holistic approach to drinking water treatment for a class of contaminants generally not studied.

In natural water, under typical pH conditions, perchlorate $\left({\mathrm{ClO}}_4\right)$ exists as a monovalent anion dissociated from perchloric acid $\left({\mathrm{HClO}}_4\right)$ . It has a unique molecular structure: a chloride atom in the center of a tetrahedral grouping of four oxygen atoms; the negative charge is evenly dispersed over the four oxygen atoms. This structure is similar to that of the methane molecule. In addition, perchlorate (molecular weight of 99.46) is larger than a chloride molecule, but smaller than sulfate.

The perchlorate ion is unreactive as a ligand and its salts are extremely soluble, even in organic solvents. Due to these properties, it does not adsorb well on most surfaces. Despite its strength as an oxidizing agent, perchlorate is very slow to react in dilute solutions because of the high activation energy necessary for perchlorate reduction. Likewise, common reducing agents do not react with perchlorate. Perchlorate also has a relatively low charge density; consequently, it does not generally form complexes with metals the way other anions do. Therefore, common cations do not precipitate it. Perchlorate is so nonreactive that it has often been used in thermodynamic, kinetic, and other solution chemical studies as an inert supporting electrolyte (Espenson 2000). Because of its fully oxidized status, perchlorate would not be subject to aerobic biological treatment. These chemical characteristics suggest unique solutions to treatments that then have unusual implications upon implementation.

Based on the literature, four processes seem to hold promise for near-term applicability for treating perchlorate: ion exchange, anaerobic biological treatment, reverse osmosis membranes, and tailored adsorbents (Speth et al. 2006). All of these technologies have implications with regard to residuals and distribution system stability. Although costs, both capital and operating, play a large role in what technology is ultimately selected, other factors such as system stability, system sustainability, impacts on other treatment systems, residual production and handling, system size, and impacts on the distribution system will affect a utility’s choice of treatment and the regulating agency’s approach for permitting the technology.

Distribution system materials can be reactors and sinks for contaminants such as arsenic, lead, vanadium, and radionuclides (Schock and Holm 2003; Lytle et al. 2004; Schock 2004). However, based on chemistry, the perchlorate ion would not be expected to be a significant electron acceptor under normal distribution system conditions. Interactions would dominantly take the form of simple surface adsorption through coulombic attraction, which in turn would be a function of pH and the chemical nature of the pipe scale/sediment surface. Even in waters with low ionic strength, bicarbonate, chloride, sulfate, nitrate, and other anions would be present at 1,000 to 100,000 times the concentration of perchlorate. Therefore, perchlorate’s ability to compete for surface sites would be severely hampered. Overall, the presence or removal of the perchlorate ion would seem to have little potential direct impact on distribution system water quality.

However, the treatment processes employed to remove perchlorate, as briefly discussed above, are capable of causing significant water chemistry changes that could disrupt the surface and scale chemistry of distribution system materials. A water quality disturbance could cause mobilization and transport of previously-adsorbed contaminants such as arsenic, lead, vanadium, and radionuclides through solubilization, destabilization of particles, and desorption mechanisms. In the absence of previously-adsorbed contaminants, a water quality disturbance could simply render water unacceptable to consumers because of the production of discolored or turbid waters. The remainder of the text will take this into account by discussing perchlorate treatment technologies and how installing such treatments may impact the entire community’s drinking water system.

Of the technologies that have near-term applicability for perchlorate removal, ion exchange is the furthest along with regard to demonstrating its ability to remove perchlorate in a stable and sustainable fashion (Tripp et al. 2003; Boodoo 2003; Aldridge et al. 2004a,b). Specific categories of resins that have been shown to remove perchlorate include Type 1 strong-base ion exchange resins, nitrate-selective resins, and recently developed perchlorate-selective resins (Gu et al. 2000). The most cost-effective resin for any given application is dependent on the site’s water quality, particularly nitrate and sulfate concentrations, and whether the column is also expected to remove other, more weakly complexing ions such as nitrate (Aldridge et al. 2004a).

The different types of anion-exchange resins will have different residual handling issues, with the perchlorate-selective resin requiring the most unique regenerate solution (Gu et al. 2001). Regardless of the resin and regeneration/disposal path chosen, serious residual issues will need to be addressed. Some of them, depending on the specifics of a chosen design, are the handling of large volumes of concentrated salt, the need to find salt-tolerant bacteria for brine treatment, the handling of unique regenerate solutions for perchlorate-selective resins, the disposal of the exhausted resin and the regulatory requirements needed to prove the resin’s appropriateness for municipal solid waste landfills, and the arrangement and permitting of incinerating the resin. These issues are complex and warrant a separate discussion; however, it suffices to say that much thought will need to be put into handling residuals for ion exchange systems used to treat for perchlorate.

Distribution system impacts from ion exchange treatment processes will largely depend on the type and specificity of the resins and the mode of use. The more specific the resin is to perchlorate, the less inadvertent coremoval of beneficial anionic species will occur over the bed-life of the resin. Anion exchange resins typically exchange chloride or hydroxide for the retained constituents, but chromatographic effects in ion exchange columns can cause a complicated sequence of accumulation and release (Clifford 1990). The balance among pH and concentrations of ions such as bicarbonate, carbonate, sulfate, and chloride is critical in achieving regulatory compliance with the Lead and Copper Rule and in maintaining stable and noncorrosive drinking water throughout the distribution system. Without adequate posttreatment, ion exchange treatment can result in a change in water quality that can lead to destabilization of existing distribution system scales. Depending on perchlorate concentration and the desired treatment concentration, blending to enable restabilization of the water may not be feasible. This situation is considerably different from the common application of ion exchange for removal of nitrate or fluoride, where only a relatively small fraction of the contaminated source water usually needs to be sent through full treatment. Some of the potentially destabilizing impacts of ion exchange product water would be:

Anaerobic biological treatment has promise for perchlorate removal (Kirisits et al. 2001; Brown et al. 2002, 2005; Min et al. 2004; Song and Logan 2004a,b; Xu et al. 2003; Coates and Achenbach 2004; Evans and Logan 2004; Adham et al. 2004; Logan et al. 2004). However, there are serious issues with this treatment option, including sustainability; stability; the multiple barrier concept; potential pathogen release; distribution system impacts; and utility, public, and primacy agency acceptance. Anaerobic biological treatment unit processes present challenges because they are not currently used to treat drinking water in the United States. Additionally, these processes require the control of any potential microbial, nutrient, and electron donor releases into the distribution system.

Currently, there are three basic designs used for biological perchlorate reduction: fixed-bed, fluidized-bed, and membrane-biofilm reactors (Nerenberg et al. 2002; Brown et al. 2003; Xu et al. 2003). As mentioned, the electron donors and nutrients added to anaerobic reactors are not reduced to zero within the anaerobic system. Failure to remove these constituents will most certainly result in distribution system regrowth issues. Other issues of concern could include health effects from the electron donor and nutrient additives, disinfection by-products (DBPs) formed from subsequent chlorination, and public rejection of their presence in tap water. The likely post-treatment method would be aeration to increase the dissolved oxygen content while stripping any hydrogen sulfide from the water, followed by aerobic biofiltration to remove excess nutrients.

Membrane biofilm reactors may be more suitable for preventing contamination of the distribution system; however, much research needs to be completed. Membrane biofilm reactors can use hydrogen gas as an electron donor for perchlorate reduction, rather than carbon compounds such as ethanol or acetate. These reactors deliver dissolved hydrogen gas through bundles of hollow core polyethylene fibers to biofilms growing on the membrane surfaces and in contact with process waters (Nerenberg et al. 2002; Adham et al. 2004). This design eliminates the need for sparging hydrogen gas through a fixed- or fluidized-bed reactor system, thereby avoiding an atmosphere of explosive gas. Bench-scale membrane biofilm reactors have been operated without problems (Nerenberg et al. 2002; Adham et al. 2004), albeit long hydraulic residence times were required for the slower–growing autotrophs. This is likely the issue of most concern with the operation of this treatment technology and much work needs to be done. Perhaps the greatest applicability of this technology is the treatment of ion exchange brine streams, where rapid treatment is not necessary.

It is now well-established that when biofilms exist on distribution system surfaces, particularly iron, their stability and interactions with the water disinfectant are intimately interrelated to the chemistry of the pipe scale and metal release (Tuovinen et al. 1980; LeChavallier et al. 1993; Camper 1996; Zacheus et al. 2001; Butterfield et al. 2002; Roberts et al. 2002b). Additionally, bacterial regrowth can be a function of the distribution system materials and their surface chemistries (Camper et al. 2003). Currently, it is not clear what by-products of microbiological treatment for perchlorate removal would enter the water distribution system. These by-products could be either chemicals used to support the microbial perchlorate removal processes (such as acetate) or even the organisms themselves. Supplemental or altogether different disinfection schemes might be warranted to ensure inactivation of any organisms that might pass into the distribution system. Through differences in the oxidation-reduction potential of the water, changes in disinfectant type or concentration could affect the mineralogy, solubility, and stability of system scales, sediments, and adsorbed contaminants.

Reverse osmosis or tight nanofiltration membranes are relatively effective for controlling perchlorate (Amy et al. 2003); however, these technologies are known to incur high treatment costs, often requiring extensive pretreatment, posttreatment, and residual treatment processes. Reverse osmosis or tight nanofiltration membranes are perhaps best suited for point-of-use applications with regard to perchlorate.

Membranes used for particulate removal, such as microfiltration or ultrafiltration, would not be practical for perchlorate removal unless the perchlorate could be sequestered or complexed to a larger molecule, particle, or adsorbent. Some data with laboratory-clean water has shown that ultrafiltration membranes could be effective, but their performance diminished significantly in the presence of natural ions (Amy et al. 2003).

Electrodialysis systems can also remove perchlorate from drinking water (Booth et al. 2000; USEPA 2005). Instead of pressure, electrodialysis systems use electric potential to separate contaminants from the product water. Because of the different driving force, electrodialysis can have advantages over pressure-driven membrane systems. For example, high levels of silica can severely foul pressure-driven membranes, but it does not interfere with the electrodialysis process. Both RO and electrodialysis treatment generate secondary waste streams that will raise residual concerns similar to those found with ion exchange technologies.

With regard to effluent water quality, membrane processes can potentially cause the same kinds of destabilization and water quality degradation issues as anion exchange treatment. The severity is magnified by the supplemental removal of other mineral constituents (e.g., Ca, Mg, ${\mathrm{SiO}}_2$ ) that may be important structural components of the protective scales and the distribution system materials themselves. The tight membranes that are optimal for perchlorate removal are likely to produce product waters that are very aggressive. The level of perchlorate contamination will dictate the extent to which blending and post-treatment will be feasible. Substantial changes in water chemistry after the implementation of membrane treatment will require extensive pilot-scale or partial-system corrosion control studies and comprehensive monitoring programs to characterize changes, protect the public against release of accumulated contaminants (Schock 2004), and meet existing Lead and Copper Rule objectives. Posttreatment will be necessary to establish that the background corrosion-controlling matrix of the water will not interfere with the existing scale. This is difficult because of the slow chemical reaction rates on the surface of pipes. Managing the blending of multiple sources of finished water of varying chemistries to minimize adverse water quality impacts of cyclical exposure to waters of substantially different corrosivity and mineral content is especially challenging for water systems using membranes.

While post-treatment challenges may abound, treatment changes are not without potential benefits. For example, substantial reductions in DBP precursor material may result in the ability to keep, or switch, to a stronger disinfectant such as chlorine without having to worry about DBP formation. This would keep the oxidation-reduction potential high, and would be especially advantageous for water systems that produce high pH waters while having to control lead dissolution in lead service lines. Although reverse osmosis systems are particularly associated with high NOM removals, this benefit would apply to any process that could remove DBP precursor material, such as anion-exchange and modified-adsorbent technologies.

Activated carbon has a certain amount of ion exchange capacity present in the functional groups on its surface. The extent and amount of these functional groups depends on how the carbon was manufactured. Therefore, many GACs will remove perchlorate to a limited extent; however, the capacity is very small when compared with ion exchange resins. To extend this capability, researchers have recently modified, or tailored, activated carbon with various materials (Parette and Cannon 2003; Cannon et al. 2004). To date, the best results came from the use of quaternary ammonia monomers (Parette and Cannon 2003). An advantage of this technology is that it can adsorb organic contaminants concurrently with perchlorate. This would be very helpful for communities that have perchlorate along with contaminants that are amenable to carbon adsorption in their raw waters.

A great deal more work needs to be conducted on tailored GACs. The results need to be verified by pilot-scale testing under different water conditions. Also, other issues need to be evaluated, such as determining the regeneration efficiency, the degree of quaternary ammonia monomers released from the carbon during operation and regeneration, the impact of this potential release, the potential of NDMA formation upon subsequent disinfection, and whether the modifying agent will be approved as a drinking water additive.

Modified GAC treatment would be expected to produce a finished water that is similar to that produced by an ion exchange resin. Therefore, all of the same residual and distribution system issues that are a concern for ion exchange treatment also pertain to modified GAC treatment. Also, the leaching of the tailoring agent could lead to corrosion issues in the distribution system through biological or chemical mechanisms. The impact of simultaneous organic matter removal via the adsorption mechanism is somewhat unknown, as the degree to which NOM may be a beneficial scale component is largely unresearched. As mentioned in the ion exchange and membrane sections, the removal of NOM will reduce DBP precursors. This could lead to a positive net effect with regard to distribution system performance. The evaluation of such trade–offs from treatment changes is currently an active area of research.

The chemical characteristics of perchlorate allow for an interesting evaluation of drinking water treatment technologies. The technologies that hold current applicability for perchlorate removal (anion exchange resin, anaerobic biological treatment, high-pressure membranes, and modified adsorbents) are not commonly used for treating drinking water. Residual production and handling would need to be addressed. The technologies can also significantly change the water quality characteristics of the finished water. This suggests that careful consideration must be made for evaluating the consequences of installing perchlorate treatment, particularly with regard to distribution systems.

All of the issues discussed herein can be addressed with proper engineering and science. Regardless of perchlorate’s future regulatory status, research into what treatment is required for it and what that implies brings an interesting dimension that will help the drinking water community’s knowledge base as we explore emerging issues such as water reuse, pharmaceuticals, endocrine disruptors, and nano-particles that seemed extremely remote only a few decades ago.

Any opinions expressed in this paper are those of the authors, and do not necessarily reflect the official positions and policies of the USEPA. Any mention of products or trade names does not constitute a recommendation for use by the USEPA.