Minimizing Ozonation By-Products in Drinking Water

About the author:

Thomas Grosvenor is an environmental engineer with Foster Wheeler Environmental Corporation, Livingston, N.J. His focus has been directed toward the areas of process/project engineering, air permitting and pollution prevention.

With the promulgation of the USEPA's restrictions on disinfection by-products (DBPs) permitted in America's drinking water attributable to chlorination, the ozonation process seems to be a more attractive disinfection method.

Ozone (O3) and its primary reactive products, the hydroxyl free radical (OH*), are strong oxidizing agents. However, it can also lead to the formation of potentially harmful by-products inclusive of bromate ions (BrO3), aldehydes and peroxides. Such by-products can be handled by either removing the DBPs themselves or by inhibiting their formation through precursor removal methods and process optimization techniques. This paper provides a detailed summary of ozonation and by-product formation chemistry, effective approaches toward the control of by-product formation and DBP/DBP precursor removal technologies.

Regulations and Health Effects

Although the disinfection of drinking water provides protection against microbial disease, it can produce chemical by-products that may pose risks to public health. Current USEPA regulations, specially applicable to ozonation disinfection by-products include two stages of the Disinfection/Disinfection By-Products (D/DBP) Rule. Stage one, proposed in 1994 and promulgated this year, provides maximum contaminant levels (MCLs) for the sum of five haloacetic acids (HAA5) at .060 mg/l, BrO3 at .010 mg/I, and brominated trihalomethanes (THMs) at .08 mg/l. Stage two, expected to be promulgated in 2002, intends to further lower the MCLs.1 In addition, the D/DBP rule calls for the implementation of an enhanced coagulation control strategy for total organic carbon (TOC) removal as a means of limiting the formation of all DBPs. Treatment facilities that achieve a specified percent removal of TOC prior to the application of a continuous disinfectant or that achieve a residual TOC concentration <2mg/l prior to the application of a continuous disinfectant are considered to be in compliance.2

Depending on the materials present in the water, several classes of DBPs can form during disinfection processes. Most DBPs have been studied toxicologically and some (particularly the ozonation DBP of BrO3) are suspected human carcinogens.3 Further, conflicting data suggests the possibility that formaldehyde is carcinogenic at concentrations produced in drinking water during ozonation. Additionally, aldehydes may be viewed as surrogates for a large number of polar organics that are formed at low levels when ozone is used in the treatment of natural water. Of major concern regarding such polar by-products is their high biodegradability. High levels of these compounds in distribution systems can promote microbial regrowth leading to operational problems and the possible exposure of consumers to gastrointenstinal disease.4 Compounds having the potential of forming include peroxides that have more serious health effects than aldehydes themselves.

Ozonation By-products

O3 is produced commercially by exposing either the oxygen in the air or pure oxygen to a high voltage electrical discharge. O3 can oxidize and volatilize organics, control algae and associated taste- and odor-producing compounds, destabilize (microflocculate) certain types of turbidity, remove color causing compounds, and oxidize iron and manganese. It is capable of decomposing and oxidizing unwanted drinking water constituents by two reaction pathways, the direct pathway (molecular O3 pathway) and the indirect pathway (hydroxyl radical or OH* pathway), with O3 decomposing to OH* by a the following chemical reaction:5

Although the use of O3 as an alternative disinfectant to chlorine will not produce chlorinated trihalomethanes (THMs), haloacetic acids (HAAs) or other chlorinated by-products; it can react with NOM to produce a variety of oxidation by-products that typically include aldehydes, aldo- and keto-acids, carboxylic acids and peroxides. NOM, a major component of TOC, is a complex matrix of organic chemicals that can be derived from partial bacterial degradation of soil, living organisms and plant detritus. Such oxidation by-products may form by the following reaction:

In addition, if bromide ion (Br) is present along with NOM, as is the case when salt water intrusion occurs, O3 has the potential to oxidize Br to hypobromous acid (HOBr), that can then oxidize NOM to brominated THMs, HAAs, haloacetonitriles (HANs) and cynogen bromides. Without NOM, O3 can still oxidize Br to HOBr or the ionized form of HOBr, hypobromite (OBr­p;). These species can react further to produce bromate ions (BrO3) and can also regenerate Br­p;. Experiments have shown that lower pH favors HOBr and organic bromide formation, whereas higher pH favors OBr­p; and BrO3 formation. O3 concentrations and contact time also affect BrO3 formation. The reactions of O3 with Br are as follows:

Summarizing the reaction scheme presented, O3 oxidizes Br­p; to form HOBr. HOBr reacts further with O3, but only in its ionized form (OBr­p;) and can be further oxidized to either BrO3 or to a species that regenerates Br­p;. The reactions of O3 with aqueous bromine (HOBr/OBr­p;) and Br are relatively slow but can cause a significant O3 demand at high concentrations and high pH.6, 7, 8

A detailed model of the reactions that take place on the ozonation of Br­p; containing solutions, incorporating both the molecular O3 and OH* mechanisms is presented in Figure 1.9 Summarizing the model, Br­p; may either be oxidized to BrO3 or to HOBr. HOBr may then react with NOM to form brominated organics such as bromoform, brominated acids, bromoacetonitriles, etc. HOBr is primarily responsible for bromination reactions with NOM. Its ionized form, or conjugate base, OBr­p;, is the primary intermediate in the oxidation of Br­p; to BrO3­p;. From the model, it is clear that a lower pH favors the reaction of HOBr with NOM, or the formation of organic bromides, while a higher pH favors the intermediate formation of OBr­p;, which inevitably leads to BrO3­p; formation.9

Illustrative of the competition between NOM and Br for O3 or NOM and HOBr/OBr­p; for OH* is the compilation of reaction rates shown in Table 1. Studying the reaction rates, one can see that NOM oxidation clearly dominates Br oxidation. NOM also scavenges OH* more aggressively than does HOBr/OBr­p;, as apparent by the high reaction rates for the reaction of OH* with NOM shown in sequence 2. Looking at sequence 3, HOBr oxidation of NOM or the formation of brominated organic compounds, appears more dominant than OBr­p; reactions to BrO3. It can be concluded from this reaction rate data that the presence of NOM significantly reduces the BrO3 formation relative to NOM-free water. Apparently, the mechanism for the decrease in BrO3 formation has two components. First, NOM can react directly with O3 and OH* to reduce the amount of oxidant available for Br and HOBr/OBr­p; oxidation. Second, NOM can react with HOBr/OBr­p;, the main intermediate in the formation of BrO3.10

Ozonation By-Product Minimization

When DBPs are a concern in drinking water treatment processes, there appear to be three measures that can be taken. First, it is viable to remove the precursors that react with the disinfectant to form the unwanted DBP. Second, water treatment processes can be optimized to control such reactions. Third, DBP removal processes can be used. All these approaches can be used in conjunction to the extent necessary to take maximum advantage of their respective benefits toward water quality enhancement.

Precursor Removal: NOM in surface water sources of many utilities can vary in type and concentration throughout the year. Such variants in influent conditions can prove a challenge for NOM removal processes. Again, NOM is a major component of TOC and is effectively removed by means of enhanced coagulation. It also can be removed by GAC adsorption and membrane filtration. GAC adsorption performance is even effectively enhanced by the upstream ozonation processes' breakdown of NOM into smaller oxidation by-products that are more biodegradable.11

Coagulation: Coagulation is a treatment process that includes chemical addition, rapid mixing and flocculation. NOM removal can be influenced by the type of coagulant used, the coagulant dosage, pH, mixing, water quality change and the order of chemical addition. Maximum NOM removal tends to occur at pH values between 5 and 6. Low alkalinity water may require the addition of lime to maintain the pH in this range. Full scale treatment plants have demonstrated that moving the location of the disinfection process to a point following coagulation and sedimentation, modifying the coagulation process for increased removal of organic materials, or both, can result in substantial reductions in DBP formation. Coagulation can be an effective pretreatment technique subsequent to GAC or membrane filtration in that it removes particles that might clog GAC beds, reducing the frequency of carbon regeneration and replacement, and it removes NOM, notorious for shortening membrane lives.12

GAC: Typically, NOM removal by GAC generally requires at least 10 to 15 minutes empty bed contact time (EBCT) to achieve reductions of any magnitude over a reasonable time. It is an effective approach for removing TOC, HANs, THMs and HAA5 precursors.12 Further, a biofiltration approach, exploiting GAC adsorbers' ability to foster a biofilm, can be taken. The irregular surface of GAC makes it especially effective for biofilm colonization, as long as chlorine is not applied continuously before filtration. O3 can be applied prior to the rapid filter or GAC adsorber, and the added O3 significantly increases the concentration of biodegradable organic matter (BOM).13

Membrane Processes: Membrane processes have been demonstrated to remove effectively and economically DBP precursors in water containing high concentrations of organic matter. Membrane treatment processes include reverse osmosis (RO), ultrafiltration (UF), nanofiltration (NF) and microfiltration (MF). Membranes are specified by membrane pore size, molecular weight cut-off (MWCO), membrane material and geometry, targeted materials to be removed, type of water quality to be treated and the treated water quality desired.

The removal or rejection characteristics of a membrane are usually rated on its nominal pore size or the MWCO. This MWCO is only a rough indication of the membrane's ability to remove a given compound because molecular shape and polarity also affect rejection. Additionally, membrane materials can be placed into two general groups: hydrophilic and hydrophobic materials. The flux of water containing NOM through a membrane and the flux recovery after backwashing appears to be greater for a hydrophilic membrane than for a hydrophobic one with similar MWCOs. Synthetic membranes have been made from organic polymers, porous carbon, ceramics and other materials. Membranes made from organic polymers are most common and offer the greatest degree of flexibility concerning rejection characteristics and module design. Such polymers include cellulose acetate, polysulfone, polyamide or polycarbonate. Commercially produced polymer membranes may be configured in bundles of hollow fibers, stacks of flat plates or spiral-wound assemblies.14

Typically, RO, or hyperfiltration, membranes are capable of removing ion-sized material such as sodium, chloride, calcium, and sulfate as well as non-polar organic molecules. UF membranes remove materials that are on the order of a nanometer in size (10­p;9m) or larger. UF and NF membranes are more frequently rated based on the smallest molecular weight of material that has been found to be removed by the membrane. MF membranes are usually rated by pore size and are capable of removing micron-sized (10­p;6m) materials from water.14

Controlling DBP Formation

Effective Ozonation: Ozonation oxidation pretreatment applications often cause rapid depletion of the residual, hampering O3 disinfection. NOM directly consumes molecular O3, scavenges OH* and promotes O3 decomposition. Among the various water quality factors that affect the rate and nature of ozone's decomposition, influent NOM concentration and pH are the most influential. Basically, the presence of NOM greatly accelerates O3 decay, while the O3 decay rate increases with increasing pH. As the rate of O3 decomposition increases due to reactions with OH­p; (pH dependent) or NOM, other solutes (e.g., Br­p;) will compete less favorably for O3 and may, as a consequence, not become oxidized. Application points to consider are downstream of either the sedimentation or filtration processes, where both the oxidant demand and turbidity levels are much lower.5, 15

The formation of BrO3­p; is strongly dependent on site-specific conditions such as pH, Br­p; concentration and O3 dosage. First, a higher pH tends to favor the increased formation of BrO3­p;. Additionally, higher O3 temperatures tend to produce more BrO3­p; due to the decrease in the pKa of HOBr and OBr­p;. Thus, to minimize BrO3­p; formation, particularly at higher pHs, O3 concentration should be kept low and contact time extended.16 Further confirmation that O3 dose affects the formation of by-products is provided from findings that the formation of low molecular weight aldehydes and carboxylic acids increases initially with the O3 dose. However, above an optimum dosage, the level of formation begins to decline. Formation is also decreased at downstream point of application as preceding processes reduce precursor content.15

AOP, the advanced oxidation process, or the PEROXONE process, incorporates H2O2 to aid in oxidation enhancement. H2O2 can initiate the O3 decomposition cycle by a single electron transfer process involving its conjugate base (HO2­p;).17

The efficiency of color removal by O3 is accelerated by H2O2. The dependence on the concentration of peroxide suggests that H2O2 is acting as a scavenger at high concentrations. Typically, the addition of H2O2 decreases BrO3­p; formation at a high pH but increases it at a low pH and leads to rapid O3 decomposition and substantially decreased CT.5

Ammonia Addition: Ammonium sulfate has proven to be an effective quenching agent for HOBr, the active brominating species resulting from the interaction of inorganic bromide with O3. Although found to increase the rate of O3 decomposition, ammonia addition has proven effective in interrupting both the molecular O3 and OH* reaction pathways, particularly in low-pH waters. It can combine with HOB4/OBr­p; to form bromamines and can act as a radical scavenger.4,5,16

Acid (H2SO4) Addition: The addition of H2SO4 to decrease pH decreases BrO3­p; formation by shifting HOBr/OBr­p; equilibrium from the reactive OBr­p; species to HOBr. Acid addition is the most predictable and potentially effective BrO3­p; minimization strategy, but may not be cost-effective in high alkalinity waters and the resulting BrO3­p; reduction is accompanied by increased total organic bromine (TOBr).5

In summary, the factors that have an effect on BrO3­p; formation include HOBr/OBr­p; concentration, O3 decomposition rates and OH* concentration. Decreasing HOBr/OBr­p; concentration decreases BrO3­p; formation. OBr­p; is more reactive than HOBr and higher OBr- concentrations lead to increased BrO3­p; formation. O3 decomposition rates affect HOBr/OBr­p; concentration and the production of OH*. Rapid O3 decomposition rates typically decrease HOBr/OBr­p; concentrations but increase OH* production. OH* plays a dominant role in oxidizing OBr­p; forward to BrO3­p;. OH* scavengers (NOM, Br­p;, t-butanol and ethanol) decrease OH* concentrations affecting both the rate and extent of BrO3­p; formation. Table 2 presents BrO3­p; control strategies in terms of how their effectiveness correlates with source water pH and alkalinity.5

DBP Removal

GAC: GAC filtration chemically reduces BrO3­p; to Br­p;. The rate of reduction is a function of EBCT and influent BRO3­p; concentration. An effective means of aldehyde removal involves the use of biologically activated granular activated carbon (BAC) filters (GAC filters supporting biomass). GAC supports a larger bacterial population and provides better removal of assimilable organic carbon (AOC) than conventional filter media.4, 16, 18

Membrane Processes: Membrane processes tend to be more effective at rejecting BRO3­p; than Br­p; because BRO3­p; has a larger molecular structure and higher molecular weight than Br­p;. Although effectively removed through membrane processes, NOM has the tendency to foul them, shortening lifetime. Thus, efforts to minimize NOM concentrations prior to membrane filtration should enhance performance with regard to both BrO3­p; and Br­p; rejection.16

Additional BrO3­p; Reduction Techniques: It has been found that the use of ultraviolet (UV) irradiation will reduce, in part, BrO3­p; to Br­p;. However, its efficiency is highly dependent on the UV wavelength and pressure utilized. Additionally, BrO3­p; is reduced to Br­p; by the addition of ferrous sulfate (FeSO4). Pilot testing has demonstrated that it is a difficult process to optimize (very pH and temperature dependent) while trying to produce filtered water with low turbidity.19

Conclusions

It is difficult to simply pave a procedural pathway toward ozonation by-product-free drinking water in the sense that each treatment plant's influent may respond differently to similar process optimization techniques, due to variations in key parameters. For certain, trade-offs exist when O3 is used. Lowering the O3 dosage may minimize the formation of BrO3­p;, but increase the formation of other DBPs. On the other hand, higher O3 dosages can lead to significant BrO3­p; formation, particularly at high Br­p; levels and at ambient pH. Further, a utility must not only measure the trade-offs associated with the control of these parameters, but also must weigh in the control of viruses and protozoan cysts. Thus, since source waters respond very differently to O3 treatment, it is essential that bench-and pilot-scale studies be performed to verify process efficiency.

Both GAC filtration and membrane filtration have proven to be effective antagonists of DBPs. GAC filtration is particularly effective in removing DBP precursors when a biofilm is fostered. It is thought that the only way in which many utilities will be able to balance the requirements for microbiological safety and minimal DBPs, while also maintaining high quality drinking water, is through such an enhancement. With minor modifications and when BOM concentrations are relatively low, conventional rapid filters and GAC adsorbers can be turned into hybrid biofilters by eliminating chlorine in the feed water.

Membrane filtration technologies also have proven effective in the removal of DBPs and precursors. However, when attempting to remove organics from a treatment plant influent, concern has to be directed toward membrane fouling. It would be beneficial in such applications to be furnished with a membrane characterized by high organics rejection and low inorganics rejection, particularly when the concentration of organics is high. Further, the use of quenching agents should not be overlooked. While ammonium sulfate can be an effective quenching agent for HOBr/OBr­p; in ozonated samples, other quenching agents such as sodium thiosulfate and sulfite have been observed to destroy some brominated organic by-products.

References

1. USEPA - Revised Final Draft for the Drinking Water Criteria Document on Ozone and Ozonation By-Product. Prepared for Health and Ecological Criteria Div., USEPA, Washington D.C., (June 1991).
2. White, M.C., Thompson, J.D., Harrington, G.W., and Singer, P.C., 1997, Evaluating Criteria for Enhanced Coagulation Compliance, Journal AWWA, 89 (5), 64.
3. Alceon Corp. An Overview of Available Information on the Toxicity of Drinking Water Disinfectants and Their By-Products. Cambridge, Mass. (Feb. 8, 1993).
4. Weinberg, H.S., and Glaze, W.H., 1996, "An Overview of Ozonation Disinfection By-Products, Disinfection By-Products in Water Treatment," CRC Press, 1996, 187.
5. AWWA Research Foundation: Amy, G.L., Westerhoff, P., Minear, R.A., and Song, R., "Formation and Control of Brominated Ozone By-Products," AWWARF, Denver, Colo., 1997.
6. Glaze, W.H., Weinberg, H.S., and Cavanagh, J.E., 1993, Evaluating the Formation of Brominated DBP's During Ozonation, Journal AWWA, 85 (1), 96.
7. Shukairy, H.M., Miltner, R.J., and Summers, R.S., Bromides Effect on DBP Formation, Speciation, and Control: Part 1, Ozonation, Journal AWWA, June 1994, 72.
8. Haag, W.R., and Hoigne, J., 1983, Ozonation of Bromide-Containing Waters: Kinetics of formation of Hypobromous Acid and Bromate, Environmental Science Technology, 23 (7), 838.
9. von Gunten, U., and Hoigne, J., 1996 ozonation of Bromide-Containing Waters: Bromate Formation Through Ozone and Hydroxyl Radicals, "Disinfection By-Products in Water Treatment," CRC Press, 1996, 187.
10. Westerhoff, P., Rengao, S., Amy, G., and Minear, R., 1998, NOM's role in Bromine and Bromate Formation During Ozonation, Journal AWWA, 89 (11), 82.
11. Drikas, M., 1997, NOM (Natural Organic Matter) The Curse of the Water Industry, Water, September/October 1997, 22.
12. Jacangelo, J.G., DeMarco, J., Owen, D.M., and Randtke, S.J., Selected Processes from Removing NOM: An Overview, Journal AWWA, January 1995, 64.
13. Rittman, D.E., 1996, Back to Bacteria: A More Natural Filtration, Civil Engineering, 66 (7), 50.
14. AWWA Membrane Technology Research Committee, 1992, Committee report: membrane Processes in Potable Water Treatment, Journal AWWA, January 1992, 59.
15. Bryant, E.A., Fulton, Gp.P., and Budd, G.C., 1992, "Disinfection Alternatives for Safe Drinking Water," Van Nostrand Reinhold, New York.
16. Marhaba, T.F., and Medlar, S.J., Treatment of Drinking Water Containing Bromate and Bromide Ions, Critical issues in Water and Wastewater Treatment - Proceedings of the 1994 National Conference of Env. Eng., July 11-13, 1994, 476.
17. Duguet, J.P., Brodard, E., Dussert, B., Mallevialle, J., Improvement in the Effectiveness of Ozonation of Drinking Water Through the Use of Hydrogen Peroxide. Ozone Science and Engineering, 7:241, 1985.
18. Krasner, S.W., Sclimenti, M.J., and Coffey, B.M., Testing Biologically Active Filters From Removing Aldehydes Formed During Ozonation, Journal AWWA, May 1993, 62.
19. Siddiqui, M., Amy, G., and McCollum, L., 1996b, Bromate Destruction by UV Irradiation and Electric Arc Discharge, Ozone Science and Engineering, 18, 271.