A New Look at Disinfection By-Products in Drinking Water
The application of disinfectants in drinking water treatment for controlling microbial quality has drawbacks. These include the formation of potentially harmful disinfection byproducts (DBPs) that can be potential carcinogens (e.g., trihalomethanes [THMs] when chlorine is used as a disinfectant). The general equation is:
ÊÊÊNOM + Disinfectant -> DBP
ÊÊÊWhere NOM = Natural Organic Matter
The United States Environmental Protection Agency (USEPA) initiated a negotiated rule-making process for the Disinfectant/Disinfection By-Products (D/DBPs) Rule in 1992. The USEPA had to draw on the expertise of others to prepare a rule due to the complexity of the issues. The regulations were proposed in two stages. Stage 1 of the D/DBPs Rule was proposed in 1994 and became effective in December 1998. It lowered the total THM (TTHM) maximum contaminant level (MCL) from 0.100 to 0.080 mg/L and MCLs for three other classes of DBPs. The rule also set maximum residual disinfectant levels (MRDL) for three disinfectants.
To provide necessary data for Stage 2 of the D/DBP regulations, the Information Collection Rule (ICR) (begun July 1, 1997, ended December 1998) was proposed in 1994 with Stage 1 of the D/DBP Rule. Stage 2 is expected to be reproposed in 2000 and require even lower MCLs for DBPs than those proposed in Stage 1. The 1996 Amendments to the Safe Drinking Water Act (SDWA) require USEPA to promulgate the Stage 2 Rule by May 2002. Stage 2 MCLs of 0.040 mg/l for TTHMs and 0.020 mg/l for HAA5 were proposed as "placeholders" until data from the ICR are available to propose new levels. The ICR requires surface water systems serving over 100,000 people and groundwater systems serving over 50,000 people to monitor for microbial contaminants and DBPs. Granular activated carbon (GAC) and membrane pilot testing are required for surface water utilities with raw water total organic carbon (TOC) greater than 4 mg/L and serving over 100,000 people as well as for groundwater with finished water TOC greater than 2 mg/L and serving over 50,000 people. Although the regulations target TOC, it is an aggregate measure of organic material and is not an indicator of the precursory character to DBPs. Table 1 summarizes the Stage 1 and anticipated Stage 2 standards on the disinfectants and the DBPs.
Bromate ions and chlorite ions are DBPs of ozone and chlorine dioxide, respectively.
Controlling and Removing Precursors
Most DBPs have not been studied toxicologically. Some of those already studied are probable or possible human carcinogens. Depending on the disinfectant used and the precursor materials present in the water, several classes of DBPs may form including THMs, HAAs, chlorate ion, chlorite ion, bromate ion and others. When DBPs are a concern in a water treatment process, there are two main approaches to solving the problem. One approach is to control the precursors that react with the disinfectant to form the unwanted DBPs. The other approach is to allow the DBPs to form and then use a separate removal process for the DBPs.
Precursor control and removal strategies are focused mainly on NOM present in water. NOM is considered to be the major precursor for DBP formation. TOC and UV-254 (ultraviolet absorption at 254 nm) typically are used as an aggregate quantitation of DBP precursors. NOM is very site specific, and the different components of NOM (e.g., fractions) are removed with varying degrees of effectiveness by different strategies. Enhanced coagulation and GAC were considered as best available technologies (BAT) by the USEPA for precursor control. Table 2 shows the common control strategies used in the removal of NOM. NOM characterization research has focused on humic content in addition to molecular weight (MW) ranges. Table 3 summarizes the viable treatment processes for five basic fractions of NOM. Limited research has been performed on the identification and removal of specific precursors.
After DBPs have formed it is possible to remove them with a subsequent treatment process. Although the USEPA considered air stripping or GAC adsorption as possible techniques (although not necessarily BAT) for the 1979 USEPA THM Rule, these technologies were not even considered for the Stage 1 D/DBP Rule. This is because DBPs will reform in the distribution system if the distribution system uses chlorine as the disinfectant. In addition, although THMs can be removed in the plant with air stripping, HAAs cannot. THMs break through GAC quickly. It is better to use GAC to remove THM precursors than the THMs themselves. When chlorinated water is passed through GAC, chlorinated dioxins are produced when the GAC is regenerated. Therefore, it is better to curtail chlorination until after GAC filtration.
Other issues not associated with removing the DBPs include alternative disinfectants that may produce DBPs of health concern and technological and economical feasibility of advanced precursor removal technologies (e.g., membranes) that can meet very low cancer risk levels.
The formation of DBPs depends on the type of disinfectant used, source water and contactor kinetic conditions. The proposed regulations are confounded by many questions including concerns that
- risks associated with many animal or epidemiological studies on DBPs remain uncertain,
- alternative disinfectants to chlorine have their own consequent problems as well as possibly additional DBPs that may be harmful to health,
- the technological and economical feasibility of technologies that can meet very low cancer risk levels as well as be applied in all geographic regions of the United States, and
- the balance between the D/DBP Rule and the ESWTR regulations will lead to higher risk of either microbial contamination or DBP formation.
Currently, aggregate parameters such as TOC and UV-254 are used as DBP precursor identification methods. However, these parameters are not indicative of the specific precursory content. Research on the identification of specific precursors to DBPs by non-aggregate parameters is needed to assess their removal and formation by water treatment unit processes for minimization of DBP formation. If minimization of DBP formation was due solely to specific removal of precursors, microbial inactivation issues would be resolved because disinfectant dosage may be increased without an increase in DBP formation. Since chlorine has been the most widely used disinfectant, it is important to focus on its DBPs.
Isolation and fractionation of NOM and subsequent DBP formation potential study has been performed by the author to determine the most problematic fraction towards the formation of chlorinated DBPs. NOM from a water source in central New Jersey was isolated and fractionated by resin adsorption methods into six fractions; hydrophilic acid, hydrophilic base, hydrophilic neutral, hydrophobic acid, hydrophobic base and hydrophobic neutral. Seven-day chlorine DBP formation potential tests were performed on all fractions. The hydrophilic acid fraction was found to be the most abundant (about 50 percent) and most problematic fraction (i.e., precursor) towards the formation of THMs and HAAs.
If the hydrophilic acid fraction can be determined in the water rapidly, utilities may be able to optimize its removal (i.e., addition of polymers, etc.) and lower the chlorine DBP formation of the water. One method that was investigated and developed by the author was the spectral fluorescent signature (SFS) technique coupled with a multiple linear regression model for rapid identification of the six NOM fractions. The research was funded by the New Jersey Department of Environmental Protection with the participation of major drinking water utilities and sewage treatment authorities. The main technique uses the fluorescent properties of organic compounds. The main advantages of fluorescent techniques over other methods of detection are high sensitivity and the allowance of the proximate diagnostics to be carried out without time-consuming pretreatment of water. In general, the SFS is the total sum of emission spectra of a sample at different excitations. It is recorded as a matrix of fluorescent intensity in coordinates of excitation and emission wavelengths in a definite spectral window. The correctness of the SFS depends on the right combination of excitation and emission wavelengths. The spectral matrix includes the main fluorescent features of organic compounds.
The multiple linear regression model was developed to predict the concentration of the fractions using specific fluorescence properties such as slopes, areas and intensities of the water sample SFS. The hydrophilic acid fraction model is as follows.
C = 0.161 + 0.358 * P + 0.000663 * A + 25.7 * S - 0.0368 * S * A
S = [P - Pi]/120
C = Predicted concentration of hydrophilic acid fraction (mg/L).
P = Intensity at excitation 225 nm and emission 345 nm (relative intensity units).
Pi = Spectrum intensity at excitation 225 nm and emission 249 nm (relative intensity units).
A = Area of emission spectrum at excitation 225 nm (relative intensity units*nm).
S = Rising slope of the excitation 225 nm spectrum (relative intensity units*nm).
Therefore, one can predict the hydrophilic acid fraction concentration in minutes using the new technique instead of days using the isolation and fractionation technique. This enables rapid on-line response beneficial for removal optimization in water treatment, since source water fractions vary according to climatic, hydraulic, geologic and other conditions pertinent to the watershed. The technique also enables rapid on-line delineation of rivers or reaches in the watershed for determining the source (point or non-point) of hydrophilic acid for source water protection and management. All this would not be possible with the use of aggregate parameters.