Report
A call for analytical techniques to determine unidentified disinfection byproducts in drinking water.
M
ost of us take for granted the ready availability of drinking water at the turn of the tap. This is, in no small measure, the result of chemicals that destroy potentially pathogenic microorganisms up to the point of ingestion. Such chemical agents are called disinfectants and are needed because of the lack of pristine water sources in today's world. However, it is inevitable, considering the pathways by which water reaches drinking water treatment plants, that a certain amount of natural organic matter (NOM) is incorporated into the water. Disinfectants are strong oxidizing agents that can engage NOM in a series of reactions, which generate disinfection byproducts (DBPs). The analysis of DBPs in water provides a serious challenge to researchers. Not only is the chemistry of the disinfection process complex but the organic constituents in water before disinfection are highly variable and are far from fully characterized Thus researchers must attempt to identify unknown products ated from incompletely defined starting materials Furthermore many of the products are not well suited to GC/MS analysis the standard forthe analvQis of drinking water l i n l / n o w n c P r r \ K a h h r tVif» Vwact wcnr t o ctr\r\rrtcknV\
the prnhlem k to (Wplnn a rancrp of rnmplpmentary tprhniniif"! that can charactpwp the-
array of compounds in the mixture.
Howard Weinberg
Disinfection
byproducts in drinking water: The analytical challenge
University of North Carolina-Chapel Hill Analytical Chemistry News & Features, December 1, 1999 8 0 1 A
Report
Structure 1
This Report presents some of the discussions that took place at a recent international workshop (i) convened to identify the appropriate analytical techniques for the determination of, as yet, unidentified DBPs in drinking water. The future regulatory drivers for protecting the quality of our drinking water and public health require analytical and aquatic chemists to collaborate on the identification of these techniques. This article is an open challenge to analytical chemists to join in this quest. We welcome ffedback and invite readers to become part of tjje research effort to answer the questions posed in this Report
tional groups associated wiih natural material, which may provide the basis for many of the byproducts following drinking water treatment. Fractionation of this material permitted surrogate measurements of the humic and fulvic materials, which may be further subdivided into hydrophilic, transphilic, and hydrophobic fractions (4). Since 1979, the U.S. Environmental Protection Agency (EPA) has regulated the acceptable levels of trihalomethanes (THMs) in tap water on the basis of toxicoIogical data and the availability of analytical techniques to measure them (5). Most of the early targeted species were thermally and chemically stable, volatile or semivolatile, neutral organic halogens—such as THMs and haloacetonitriles (HANs). These compounds could be easily isolated from the aquatic matrix by purge-and-trap or liquid-liquid extraction and then detected on a gas chromatograph equipped with an electron-capture detector.
The missing mass
Chlorination, the most widely practiced disinfection technique, is known to produce halogenated, volatile byproducts in drinking waters, which may have public health implications (2). The natural decaying organic materials present in surface waters have been identified as the primary precursor materials to organic DBPs, but there is no single, universally accepted model of the precursor material. The degree of complexity of this subject was outlined by Christman and Gjessing (3), who presented several model structures of NOM components based on various spectroscopic measurements. An example is Structure 1. This model like many others presented in the literature indicates the type of func802 A
However, a mass balance of the halogen component of the identified DBPs compared with the total organic halogen content of the chlorinated water indicated that a large fraction of fhe halogenated DBPs in nrinking water were not identified (6). As drinking water treatment plants incorporated new processes, such as the use of chlorine dioxide or ozonation and subsequent chloramination (in part, to lower the levels of these DBPs in their finished product) alternative byproducts became a concern Ozonation as an oxidizer appears to generate more polar materials than chlorine but without the "fingerprint" of the halogen atom. With currently available techniques, less than 40% of the byproducts of ozonation have been identified (Figure 1). It has thus become clear that more sophisticated analytical approaches are necessary to isolate such byproducts from the aquatic matrix and convert them into a form that permits sensitive detection. Furthermore, more than 50% of the organic halides produced by chlorine disinfection are currently uncharacterized, with large portions thought to be nonvolatile and/or polar, and, therefore, not readily amenable to the traditionally practiced techniques of fxtraction and GC. .pecies with reactive (labile) functional groups such as haloacetic acids and aldehydes have been targeted through derivatization techniques (EPA M e t h o d WW 9 \7X\ Knt ciir^Vi an annroar-Vi Hnfic
"broad spertrum" analysis of nnten I*
ii
hundreds ifnotthnn
sands, of unidentified „ „ „
n
'.,
,,
DBPs that could affect human xiGcUui 3TG idcntiFigure 1 . Distribution of DBPs in drinking w a t e r .
clIlU. IdSlcl mcOlOQo TXlciL
(a) Relative composition of ozonation byproducts in drinking water as a proportion of the assimilable organic carbon. (Data collected by Stuart W. Krasner, Metropolitan Water District of Southern California. Adapted with permission.) (b) Distribution of halogenated byproducts as a proportion of total organic halogen in chlorinated drinking water.
analysts at drinking water utiuues can use wiuiout high SKiti levels.
Analytical Chemistry News & Features, December 1, 1999
Predictive chemistry
Chlorine reacts with NOM through a variety of mechanisms, including electrophilic aromatic substitution, electrophilic addition, oxidation, and free radical reactions (8). Chlorine will react with both phenolic and nonphenolic, including aromatic and nonaromatic, structures in NOM, generating many of the chlorinated byproducts produced during disinfection (9,10). Deetherification of the fJ-arylether linkages of the NOM polymer produces smaller molecular weightfragments,which are more water soluble. Demethylation reactions form ionizable phenolic that also facilitate the base solubility of the NOM fragments The more water-soluble the byproducts of chlorination the greater the challenge it is to isolate themfromwater Chlorine dioxide is more selective in its reactions with NOM. The primary reaction mechanism appears to be the cleavage of the (3-arylether linkages, with the subsequent formation of hydrophilic quinones, muconic acids, and muconic acid monomethyl ethers {11). This creates NOM fragments with higher water solubility, providing an analytical challenge for isolation and identification. On the other hand, chlorine dioxide forms chlorinated products only to a very limited extent, which are usually related to the amount of chlorine impurity in the feedstock Understanding ozone reactions and predicating the byproducts are complicated by the many sites of attack on NOMs. The reaction products themselves are even susceptible to attack. For example, ozone is a powerful oxidant that, like chlorine dioxide, reacts with free phenolic groups in the NOM structure (12). Although reaction rates with alkene groups are more rapid than with the aromatic structures, ozone will open aromatic rings to form muconic acid structures. To date identified ozone byproducts include a homologous series of straight chain aldehydes short chain aldoand di-acids various hydroxylated and keto acids vanillin vanillic acid glyoxal and glyoxvlic acid (and possibly the methyl ester) A comprehensive summary of confirmed and tentatively identified DBPs been compiled by Richardson There are, therefore, a wide diversity and number of potential byproducts that result
Table 1 . Priority list of DBPs that currently have inadequate or no analytical methodologies. Brominated furanones Bromopicrins lodinated chloromethanes Cis- and frans-2,3,4-trichloro-2-butene nitrile Halopropanones 3,3-Dichloropropenoic acid (3,3-dichloroacrylic acid) Brominated acetonitriles Halogenated aldehydes Halomethyl-5-hydroxy-2(5H)-furanones 2,3-Dichloro-4-oxobutenoic acid 2,2-Dichloroacetamide 2-Hexenal 5-Keto-1-hexanal 2-Butanone Cyanoformaldehyde 6-Hydroxy-2-hexanone
from drrnking water disinfectton, many oo which have been hypothesized or tentatively identified only through laboratory-controlled synthetic reactions at a high disinfectant dose. Methods for their identification are severely hindered by some of the limitations previously noted. Yet, the design of new analytical techniques to further our knowledge in this area would similarly be distracted if a priority list of DBPs were not established. Recent discussions havefocusedon such a list which is summarized in Table 1. Thii list is certainly incomplete, but it begins a drive to evaluate new analytical approaches that extend beyond those currently practiced Extraction techniques
The successful analysis of organic micropollutants in water depends on the effective recovery of the targeted analyte from the matrix. DBP history actually started on its analytical road with THM analyses performed by direct injection of aqueous samples into a gas chromatograph. It is surprising that so little emphasis has been placed on the isolation of organic micropollutants by direct analysis of water using reversedphase HPLC. The only approved methodology that minimizes sample pretreatment which is in routine use for nonvolatile DBPs is an ion chromatographic analysis of oxyhalides (EPA Method 300 0) In the past 20 years, a wide range of additional DBPs have been identified that require varying degrees of sample preparation to isolate themfromthe aquatic matrix and present them in an appropriate form for sep-
aration and detection. The isolation and con centration procedures employed today take into account the interactive forces binding a analyte to the aquatic matrix and attempt to eliminate sources of interference while retaining analyte integrity. Finally, the analyte has to be prepared in a way that ensures enhanced separation, column efficiency and lifetime, detector sensitivity, and chromatographic peak resolution. Health-effects data promoting the moni toring of part-per-trillion levels of DBPs ar pushing our sample-handling techniques t new levels of sensitivity and sophistication Solid-phase microextraction (SPME) and supercritical fluid extraction (SFE) afford die most promise for organic solvent-free sample handling and full automation, with die potential for on-line monitoring of wau. streams. Analysis options
In addition to GC/MS and GC witii electron capture detection (GC/ECD) for identifying DBPs, the atomic emission detector (AED), with sensitivity similar to that of an ECD, ha been used to provide elemental information with chromatographically resolved components (14). Thus, the AED could iddntify polar moieties in DBP extracts; however, ,ts use has not been reported in recent DBP research. On the other hand, multispectral methods have been employed to identify bromohydrins generated through ozonation of waters containing elevated levels of bromide The approach involved a combination of GC/MS witii electron impact (ED and chemical ionization (C\) and FT-IR spectro* coov (15) This same suite of SDectrometric methods was used to studv the dioxide disin fection with Ti0 in the presence of IIV liu-ht (If!) Other novel approaches to DBP analyses include the use of HPLC coupled with electrospray MS, membrane concentration capillary electrophoresis/MS, and membrane introduction MS for continuous in situ water analysis. The potential of some of these methods for targeting a wide variety of DBPs follows. A summary of successful approaches used for DBP analysis is presented in Table 2. Solid-phase extraction (SPE)/ HPLC. HPLC has the potential lo analyze
Analytical Chemistry News & Features, December 1, 1999 8 0 3 A
Report
Table 2. Analytical approaches for the recovery of organic DBPs. Extraction method
Separation
Detector
Detection limit (ug/L)
Analytes identified to date
LLE
GC
ECD
0.1
Headspace PFBHA, LLE
GC GC
ECD, MS ECD, MS (El and CI)
0.1 1
PFBHA, SPE
GC
ECD, MS
0.1
LLE, diazomethane PFBHA, LLE, diazomethane PFBHA, LLE, MTBSTFA DNPH.SPE
GC GC GC HPLC
ECD ECD, MS ECD, MS ESI, DAD
1 1 1 1
Peroxidase/catalase None LLE, BF3/methanol
Fluorescence Flow injection Ion chromatography Conductivity GC ECD, MS
THMs, HANs, VOCs, HKs, chloral hydrate, bromonitromethanes CNCI, CNBr Ci—C10 aliphatic aldehydes, benzaldehyde, glyoxal, methyl glyoxal, dimethyl glyoxal Glyoxal, methyl glyoxal, dimethyl glyoxal, C1-C10 aliphatic aldehydes 9-Haloacetic acids Pyruvic-, glyoxylic-, and ketomalonic acids Hydroxyacetone Same compounds as for PFBHA, SPE plus 2butanone, 2-pentanone, acetone, 5-ketohexanal, 2-ketobutanedioic acid, 6-hydroxy-2-hexanone Organic peroxides Formic, acetic, oxalic acids MX, EMX, BMX1-3, mucochloric acid
1 20 0.05
LLE = liquid-liquid extraction; PFBHA = pentafluorobenzylhydroxylamine; DNPH = 2,4-dinitrophenylhydrazine; ECD = electron capture detector; El = electron impact; CI = chemical ionization; DAD = diode-array detector; ESI • electrospray ionizatton; THMs = trihalomethanes; VOCC = volatile organic compounds; HANs = haloacetonitriies; HKs = haloketones; MTBSTFA = N-f-butyldimethylsilyl-W-methyltrifluoroacetamide; MX = 3-chloro-4-(dichloromethyl)-5-hydroxy-2(5H)-furanone; EMX = (£)-2chloro-3-(dichloromethyl)-4-oxo-butenoic acid; BMX1 = 3-chloro-4-(bromochloromethyl)-5-hydroxy-2(5H)-furanone; BMX2 = 3-chloro-4-(dibromomethyl)-5-hydroxy-2(5H)-furanone; and BMX3 = 3-bromo-4-(dbromomethyl)-5-hydroxy-2(5/-/)-furanone.
water directly without isolating analytes. To preconcentrate the targeted components of the aquatic matrix to levels amenable to detection, the instrument can be modified so that the sample loop on the injection valve is replaced by a miniaturized concentration column (e.g., 3-mm i.d. x 20 mm), which is filled wiih a suiiable sorbent. Porous polymers, for example, have been applied to the analysis of various groups of polar pesticides and some of their major metabolites (17). With UV detection and a sample volume of 5-10 mL sensitivity is as low as 0 03 ug/L and recoveries are >90% with a coefficient of variation of
