216 Disinfectants Part 2

2.
CHEMISTRY OF DISINFECTANTS AND

DISINFECTANT BY-PRODUCTS
2.1 Background
The use of chlorine (Cl2) as a water disinfectant has come under

scrutiny because of its potential to react with natural organic matter
(NOM) and form chlorinated disinfectant by-products (DBPs). Within
this context, NOM serves as the organic DBP precursor, whereas bro-
mide ion (Br–) serves as the inorganic precursor. Treatment strategies
generally available to water systems exceeding drinking-water standards
include removing DBP precursors and using alternative disinfectants
for primary and/or secondary (distribution system) disinfection.
Alternative disinfectant options that show promise are chloramines
(NH2Cl, monochloramine), chlorine dioxide (ClO2) and ozone (O3).
While ozone can serve as a primary disinfectant only and chloramines
as a secondary disinfectant only, both chlorine and chlorine dioxide can
serve as either primary or secondary disinfectants.
Chloramine presents the significant advantage of virtually elimi-

nating the formation of chlorination by-products and, unlike chlorine,
does not react with phenols to create taste- and odour-causing com-
pounds. However, the required contact time for inactivation of viruses
and Giardia cysts is rarely obtainable by chloramine post-disinfection
at existing water treatment facilities (monochloramine is significantly
less biocidal than free chlorine). More recently, the presence of
nitrifying bacteria and nitrite (NO2–) and nitrate (NO3–) production in
chloraminated distribution systems as well as the formation of organic
chloramines have raised concern.
The use of chlorine dioxide, like chloramine, can reduce the

formation of chlorinated by-products during primary disinfection.
However, production of chlorine dioxide, its decomposition and
reaction with NOM lead to the formation of by-products such as
chlorite (ClO2–), a compound that is of health concern.
If used as a primary disinfectant followed by a chloramine residual

in the distribution system, ozone can eliminate the need for contact
between DBP precursors and chlorine. Ozone is known to react both
27
EHC 216: Disinfectants and Disinfectant By-products
with NOM to produce organic DBPs such as aldehydes and increase

levels of assimilable organic carbon and with bromide ion to form
bromate.
A thorough understanding of the mechanisms of DBP formation

allows microbial inactivation goals and DBP control goals to be
successfully balanced. This chapter examines a range of issues affecting
DBP formation and control to provide guidance to utilities considering
the use of various disinfecting chemicals to achieve microbial
inactivation with DBP control.
2.2 Physical and chemical properties of common

disinfectants and inorganic disinfectant by-
products
The important physical and chemical properties of commonly used

disinfectants and inorganic DBPs are summarized in Table 1.
2.2.1 Chlorine
Chlorine, a gas under normal pressure and temperature, can be

compressed to a liquid and stored in cylindrical containers. Because
chlorine gas is poisonous, it is dissolved in water under vacuum, and
this concentrated solution is applied to the water being treated. For
small plants, cylinders of about 70 kg are used; for medium to large
plants, tonne containers are common; and for very large plants, chlorine
is delivered by railway tank cars or road (truck) tankers. Chlorine is also
available in granular or powdered form as calcium hypochlorite
(Ca(OCl)2) or in liquid form as sodium hypochlorite (NaOCl; bleach).
Chlorine is used in the form of gaseous chlorine or hypochlorite

(OCl–). In either form, it acts as a potent oxidizing agent and often
dissipates in side reactions so rapidly that little disinfection is
accomplished until amounts in excess of the chlorine demand have been
added. As an oxidizing agent, chlorine reacts with a wide variety of
compounds, in particular those that are considered reducing agents
(hydrogen sulfide [H2S], manganese(II), iron(II), sulfite [SO 32–], Br–,
28
Table 1. Physical and chemical properties of commonly used disinfectants and inorganic disinfectant by-products
Chemicala Eo (V)b Oxidation number of Cl or Br 8max (nm)c e (mol–1 litre–1 cm–1)d p, o e pKa f
–
HOCl/Cl +1.49 +1 254 60 +25.2 7.5
292 (OCl–) 419
ClO2/ClO2– +0.95 +4 359 1250 +16.1 –
NH2Cl – +1 245 416 – –
O3/O2 +2.07 – 254 3200 +35.0
HOBr/Br– +1.33 +1 330 50 +22.5 8.7
ClO2–/Cl– +0.76 +3 262 – +12.8 1.96
ClO3–/Cl– +0.62 +5 360 – +10.5 1.45
– –
BrO3 /Br +0.61 +5 195 – 0.72
a
Half-cell reactants/products.
b
Eo = standard electrode potential (redox potential) in water at 25 °C. The oxidation–reduction state of an aqueous environment at equilibrium can be stated
in terms of its redox potential. In the chemistry literature, this is generally expressed in volts, E, or as the negative logarithm of the electron activity, p,. When
p, is large, the electron activity is low and the system tends to be an oxidizing one: i.e., half-reactions tend to be driven to the left. When p, is small, the
system is reducing, and reactions tend to be driven to the right.
c
8max = maximum absorbance wavelength of that particular solution in nm.
d
e = molar absorptivity (molar extinction coefficient), in mol–1 litre–1 cm–1. This can be used for quantitative determination of the various species of chemicals
and is the only direct physical measurement. There is often some background absorbance that may interfere with the measurement in natural waters that
should be considered.
e
p,o = ! log {e–} where {e–} = electron activity.
f
pKa = negative logarithm of the acid ionization constant (e.g., at pH 7.5, the molar concentration of HOCl is same as that of OCl–). As this parameter is
dependent upon temperature, the values listed were determined at 25 °C.
iodide [I–], nitrite). From the point of view of DBP formation and
disinfection, these reactions may be important because they may be fast
and result in the consumption of chlorine.
Chlorine gas hydrolyses in water almost completely to form

hypochlorous acid (HOCl):
Cl2 + H2O 6 HOCl + H+ + Cl–
The hypochlorous acid dissociates into hydrogen ions (H+) and

hypochlorite ions in the reversible reaction:
HOCl : H+ + OCl–
Hypochlorous acid is a weak acid with a pKa of approximately 7.5

at 25 °C. Hypochlorous acid, the prime disinfecting agent, is therefore
dominant at a pH below 7.5 and is a more effective disinfectant than
hypochlorite ion, which dominates above pH 7.5.
The rates of the decomposition reactions of chlorine increase as

the solution becomes more alkaline, and these reactions can theoreti-
cally produce chlorite and chlorate (ClO3–); they occur during the
electrolysis of chloride (Cl–) solutions when the anodic and cathodic
compartments are not separated, in which case the chlorine formed at
the anode can react with the alkali formed at the cathode. On the other
hand, hypochlorous acid/hypochlorite (or hypobromous acid/hypo-
bromite, HOBr/OBr–) can be formed by the action of chlorine (or
bromine) in neutral or alkaline solutions. The decomposition of hypo-
halites (XO–) is favoured in alkaline solutions (2XO– 6 X– + XO2–) and
is such that there is no longer any domain of thermodynamic stability
for the hypohalite ions. These oxyhalites are further converted to stable
oxyhalates as follows:
XO– + XO2– 6 X– + XO3–
Another reaction that occurs in waters containing bromide ion and

hypochlorite is the production of hypobromous acid:
HOCl + Br– 6 HOBr + Cl–
This reaction is irreversible, and the product hypobromous acid is a

better halogenating agent than hypochlorous acid and interferes with
30
Chemistry of Disinfectants and Disinfectant By-products
common analytical procedures for free chlorine. The presence of

bromide in hypochlorite solutions can ultimately lead to the formation
of bromate (BrO3–).
Hypobromous acid is a weak acid (pKa = 8.7); like hypochlorite,

hypobromite is metastable. In alkaline solution, it decomposes to give
bromate and bromide:
3OBr– 6 BrO3– + 2Br–
Bromic acid (HBrO3) is a strong acid (pKa = 0.7). Bromic acid and
bromate can be obtained by the electrolytic oxidation of bromide
solutions or bromine water using chlorine. Bromic acid and bromate
are powerful oxidizing agents, but the speed of their oxidation reactions
is generally slow (Mel et al., 1953).
2.2.2 Chlorine dioxide
Chlorine dioxide is one of the few compounds that exists almost

entirely as monomeric free radicals. Concentrated chlorine dioxide
vapour is potentially explosive, and attempts to compress and store this
gas, either alone or in combination with other gases, have been
commercially unsuccessful. Because of this, chlorine dioxide, like
ozone, must be manufactured at the point of use. Chlorine dioxide in
water does not hydrolyse to any appreciable extent. Neutral or acidic
dilute aqueous solutions are quite stable if kept cool, well sealed and
protected from sunlight.
Chlorine dioxide represents an oxidation state (+4) intermediate

between those of chlorite (+3) and chlorate (+5). No acid or ion of the
same degree of oxidation is known. Chlorine dioxide is a powerful
oxidizing agent that can decompose to chlorite; in the absence of
oxidizable substances and in the presence of alkali, it dissolves in water,
decomposing with the slow formation of chlorite and chlorate:
2ClO2 + H2O 6 ClO2– + ClO3– + 2H+
Chlorine dioxide has an absorption spectrum with a maximum at

359 nm, with a molar absorptivity of 1250 mol–1 litre–1 cm–1. This
extinction coefficient is independent of temperature, pH, chloride and
ionic strength. Chlorine dioxide is readily soluble in water, forming a
31
greenish-yellow solution. It can be involved in a variety of redox

reactions, such as oxidation of iodide ion, sulfide ion, iron(II) and
manganese(II). When chlorine dioxide reacts with aqueous contami-
nants, it is usually reduced to chlorite ion. The corresponding electron
transfer reactions are comparable to those occurring when singlet
oxygen acts as an oxidant (Tratnyek & Hoigne, 1994).
Bromide (in the absence of sunlight) is not oxidized by chlorine

dioxide. Therefore, water treatment with chlorine dioxide will not
transform bromide ion into hypobromite and will not give rise to the
formation of bromoform (CHBr3) or bromate. This is an important
difference between the use of chlorine dioxide as an oxidant and the use
of chlorine or ozone as an oxidant.
2.2.3 Ozone
Ozone is a strong oxidizing agent (Eo = 2.07 V). Oxidation

reactions initiated by ozone in water are generally rather complex; in
water, only part of the ozone reacts directly with dissolved solutes.
Another part may decompose before reaction. Such decomposition is
catalysed by hydroxide ions (OH–) and other solutes. Highly reactive
secondary oxidants, such as hydroxyl radicals (OH@), are thereby
formed. These radicals and their reaction products can additionally
accelerate the decomposition of ozone. Consequently, radical-type
chain reactions may occur, which consume ozone concurrently with the
direct reaction of ozone with dissolved organic material.
Many oxidative applications of ozone have been developed,

including disinfection, control of algae, removal of tastes and odours,
removal of colour, removal of iron and manganese, microflocculation,
removal of turbidity by oxidative flocculation, removal of organics by
oxidation of phenols, detergents and some pesticides, partial oxidation
of dissolved organics and control of halogenated organic compounds.
For disinfection and for oxidation of many organic and inorganic
contaminants in drinking-water, the kinetics of ozone reactions are
favourable; on the other hand, for many difficult-to-oxidize organic
compounds, such as chloroform (CHCl3), the kinetics of ozone
oxidation are very slow (Hoigne et al., 1985).
32
2.2.4 Chloramines
Monochloramine has much higher CT values1 than free chlorine

and is therefore a poor primary disinfectant. Additionally, it is a poor
oxidant and is not effective for taste and odour control or for oxidation
of iron and manganese. However, because of its persistence, it is an
attractive secondary disinfectant for the maintenance of a stable distri-
bution system residual. The use of disinfectants such as ozone or
chlorine dioxide combined with chloramines as a secondary disinfectant
appears to be attractive for minimizing DBP formation (Singer, 1994b).
Monochloramine is the only useful ammonia-chloramine disin-

fectant. Dichloramine (NHCl2) and nitrogen trichloride (NCl3) are too
unstable to be useful and highly malodorous. Conditions practically
employed for chloramination are designed to produce only mono-
chloramine.
2.3 Analytical methods for disinfectant by-products

and disinfectants
Analytical methods for various DBPs and their detection limits are
summarized in Table 2. Methods for disinfectants are summarized in
APHA (1995).
2.3.1 Trihalomethanes, haloacetonitriles, chloral hydrate,

chloropicrin and haloacetic acids
Gas chromatographic (GC) techniques are generally employed

for organic DBPs. Detection and quantification of haloacetonitriles
(HANs) and chloral hydrate in chlorinated natural waters are compli-
cated by (i) hydrolysis of dihaloacetonitriles and chloral hydrate to
dihaloacetic acids and chloroform, respectively; (ii) degradation of
HANs by dechlorinating agents such as sodium sulfite and sodium
thiosulfate; (iii) low purge efficiency for the HANs and chloral hydrate
1
The CT value is the product of the disinfectant concentration C in mg/litre and
the contact time T in minutes required to inactivate a specified percentage (e.g.,
99%) of microorganisms.
33
Table 2. Summary of analytical methods for various DBPs and their minimum detection limits
DBPs Analytical method APHA a Minimum detection Major References

method limit (:g/litre) interferences
THMs MTBE extraction – 0.4 None AWWARF (1991)
Pentane extraction 6232B 0.1
HAAs Salted MTBE extraction and derivatization 6233B 0.5–1.0 None AWWARF (1991)
with diazomethane
HANs Pentane extraction 6232B 0.05 None Koch et al. (1988)
Cyanogen chloride MTBE extraction 6233A 0.5 None AWWARF (1991)
Chloramine Derivatization with 2-mercaptobenzothiazole – – None Lukasewycz et al. (1989)
Haloketonesb Pentane extraction 6232B 0.2 None Krasner et al. (1995)
AWWARF (1991)
Chloral hydrate MTBE extraction – 0.5 None AWWARF (1991)
Aldehydes Extraction with hexane and derivatization with – 1.0 PFBHA sulfate Sclimenti et al. (1990)
PFBHA
Bromate Ion chromatography (H3BO3/NaOH) 4500 2.0 c Cl– Siddiqui et al. (1996a)
Krasner et al. (1993)
0.2 Weinberg et al. (1998)
Chlorate Ion chromatography (H3BO3/NaOH) 4500 5 Cl–, acetate Siddiqui (1996)
Chlorite Ion chromatography (NaHCO3/Na2CO3) 4500 10 Cl–, acetate AWWARF (1991)
TOC UV/persulfate or combustion 5310 200 Metals APHA (1995)
a
American Public Health Association.
b
Sum of 1,1-DCPN and 1,1,1-TCPN.
c
1.0 :g/litre with high-capacity column.
in the purge-and-trap technique; and (iv) low extraction efficiency for

chloral hydrate with pentane in the liquid–liquid extraction normally
used. Although chloral hydrate is not efficiently extracted from water
with pentane, it can be extracted with an efficiency of approximately
36% when the ratio by volume of methyl tert-butyl ether (MTBE) to
water is 1 : 5 (Amy et al., 1998). MTBE quantitatively extracts HANs,
trihalomethanes (THMs), chloral hydrate and chloropicrin, permitting
simultaneous analysis for all of these DBPs. Chloral hydrate decom-
poses on packed columns to trichloroacetaldehyde, resulting in con-
siderable band broadening, although this does not appear to be a
significant problem with DB-1 and DB-5 columns.
The extraction of THMs can be accomplished using MTBE (EPA

Method 551) or pentane. Method 551 also permits simultaneous
extraction and measurement of chloral hydrate, HANs, THMs,
chloropicrin and haloketones. The pentane method can be used to
extract THMs, HANs, haloketones and chloropicrin but not chloral
hydrate in the same run (APHA, 1995).
The haloacetic acid (HAA) analytical method involves using an

acidic salted ether or acidic methanol liquid–liquid extraction, requiring
esterification with diazomethane prior to analysis on a gas chromato-
graph equipped with an electron capture detector (ECD). THMs and
HANs can be analysed by extraction with pentane prior to analysis on
a capillary-column GC equipped with an ECD. The analysis of cyano-
gen compounds involves extraction with MTBE prior to injection into
GC–ECD. Aldehydes require derivatization with O-(2,3,4,5,6-penta-
fluorobenzyl)-hydroxylamine (PFBHA) (to form an oxime), extraction
with hexane and GC–ECD analysis [(C6F5)-CH2ONH2 + RCHO 6
(C6F5)–CH2ON=CHR + H2O]. It should be noted that PFBHA peaks are
very large relative to other peaks in the chromatogram from a purge-
and-trap system, whereas the peaks are comparable to other peaks in a
GC–ECD chromatogram (Trehy et al., 1986).
2.3.2 Inorganic disinfectant by-products
An ion chromatography (IC) method (EPA Method 300) has been

developed to determine inorganic by-products. The elution order is
fluoride, chlorite, bromate, chloride, nitrite, bromide, chlorate, nitrate
and sulfate ion. The eluent is a carbonate buffer. Ethylenediamine is
used to preserve chlorite samples and to minimize the potential for
chlorite ion reaction on the IC separating column. EPA Method 300
35
involves measurement by an IC system using a separating column (e.g.,

Ion Pac AS9-SC) fitted with an anion micromembrane suppressor
column. An eluent containing 2.0 mmol of sodium carbonate (Na2CO3)
per litre / 0.75 mmol of sodium bicarbonate (NaHCO3) per litre is used
for bromide determination, and an eluent containing 40 mmol of boric
acid (H3BO3) per litre / 20 mmol of sodium hydroxide (NaOH) per litre
is used for bromate and chlorate determination. The analytical mini-
mum detection limits for bromate and chlorate using a borate eluent
have been reported as 2 :g/litre and 5 :g/litre, respectively (Siddiqui,
1996; Siddiqui et al., 1996a). For samples with high chloride ion
content, a silver cartridge can be used to remove chloride prior to IC
analysis to minimize its interference with bromate measurement. It
should be noted that for natural sources and waters with high total
organic carbon (TOC) levels, detection limits will be slightly different
because of the masking effect of NOM and high concentrations of
carbonate/bicarbonate ions that may interfere with bromate/chlorate
measurement.
2.3.3 Total organic carbon and UV absorbance at 254 nm
TOC is the primary surrogate parameter for the measurement of

NOM in water supplies. Several investigators have reported that the
ultraviolet (UV)/persulfate oxidation method underestimates the TOC
concentration in natural waters as compared with the combustion
method because of the inability of the persulfate method to oxidize
highly polymerized organic matter. It is generally assumed that the
calibration of a TOC analyser with a potassium hydrogen phthalate
standard is sufficient for the measurement of TOC in natural waters,
but potassium hydrogen phthalate has a simple molecular structure and
is easy to oxidize. Dissolved organic carbon (DOC) is operationally
defined by a (0.45-:m) filtration step. UV absorbance at 254 nm
(UVA254) is used to describe the type and character of NOM, whereas
TOC describes just the amount of NOM.
2.3.4 Chloramines
Knowledge of the amine content of the water during water treat-

ment processes involving chloramination is important to define more
adequately the content of a matrix described only as a combined
chlorine residual. The presence of organic nitrogen and the instability
36
of many organic chloramines continue to challenge the analyst.

Lukasewycz et al. (1989) developed a technique for the analysis of
chloramines and organic chloramines present in water using 2-mercap-
tobenzothiazole as a derivatizing agent. The resulting sulfanilamides are
stable and can be conveniently analysed by high-performance liquid
chromatography (HPLC) using UV or electrochemical detection. This
method appears to be superior to the use of diazotization or phenyl-
arsine oxide as a method of detection. Organic chloramines are much
weaker disinfectants than inorganic monochloramine but are indistin-
guishable by the common analytical methods.
2.4 Mechanisms involved in the formation of disinfectant

by-products
2.4.1 Chlorine reactions
Chlorine reacts with humic substances (dissolved organic matter)

present in most water supplies, forming a variety of halogenated DBPs,
such as THMs, HAAs, HANs, chloral hydrate and chloropicrin, as
follows:
HOCl + DOC 6 DBPs
It is generally accepted that the reaction between chlorine and

humic substances, a major component of NOM, is responsible for the
production of organochlorine compounds during drinking-water treat-
ment. Humic and fulvic acids show a high reactivity towards chlorine
and constitute 50–90% of the total DOC in river and lake waters
(Thurman, 1985). Other fractions of the DOC comprise the hydrophilic
acids (up to 30%), carbohydrates (10%), simple carboxylic acids (5%)
and proteins/amino acids (5%). The reactivity of carbohydrates and
carboxylic acids towards chlorine is low, and they are not expected to
contribute to the production of organochlorine compounds. However,
hydrophilic acids such as citric acid and amino acids will react with
chlorine to produce chloroform and other products and may contribute
towards total organochlorine production (Larson & Rockwell, 1979).
Free chlorine reacts with water constituents by three general

pathways: oxidation, addition and substitution (Johnson & Jensen,
1986). Chlorine can undergo an addition reaction if the organic
compound has a double bond. For many compounds with double bonds,
37
this reaction is too slow to be of importance in water treatment. The

oxidation reactions with water constituents such as carbohydrates or
fatty acids (e.g., oleic acid) are generally slow.
Most chlorine DBPs are formed through oxidation and substitu-

tion reactions. THMs have the general formula CHX3, where X can be
Cl or Br. Chloroform may be produced through a series of reactions
with functional groups of humic substances. The major functional
groups of humic substances include acetyl, carboxyl, phenol, alcohol,
carbonyl and methoxyl. The reactions proceed much more rapidly at
high pH than at low pH.
Rook (1977) proposed resorcinol structures to be the major

precursor structure in humic material for chloroform formation. In
accordance with this hypothesis in the chlorination of terrestrial and
aquatic humic substances, a series of intermediates were detected that
contained a trichloromethyl group and that could be converted to
chloroform by further oxidation or substitution reactions (Stevens et
al., 1976).
However, the production of chlorinated compounds such as

dichloropropanedoic acid, 2,2-dichlorobutanedoic acid, cyanogen
chloride (CNCl), HANs or the cyano-substituted acids cannot be
explained on the basis of resorcinol structures, and possible production
pathways require protein-type precursors (De Leer et al., 1986). The
reaction pathway for amino acids involves initial rapid formation of the
monochloramine and dichloramine, which can react further to form
aldehyde or HANs, respectively. Trehy et al. (1986) demonstrated the
formation of chloral hydrate along with HANs after chlorination of
amino acids by substitution reactions, and aldehydes were shown to be
the oxidation products. Luknitskii (1975) provided a detailed chemistry
of chloral hydrate formation.
Christman et al. (1983) also identified chloroform, chloral

hydrate, dichloroacetic acid (DCA), trichloroacetic acid (TCA) and 2,2-
dichlorobutanedoic acid as the major products, accounting for 53% of
the total organic halogen (TOX). A number of other minor products
have been detected, including several chlorinated alkanoic acids and
non-chlorinated benzene carboxylic acids. De Leer et al. (1985)
extended these studies to incorporate chloroform intermediates,
chlorinated aromatic acids and cyano-compounds as potential products
38
in drinking-water. The presence of unhalogenated aldehydes and HANs

in chlorinated natural waters can be attributed in part to the presence of
amino acids or peptides in natural waters. Humic acids may also
contribute to the presence of amino acids in natural waters, as they have
amino acids associated with them either in a free or in a combined
form. Several studies regarding the chlorination of amino acids have
shown that the primary amino group on the amino acids can be
converted to either an aldehyde or a nitrile group (Morris et al., 1980;
Isaac & Morris, 1983). These studies indicate that with an equimolar
amount of halogenating agent, the major product is an aldehyde.
However, if an excess of halogenating agent is added, then the
corresponding nitrile can also be formed, with the ratio of the aldehyde
to nitrile formed increasing with pH.
Many treated waters contain not only chlorinated but also

brominated compounds, such as bromoform. These compounds form
because aqueous chlorine converts bromide in the water to hypo-
bromous acid. The bromine can then react with the organic matter in the
same way as hypochlorous acid to form various bromochlorinated
DBPs. However, compared with hypochlorous acid, hypobromous acid
is a weaker oxidant and stronger halogenating agent.
Chlorate, an inorganic by-product of chlorine, is formed in

concentrated hypochlorite solutions during their production and storage
through the following reactions (Gordon et al., 1997):
OCl– + OCl– 6 ClO2– + Cl–

OCl– + ClO2– 6 ClO3– + Cl–
The first reaction proceeds at a much slower rate and is rate

limiting, hence the generally observed second-order kinetics. Sodium
hypochlorite is stored at pH greater than 12 to prevent rapid decom-
position, and most of the sodium hypochlorite is present as hypo-
chlorite ion. The average rate constant for the formation of chlorate is
85 × 10–5 mol–1 litre–1 d–1 (Gordon et al., 1995).
2.4.2 Chlorine dioxide reactions
The major chlorine dioxide by-products of concern are chlorite and

chlorate. Chlorine dioxide reacts generally as an electron acceptor, and
hydrogen atoms present in activated organic C–H or N–H structures are
39
thereby not substituted by chlorine (Hoigne & Bader, 1994). Moreover,

in contrast to chlorine, chlorine dioxide’s efficiency for disinfection
does not vary with pH or in the presence of ammonia, and it does not
oxidize bromide. As opposed to chlorine, which reacts via oxidation
and electrophilic substitution, chlorine dioxide reacts only by oxidation;
this explains why it does not produce organochlorine compounds. In
addition to this, chlorine dioxide is more selective in typical water
treatment applications, as evidenced by its somewhat lower disinfectant
demand as compared with chlorine.
Chlorine dioxide is generally produced by reacting aqueous

(sodium) chlorite with chlorine (Gordon & Rosenblatt, 1996):
2ClO2– + HOCl + H+ 6 2ClO2(aq) + Cl– + H2O
However, under conditions of low initial reactant concentrations

or in the presence of excess chlorine, the reactant produces chlorate ion:
ClO2– + HOCl 6 ClO3– + Cl– + H+
This reaction scenario is common in generators that overchlorinate

to achieve high reaction yields based on chlorite ion consumption.
An alternative approach to chlorine dioxide generation is with

hydrochloric acid (HCl), a process that results in less chlorate during
production:
5NaClO2 + 4HCl 6 4ClO2 + 5NaCl + 2H2O
Chlorite ion is also produced when chlorine dioxide reacts with

organics (Gordon & Rosenblatt, 1996):
ClO2 + NOM 6 Products + ClO2–
Chlorine dioxide can also undergo a series of photochemically

initiated reactions resulting in the formation of chlorate ion (Gordon et
al., 1995).
While bromide is not generally oxidized by chlorine dioxide,

bromate can be formed in the presence of sunlight over a wide range of
40
pH values (Gordon & Emmert, 1996). Utilities need to be concerned

with bromate ion in the chlorine dioxide treatment of drinking-water if
the water contains bromide and is exposed to sunlight. Practically, this
means minimizing exposure to sunlight when chlorine dioxide is
applied in the presence of bromide ion. There appears to be a problem
with chlorine dioxide producing odour-causing compounds at the tap.
This has been linked to chlorine dioxide reacting with volatile organic
compounds derived from new carpets and office products (Hoehn et al.,
1990).
Hoigne & Bader (1994) described the kinetics of reaction between

chlorine dioxide and a wide range of organic and inorganic compounds
that are of concern in water treatment. Measured rate constants were
high for nitrite, hydrogen peroxide, ozone, iodide, iron(II), phenolic
compounds, tertiary amines and thiols. Bromide, ammonia, structures
containing olefinic double bonds, aromatic hydrocarbons, primary and
secondary amines, aldehydes, ketones and carbohydrates are unreactive
under the conditions of water treatment. Chlorine dioxide rapidly
oxidizes substituted phenoxide anions and many phenols, and second-
order rate constants have been measured (Rav-Acha & Choshen, 1987).
2.4.3 Chloramine reactions
Chloramination of drinking-water produces THMs (if chloramine

is formed by chlorination followed by ammonia addition), HAAs,
chloral hydrate, hydrazine, cyanogen compounds, nitrate, nitrite, organic
chloramines and 1,1-dichloropropanone (1,1-DCPN) (Dlyamandoglu
& Selleck, 1992; Kirmeyer et al., 1993, 1995).
In the presence of even small quantities of organic nitrogen, it is

possible for chloramination to produce organic chloramines. Several
researchers have shown that monochloramine readily transfers its
chlorine at a comparatively rapid rate to organic amines to form
organohalogen amines (Isaac & Morris, 1983; Bercz & Bawa, 1986).
Monochloramine was shown to cause binding of radiolabelled halogen
to nutrients such as tyrosine and folic acid; the amount of binding
varied with pH but was generally less at neutral pH than at higher pH
(Bercz & Bawa, 1986). Organic chloramines are much weaker
disinfectants than inorganic monochloramine but are indistinguishable
by common analytical methods. Organic chloramine formation may
41
necessitate changing chloramination conditions (e.g., ammonia and

chlorine addition order, chlorine-to-ammonia ratios and contact time).
HANs and non-halogenated acetonitriles are produced when

chloramines are reacted with humic materials and amino acids (Trehy
et al., 1986). The reaction pathway for these products is quite
complicated and very similar to that for chlorine, with many
intermediates and by-products formed. In the case of aspartic acid, De
Leer et al. (1986) demonstrated the presence of at least 11 other
significant products.
2.4.4 Ozone reactions
Ozone has been shown to oxidize bromide to hypobromite and

bromate, and hypochlorite to chlorate (Glaze et al., 1993; Siddiqui et
al., 1995; Siddiqui, 1996).
Bromate generally forms through a combination of molecular

ozone attack and reaction of bromide with free radical species. The
molecular ozone mechanism does not account for hydroxyl radicals
always formed as secondary oxidants from decomposed ozone during
water treatment. Siddiqui et al. (1995) indicated that there is a radical
pathway that is influenced by both pH and alkalinity. The hydroxyl
radical and, to a lesser degree, the carbonate radical (CO3@) pathway may
be more important than the molecular ozone pathway. Oxidants such as
hydroxyl and carbonate radicals may interact with intermediate bromine
species, leading to the formation of hypobromite radicals (BrO@), which
eventually undergo disproportionation to form hypobromite and
bromite (BrO2–). Bromate is then formed through oxidation of bromite
by ozone. The radical mechanism for the formation of bromate includes
two decisive reaction steps still involving molecular ozone: the
formation of hypobromite and oxidation of bromite.
Bromate ion formed through reactions with molecular ozone

contributes in the range of 30–80% to the overall bromate ion
formation in NOM-containing waters (von Gunten and Hoigne, 1994).
Siddiqui et al. (1995) reported up to 65% and 100% bromate ion
formation through the radical pathway in NOM-free and NOM-
containing waters, respectively. Differences in NOM-containing waters
can be attributed to differences in the characteristics of the NOM
present. A change in mechanism as a function of pH and the competitive
42
roles of the free radical (one electron transfer) mechanism above pH 7

versus oxygen atom (two electron transfer) mechanism help explain
both the large variations in bromate ion yield and the sensitivity to
reactor design, concentration of organic precursors and ozone/bromide
ion concentrations (Gordon, 1993).
The presence of bromide ion in a source water further complicates

the reaction of ozone and leads to the formation of additional DBPs,
such as bromoform, dibromoacetonitrile (DBAN) and dibromoacetone
(DBAC) (Siddiqui, 1992; Amy et al., 1993, 1994).
2.5 Formation of organohalogen disinfectant by-products
Table 3 summarizes the DBPs identified as being formed from the

use of chlorine, chlorine dioxide, chloramine and ozone.
The formation of organochlorine and organobromine compounds

during drinking-water treatment is a cause of health concern in many
countries. These compounds include THMs, HAAs, HANs, chloral
hydrate, chloropicrin, acetohalides, halogenated furanones and other
compounds.
2.5.1 Chlorine organohalogen by-products
Table 4 summarizes the range of concentrations of chlorinated

DBPs formed from the reaction of chlorine with NOM, from various
sources.
The major chlorination DBPs identified are THMs, HAAs, HANs,

haloketones, chloropicrin and chloral hydrate. HAAs represent a major
portion of the non-THM halogenated organic compounds (Miller &
Uden, 1983; Reckhow & Singer, 1985). Many researchers have identi-
fied HANs and haloketones as other important DBPs (Trehy & Bieber,
1981; Miller & Uden, 1983; Oliver, 1983; Reckhow & Singer, 1985).
According to an AWWARF (1991) study, for all eight utilities tested,
1,1,1-trichloropropanone (1,1,1-TCPN) was the more prevalent of the
two measured haloketone compounds. In addition, Kronberg et al.
(1988) identified the extremely mutagenic compound, MX.
43
Table 3. Disinfectant by-products present in disinfected waters
Disinfectant Significant organo- Significant Significant non-

halogen products inorganic halogenated
products products
Chlorine/ THMs, HAAs, Chlorate (mostly Aldehydes,

hypochlorous HANs, chloral from hypo- cyanoalkanoic
acid hydrate, chlorite use) acids, alkanoic
chloropicrin, acids, benzene,
chlorophenols, carboxylic acids
N-chloramines,
halofuranones,
bromohydrins
Chlorine chlorite, chlorate unknown

dioxide
Chloramine HANs, cyanogen nitrate, nitrite, aldehydes,

chloride, organic chlorate, ketones
chloramines, hydrazine
chloramino acids,
chloral hydrate,
haloketones
Ozone bromoform, MBA, chlorate, iodate, aldehydes,

DBA, DBAC, bromate, ketoacids,
cyanogen bromide hydrogen ketones,
peroxide, carboxylic acids
hypobromous
acid, epoxides,
ozonates
Despite the fact that HAA formation and THM formation have
very different pH dependencies, HAA formation correlates strongly
with THM formation when treatment conditions are relatively uniform
and when the water has a low bromide concentration (Singer, 1993).
DBP formation and requisite chlorine dosage for disinfection strongly
correspond to the concentration of TOC at the point of chlorine
addition, suggesting that optimized or enhanced removal of organic
carbon prior to chlorination will decrease the formation of DBPs.
HAA formation can be appreciable when drinking-water is

chlorinated under conditions of slightly acidic pH and low bromide
concentrations. The concentrations of DCA and TCA are similar to the
concentrations of chloroform, and the total HAA concentration can be
as much as 50% greater than the THM concentration in the finished
water on a weight basis.
44
Table 4. Concentration range of chlorinated disinfectant by-products in drinking-watera
DBPs Peters et al. (1990); Krasner et al. (1989) Nieminski et al. (1993) Koch et al. (1991) Reckhow et al. (1990)
Peters (1991)
THMs 3.1–49.5 30.0–44.0 17.0–51.0 49.0–81.0 201–1280
HAAs <0.5–14.7 13.0–21.0 5.0–25.0 22.0–32.0 118–1230
HANs 0.04–1.05 2.5–4.0 0.5–5.0 2.0–2.6 3.0–12.0
Haloketones – 0.9–1.8 0.2–1.6 1.0–2.0 4.8–25.3
Chlorophenols – – 0.5–1.0 – –
Chloral hydrate – 1.7–3.0 – – –
Chloropicrin – 0.1–0.16 <0.1–0.6 – –
TOC 1.7–5.6 2.9–3.2 1.5–6.0 2.5–3.0 4.8–26.6
Bromide 100–500 70–100 – 170–420 –
a
All values shown in :g/litre, except TOC (mg/litre).
McGuire & Meadow (1988) reported that the national average

THM concentration in the USA was 42 :g/litre for drinking-water
utilities serving more than 100 000 persons, and only 3% of systems
were above the US maximum contaminant level of 100 :g/litre. Amy
et al. (1993) estimated that the national average THM concentration in
the USA was 40 :g/litre, with an average TOC concentration of
3.0 mg/litre. The median annual average THM concentration found for
utilities among the American Water Works Association’s (AWWA)
Water Industry Database was 35 :g/litre, as compared with 50 :g/litre
for the non-database utilities (Montgomery Watson, Inc., 1993).
In Germany, 10% of the utilities produced disinfected drinking-

water with a THM concentration above 10 :g/litre; the median annual
average concentration was between 1 and 4 :g/litre, depending on raw
water quality and size of facility (Haberer, 1994).
Total THM levels in treated drinking-water were reported in one

survey in the United Kingdom (Water Research Centre, 1980):
chlorinated water derived from a lowland river contained a mean level
of 89.2 :g/litre, and that from an upland reservoir, 18.7 :g/litre. The
study also showed that chlorinated groundwater was contaminated by
THMs to a significantly lesser extent than chlorinated surface waters.
In a national survey of the water supplies of 70 communities

serving about 38% of the population in Canada, conducted in the
winter of 1976–1977, chloroform concentrations in treated water of
the distribution system 0.8 km from the treatment plant, determined by
the gas sparge technique, averaged 22.7 :g/litre. Levels of the other
THMs were considerably lower, averaging 2.9 :g/litre for bromo-
dichloromethane (BDCM), 0.4 :g/litre for dibromochloromethane
(DBCM) and 0.1 :g/litre for bromoform. Using direct aqueous
injection techniques, average concentrations of most of the THMs were
higher (Health Canada, 1993).
Samples collected from the distribution systems of eight major

cities in Saudi Arabia showed that THMs occurred in all the water
supplies, at concentrations ranging between 0.03 and 41.7 :g/litre.
Median total THM concentrations in several cities were higher during
the summer than during the winter. In addition, THM concentrations
were low in cities that did not mix groundwater and desalinated water.
46
Brominated THMs dominated (with bromoform the most abundant) and

existed at the highest concentration levels, whereas chloroform was the
least prevalent compound. This is the opposite of the occurrence pattern
found in almost all water distribution systems worldwide (Fayad,
1993).
The concentrations of chloral hydrate in drinking-water in the USA

were summarized by IARC (1995) and varied from 0.01 to 28 :g/litre.
The highest values were found in drinking-water prepared from surface
water.
Chlorination of water as well as the combination of ozonation and

chlorination can lead to the formation of chloropicrin (Merlet et al.,
1985). In a study conducted for over 25 utilities, very low levels of
chloropicrin were observed, and chlorination produced maximum
concentrations of less than 2 :g/litre (AWWARF, 1991). The chloro-
picrin appeared to form slowly during the incubation period, with
concentrations tending to level off at approximately 40 h.
Dichloroacetonitrile (DCAN) is by far the most predominant HAN

species detected in water sources with bromide levels of 20 :g/litre or
less. For sources with higher bromide levels (50–80 :g/litre), bromo-
chloroacetonitrile (BCAN) was the second most prevalent compound.
However, none of these sources had a DBAN concentration exceeding
0.5 :g/litre, including one source water that had a much higher bro-
mide level, 170 :g/litre. Thus, it appears that ambient bromide concen-
tration is not the only factor influencing the speciation of HAN
compounds.
Chlorine can react with phenols to produce mono-, di- or

trichlorophenols, which can impart tastes and odours to waters. The
control of chlorophenolic tastes and odours produced when phenol-
laden water is treated with chlorine is essential. The sources of phenolic
compounds in water supplies are reported to be industrial wastes.
In natural waters, one of the most important sources of organic

nitrogen is proteins and their hydrolysis products. The reaction of
aqueous chlorine or monochloramine with organic nitrogen may form
complex organic chloramines (Feng, 1966; Morris et al., 1980; Snyder
47
& Margerum, 1982). The formation of N-chloramines resulting from

the reaction of amines and chlorine has been reported (Weil & Morris,
1949; Gray et al., 1979; Morris et al., 1980). Likewise, the chlorination
of amides has been reported (Morris et al., 1980).
Nieminski et al. (1993) reported the occurrence of DBPs for Utah

(USA) water treatment plants. All plants used chlorine for primary and
secondary disinfection purposes. Overall, THMs and HAAs represented
75% of the total specific DBPs analysed for the survey; however, total
DBPs represented only 25–50% of the TOX concentration. THMs
constituted 64% of the total DBPs by weight; HAAs were 30% of the
total DBPs by weight and approximately one-half of the total THM
concentrations. (However, in some waters, HAA concentrations may
approach or possibly exceed THM concentrations.) HANs, haloketones,
chlorophenols and chloropicrin represented 3%, 1.5%, 1.0% and 0.5%,
respectively, of the total surveyed DBPs.
The occurrence of DBPs in drinking-waters in the USA was

evaluated at 35 water treatment facilities that had a broad range of
source water qualities and treatment processes (Krasner et al., 1989).
THMs were the largest class of DBPs, and HAAs were the next most
significant class. Aldehydes, by-products of ozonation, were also pro-
duced by chlorination. Over four quarterly sampling periods, median
total THM concentrations ranged from 30 to 44 :g/litre, with chloro-
form, BDCM, DBCM and bromoform ranges of 9.6–15, 4.1–10,
2.6–4.5 and 0.33–0.88 :g/litre, respectively. Median total HAA
concentrations ranged from 13 to 21 :g/litre, with TCA, DCA, mono-
chloroacetic acid (MCA), dibromoacetic acid (DBA) and monobromo-
acetic acid (MBA) ranges of 4.0–6.0, 5.0–7.3, <1–1.2, 0.9–1.5 and
<0.5 :g/litre, respectively.
Concentrations of DCA and TCA measured in various water

sources have been summarized by IARC (1995): in Japan, chlorinated
drinking-water contained 4.5 and 7.5 :g of DCA and TCA per litre,
respectively; rainwater in Germany contained 1.35 :g of DCA per litre
and 0.1–20 :g of TCA per litre, whereas groundwater contained
0.05 :g of TCA per litre; in Australia, a maximum concentration of
200 :g/litre was found for DCA and TCA in chlorinated treated water;
and chlorinated water in Switzerland contained 3.0 :g of TCA per litre.
48
In a survey of 20 drinking-waters prepared from different source

waters in the Netherlands, HAAs were found in all drinking-waters
prepared from surface water, whereas they could not be detected in
drinking-waters prepared from groundwaters. Brominated acetic acids
accounted for 65% of the total acid concentration (Peters et al., 1991).
In another survey of Dutch drinking waters, the average concentration
of dihaloacetonitriles was about 5% of the average THM concentration
(Peters, 1990).
2.5.2 Chloramine organohalogen by-products
Chloramine treatment practice involves three potential approaches:

free chlorine followed by ammonia addition, ammonia addition
followed by chlorine addition (in situ production) and pre-formed (off-
line formation) chloramines. Generally, the objective is
monochloramine formation. Chlorine followed by ammonia is a
common approach, and, during the free-chlorine period, DBP formation
may mimic that of chlorine. Chloramination results in the production
of THMs (predominantly formed by chlorination followed by ammonia
addition), HAAs, chloral hydrate, hydrazine, cyanogen compounds,
organic chloramines and 1,1-DCPN (Dlyamandoglu & Selleck, 1992;
Singer, 1993; Kirmeyer et al., 1993, 1995). Chloramination
significantly reduces but does not eliminate THM formation; cyanogen
chloride and TOX represent the major DBP issues with respect to
chloramines.
Scully et al. (1990) identified chloramino acids such as N-chloro-

glycine, N-chloroleucine and N-chlorophenylalanine as by-products
after chlorination of water containing nitrogen or after chloramination.
2.5.3 Chlorine dioxide organohalogen by-products
Chlorine-free chlorine dioxide does not form THMs (Noack &

Doerr, 1978; Symons et al., 1981). Several studies show that the TOX
formed with chlorine dioxide is 1–15% of the TOX formed with
chlorine under the same reaction conditions (Chow & Roberts, 1981;
Symons et al., 1981; Fleischacker & Randtke, 1983).
Treatment of phenol-laden source waters with chlorine dioxide

does not produce the typical chlorophenolic taste and odour compounds
49
that are produced when the water is treated using chlorine and is
effective in removing existing tastes and odours of this type.
2.5.4 Ozone organohalogen by-products
Ozonation of drinking-water containing bromide ion has been

shown to produce hypobromous acid/hypobromite, with hypobromite
ion serving as an intermediate to bromate formation. In the presence of
NOM, hypobromous acid produces a host of brominated organic
compounds, such as bromoform, MBA, DBA, DBAN, cyanogen
bromide and DBAC (Glaze et al., 1993; Siddiqui & Amy, 1993).
Cavanagh et al. (1992) and Glaze et al. (1993) reported the identi-
fication of bromohydrins, a new group of labile brominated organic
compounds from the ozonation of a natural water in the presence of
enhanced levels of bromide. However, results by Kristiansen et al.
(1994) strongly suggest that the bromohydrins, such as 3-bromo-2-
methyl-2-butanol, in extracts of unquenched disinfected water are
artefacts formed from the reaction of excessive hypobromous acid with
traces of olefins in the extraction solvents and not novel DBPs.
Table 5 compares the median concentrations of various DBPs after

ozonation and chlorination.
2.6 Formation of inorganic disinfectant by-products
Although organic DBPs have been the subject of study over a

longer time frame, the formation of many inorganic by-products is
coming under increasing scrutiny.
2.6.1 Chlorine inorganic by-products
Chlorite and chlorate are inorganic by-products formed in some

chlorine solutions. This is of interest because many small drinking-
water utilities use hypochlorite solutions as a source of free chlorine
for disinfection. Bolyard & Fair (1992) examined the occurrence of
chlorate in samples of untreated source water, drinking-water and
hypochlorite solutions from 14 sites that use hypochlorite solutions.
The hypochlorite solutions used were found to contain significant
levels of chlorate. Chlorite and bromate were also found in
hypochlorite solutions from these same water utilities. Chlorate was
present in drinking-water, either as a manufacturing by-product or from
50
Table 5. Median concentrations of organic disinfectant by-products in drinking-water
DBPs Median concentration Median concentration

(:g/litre): chlorinationa (:g/litre): ozonationb
THMs 40 <1.0
Chloroform 15 –
BDCM 10 –
DBCM 4.5 –
Bromoform 0.57 <1.0
HANs 2.5 <1.0

TCAN <0.012 –
DCAN 1.1 –
BCAN 0.58 –
DBAN 0.48 <1.0
Haloketones 0.94 –
DCPN 0.46 –
TCPN 0.35 –
HAAs 20 <5.0
MCA 1.2 –
DCA 6.8 –
TCA 5.8 –
MBA <0.5 <1.0
DBA 1.5 <5.0
Aldehydes 7.8 45
Formaldehyde 5.1 20
Acetaldehyde 2.7 11
Glyoxal – 9
Methylglyoxal – 5
Chloral hydrate 3.0 –
Ketoacids – 75
Trichlorophenol <0.4 –
a
Krasner et al. (1989).
b
Siddiqui et al. (1993).
decomposition reactions occurring during storage. Approximately

0.2 mg of chlorate per litre was observed in water following the
addition of chlorine as sodium hypochlorite at a dose sufficient to
maintain a residual of 0.45 mg/litre (Andrews & Ferguson, 1995). The
concentration of chlorite in commercial bleach solutions typically
ranges from 0.002 to 0.0046 mol/litre; similarly, the chlorate concen-
tration ranges from about 0.02 to 0.08 mol/litre (Gordon et al., 1995).
51
A detailed study by Bolyard & Fair (1992) demonstrated that

hypochlorite solutions used to disinfect drinking-water contain
significant levels of chlorite and chlorate. The concentration of chlorite
ranged from <2 to 130 mg/litre for free available chlorine concentra-
tions ranging from 3 to 110 g/litre. The concentration of chlorate varied
over the range 0.19–50 g/litre, with a median of 12 g/litre. These
solutions also contained bromate levels ranging from <2 to 51 mg/litre.
The concentrations of chlorate in treated source waters ranged from 11
to 660 :g/litre. In another study involving 25 samples from plants
using gaseous chlorine, no chlorate was detected, indicating that the use
of gaseous chlorine does not produce chlorate (Bolyard & Fair, 1992).
Nieminski et al. (1993) measured chlorate and chlorite for six water
treatment plants that use liquid chlorine (i.e., hypochlorite) and found
chlorate concentrations ranging from 40 to 700 :g/litre, with no
chlorite or bromate detected in finished waters. These chlorate concen-
trations may be attributed to high concentrations of chlorate, ranging
from 1000 to 8000 mg/litre, detected in a bleach used for disinfection
and resulting from the decomposition of hypochlorite stock solution.
However, no chlorite or chlorate was detected in any of the samples of
finished water of the treatment plants that apply gaseous chlorine.
Chlorate formation is expected to be minimal in low-strength hypo-
chlorite solutions freshly prepared from calcium hypochlorite, because
of the low hypochlorite concentration and only mildly alkaline pH.
2.6.2 Chloramine inorganic by-products
Inorganic by-products of chloramination include nitrate, nitrite,

hydrazine and, to some extent, chlorate (Dlyamandoglu & Selleck,
1992; Kirmeyer et al., 1995).
2.6.3 Chlorine dioxide inorganic by-products
The major inorganic by-products of chlorine dioxide disinfection

have been identified as chlorite and chlorate. Andrews & Ferguson
(1995) measured a chlorate concentration of 0.38 mg/litre when a
chlorine dioxide residual of 0.33 mg/litre was maintained. The appli-
cation of chlorine dioxide produces about 0.5–0.7 mg of chlorite and
0.3 mg of chlorate per mg of chlorine dioxide consumed or applied
(Andrews & Ferguson, 1995).
52
2.6.4 Ozone inorganic by-products
When bromide or iodide ions are present in waters, some of the

halogen-containing oxidants that can be produced during ozonation
include free bromine, hypobromous acid, hypobromite ion, bromate
ion, free iodine, hypoiodous acid and iodate ion.
During the oxidation or chemical disinfection of natural waters

containing bromide ion with ozone, bromate ion can be formed at
concentrations ranging from 0 to 150 :g/litre under normal water
treatment conditions (Siddiqui, 1992). Chlorate formation with an
initial total chlorine concentration of 0.6 mg/litre was evaluated at pH
levels of 8.0, 7.0 and 6.0, and chlorate concentrations ranging from 10
to 106 :g/litre were formed after ozonation (Siddiqui et al., 1996a).
It has been reported that ozone reacts with many metal ions and
with cyanide ion (Hoigne et al., 1985; Yang & Neely, 1986). Bailey
(1978) discussed the formation of ozonates, compounds of metal
cations having the general formula M+O3–. Hydrogen peroxide has been
identified as a by-product of ozonation of organic unsaturated
compounds (Bailey, 1978).
Table 6 provides the range of bromate concentrations normally

encountered in drinking-waters with a variety of source water
characteristics after ozonation.
2.7 Formation of non-halogenated organic disinfectant

by-products
2.7.1 Chlorine organic by-products
Lykins & Clark (1988) conducted a 1-year pilot plant study of the
effects of ozone and chlorine and determined that the concentration of
aldehydes increased by 144% upon ozonation. In the chlorinated
stream, the concentration of these aldehydes increased by 56%. This
study indicates that aldehyde formation, although greater with ozone,
is not unique to ozonation, but is associated with chlorination and other
oxidants as well.
53
Table 6. Summary of bromate ion formation potentials in different source waters under different conditions following ozonation
Na Bromide (:g/litre) Ozone (mg/litre) pH Alkalinity (mg/litre) DOC (mg/litre) Bromate (:g/litre) Reference
18 10–800 1–9.3 5.6–9.4 20–132 2.2–8.2 <5–60 Krasner et al. (1992)
4 60–340 3–12 6.5–8.5 90–230 3–7 <5–40 Siddiqui & Amy (1993)
28 10–100 2–4 6.8–8.8 20–120 0.3–11 <5–100 Amy et al. (1993, 1994)
4 12–37 0–3.97 7.8 N/A N/A <7–35 Hautman & Bolyard (1993)
1 500 2.3–9.5 7.2–8.3 N/A N/A 13–293 Yamada (1993)
23 12–207 0.3–4.3 5.7–8.2 14–246 0.5–6.8 <2–16 Legube et al. (1993)
8 107–237 1–5 6.8–8.0 N/A 2–5 <5–50 Kruithof & Meijers (1993)
a
N = number of sources studied.
2.7.2 Chloramine organic by-products
When Suwannee River (USA) fulvic acid was reacted with

aqueous solutions of 15N-labelled chloramine and 15N-labelled ammo-
nia, lyophilized products exhibiting nuclear magnetic resonances
between 90 and 120 ppm were observed, denoting the formation of
amides, enaminones and aminoquinones (Ginwalla & Mikita, 1992).
This represents evidence for the formation of nitrogen-containing
compounds from the chloramination of NOM in natural waters.
Amino acids, peptides and amino sugars were chlorinated under

various chlorine/nitrogen ratios (Bruchet et al., 1992). Six natural
amino acids (alanine, methionine, valine, phenylalanine, leucine and
isoleucine) were shown to induce tastes and odours at concentrations
in the range of 10–20 :g/litre. Detectable odours were consistently
induced in a multicomponent mixture containing each of these amino
acids after a 2-h contact time with chlorine. Investigation of the by-
products indicated that the odours generated were systematically linked
to the aliphatic aldehydes formed. The peptides investigated had varying
degrees of odour formation potential, while the amino sugars did not
impart any odour. Chlorinous odours occasionally detected during
these experiments were found to be due to organic chloramines and
other oxidation by-products.
2.7.3 Chlorine dioxide organic by-products
Gilli (1990) showed the formation of carbonyl compounds

(34 :g/litre) such as n-valeraldehyde (7–15 :g/litre), formaldehyde
(3.4–9 :g/litre), acetaldehyde (4.5 :g/litre) and acetone (3.2 :g/litre)
after using chlorine dioxide.
2.7.4 Ozone organic by-products
Ozone aliphatic oxidation products from organic impurities in

water are usually acids, ketones, aldehydes and alcohols. So-called
ultimate oxidation products of organic materials are carbon dioxide,
water, oxalic acid and acetic acid. However, ozonation conditions
generally employed in treating drinking-water are rarely sufficient to
form high concentrations of these ultimate products.
55
When source waters containing NOM and unsaturated organic

compounds are ozonated, ozonides, peroxides, diperoxides, triperoxides
and peroxy acids, for example, can be produced. The limited research
that has been conducted in aqueous solutions indicates that these
intermediates decompose readily in water to form products such as
aldehydes, ketones, carboxylic acids and ketoacids.
Coleman et al. (1992) identified numerous compounds in addition

to the following in ozonated humic samples: monocarboxylic acids up
to C-24, dicarboxylic acids up to C-10, ketoacids, furan carboxylic
acids, and benzene mono-, di- and tricarboxylic acids. Among the
various aldehydes, Paode et al. (1997) found four (formaldehyde, acet-
aldehyde, glyoxal and methylglyoxal) to be dominant. Table 7 provides
a range of concentrations for aldehydes from the ozonation of a variety
of source waters.
2.8 Influence of source water characteristics on the

amount and type of by-products produced
The extensive literature pertaining to DBP levels in disinfected

source waters and control of DBPs by various treatment processes
attests to the wide variety of factors influencing DBP formation and the
complex interrelationships between these factors. Variables including
the concentration and characteristics of precursor material, pH, chlorine
concentration, bromide level, presence of chlorine-demanding
substances such as ammonia, temperature and contact time all play a
role in DBP formation reactions.
2.8.1 Effect of natural organic matter and UV absorbance at 254 nm
NOM consists of a mixture of humic substances (humic and fulvic

acids) and non-humic (hydrophilic) material. Both the amount (as
indicated by TOC or UVA254) and the character (as described by
UVA254) of NOM can affect DBP formation. NOM provides the
precursor material from which organic DBPs are formed; consequently,
increasing concentrations of NOM lead to increasing concentrations of
by-products. This relationship has led to the use of TOC and UVA254
measurements as surrogate parameters for estimating the extent of DBP
formation.
56
Table 7. Effect of ozone dose and TOC on non-halogenated organic by-products
Ozone dose TOC Formal Acetal Glyoxal Methyl-glyoxal Reference

(mg/litre) (mg/litre) (:g/litre) (:g/litre) (:g/litre) (:g/litre)
1.2–4.4 2.66 8–24 2–4 4–11 4–15 Miltner et al. (1992)
1.0–9.2 1.0–25.9 3–30 7–65 3–15 3–35 Weinberg et al. (1993)
5.5–28.5 5.4–17.4 58–567 6–28 15–166 17–54 Schechter & Singer (1995)
The removal of NOM is strongly influenced by those properties

embodying the size, structure and functionality of this heterogeneous
mixture. The humic acids are more reactive than fulvic acids with
chlorine (Reckhow et al., 1990) and ozone, in terms of both oxidant/
disinfectant demand and DBP formation. Processes such as coagula-
tion, adsorption and membrane filtration are separation processes that
remove NOM intact, while ozonation transforms part of the NOM into
biodegradable organic matter, potentially removable by biofiltration.
Coagulation preferentially removes humic/higher molecular weight
NOM; the selectivity of membranes for NOM removal is largely
dictated by the molecular weight cutoff of the membrane; the use of
granular activated carbon (GAC) requires a significant empty bed
contact time; biofiltration can remove only the rapidly biodegradable
NOM fraction.
In an investigation of the nature of humic and fulvic acids isolated

from a variety of natural waters, Reckhow et al. (1990) found that the
fulvic fractions had a lower aromatic content and smaller molecular
size than the humic fractions. UV absorbance was correspondingly
higher for the humic fractions, owing to the higher aromatic content
and larger size. These researchers also found that for all of the organic
material investigated, the production of chloroform, TCA, DCA and
DCAN was higher upon chlorination of the humic fractions than upon
chlorination of the corresponding fulvic fractions. These findings
support the findings of other researchers and show that the UV absor-
bance measurement is an indicator of the nature of the precursor
material present in a sample. This measurement, in conjunction with the
TOC (or DOC) measurement, can be employed in the evaluation of data
to provide an indication of the reactivity of NOM towards forming
DBPs.
The reaction of ozone with NOM can occur directly or by radical

processes. The disappearance of disinfecting chemical is influenced by
the type and concentration of NOM present in natural waters. Direct
consumption of these chemicals is greater when the UV absorbance
(due to electrophilic and nucleophilic sites of NOM) of the source
water is significant, resulting in decreased DBP formation potential.
It appears that the nature of the organic material in a source water

may have some impact on the relative concentrations of THMs and
HAAs formed upon chlorination. Treatment techniques that lower the
58
levels of DOC without affecting bromide levels have been implicated

in a shift from chlorinated to brominated THM compounds. This is of
concern because the theoretical risk to humans varies for the individual
THMs, with the brominated species generally being of more concern
(Bull & Kopfler, 1991).
2.8.2 Effect of pH
The impact of pH on THM concentrations has been reported by a

number of researchers since THMs in drinking-water first came to the
attention of the water industry (Stevens et al., 1976; Lange &
Kawczynski, 1978; Trussell & Umphres, 1978). More recently, the
impact of pH on a number of other chlorination by-products has been
reported (Miller & Uden, 1983; Reckhow & Singer, 1985). The rate of
THM formation increases with the pH (Stevens et al., 1976).
Kavanaugh et al. (1980) reported a 3-fold increase in the reaction rate
per unit pH.
In general, increasing pH has been associated with increasing

concentrations of THMs and decreasing concentrations of HAAs (pH
primarily impacting TCA), HANs and haloketones. The concentrations
of TCA tend to be higher in waters with pH levels less than 8.0 than in
waters with pH levels greater than 8.0; a less marked trend is observed
for DCA. Other researchers have reported similar findings with respect
to the pH dependency of HAA concentrations. For example, Stevens et
al. (1976) found that TCA concentrations were significantly lower at
a pH of 9.4 than at pH levels of approximately 5 and 7. TCA was by far
the most predominant of the measured HAA species at six of the eight
utilities surveyed. Carlson & Hardy (1998) reported that at pH levels
greater than 9.0, THM formation decreased with increasing pH. It is
possible that the shift in chlorine species from hypochlorous acid to
hypochlorite affects THM formation during short reaction times.
AWWARF (1991) observed no relationship between pH and the

concentrations of THMs at eight utilities over time, suggesting that
although THM concentrations for a particular water are known to be
pH dependent, factors other than pH influence THM concentrations
over a variety of source waters. Nieminski et al. (1993) reported that
treatment plants with a pH of about 5.5 in finished water produced
equal amounts of THMs and HAAs, whereas plants with pHs greater
59
than 7.0 in finished water produced higher amounts of THMs as

compared with HAAs.
No strong relationship has been observed between HAN

concentration and pH over time. Within the approximate pH range
7–8.5, HAN concentrations increased slightly over time. In general, a
trend of decreasing HAN concentrations with increasing pH would be
expected, since these compounds are known to undergo base-catalysed
hydrolysis and have been identified as intermediates in the formation of
chloroform (Reckhow & Singer, 1985). Therefore, these compounds
may be unstable in the presence of free chlorine or under basic
conditions. In general, after an initial formation period, HAN and
haloketone concentrations level off or begin to decline over the
remainder of the reaction period. This indicates that base-catalysed
hydrolysis may not be a significant mechanism of reaction for the
relatively low pH sources.
Stevens et al. (1989) evaluated the effects of pH and reaction time

(4, 48 and 144 h) on the formation of chloral hydrate. Chloral hydrate
formation increased over time at pH 5 and 7, whereas chloral hydrate
that had formed within 4 h at pH 9.4 decayed over time at the elevated
pH.
The pH of the source water can also affect the formation of by-
products after chloramine addition. The disproportionation of mono-
chloramine, which is an important reaction leading to an oxidant loss,
has been shown by several researchers to be catalysed by hydrogen ion,
phosphate, carbonate and silicate (Valentine & Solomon, 1987).
Humic acids have shown reaction rates with chlorine dioxide that
increased by a factor of 3 per pH unit (pH 4–8) (Hoigne & Bader,
1994).
In addition to the impact of pH on THM and HAA formation noted

above, overall TOX formation decreases with increasing pH. Many of
the halogenated DBPs tend to hydrolyse at alkaline pH levels (>8.0)
(Singer, 1994a). This has significant implications, for example, for
precipitative softening facilities.
pH has a strong effect on aldehyde formation (Schechter & Singer,

1995). Higher ozonation pH values produced lower amounts of
60
aldehydes, supporting the theory that these DBPs are formed primarily
through the direct molecular ozone reaction pathway, as opposed to the
radical pathway. These results may also reflect greater destruction of
aldehydes by hydroxyl radicals at elevated pH levels.
2.8.3 Effect of bromide
The presence of bromide ion during water treatment disinfection

can lead to the formation of DBPs such as brominated organics and
bromate ion. Low but significant levels of bromide, the ultimate
precursor to bromate and other brominated compounds, may occur in
drinking-water sources as a result of pollution and saltwater intrusion
in addition to bromide from natural sources. An understanding of the
sources and levels of bromide ion in different source waters is crucial
for an understanding of the bromate ion formation potential in
drinking-waters. There are no known treatment techniques available for
economically removing bromide ion present in source waters during
drinking-water treatment.
The impact of bromide on the speciation of DBPs within a class

of compounds such as THMs or HAAs has been discussed by Cooper
et al. (1983, 1985) and Amy et al. (1998). Rook et al. (1978) reported
that bromine is more effective than chlorine in participating in
substitution reactions with organic molecules; furthermore, precursor
materials may differ in their susceptibility to bromination versus
chlorination reactions. Hypobromous acid formed from bromide may
also react with ammonia to form bromamines (Galal-Gorchev &
Morris, 1965).
2.8.4 Effect of reaction rates
After chlorine addition, there is a period of rapid THM formation

for the initial few hours (e.g., 4 h), followed by a decline in the rate of
THM formation, suggesting fast and slow NOM reactive sites. Many
authors have indicated that the concentration of chloroform appears to
increase slowly even after 96 h, suggesting that as long as low
concentrations of free chlorine are present, chloroform continues to
form. Bromochlorinated THM species have been found to form more
rapidly than chloroform. Further data from many sources indicate that
bromoform formation slows at approximately 7–8 h and levels off
almost completely after 20 h (AWWARF, 1991; Koch et al., 1991).
61
The same general kinetic trend observed for THMs also appears to apply
to HAAs. A period of rapid formation occurs during the first 4–8 h,
followed by a reduction in the formation rate. In general, for most
sources, concentrations of chlorinated HAAs appear to slowly increase
even after 96 h, while the formation of DBA levels off after about
18–20 h.
Miller & Uden (1983) observed that nearly 90% of the final
concentrations of THMs, TCA and DCA form within the first 24 h of
chlorine addition to waters containing NOM. Reckhow et al. (1990)
found that although waters containing precursor materials isolated from
six different water sources differed in their yields of chlorinated organic
by-products, the formation curves for chloroform, TCA and DCA had
the same general shapes for all six precursor materials. Some
researchers have suggested that DCA may be an intermediate in TCA
formation; however, for all eight source waters studied, both DCA and
TCA concentrations increased or remained stable throughout the 96-h
reaction period, suggesting that DCA was an end-product (AWWARF,
1991). Carlson & Hardy (1998) indicated that HAA formation fol-
lowed a pattern similar to that of THM formation. As with the THMs,
HAA formation rate appeared to be rapid for the first 30 min; after
30 min, HAAs formed at nearly a constant rate in four of the source
waters studied.
Different trends were observed in the HAN concentrations of

different source waters. For two source waters, HAN levels formed
rapidly for the first 8 h and continued to increase slowly or levelled off
after 96 h (AWWARF, 1991). DBAN levels remained relatively stable
over the 96 h, as did BCAN and DCAN levels. For other sources, levels
of HANs consisting mostly of DCAN increased rapidly up to 4–8 h and
began to decline by the end of the 96-h period. For these sources,
BCAN appeared to be slightly more stable than DCAN.
Very low levels of chloropicrin formation have been observed by

many researchers (AWWARF, 1991). The highest concentration
observed was 4.0 :g/litre. Chloropicrin appears to form slowly during
the incubation period, with concentrations tending to level off at
approximately 40 h.
62
2.8.5 Effect of temperature
The formation rates of THMs, HAAs, bromate ion and HANs have
been shown to increase with temperature (AWWARF, 1991; Siddiqui
& Amy, 1993). Both haloketone and chloropicrin levels were found to
be higher at a lower temperature, while the concentrations of other
DBP species were similar or not significantly different. These results
suggest that a higher temperature allows for more rapid progression of
the transformation of haloketones to other by-products. In studies on
the effect of temperature on THMs, Peters et al. (1980) found an
Arrhenius dependency between the rate constant and temperature with
an activation energy of 10–20 kJ/mol.
The impact of temperature on THMs was strongest at longer

contact times (Carlson & Hardy, 1998). On a conceptual basis, it may
be that rapidly forming compounds are more reactive and form DBPs
regardless of temperature. On the other hand, slowly forming
compounds require higher activation energy, and an increase in the
temperature supplies the energy. In addition to reaction kinetics, the
temperature of a source water can also affect disinfection efficiency.
The biocidal effectiveness of monochloramine is significantly less than
that of free chlorine and is dependent on temperature, pH and residual
concentration.
2.8.6 Effect of alkalinity
Although pH is a very influential variable and alkalinity affects pH,

alkalinity itself does not appear to directly affect the formation of
THMs and HAAs (by chlorination) and has only a slight effect on
aldehydes and other organic by-products following ozonation (Andrews
et al., 1996). However, the majority of studies on the effect of
alkalinity on the formation of bromate during ozonation indicate that
increased alkalinity increases bromate formation (Siddiqui et al., 1995).
The quantity of aldehydes produced remains approximately constant for
similar changes in alkalinity and pH; however, deviation from
equivalent changes in pH and alkalinity results in increased aldehyde
concentrations. Therefore, conditions of high alkalinity and low pH or
low alkalinity and high pH produce greater quantities of aldehydes than
do intermediate values of these parameters (Andrews et al., 1996).
63
2.9 Influence of water treatment variables on the

amount and type of by-products produced
Since DBPs are formed by all of the above chemical disinfectants,

the adoption of alternative disinfectants for DBP control often means
only a trade-off between one group of DBPs and another. The most
effective DBP control strategy is organic precursor (NOM) removal
through enhanced coagulation, biofiltration, GAC or membrane
filtration. There has been little success with bromide removal. Other
DBP control options include water quality modifications — for
example, acid or ammonia addition for bromate minimization.
2.9.1 Effect of ammonia
The presence of ammonia in source waters during disinfection can

cause chlorine and ozone demand and participation in the formation of
by-products such as nitrate, cyanogen chloride and other nitrogenous
compounds.
Ammonia also does not consume chlorine dioxide. In contrast to

chlorine, chlorine dioxide can therefore be considered as a virucide
when ammonia is present. This might be one of the historical reasons
why chlorine dioxide has been adopted as a disinfectant by some
treatment plants using well oxidized waters but containing changing
ammonia concentrations. The addition of ammonia has been shown to
reduce the formation of bromate after ozonation (Siddiqui et al., 1995),
and the ammonia has been shown to participate in the formation of
HANs and cyanogen bromide (CNBr) (Siddiqui & Amy, 1993).
The growth of nitrifying bacteria is a potential problem in

chloraminated water supplies or chlorination of sources containing
nitrogen. In a study conducted by Cunliffe (1991), nitrifying bacteria
were detected in 64% of samples collected from five chloraminated
water supplies in South Australia and in 21% of samples that contained
more than 5 mg of monochloramine per litre. Increased numbers of the
bacteria were associated with monochloramine decay within the
distribution systems.
64
2.9.2 Effect of disinfectant dose
Chlorine dose is a factor affecting the type and concentration of

DBPs formed. The THM level rises with increasing chlorine dose
(Kavanaugh et al., 1980). However, there is some disagreement
regarding the quantitative relations between chlorine concentration and
THM levels (or the rate of THM production). Most investigators found
a linear relationship between chlorine consumption and THM
production, with an order of reaction greater than or equal to unity
(Trussell & Umphres, 1978; Kavanaugh et al., 1980). However, it is
also possible that the order of reaction changes during the course of the
reaction.
Reckhow & Singer (1985) found that the concentration of DBP

intermediates such as DCAN and 1,1,1-TCPN formed after 72 h of
reaction time was dependent on chlorine dose. DCAN, which was
measured at a concentration of approximately 5 :g/litre at a chlorine
dose of 10 mg/litre, was not detected in samples dosed with 50 mg of
chlorine per litre. The concentration of chloroform was about
150 :g/litre in a sample dosed with 10 mg of chlorine per litre but was
approximately 200 :g/litre in a sample dosed with 20 mg of chlorine
per litre. Thus, it is imperative to have uniform chlorine doses for
performing DBP formation kinetic measurements.
Since chloramine residuals are longer-lasting than free chlorine

residuals, the doses for each set of chlorinated and chloraminated
samples will be different in order to achieve the prescribed target
residual. The disappearance of chloramines can be explained approxi-
mately by a second-order reaction. However, as the chlorine dose
increased, the observed rate constant was found to decrease, then
increased after reaching a minimum value (Dlyamandoglu & Selleck,
1992). Below the chlorine dose at the minimum value of the observed
rate constant, the rate constant was proportional to the 1.4 power of the
chlorine dose, regardless of the ammonia concentration (Yamamoto et
al., 1988).
2.9.3 Effect of advanced oxidation processes
Water utilities can add treatment processes that remove DBP

precursors or DBPs. Many utilities will be using both approaches. The
65
hydrogen peroxide/UV process, an advanced oxidation process, offers

small water utilities a treatment process with the potential to provide
primary disinfection and a method of DBP control (Symons & Worley,
1995). This process has been shown to oxidize dissolved organic
halogens and decrease TOC. TOC removal as a function of UV dose
has also been demonstrated by Worley (1994), with TOC removals of
between 0% and 80%. Andrews et al. (1996) evaluated the effect of the
hydrogen peroxide/UV process on THM formation and concluded that
this process is only slightly effective in reducing the formation of
DBPs. However, using hydrogen peroxide at 1 mg/litre in combination
with UV effectively reduced or prevented the formation of aldehydes.
Other advanced oxidation processes (e.g., hydrogen peroxide/ozone,
ozone/UV) involving hydroxyl (and hydroperoxyl) radical formation
may provide similar opportunities.
2.9.4 Effect of chemical coagulation
Enhanced coagulation and softening will remove TOC. Enhanced

coagulation is characterized by coagulant doses greater than those
required for optimum turbidity removal; as an alternative to higher
doses, a combination of acid (pH depression) and coagulant addition
can be practised.
All organic DBPs were reduced by the addition of commonly used

coagulants. Iron-based coagulants, such as ferric chloride, were
consistently more effective than alum in removing NOM (Crozes et al.,
1995). Alum coagulation removed all DBP precursors to a significant
extent. The percentage removals showed the same trends as, but were
not identical to, the percentage removals of TOC and UV absorbance.
UV absorbance was removed to a somewhat greater extent than TOC.
Hence, TOC and UV absorbance can serve as surrogate parameters for
DBP formation potential. A fairly good correlation was observed
between the ratio of HAAs to THMs and the ratio of UV absorbance to
TOC, indicating that the relative concentrations of HAAs and THMs do
to some extent depend on the nature of the precursor material.
However, more data from waters of different qualities would be
required to evaluate the validity of this relationship.
The effectiveness of coagulants in removing DBP precursors is

dependent upon the molecular size of the dissolved organic matter.
Normally, higher molecular weight fractions are effectively removed
66
through coagulation. In a study conducted by Teng & Veenstra (1995),

water containing dissolved organic matter with molecular weights in
the range 1000–10 000 daltons generally produced the largest amounts
of THMs and HAAs under conditions of free chlorination. Coagulation
and ozonation shift a proportionately greater amount of the THM and
HAA formation potential to the smallest molecular weight range
(<1000 daltons).
Coagulation and filtration remove NOM but not bromide, hence

increasing the ratio of bromide to TOC. As a result, the subsequent use
of chlorine generally favours the formation of brominated organic
DBPs.
2.9.5 Effect of pre-ozonation
Several studies of ozone oxidation followed by chlorination

showed increased, rather than decreased, levels of THMs (Trussell &
Umphres, 1978). This is attributed, at least partially, to the formation
of aldehydes by ozonation. Another possibility is hydroxylation of
aromatic compounds to produce m-dihydroxy aromatic derivatives,
which are known THM precursors (Lykins & Clark, 1988). Although
the aldehydes produced contain polar groupings, they are nevertheless
not easily removed during the flocculation step by complexation with
aluminium or iron salts. A convenient and more appropriate method for
the removal of the aldehydes formed during ozonation is the incor-
poration of a biological treatment step (biofiltration) in the water
treatment process following ozone oxidation.
Pre-ozonation can have both positive and negative effects on DBP

formation. Pre-ozonation with typical water treatment dosages and
bicarbonate levels has been shown to remove TCA and DCAN
precursors. However, such treatment can result in no net change in the
DCA precursors and may lead to an increase in 1,1,1-TCPN precursors
(Reckhow & Singer, 1985). According to Teng & Veenstra (1995), pre-
ozonation resulted in enhanced formation of DCA during chlorination
and chloramination in the presence of precursors in the <1000 dalton
molecular weight range. They also indicated that pre-ozonation plus
chloramination controlled the overall production of THMs and HAAs.
However, the use of pre-ozonation coupled with free chlorination
67
increased the yield of DCA for both the hydrophilic and hydrophobic
fractions of NOM as compared with free chlorination alone.
With ozone–chlorine treatment, chloral hydrate formation can be

enhanced. This behaviour, which has also been observed for DCA,
suggests that the reaction that produces chloral hydrate is accelerated
under the conditions of ozonation in combination with prechlorination
and warm water temperatures (LeBel et al., 1995).
Ozonation in the presence of traces of hypochlorite ion can form

inorganic by-products such as chlorate. Siddiqui et al. (1996a) showed
that if there is any residual chlorine present, ozone can potentially
oxidize hypochlorite ion to chlorate.
Coleman et al. (1992) suggested that brominated MX analogues

and other mixed bromochlorinated by-products formed after ozonation
and chlorination can possibly increase mutagenic activity.
2.9.6 Effect of biofiltration
Biofiltration (ozone–sand filtration or ozone–GAC) can poten-

tially reduce TOC, organic by-products and the formation of halo-
genated DBPs.
Passage of ozonated water samples through a rapid sand filter

reduced the concentration of aldehydes by 62% (Lykins & Clark, 1988).
Chlorinated samples experienced a 26% reduction in aldehyde
concentrations under the same conditions. These reductions in aldehyde
levels are attributable to biological activity in the sand filters. If GAC
filtration follows sand filtration, ozone oxidation can be expected to
promote more bioactivity in the GAC filter, because a better
colonization environment is provided for microorganisms on GAC
particles than on sand. Thus, the biological conversion of oxidized
water impurities to carbon dioxide and water will be greater during
passage through GAC media. Similar aldehyde removals have been
observed by several researchers (Van Hoof et al., 1985; Sketchell et al.,
1995).
Drinking-water treatment techniques that remove organic contami-

nants without affecting bromide concentrations cause a shift in the
formation of DBPs towards brominated DBPs. Sketchell et al. (1995)
68
studied three sources containing three different DOC levels and ambient
bromides, which were filtered through biologically active GAC filters.
Analysis of treated waters showed no removal of bromide ion and a
shift towards more brominated organo-DBPs. THM levels after
treatment with GAC with no added ozone decreased from 900–1700 to
100–700 :g/litre. These water sources contained DOC levels ranging
from 10 to 25 mg/litre and high concentrations of biodegradable DOC
(DOC removals ranged from 60% to 80% after GAC treatment).
Table 8 summarizes the effects of ozonation and biofiltration on

the formation of DBPs from various sources.
2.10 Comparative assessment of disinfectants
A comparative assessment (Table 9) of various disinfecting

chemicals for pre-disinfection (or oxidation) and post-disinfection and
maintaining a residual for 5 days to simulate concentrations in the
distribution system showed that the use of free chlorine produces the
largest concentration of halogenated DBPs (Clark et al., 1994). The
concentration of DBPs may be reduced by adding ozone or chlorine
dioxide as a preoxidant, although enhanced formation has been
observed.
Table 10 summarizes the effects of water quality and treatment

variables on the formation of DBPs.
2.11 Alternative strategies for disinfectant by-product

control
The concern about chlorite, bromate, chlorate and other DBPs in

drinking-water following treatment with disinfectants has stimulated
research into ways to eliminate the production or enhance the removal
of DBPs. Strategies for DBP control include source control, precursor
removal, use of alternative disinfectants and removal of DBPs by
technologies such as air stripping, activated carbon, UV light and
advanced oxidation technologies. For DBPs that can arise in
hypochlorite solutions (e.g., chlorate), the purity and storage conditions
of the solutions are important concerns.
69
Table 8. Effects of ozonation and biofiltration on chlorine organic by-products
DBPs Ozonation Biofiltration Ozonation + biofiltration Reference

(% change) (% change) (% change)
THMs !20 !20 !40 (chlorine) Speitel et al. (1993)
HAAs !10 !13 !25 (chlorine) Speitel et al. (1993)
Chloropicrin +50 to +250 !50 to !100 (chlorine) Miltner et al. (1992)
Aldehydes +425 to +1300 !40 to !50 !92 to !98 Miltner et al. (1992)
TOX !30 !51 (chlorine) Miltner et al. (1992)
TOX !10 !38 !47 (chlorine) Shukairy & Summers (1992)
TOX +32 !69 !60 (monochloramine) Shukairy & Summers (1992)
Table 9. Comparative assessment of organic disinfectant by-products (:g/litre) in distribution systemsa,b
DBPs Sand–Cl2 Cl2–Sand–Cl2 O3–Sand–Cl2 NH2Cl–Sand–NH2Cl O3–Sand–NH2Cl ClO2–Sand–Cl2

THMs 236.0 225.0 154.0 9.0 3.2 138.0
HAAs 60.0 146.0 82.0 14.0 9.0 44.0
HANs 3.1 2.9 2.7 <0.1 <0.1 <0.1
Haloketones 2.1 2.6 2.6 <0.1 <0.1 4.2
Chloropicrin 1.3 1.3 7.7 <0.1 <0.1 1.4
Chloral hydrate 79.0 75.0 55.0 <0.1 <0.1 45.0
TOX 557.0 540.0 339.0 59.0 27.0 379.0
a
Clark et al. (1994).
b
TOC = 3.0 mg/litre; pH = 7.6.
Table 10. Summary of impact of water quality and treatment variables on disinfectant by-product formation
Variable Impact on THMs Impact on HAAs Impact on aldehydes Impact on chlorate/chlorite Impact on bromate
Contact time Curvilinear increase Curvilinear increase with Linear increase as long Linear increase in bleach Curvilinear increase with
with increasing contact increasing contact time as residual chemical solutions most bromate forming in
time Rapid formation <5 h present No discernible effects in <5 min
Rapid formation <5 h 90% formation in 24 h Secondary reactions dilute solutions Formation is a function of
90% formation in 24 h Levels off at 150 h between disinfectants If oxidation of ozone residual and
Levels off at 96 h and aldehydes possible hypochlorite, contact time bromide
has a positive effect
Disinfectant Rapid and curvilinear Curvilinear increase Curvilinear with Concentrations related to Linear increase after
dose increase after TOC after TOC demand with increasing ozone dose hypochlorite doses TOC demand and then
demand with dose, increasing dose, or chlorine dose applied levelling off after ozone
levelling off at 2.0 levelling off at 2.0 No appreciable effect Ozone oxidation of residual disappearance
mg/litre for TOC of 2.0 mg/litre after ozone/DOC = 2 : 1 hypochlorite increases
mg/litre with dose
pH Curvilinear increase Mixed, possible pH Negative effect (forms Positive effect Strong linear positive
with increasing pH to maximum for DCAA and mostly through molecular Decomposition of hypo- effect
pH 7.0 and possible DBAA ozone) chlorite increases with pH Hydroxyl radical
pH maximum TCAA decreases up to 25% decrease for pH Oxidation of hypochlorite generation efficiency
No positive effect at pH pH > 9 7–8.5 by ozone increases increases
> 9.5 DCAA maximum at pH
7–7.5
Table 10 (Contd).
Temperature Linear increase with Linear increase with Terminal products such Positive effect Curvilinear increase
increasing temperature increasing temperature as carbon dioxide Decomposition of 20–30% increase for
(10–30 °C; 15–25% (10–30 °C; 20–30% increase and total hypochlorite increases 15–25 °C
increase) increase) aldehydes slightly
decrease
TOC Increase with Increase with increasing Positive effect Negative effect if ozone is Decreases with
increasing TOC; TOC; precursor content (hydrophobic fraction used for hypochlorite increasing TOC;
precursor content important mostly responsible) oxidation precursor content
important Humic acids more Doubles for every 2 Most likely no effect with important
Humic acids more reactive than fulvic acids mg/litre hypochlorite Non-humic acid being
reactive than fulvic less reactive with ozone
acids
UVA254 Increase with Increase with increasing Positive effect Negative effect if ozone is Decreases with
increasing UV UV absorbance; Ozone demand used for hypochlorite increasing UV
absorbance; precursor precursor content increases with UV (UV oxidation absorbance; precursor
content important important absorbance is mostly Probable negative effect content important
Aromaticity of TOC Aromaticity of TOC due to aromaticity and with hypochlorous acid Humic acid being more
being more important being more important hydrophobic fraction) reactive with ozone
Bromide Shift towards Shift towards Independent of bromide Shift towards more toxic Bromide threshold
brominated species brominated species at <0.25 mg/litre bromate in hypochlorite Curvilinear increase and
At >0.25 mg/litre, solutions dependent upon ozone
aldehydes can decrease residual
due to ozone–bromide
oxidation
Alkalinity No discernible effect No discernible effect Slight positive effect Unknown Positive effect
91
Table 10 (Contd).
Minimization TOC removal, TOC removal, pH control, TOC removal Avoid hypochlorite dosing pH depression, ammonia
strategies minimizing chlorine minimizing chlorine by coagulation, GAC, solution Minimize storage addition, radical
residual, alternative residual, alternative optimizing doses, Properly tune generators scavengers, minimizing
disinfectants, pH disinfectants, pH control, contact time Use freshly made and optimizing ozone
control, minimizing minimizing contact time solutions residual
contact time
Removal GAC, electron beam, GAC, electron beam Biofiltration, advanced Ferrous sulfate, GAC, Ferrous sulfate, UV
strategies air stripping oxidation, GAC, electron beam, UV irradiation, high-energy
nanofilters irradiation, nanofilters electron beam, GAC
2.11.1 Source control
Source control options involve controlling nutrient inputs to

waters (e.g., algae growth control) (Hoehn et al., 1990) that are used as
drinking-water sources, watershed management (e.g., constructing
stormwater detention basins), saltwater intrusion control (e.g., develop-
ment of structural or hydrodynamic barriers to control TOC and
bromide), and using the concept known as aquifer storage and recovery
(e.g., drawing water during seasons when the quality of the water is
best) (Singer, 1994a).
2.11.2 Organohalogen by-products
Strategies for control of organohalogen by-products include

removal of DBPs that are formed using technologies such as oxidation,
aeration and carbon adsorption (Clark et al., 1994); and removal of
precursors using treatment techniques such as conventional treatment,
oxidation, membrane processes, carbon adsorption and biological
degradation. For many organic compounds that are difficult to oxidize,
such as chloroform, the kinetics of ozone oxidation are generally very
slow but are faster if used in combination with UV irradiation. GAC
adsorption and membrane filtration are relatively expensive processes;
moreover, NOM removal by GAC cannot be accomplished to any
significant degree in a filter/adsorber (i.e., GAC filter cap) mode but
requires a separate post-filtration adsorber bed. The use of membranes
requires pretreatment to prevent fouling, as well as processing of waste
brine. The use of ozone in combination with biologically active GAC
filters is a promising alternative to reduce DBP precursors.
2.11.3 Inorganic by-products
Properly designed and operated chlorine and chlorine dioxide

generator systems can minimize some of the production of chlorate ion.
Removal of chlorite and chlorate has been reported using reduction by
Fe2+ or sulfite or by GAC (Voudrias et al., 1983; Lykins & Clark,
1988). GAC is seen as problematic because of chlorate production and
a short bed life. A chemical process using an appropriate agent such as
reduced iron (e.g., ferrous sulfate) appears to be a more promising
approach (Kraft & van Eldick, 1989; Gordon et al., 1990).
75
If bromate is present in treated water entering the coagulation

process (i.e., formed during pre-ozonation), several options exist for its
removal. An aqueous-phase reducing agent (e.g., Fe2+) can be added at
the rapid mix step. Powdered activated carbon can likewise be added as
a solid-phase reductant to remove bromate and DBP precursors. Not all
utilities contemplating ozone application intend to employ pre-
ozonation. Rather, they may use intermediate ozonation prior to the
filtration process; in this situation, removal of bromate by activated
carbon is possible. This approach has potential relevance to integration
of GAC columns into a process train or, more realistically, to retro-
fitting of rapid sand filters with GAC filters. For groundwaters that
require no coagulation, bromate can be removed after ozonation using
a GAC filter, UV irradiation or high-energy electron beam irradiation
(Siddiqui et al., 1994, 1996a,b,c).
Brominated or bromochlorinated amines formed during the oxida-

tion step of the process train using chlorine can potentially be removed
using a suitable activated carbon before terminal chlorination. How-
ever, carbon that has an accumulation of surface oxides, which develop
through reaction of amines, will have a diminished capacity to reduce
halogenated amines to nitrogen. Organic amines can potentially be
removed by activated carbon adsorption.
2.11.4 Organic by-products
There are some technologies for removing organic contaminants

formed after chlorination and chloramination, a less viable option than
minimizing their formation through DBP precursor removal or use of
alternative disinfectants. Studies of ozone oxidation have shown that
aromatic compounds, alkenes and certain pesticides (some of which
have structural similarities to certain organic DBPs) are removed well
by ozone treatment, but that alkanes are poorly removed. Also, removal
efficiency improves for the alkenes and aromatic compounds with
increasing ozone dosage and for some alkanes with increasing pH. For
most compounds, the efficacy of ozone is not affected by the back-
ground water matrix if the ozone is used after coagulation. Andrews et
al. (1996) showed that using hydrogen peroxide at 1.0 mg/litre in
combination with UV effectively reduced or prevented the formation
of aldehydes.
76
2.12 Models for predicting disinfectant by-product

formation
The regulation of THMs and other halogenated DBPs has been

complicated by findings that alternative disinfectants to free chlorine
may also form by-products that are of potential health concern.
Additional complicating factors impacting the regulation of DBPs have
been the emergence of Giardia and Cryptosporidium as major water-
borne pathogens.
In view of the finding that water chlorination produces DBPs,

some of which are carcinogenic, mutagenic or possibly teratogenic,
several countries have recently laid down standards for various DBP
levels. This stimulated the search for mathematical models to describe
or predict DBP formation in disinfected water and to evaluate the
effectiveness of water treatment technologies designed to reduce DBP
levels so as to comply with the standards. Most of these models are
based on fitting mathematical equations to various empirical observa-
tions, rather than mechanistic and kinetic considerations. This is mainly
due to the complexity of the reactions between organic precursors and
disinfecting chemicals, which usually involve several parallel pathways
leading to a great variety of products. The complexity of the DBP
formation reactions also makes it difficult to develop universally
applicable models for simulating DBP formation potential associated
with disinfection of a diverse array of natural source waters. However,
the analysis presented by many models suggests that many waters
exhibit comparable general responses to changes in a given parameter
(i.e., responses lending themselves to simulation by a particular
mathematical functionality), although specific responses associated
with individual waters may vary. The multiple regression models
developed by many researchers represent a rational framework for
modelling DBP formation in many sources. Another potential appli-
cation is the modelling of DBP mixtures, e.g., predicting HAA levels
from THM and water quality data.
2.12.1 Factors affecting disinfectant by-product formation and

variables of interest in disinfectant by-product modelling
The information on the factors controlling DBP formation, which

is available in the literature, is briefly summarized below.
77
The extent of formation of DBPs is dependent on several water

quality parameters, such as TOC concentration, UVA254, bromide
concentration and temperature. It is also dependent on chlorination
conditions, such as chlorine dose, pH, ammonia concentration and
contact time. After the various statistically significant factors were
identified, mathematical equations were developed to describe the
formation of various DBPs. A least squares method was used to
determine the optimum equation coefficients that best describe the
experimental data. The optimum coefficients have been defined as those
that produce a minimum residual error between the mathematical
predictions and the experimental data.
2.12.2 Empirical models for disinfectant by-product formation
Numerous models for predicting THM formation through chlori-

nation have been reported (Moore et al., 1979a; Kavanaugh et al., 1980;
Engerholm & Amy, 1983; Urano et al., 1983; Amy et al., 1987, 1998;
Morrow & Minear, 1987; AWWARF, 1991; Hutton & Chung, 1992).
Of these, models reported by AWWARF (1991) and Amy et al. (1998)
are more recent and were derived from a variety of natural source
waters and more realistic treatment conditions. Not much information
has been reported on the formation of other chlorination DBPs. Only
Amy et al. (1998) summarized empirical models for THMs, HAAs and
chloral hydrate. These chlorination by-product models can be used to
assess both in-plant and distribution system formation of THMs, HAAs
and chloral hydrate. Water quality conditions such as DOC, pH,
temperature and bromide are needed as inputs to the models; such data
then allow assessment of chlorination DBP formation as a function of
reaction time:
DBP concentration (total THMs or THM species, total HAAs or

HAA species, or chloral hydrate) =
f(TOC, bromide, chlorine, pH, temperature, time)
Relatively little is known about the kinetics of the formation of

bromate and other DBPs during ozonation and the quantitative effects
of water quality factors (temperature, pH, etc.); such an understanding
is crucial for evaluating various bromate control strategies. Siddiqui &
Amy (1993) and Amy et al. (1998) developed statistical relations to
predict the concentrations of various ozone DBPs, including bromate,
as a function of water treatment variables. Correlation matrix analysis
78
has shown that ozone dose, dissolved ozone concentration, bromide

concentration, pH and reaction time all have a positive influence on
bromate formation. Von Gunten & Hoigne (1994) developed kinetic
models for bromate formation.
Ozone, as a result of its strong oxidizing power, produces a variety

of organic by-products, such as aldehydes and ketoacids, when used to
treat natural source waters. These by-products — especially aldehydes
— are highly biodegradable, and there is concern for regrowth of
microorganisms following ozone treatment. They are also potentially
hazardous and may produce increased amounts of chlorinated by-
products upon chlorination. Siddiqui et al. (1997) developed a model
to estimate the potential for total aldehyde formation in source waters
upon ozonation.
2.12.3 Models for predicting disinfectant by-product precursor

removal
It is recognized that chlorination will continue to be the most

common disinfection process; hence, enhanced removal of DBP
precursors present in raw sources represents a valuable option for
reducing the potential for by-product formation. The removal of NOM
can be achieved either by providing additional processes, such as GAC
and nanofiltration, or by enhancing the existing coagulation, floccu-
lation and sedimentation processes. Predictive models have been dev-
eloped for assessing coagulation efficiency in removing NOM and
reducing DBP precursor levels (AWWARF, 1991; Amy et al., 1998).
Coagulation can reduce DOC and DBP precursors but not

bromide levels; hence, a greater proportion of brominated DBP species
can potentially be produced in the finished water.
The effects of precursor removal by chemical coagulation can be

assessed through the use of treated water models. One can either predict
DBPs formed under a given degree of precursor removal or define the
degree of precursor removal required to meet DBP regulations. The
impact of bromide ion on meeting regulations can also be assessed. If
one makes the assumption that precursor reactivity (i.e., DBP/DOC)
does not change, one can also assess other precursor removal processes,
such as GAC or membrane processes, through use of the raw/untreated
79
water models. Care should be exercised when using models to

approximate post-chlorination DBPs following an ozonation step.
2.13 Summary
• The primary and most important role of drinking-water treatment

is to remove or inactivate harmful microorganisms. Another role
is to minimize the concentrations of disinfectants and DBPs
without compromising in any way the removal or inactivation of
pathogens.
• Drinking-water utility managers must be more knowledgeable

about options to meet regulations. It is often more practical to use
treatment methods that control the concentration of several
contaminants than to modify treatment practices for each new
standard that is promulgated.
• A thorough understanding of DBP formation would help the

successful balancing of appropriate microbial inactivation with the
minimization of DBPs. Water quality variables affect DBP
formation and must be considered when developing a strategy to
control DBPs with various disinfectants.
• The chemistry of chlorine and its by-products has been well

studied, and ozone and its by-products have recently received
much attention. Studies of chlorine dioxide and chloramines and
their by-products are relatively few, although more work in these
areas is now being undertaken.
• One of the simplest processes to minimize halogenated DBP

formation is limiting the free chlorine contact time by using
monochloramine to maintain a distribution system residual
following primary disinfection by chlorine or ozone. Chloramines
are an effective means of controlling DBPs. However, the growth
of nitrifying bacteria (and related production of nitrite) is a
potential problem in chloraminated water supplies.
• Various nitrogen-containing organic compounds may be present

in source waters after chlorination and chloramination. Because
of analytical complexities, very few detailed studies have been
undertaken to
80
determine the individual compounds present and their concentrations.
• Many factors between the source and the tap can influence the
DBPs to which consumers are exposed. Although THMs and
HAAs continue to form with increasing contact time, some other
halogenated DBPs, such as HANs and haloketones, form rapidly
but then decay in the distribution system as a result of hydrolysis.
This has major implications regarding exposure to these DBPs,
depending upon their proximity to the treatment plant. For treated
source waters, median levels of HAAs are often approximately
one-half of the median THM levels.
• For low-bromide source waters, chloroform is normally the

dominant THM species; DCA and TCA are the most prevalent
HAA species; DCAN is the most prevalent HAN species; and
1,1,1-TCPN is the most prevalent of the two measured halo-
ketones. Very low levels of chloropicrin have been observed by
various researchers; this compound appears to form slowly during
the incubation period, with concentrations tending to level off at
40 h. For high-bromide waters, increased levels of brominated
DBPs are observed.
• Chlorine dioxide is a strong oxidant that under certain conditions

surpasses chlorine in its ability to destroy pathogenic organisms.
When chlorine dioxide is prepared and administered without
excess free chlorine, THMs and other chlorinated by-products are
not produced, but inorganic by-products are formed.
• TOC levels have been found to be correlated with halogenated

DBP formation. The nature of this relationship varies with the
source. TOC removal can be used as a surrogate for the reduction
of DBP formation.
• Although the presence of chloral hydrate and HANs in chlorinated

samples may be attributed to precursors other than amino acids,
the potential for amino acids to be present in natural sources is
well documented. Surface waters, but not groundwaters, tend to
contain amino acids. However, the removal of these precursors
during conventional water treatment is not well understood.
81
• The amount of chlorate that is present in delivered hypochlorite

solutions depends on many factors. Freshly made hypochlorite
solutions will contain less chlorate than hypochlorite that is stored
without concern for temperature and pH. If a utility is using a
single tank to store hypochlorite, it is likely that the level of
chlorate is increasing in the tank. Thus, storage tanks should be
periodically flushed and cleaned, and, if possible, the storage time
should be reduced.
• Models have been developed that can be used to simulate the fate
and movement of DBP precursors in distribution systems. The
models can be designed as a planning tool for evaluating the
impacts of source water management strategies and estimating
DBP exposures. Some limitations of existing models include
calibration with a limited database, application to only a specific
water source or group of related sources, lack of terms to simulate
important parameters, such as reaction time, and inadequate
validation.
82
3. TOXICOLOGY OF DISINFECTANTS
In assessing the hazards associated with drinking-water dis-

infection, it is important not to neglect the disinfectants themselves.
Adding disinfectant in excess of the demand has several practical
benefits. First, it ensures that reaction of the disinfectant with DBP
precursors (largely organic material and ammonia) does not shorten
contact time to the point of ineffective disinfection. Second, residual
disinfectant helps to prevent regrowth of organisms in the remaining
portions of the treatment and distribution systems.
The result of this practice, however, is that one of the chemicals

that is present in the finished water at the highest concentration is the
disinfectant. In the present regulatory climate in many countries,
chemicals that are introduced as direct additives to food would be
subjected to a significant amount of toxicological screening before
they could be used. Since the major disinfectants were introduced
almost 100 years ago, they were subjected to much less thorough
toxicological evaluations than would be required today. However,
many of these data gaps have been addressed in the past decade.
3.1 Chlorine and hypochlorite
3.1.1 General toxicological properties and information on dose–

response in animals
Chlorine gas has long been recognized as a lung irritant. This

topic will not be reviewed in the present document, as it appears to be
largely irrelevant to the small amounts of chlorine that are volatilized
from chlorinated water in showers or other points of use in the house-
hold. In water treatment plants, however, there is a possibility of occu-
pational exposures that could have severe sequelae. For information
on these higher-level exposures, the interested reader is referred to a
recent review by Das & Blanc (1993). The effects of chlorine gas that
have been observed in humans will be discussed in section 3.1.3.
Sodium hypochlorite (NaOCl) or calcium hypochlorite

(Ca(OCl)2) solutions have also been utilized extensively in the
disinfection of drinking-water. The stock solutions used for this
83
purpose are highly caustic and are a clear concern for occupational
exposures. The concentration required to produce irritation and
decreased basal cell viability in the skin of guinea-pigs after an
application period of 2 weeks was 0.5% sodium hypochlorite (Cotter
et al., 1985). Reducing the concentration to 0.1% resulted in no effect
on basal cell viability relative to control animals. Yarington (1970)
demonstrated that instillation of bleach into the oesophagus of dogs
produced irritation. The minimal exposure that produced oesophageal
burns was 10 ml of commercial bleach with a 5-min exposure. It
should be noted that the highly alkaline pH (about pH 11) of sodium
hypochlorite is not likely to be encountered in drinking-water.
There have been relatively few evaluations of the effects of

chlorine or hypochlorite in drinking-water. The present review will
focus on studies with treatment periods longer than 4 weeks where
drinking-water was the primary route of exposure. Reference to earlier
studies of shorter duration and less general applicability to a safety
evaluation can be found in previous reviews (Bull, 1980, 1982a,b,
1992; Bull & Kopfler, 1991).
Daniel et al. (1990a) evaluated the toxicity of solutions of

chlorine prepared by bubbling chlorine gas into distilled water and
adjusting the pH to 9.4. The nominal concentrations of chlorine used
were 0, 25, 100, 175 or 250 mg/litre in distilled water (approximately
0, 3, 10, 16 or 21 mg/kg of body weight per day). These solutions
were provided as drinking-water to both male and female Sprague-
Dawley rats (10 per sex per dose) for 90 days. No deaths occurred in
any treatment group. However, there were statistically significant
decreases in drinking-water consumption in females treated with 100
mg/litre and higher, probably due to decreased palatability. There
were no consistent effects of chlorine treatment on organ to body
weight ratios or clinical chemistry parameters. A no-observed-effect
level (NOEL) of 10 mg/kg of body weight per day was identified by
the authors based on reduced body weight gain. However, since this
was associated with reduced palatability of the drinking-water, it is
not considered to be a true toxicological end-point.
The study in rats was followed up with another study in B6C3F1

mice (Daniel et al., 1991a). Male and female B6C3F1 mice (10 per sex
per group) were administered 12.5, 25, 50, 100 or 200 mg of chlorine
per litre of drinking-water for 90 days (calculated mean daily doses
84
Toxicology of Disinfectants
were 2.7, 5.1, 10.3, 19.8 or 34.4 mg/kg of body weight in males and
2.8, 5.8, 11.7, 21.2 or 39.2 mg/kg of body weight in females). Spleen
and liver weights were depressed in males, but not in females, at the
highest dose rates (100 and 200 mg/litre). There were no other con-
sistent indications of target organ effects based on serum enzyme
concentrations. No gross or microscopic lesions could be related to
treatment with chlorine.
Several of the following studies utilized solutions of sodium

hypochlorite as the treatment chemical. It is now known that such
solutions can contain very high concentrations of chlorate within a
short time of their preparation (Bolyard et al., 1993). The extent of
this contamination has not been reported.
Hasegawa et al. (1986) examined the effects of much higher

concentrations (0.025, 0.05, 0.1, 0.2 or 0.4%) of sodium hypochlorite
(equivalent to 7, 14, 28, 55 and 111 mg/kg of body weight per day)
administered in drinking-water to male and female F344 rats for
13 weeks. Twenty rats of each sex were assigned to each experimental
group. Significant suppression of body weight (as a result of decreased
consumption of water and food) occurred at 0.2% and above. The
authors noted slight damage to the liver as indicated by increased
levels of serum enzymes (not specified) at 0.2% and 0.4% sodium
hypochlorite in both sexes. No evidence of treatment-related
pathology was observed in this study or in a 2-year study in which
males were subjected to 0.05% or 0.1% (13.5 or 27.7 mg/kg of body
weight per day) and females to 0.1% or 0.2% (34 or 63 mg/kg of body
weight per day) sodium hypochlorite. The extended exposures were
conducted with 50 animals of each sex per treatment group. Analysis
of dosing solutions was not reported.
In a 2-year bioassay, the National Toxicology Program (NTP)

examined chlorine at 0, 70, 140 or 275 mg/litre (expressed as atomic
chlorine, Cl) in drinking-water of F344 rats and B6C3F1 mice (70 per
sex per group) (NTP, 1992). These solutions were prepared from
gaseous chlorine and neutralized to pH 9 by the addition of sodium
hydroxide. Stability studies indicated that 85% of the initial target
concentration remained after 3 days of preparation. Stock solutions
(concentrations not specified) were prepared once weekly, and solu-
tions for drinking were prepared 4 times weekly. Based on body
weight and water consumption, doses in these studies were
85
approximately 0, 4, 7 or 14 mg/kg of body weight per day for male

rats; 0, 4, 8 or 14 mg/kg of body weight per day for female rats; 0, 7,
14 or 24 mg/kg of body weight per day for male mice; and 0, 8, 14 or
24 mg/kg of body weight per day for female mice. The only treatment-
related non-tumour pathology was found to be a dilatation of renal
tubules in male mice receiving 275 mg/litre for more than 66 weeks.
No non-neoplastic lesions were observed in either male or female rats.
A number of immunological changes have been associated with

the treatment of rodents with sodium hypochlorite in drinking-water.
Water containing 25–30 mg of sodium hypochlorite per litre was
found to reduce the mean number of peritoneal exudate cells
recovered from female C57BL/6N mice after 2 weeks of treatment.
This was, in turn, associated with a significant decrease in
macrophage-mediated cytotoxicity to melanoma and fibrosarcoma cell
lines (Fidler, 1977). The treatment period was increased to 4 weeks
in a subsequent study, which demonstrated that 25 mg of sodium
hypochlorite per litre decreased the ability of peritoneal macrophages
to phagocytose 51Cr-labelled sheep red blood cells. Macrophages
obtained from the mice treated with hypochlorite were found to be less
effective in destroying B16-BL6 melanoma cells in vitro. Mice so
treated were also found to have increased pulmonary metastasis of
B16-BL6 cells when they were introduced by subcutaneous injection
(Fidler et al., 1982).
Exon et al. (1987) examined the immunotoxicological effects of

sodium hypochlorite at 5, 15 or 30 mg/litre (0.7, 2.1 or 4.2 mg/kg of
body weight per day) in the drinking-water of male Sprague-Dawley
rats (12 per dose) for 9 weeks. Delayed hypersensitivity reaction to
bovine serum albumin was observed at the highest dose administered.
Oxidative metabolism by adherent resident peritoneal cells was
decreased at 15 and 30 mg/litre, and the prostaglandin E2 levels of
these cells were found to be significantly elevated. No effects on
natural killer cell cytotoxicity, antibody responses, interleukin 2
production or phagocytic activity were observed. The effects on
macrophage activity suggest that some impairment does occur at
relatively low levels of sodium hypochlorite. As pointed out by the
authors, these were relatively mild effects, the significance of which
was unknown. It is not clear that these effects would be translated into
a significant impairment of the immune response to a particular
infectious agent. However, modification of macrophage function
appears to be one of the most sensitive responses identified in studies
86
of chlorine or hypochlorite in experimental animals. A study in which

female C57BL/6 mice were administered hypochlorite in their drink-
ing-water (7.5, 15 or 30 mg of hypochlorite per litre) for 2 weeks
showed no effects on the immune system as measured by spleen and
thymus weight, plaque-forming cell response, haemagglutination titre
and lymphocyte proliferation (French et al., 1998).
Altered liver lipid composition has been observed as a result of

acute intragastric administration of sodium hypochlorite (5 ml of a
1% solution) to rats (Chang et al., 1981). These data do not provide
a clear indication of whether these effects might give rise to
pathology. The concentrations of hypochlorite utilized were much
greater than those that would be encountered in drinking-water.
The effects of hypochlorous acid and hypochlorite on the skin

have received relatively little attention despite the current interest in
bathing as a significant source of chemical exposure from drinking-
water. Robinson et al. (1986) examined the effects of both
hypochlorite and hypochlorous acid solutions applied to the skin of
the entire body of female Sencar mice except for the head. Exposures
were to 1, 100, 300 or 1000 mg/litre as hypochlorous acid at pH 6.5
for 10 min on 4 consecutive days. Hypochlorite (formed by raising the
pH to 8.5) was studied only at 1000 mg/litre. Significant increases in
epidermal thickness and cell counts within the epidermal layer were
observed at concentrations of hypochlorous acid (pH 6.5) of 300
mg/litre and above, but the thickness of the skin was not significantly
different from that in animals at 100 mg/litre. The increases in skin
thickness were associated with an epidermis whose thickness was
increased to 4–6 cells as compared with the normal 1–2 cells seen in
mice. The effects of hypochlorite were much less marked. Following
a single application, the increased thickness of the skin observed in
mice exposed to hypochlorous acid (i.e., pH 6.5) did not appear until
4 days after the treatment. This differed from hypochlorite, other
disinfectants and the positive control, 12-O-tetradecanoylphorbol-13-
acetate (TPA). In the latter cases, the maximal response was observed
within 24–48 h after treatment. The hyperplastic response to
hypochlorous acid required 12 days to return to normal. This study
suggests a considerable margin of safety between the concentrations
of chlorine required to produce hyperplasia and those that are found
in drinking-water.
87
The reactive nature of chlorine always raises questions of

whether it is chlorine or a by-product that is responsible for any effect.
Several studies have examined the formation of by-products in the
gastrointestinal tract following the administration of chlorine.
Invariably, these studies have involved the administration of chlorine
or hypochlorite by gavage at very high concentrations relative to the
amounts that would be encountered in chlorinated drinking-water. As
a consequence, the by-products formed following gavage dosing of
high concentrations may not be representative of the by-products that
would be seen following the consumption of modest to moderate
levels of chlorine in larger volumes of water. A particular issue is that
the high organic carbon concentration relative to chlorine that would
be encountered in the gastrointestinal tract when water is consumed
at low concentrations should dissipate disinfectant before sufficient
oxidative power would be present to break down substrates to small
molecules. Despite these design flaws, the data do indicate that by-
products are formed. The bulk of them remain as higher molecular
weight products, which may have little toxicological importance.
Vogt et al. (1979) reported that chloroform could be measured in

the blood, brain, liver, kidneys and fat of rats to which sodium hypo-
chlorite was administered by gavage at doses of 20, 50 or 80 mg in
5 ml of water. Thus, the by-product chloroform can be formed by the
reaction of chlorine with stomach contents.
Mink and co-workers (1983) pursued this observation and found

that other by-products could be detected in the stomach contents and
plasma of rats that had been administered sodium hypochlorite
solutions neutralized to pH 7.9. In addition to chloroform, DCAN,
DCA and TCA were detected in the stomach contents analysis. DCA
and TCA were also detected in blood plasma.
The third group of compounds identified as by-products of

chlorination in stomach contents of the rat are the organic N-
chloramines (Scully et al., 1990). N-Chloroglycine, N-chloroleucine
or N-chloroisoleucine and N-chlorophenylalanine were confirmed
products of reactions with normal amino acids that would ordinarily
be found in the gastrointestinal tract. N-Chlorovaline and N-
chloroserine were also tentatively identified. Organic chloramines are
reactive and could be responsible for toxic effects that may be
attributed to chlorine in toxicological studies. The chlorine demand
88
of free amino acids in stomach contents was found to be only about

4% of the total. Consequently, this process may be substrate-limited
at concentrations of chlorine found as residuals in drinking-water.
However, use of higher concentrations of chlorine would also lead to
breakdown of proteins present in the stomach fluid. Thus, as
concentrations are increased to levels that would be used in animal
studies, these products would be formed at a much higher
concentration, similar to the phenomena noted with THM and HAA
by-products.
3.1.2 Reproductive and developmental toxicity
In general, animal studies have demonstrated no reproductive or

teratogenic effects of chlorine. Druckrey (1968) examined the effects
of water chlorinated to a level of 100 mg/litre (approximately 10
mg/kg of body weight per day) in BDII rats for seven generations. No
effects were observed on fertility, growth or survival.
A number of subsequent studies have studied the effects of

chlorine or hypochlorite on more specific aspects of reproduction or
development. Meier et al. (1985b) reported that oral administration
of sodium hypochlorite (pH 8.5) prepared from chlorine gas and
administered at 4 or 8 mg/kg of body weight per day for 5 weeks
increased the incidence of sperm head abnormalities in B6C3F1 mice
(10 animals per group). The effect was not observed when the
solutions were administered at pHs at which hypochlorous acid was
the predominant species (pH 6.5). However, other studies have not
been able to associate adverse reproductive outcomes with the
administration of chlorine or sodium hypochlorite.
Carlton et al. (1986) found no evidence of sperm head abnor-

malities or adverse reproductive outcomes in Long-Evans rats. Male
rats were treated for 56 days prior to mating and female rats from
14 days prior to mating through gestation. Each experimental group
consisted of 11–12 males and 23–24 females. Solutions of chlorine
were prepared at pH 8.5, so the study evaluated hypochlorite as the
dominant form in the drinking-water. Doses were as high as 5 mg/kg
of body weight per day.
89
3.1.3 Toxicity in humans
There have been significant human exposures to chlorine and

hypochlorite solutions. Much of that experience is with inhalation of
chlorine gas, which is known to be a strong respiratory irritant.
Chlorine gas is also the largest single component involved in toxic
release incidents. A third major source of exposure is solutions of
sodium hypochlorite, usually marketed as bleach. Bleach is frequently
involved in human poisonings. These exposures are not particularly
relevant to exposures to chlorine or hypochlorite in drinking-water.
Therefore, only a few case reports are identified that illustrate the
types of problems that have been encountered. There was no attempt
to make this review comprehensive.
The irritating effects of chlorine gas have been well documented

because of its use as a chemical warfare agent during World War I
(Das & Blanc, 1993). In a follow-up of survivors of gassing, it was
concluded that there was no evidence of permanent lung damage;
however, these studies clearly indicated that survivors had breath
sounds that suggested bronchitis and limited chest and diaphragmatic
movement, even emphysema. Most studies suggested that there were
high incidences of acute respiratory disease and a lesser prevalence of
chronic sequelae. Similar sequelae have been identified following
exposure of humans to accidental releases of chlorine gas. In these
more modern characterizations, the acute signs and symptoms
included a high incidence of pulmonary oedema and severe
bronchitis. These signs and symptoms are of generally short duration
and resolve themselves over the course of about 1–4 weeks. However,
chronic sequelae are observed in some individuals, depending in part
upon the severity of the exposure. In such cases, a decrease in the
forced expiratory volume is the most consistently reported clinical
sign.
Two recent reports suggest that chronic sequelae to acute expo-

sures to chlorine gas may be more prevalent than previously
appreciated. Moore & Sherman (1991) reported on an individual who
was previously asymptomatic and who developed chronic, recurrent
asthma after exposure to chlorine gas in an enclosed place. Schwartz
et al. (1990) followed 20 individuals who had been exposed to
chlorine gas in a 1975 incident. The prevalence of low residual lung
volume was increased during the follow-up period. Sixty-seven per
cent of those tested were found to have residual volumes below 80%
90
of their predicted values. Five of 13 subjects tested for airway

reactivity to methacholine were found to have a greater than 15%
decline in forced expiratory volume.
Controlled studies have been conducted in healthy, non-smoking

men exposed to chlorine gas at 1.5 and 2.9 mg/m3 (0.5 and 1.0 ppm)
for 4 or 8 h (reviewed in Das & Blanc, 1993). Four hours of exposure
to 2.9 mg/m3 (1.0 ppm) produced significant decreases in the forced
expiratory volume. One individual who was found to be experiencing
more difficulty than other subjects at this dose and who was later
exposed to 1.5 mg/m3 (0.5 ppm) experienced a significant decrease in
forced expiratory volume. While 1.5 mg/m3 (0.5 ppm) appears protec-
tive for most people, some more sensitive individuals may in fact have
more significant responses to chlorine gas.
The effects of chronic exposure to chlorine gas have received

only limited study. In one study of paper mill workers, a more rapid
age-related decrease in lung volumes of workers exposed to chlorine
relative to those exposed to sulfur dioxide was noted, but the trend
was not statistically significant (Das & Blanc, 1993). Other studies
failed to identify chronic sequelae.
There are frequent reports of human poisonings from bleach.

Most often these exposures result from the mixing of bleach with
acidic products or ammonia. Acidification converts hypochlorite to
hypochlorous acid, which dissociates to chlorine gas, offgasses very
rapidly from the solutions and presents an inhalation exposure
(MMWR, 1991). Mixing bleach with ammonia results in the
formation of monochloramine and dichloramine, both of which are
effective respiratory irritants (MMWR, 1991).
Any potential effects of chlorine or hypochlorite in drinking-

water are obscured by the fact that by-products inevitably coexist with
the residual chlorine. One series of studies in which by-products
formed were minimized by dissolving chlorine in distilled water
attempted to identify effects of chlorine in drinking-water on humans.
Chlorine in drinking-water was administered in a rising-dose
tolerance study beginning with 0.1 mg/litre in two 500-ml portions
and rising to a concentration of 24 mg/litre, equivalent to 0.34 mg/kg
of body weight per day (Lubbers & Bianchine, 1984). No clinically
important changes were observed. No findings of clinical importance
91
were identified in a follow-up treatment with repeated dosing with

500-ml portions of a solution containing 5 mg of chlorine per litre for
a 12-week period (Lubbers et al., 1984a).
Another study attempted to determine whether consumption of

chlorinated drinking-water affected blood cholesterol levels (Wones
et al., 1993a). The impetus for this study was a toxicological study in
pigeons that suggested that chlorine raised blood cholesterol levels
and modified serum thyroid levels (Revis et al., 1986a,b) and an
epidemiological study that associated small increases in cholesterol of
women with residence in communities having chlorinated water
(Zeighami et al., 1990a,b; described in detail in section 5.2.2). A prior
study (quoted in Wones et al., 1993a) was conducted that examined
men who consumed water containing 2, 5 or 10 mg of chlorine per
litre and found a small increase in serum cholesterol levels at the
highest dose group. However, no control group was studied, so the
changes could have been attributed to the change in diet imposed as
part of the study protocol (Wones & Glueck, 1986). The longer-term
study was composed of 30 men and 30 women who received a
controlled diet for the duration of the study. The first 4 weeks
represented an acclimatization period during which all subjects
received distilled water. Half the subjects were assigned to a group
that consumed 1.5 litres of water containing 20 mg of chlorine per
litre for the following 4 weeks. At the end of each 4-week period,
blood was analysed for cholesterol, triglycerides, high-density
lipoprotein (HDL) cholesterol, low-density lipoprotein (LDL)
cholesterol or apolipoproteins A1, A2 and B. There were no
significant effects. There was a slight trend towards lower thyroid
hormone levels in men consuming chlorine, but this was not clinically
significant (Wones et al., 1993a). These data suggest that observations
obtained previously in pigeons could not be repeated under
comparable conditions of chlorine consumption. It is notable that the
animals utilized in the original pigeon study had consumed a
modified diet (Revis et al., 1986a) that was deficient in calcium and
other trace metals. A subsequent study failed to replicate the previous
results in pigeons (Penn et al., 1990).
3.1.4 Carcinogenicity and mutagenicity
The International Agency for Research on Cancer (IARC) has

evaluated the carcinogenicity of hypochlorite salts and concluded that
there were no data available from studies in humans on their
92
carcinogenicity and inadequate evidence for their carcinogenicity in

experimental animals. Hypochlorite salts were assigned to Group 3:
the compounds are not classifiable as to their carcinogenicity to
humans (IARC, 1991).
Several studies have shown that sodium hypochlorite produces

mutagenic responses in bacterial systems and mammalian cells in
vitro. However, there is no evidence of activity in mammalian test
systems in vivo. It is not clear to what extent this is influenced by the
formation of mutagenic by-products as a result of reactions with
components of the incubation media. Wlodkowski & Rosenkranz
(1975) used short-term exposures of Salmonella typhimurium strain
TA1530 followed by ascorbic acid-induced decomposition to reduce
the cytotoxic effects of hypochlorite. The investigators applied 0.14
:mol per tube and added ascorbic acid after intervals of 5, 10 and 15
min. At 5 min, a clear positive response was observed with minimal
cytotoxicity. Significant responses were also observed in strain
TA1535, but not in strain TA1538.
Rosenkranz (1973) and Rosenkranz et al. (1976) also demon-

strated a positive mutagenic response in DNA polymerase A deficient
Escherichia coli to 0.006 :mol of sodium hypochlorite. This response
was unaffected by the addition of catalase, suggesting that the
response was not related to the generation of hydrogen peroxide.
Matsuoka et al. (1979) reported that sodium hypochlorite at a

concentration of 6.7 mmol/litre (0.5 mg/ml) produced chromosomal
aberrations in Chinese hamster ovary (CHO) cells in the presence of
S9 mix. This concentration was cytotoxic in the absence of S9. Some
concern must be expressed about whether responses observed with
such high and clearly cytotoxic concentrations in an in vitro system
represent specific clastogenic effects. The authors report only one
concentration tested with and without S9. It is probable that the
positive response in the presence of S9, if it is a specific response, was
a result of detoxifying hypochlorite. This protection could be non-
specific as well, in that it may not have depended upon any catalytic
activities present in the S9 fraction (i.e., the added protein may have
acted as a reactive sink to dissipate excess hypochlorite). Conse-
quently, it is difficult to use these data in interpreting the effects of
chlorine or hypochlorite in vivo.
93
Ishidate (1987) studied the induction of chromosomal aberrations

in cultures of Chinese hamster CHL cells at sodium hypochlorite
concentrations ranging from 125 to 500 :g/ml without exogenous
metabolic activation and from 31 to 125 :g/ml with and without rat
liver S9 mix. A clear increase in the number of cells with structural
chromosomal aberrations was observed at 500 :g/ml without S9 mix,
while the results obtained in the other series, showing weakly positive
responses, were considered inconclusive.
Meier et al. (1985b) evaluated the ability of hypochlorite and

hypochlorous acid to induce chromosomal damage or micronuclei in
the bone marrow of CD-1 mice. The samples to be tested were
generated by bubbling chlorine gas into water and then adjusting the
pH to 6.5 (predominantly hypochlorous acid) or 8.5 (predominantly
hypochlorite). The doses administered were 1.6, 4 or 8 mg/kg of body
weight for 5 consecutive days. There was no evidence of increased
micronuclei or chromosomal abnormalities in bone marrow cells.
Significant positive responses were observed with positive control
chemicals in both assays. As reported in section 3.1.2, these authors
detected a positive response in the sperm head abnormality assay in
mice treated at these same doses of hypochlorite in two separate
experiments. This assay is used primarily as a mutagenicity assay
rather than as an assay for reproductive toxicities. Hypochlorous acid
had no effect in the sperm head abnormality assay.
Tests of the ability of hypochlorite to induce cancer in rodents

were conducted in F344 rats by Hasegawa et al. (1986). Sodium
hypochlorite concentrations of 0, 500 or 1000 mg/litre (males) and 0,
1000 or 2000 mg/litre (females) were administered in the drinking-
water for 104 weeks (equivalent to 13.5 and 27.7 mg/kg of body
weight per day for males and 34.3 and 63.2 mg/kg of body weight per
day for females). There were 50 male and 50 female rats assigned to
each experimental group. No tumours could be attributed to sodium
hypochlorite administration.
NTP (1992) conducted a 2-year bioassay of chlorine in F344 rats

and B6C3F1 mice. The concentrations administered in drinking-water
were 0, 70, 140 or 275 mg/litre, and there were 70 animals of each
sex assigned to each group (approximately 0, 4, 8 or 14 mg/kg of body
weight per day for rats and 0, 7, 14 or 24 mg/kg of body weight per
94
day for mice). There was an apparent positive trend in the induction
of stromal polyps of the uterus of female mice treated with chlorine,
but this was considered unlikely to be treatment-related because the
incidence was below those observed in historical controls. In female
rats, there was an increase in mononuclear cell leukaemia at both 140
and 275 mg/litre (8 and 14 mg/kg of body weight per day). However,
the response was not considered treatment-related because it fell
within the range of historical controls, there was no apparent
dose–response, and there was no evidence for such an increase in
male F344 rats.
A single study suggested that sodium hypochlorite could act as

a promoter of skin tumours following initiation with 4-nitroquinoline-
1-oxide in female ddN mice (Hayatsu et al., 1971). A solution of
sodium hypochlorite that contained 10% effective concentrations of
chlorine was utilized. Skin tumours were produced in 9 of 32 mice
given 45 applications of sodium hypochlorite following initiation.
Sodium hydroxide solutions were utilized as a control for the alkaline
pH of sodium hypochlorite and produced no tumours. No tumours
were observed with 60 applications of sodium hypochlorite solution
in non-initiated mice. Pfeiffer (1978) conducted a much larger
experiment that utilized 100 mice per group. This author found that
a 1% sodium hypochlorite solution applied alternately with
benzo[a]pyrene for 128 weeks was ineffective in producing skin
tumours in female NMRI mice above those that had been initiated
with benzo[a]pyrene alone at doses of 750 or 1500 :g. Pretreatment
with the sodium hypochlorite solution before application of the
benzo[a]pyrene actually reduced tumour yields at 128 weeks with
doses of either 750 or 1500 :g of benzo[a]pyrene. Sodium
hypochlorite used in a more traditional initiation/promotion study
(i.e., sodium hypochlorite treatment following initiation with
benzo[a]pyrene) produced a decrease in the tumour yield with the 750
:g dose of benzo[a]pyrene, but had no effect following 1500 :g.
Thus, the ability of sodium hypochlorite to act as a tumour promoter
may depend upon the initiator used, or the smaller experiment of
Hayatsu et al. (1971) may simply be a false result.
As pointed out in section 3.1.1, application of solutions of hypo-

chlorous acid to the skin of Sencar mice results in the development of
hyperplasia. The concentrations required are considerably lower
(300 mg/litre) (Robinson et al., 1986) than those used in the studies
of either Hayatsu et al. (1971) or Pfeiffer (1978). Sodium hypochlorite
95
was also effective at lower doses, but less so than equivalent con-
centrations of hypochlorous acid. These results suggest that these
prior evaluations may have been conducted at too high a dose. There
appear to be no reports on the effectiveness of hypochlorous acid as a
tumour promoter, but the lack of activity at doses of less than 300
mg/litre would suggest that this is of no concern.
3.1.5 Comparative pharmacokinetics and metabolism
A series of pharmacokinetic studies using 36Cl-labelled hypo-

chlorous acid were conducted by Abdel-Rahman and co-workers
(1983). These studies are of limited value because the form of 36Cl
could not be determined in various body compartments.
3.1.6 Mode of action
There are no specific toxicities of chlorine for which a

mechanism needs to be proposed. It is a strong oxidizing agent, and
it must be presumed that damage induced at high doses by either
gaseous chlorine or solutions of hypochlorite is at least partially
related to this property. In studies in which sodium hypochlorite is
used without neutralization, a strong alkaline pH can also contribute
to its effects. There is always the possibility that chlorine is inducing
subtle effects by virtue of its reaction with organic compounds that are
found in the stomach. Such reactions have been demonstrated, but
there is no convincing evidence to date that any specific toxicity can
be attributed to these by-products.
3.2 Chloramine

response in animals
There have been relatively few evaluations of the toxic properties

of chloramine in experimental animals. In large part this is because
it is not marketed as a product but is created for disinfection purposes
on-site and in situ. Chloramine is primarily used as a residual
disinfectant in the distribution system. The final solution consists of
mostly monochloramine, with traces of other chloramines, such as
dichloramine. Chloramines, as a group, are generally recognized as
potent respiratory irritants, because the formation of these compounds
when household bleach and ammonia are mixed results in a number
96
of poisoning cases each year (MMWR, 1991). In spite of this, there

has been no attempt to quantify dose–response relationships in
animals.
Eaton et al. (1973) investigated concerns about chloramine-

induced methaemoglobin formation in kidney patients dialysed with
chloramine-containing water. This was done by examining the ability
of relatively large volumes of tapwater to oxidize haemoglobin in
dilute suspensions of red blood cells. This circumstance is reflective
of dialysis, but not of the concentrated suspension of these cells in
vivo. Nevertheless, the authors were able to show that
methaemoglobin formation occurred in a dose-related manner when
1 volume of red blood cells (human) was suspended in 100 volumes
of tapwater containing 1 mg of chloramine per litre or above. This
effect was not produced by comparable concentrations of sodium
hypochlorite. The ability to induce methaemoglobin formation was
eliminated by treating the water by reverse osmosis followed by
carbon filtration. Clearly, chloramine is capable of inducing
methaemoglobinaemia at low concentrations when there is a large
reservoir of chloramine. This is a decidedly different exposure pattern
from that of normal humans and other mammals, as they consume
small volumes of water relative to the volume of red blood cells that
are exposed.
Moore et al. (1980a) studied alterations of blood parameters in

male A/J mice treated with 0, 2.5, 25, 50, 100 or 200 mg of mono-
chloramine per litre in carbonate/bicarbonate-buffered (pH 8.9) drink-
ing-water. Twelve animals were assigned to each group, and
treatments were maintained over a 30-day period. Consistent with the
interpretation provided above, there were no treatment-related effects
on osmotic fragility, methaemoglobin levels, haemoglobin concen-
trations, reticulocyte counts or a number of other derived parameters.
Haematocrits of mice treated with 50, 100 or 200 mg/litre were
actually higher than those observed in control mice. White blood cell
counts were not altered in these animals.
Daniel et al. (1990a) conducted a more traditional 90-day study

of monochloramine in Sprague-Dawley rats (10 animals per sex per
dose). Treatment concentrations were 0, 25, 50, 100 or 200 mg/litre,
corresponding to doses of 0, 1.8, 3.4, 5.8 or 9.0 mg/kg of body weight
per day in males and 0, 2.6, 4.3, 7.7 or 12.1 mg/kg of body weight per
day in females. Controls received carbonated, pH-adjusted drinking-
water. A large number of haematological and clinical chemistry
97
measures were included in the evaluation. Body weights were signi-

ficantly depressed in both sexes at treatment concentrations in the
50–200 mg/litre range, but this appeared to be related to depressed
water and food consumption. There were minor changes in organ to
body weight ratios at the highest dose, but no evidence of treatment-
related pathology was observed. Male rats were found to have
decreased haematocrits at 100 mg/litre, and red blood cell counts were
slightly depressed at 100 and 200 mg/litre. The authors concluded
that monochloramine was more toxic than chlorine or chlorine
dioxide. However, it must be noted that the changes in blood
parameters were small and in themselves of no clinical significance.
Other measures were not related to specific toxic reactions. Based on
the decrease in organ and body weights observed in both sexes, the
authors concluded that the no-observed-adverse-effect level (NOAEL)
was 100 mg/litre, equivalent to 5.8 mg/kg of body weight per day.
The work in rats was followed up with a second study in B6C3F1

mice (Daniel et al., 1991a). Male and female B6C3F1 mice (10 per sex
per group) were administered 0, 12.5, 25, 50, 100 or 200 mg of chlor-
amine per litre of drinking-water for 90 days (calculated mean daily
dose was 0, 2.5, 5.0, 8.6, 11.1 or 15.6 mg/kg of body weight for males
and 0, 2.8, 5.3, 9.2, 12.9 or 15.8 mg/kg of body weight for females).
Water consumption significantly decreased at 100 and 200 mg/litre
in males and at 25–200 mg/litre in females. Weight gain was
significantly depressed in both sexes at 100 and 200 mg/litre.
Neutrophil concentrations in blood were significantly depressed in
both male and female mice at the two highest doses, but other white
blood cell counts were unaltered. Absolute and relative spleen and
liver weights were depressed at both 100 and 200 mg/litre. No gross
or microscopic evidence of target organ toxicity was observed that
could be related to treatment. Based on decreased organ weights,
weight gain, and food and water consumption, the authors concluded
that the NOAEL was 50 mg/litre, equivalent to 8.6 mg/kg of body
weight per day.
A 13-week study in which groups of 10 Sprague-Dawley rats

were given drinking-water containing 200 mg of monochloramine per
litre or buffered water as a control, ad libitum or restricted to the same
consumption as the monochloramine group, was designed to resolve
some of the outstanding toxicological questions. The results of this
study indicated that the reduced body weight gain and the minor
biochemical, haematological, immunological and histopathological
98
changes associated with exposure to 200 mg of monochloramine per

litre (equivalent to 21.6 mg/kg of body weight per day) in drinking-
water were largely related to reduced water and food consumption
(Poon et al., 1997).
In a 9-week study, Exon et al. (1987) examined the ability of

monochloramine to modify immunological parameters in male
Sprague-Dawley rats (12 per dose) exposed to concentrations of 0, 9,
19 or 38 mg of monochloramine per litre, equivalent to 0, 1.3, 2.6 or
5.3 mg/kg of body weight per day. At the middle and highest dose,
chloramine treatment was observed to increase prostaglandin E2 syn-
thesis by adherent resident peritoneal cells (which include macro-
phages) in response to lipopolysaccharide stimulation. No attempt was
made to relate this finding to other indices of modified macrophage
function. A small depression in spleen weights was observed at the
highest dose. The implications of these data for immune function are
not clear. Other measures of immune function did not reveal statis-
tically significant changes with treatment. These included a decrease
in antibody formation at the lowest and middle doses in response to
keyhole limpet haemocyanin injection or delayed-type hypersensitivity
reactions to bovine serum albumin injected into the footpad.
The effect of monochloramine on skin irritation was tested by

immersing Sencar mice into water containing chloramine at concen-
trations ranging from 1 to 1000 mg/litre for 10 min a day (Robinson
et al., 1986). Unlike hypochlorous acid (pH 6.5) or hypochlorite (pH
8.5), chloramine did not produce hyperplasia of the skin.
Studies in laboratory animals have indicated no reproductive or

developmental effects associated with chloramine. Abdel-Rahman et
al. (1982a) administered monochloramine to female Sprague-Dawley
rats at concentrations of 0, 1, 10 or 100 mg/litre (0, 0.15, 1.5 or
15 mg/kg of body weight per day) for 2.5 months prior to breeding
and through gestation. Only six animals were assigned to each
treatment group. Reproductive performance was comparable between
groups, and fetal weights were not adversely affected by treatment.
Between 50 and 60 fetuses were available (male and female
combined) for evaluation. There was no evidence of treatment-related
skeletal or soft tissue anomalies.
99
Carlton et al. (1986) examined the effects of monochloramine

administered by gavage at doses of 0, 2.5, 5 or 10 mg/kg of body
weight per day on the reproductive performance of Long-Evans rats.
Males (12 per group) were treated from 56 days prior to and through
mating, and females (24 per group) from 14 days prior to mating and
throughout the mating period. No statistically significant effects on
sperm morphology, concentration or motility were observed, nor were
there any effects on fertility, viability, litter size, pup weights, day of
eye opening or day of vaginal patency.
The primary harmful effects of chloramine have been

documented in humans poisoned by chloramine formed when
household bleach was mixed with ammonia for use as a cleaning
solution. Chloramine is a strong respiratory irritant. These effects
were discussed in section 3.1.3.
Forty-eight men completed an 8-week protocol during which diet

and other factors known to affect lipid metabolism were controlled.
During the first 4 weeks of the protocol, all subjects consumed
distilled water. During the second 4 weeks, one-third of the subjects
were assigned randomly to drink 1.5 litres of water containing 0, 2 or
15 mg of monochloramine per litre each day. At 2 mg/litre, no
significant changes were observed in total, HDL or LDL cholesterol,
triglycerides or apolipoproteins A1, A2 or B. Parameters of thyroid
function were unchanged. However, an increase in the level of
apolipoprotein B was observed at 15 mg/litre (Wones et al., 1993b).
Shih & Lederberg (1976) first demonstrated that

monochloramine induced mutation in a Bacillus subtilis reversion
assay. The concentration range studied extended from 18 to 74
:mol/litre. A positive dose–response was observed through 56
:mol/litre, but 74 :mol/litre was cytotoxic. Repair-deficient mutants
of B. subtilis, rec3, recA and polyA5, were consistently more sensitive
to the cytotoxic effects of chloramine, while the uvr and recB mutants
were not. The sensitivity of the polyA5 mutants parallels the
observations of Rosenkranz (1973) with sodium hypochlorite and
suggests that DNA polymerase A is involved in the repair of DNA
lesions produced by both chemicals. Thus, it is possible that a
100
common intermediate or mechanism is involved in the mutagenic

effects of hypochlorite and chloramine.
A broader list of chloramines was tested by Thomas et al. (1987)

in Salmonella typhimurium tester strains TA97a, TA100 and TA102.
The chloramines tested included those that could be formed at low
levels from natural substrates in drinking-water or in the stomach.
TA100 was found to consistently be the most sensitive strain. The
most potent mutagens were the lipophilic dichloramines formed with
histamine, ethanolamine and putrescine. The corresponding mono-
chloramines were less potent. The more hydrophilic chloramines,
such as taurine-chloramine, had little activity. Monochloramine was
active in the 50 :mol/litre range, remarkably consistent with the data
of Shih & Lederberg (1976). Hypochlorous acid was inactive at all
concentrations that were tested, up to and including concentrations
that induced cytotoxicity.
Ashby and co-workers (1987) were unable to induce clastogenic

effects in the mouse bone marrow micronucleus assay when
chloramine was administered orally. They suggested that the in vitro
clastogenic effects were probably attributable to non-specific cytotoxic
effects that are secondary to the release of hypochlorite to the media.
Meier et al. (1985b) found that intraperitoneal administration of

monochloramine to CD-1 mice at doses of up to 8 mg/kg of body
weight was without significant effect on either micronuclei or
chromosomal aberrations in the bone marrow. These data would
appear to be consistent with the findings of Ashby et al. (1987).
Studies on the carcinogenicity of chloramine are limited to a

single set of 2-year experiments conducted by the NTP (1992).
Drinking-water containing 0, 50, 100 or 200 mg of chloramine per
litre was provided to F344 rats and B6C3F1 mice. Seventy animals of
each species and of each sex within a species were assigned to each
experimental and control group. Doses in rats were 0, 2.1, 4.8 or
8.7 mg/kg of body weight per day in males and 0, 2.8, 5.3 or 9.5
mg/kg of body weight per day in females; doses in mice were 0, 5.0,
8.9 or 15.9 mg/kg of body weight per day in males and 0, 4.9, 9.0 or
17.2 mg/kg of body weight per day in females. Of some interest was
the finding that two renal cell adenomas were found in male B6C3F1
101
mice treated with the high dose of chloramine. In addition, one renal
adenoma was found in one male mouse treated with 100 mg/litre and
in one female mouse treated with 200 mg/litre. While this tumour site
is rare in both species, there was no real dose–response trend, nor
were the differences between the control and treatment groups
statistically significant. A second finding of some concern was an
increase in the incidence of mononuclear cell leukaemia in F344 rats.
This pathology was increased in rats treated with chloramine or
hypochlorite, although the effects were not clearly dose-dependent.
The incidence of mononuclear cell leukaemia was significantly
greater than in concurrent controls and was elevated above the
historical incidence as well. Nevertheless, these increases were not
considered to be treatment-related. In part, this conclusion arose from
the lack of a clear dose–response. It was also based on the fact that
there was no comparable trend in male rats.
The pharmacokinetics of 36Cl derived from monochloramine have

been examined in male Sprague-Dawley rats (Abdel-Rahman et al.,
1983). These data are difficult to interpret because the specific form
of the label is not known.
3.3 Chlorine dioxide

response in animals
Despite its use as a disinfectant, there have been very few general
toxicological evaluations of chlorine dioxide, because most studies
have focused on its major by-product, chlorite, which is considered in
section 4.6. The present review will first focus on the limited charac-
terization of chlorine dioxide’s general toxicology, then follow up
with a discussion of its haematological and thyroid effects.
Some very cursory investigations of chlorine dioxide’s effects as

a respiratory irritant were published by Haller & Northgraves (1955)
in an article dealing with the general chemical properties of the
compound. In essence, these data suggested that exposure to chlorine
dioxide in air at a concentration of more than 420 mg/m3 (150 ppm)
for
102
longer than 15 min was fatal to guinea-pigs. The total study involved
six guinea-pigs.
Rats (3–5 per group) were exposed to chlorine dioxide in various

concentrations (0.28–9520 mg/m3 [0.1–3400 ppm]) and for various
periods (3 min–10 weeks). All the rats exposed daily to chlorine
dioxide at 28 mg/m3 (10 ppm) died in less than 14 days. Purulent
bronchitis and disseminated bronchopneumonia were found at necros-
copy. No such changes were demonstrable in rats exposed to approxi-
mately 0.28 mg/m3 (0.1 ppm) for about 10 weeks (Dalhamn, 1957).
The LC50 of chlorine dioxide in rats (5 per sex per group)

exposed by inhalation for 4 h was 90 mg/m3 (32 ppm) (Ineris, 1996).
Robinson et al. (1986) studied the ability of chlorine and alter-

native disinfectants to induce epidermal hyperplasia in the skin of
Sencar mice. The thickness of the interfollicular epidermis was signi-
ficantly increased by 10-min daily exposures to water containing up
to 1000 mg of chlorine dioxide per litre for 4 days. The thickness of
the epidermis was similar to that induced by an equivalent dose of
hypochlorous acid. Unlike hypochlorous acid, however, there was no
significant increase in skin thickness at concentrations of 300 mg/litre
or less.
The study of Daniel et al. (1990a) was the first subchronic study
that adhered to modern expectations of toxicological studies. These
authors provided male and female Sprague-Dawley rats (10 per sex
per treatment group) with 0, 25, 50, 100 or 200 mg of chlorine
dioxide per litre of drinking-water for 90 days, equivalent to 0, 2, 4,
6 or 12 and 0, 2, 5, 8 or 15 mg/kg of body weight per day for males
and females, respectively. Conventional measures of body weight,
organ weights, a broad battery of clinical chemistry parameters and
histopathological examinations were all included in the study design.
Body and organ weights were significantly depressed at 200 mg/litre
in both sexes. This appeared to be secondary to depressed water
consumption, which is known to be tightly coupled to food
consumption in rats. The only significant histopathological damage
found was goblet cell hyperplasia and inflammation. This was
observed at all doses of chlorine dioxide in both male and female rats.
Presumably this inflammation occurs as a result of volatilization of
103
chlorine dioxide from the water bottle. The amount of chlorine

dioxide actually inhaled as a result of volatilization from the drinking-
water containing the lowest dose of chlorine dioxide (25 mg/litre)
must be extremely low. This suggests that there might be some
concern for sensitive individuals showering with water containing
chlorine dioxide.
The ability of chlorine dioxide to induce methaemoglobinaemia

and haemolytic anaemia has received extensive study. Abdel-Rahman
et al. (1980) found decreased red blood cell glutathione (GSH)
concentrations and decreased osmotic fragility in Sprague-Dawley rats
and white leghorn chickens given drinking-water containing chlorine
dioxide concentrations of 1, 10, 100 and 1000 mg/litre for up to
4 months, but the changes were not consistently dose-related.
However, the authors found that the morphology of red blood cells
was modified (codocytes and echinocytes) in all dose groups, the
severity increasing with increased treatment concentration.
Methaemoglobin was not detected throughout these studies. However,
there was no formal statistical evaluation of these results, and only
four rats were assigned to each experimental group. Administration
of acute doses of as little as 1 mg/kg of body weight by gavage
decreased red blood cell GSH concentrations. This response was not
increased as dose was increased to 4 mg/kg of body weight.
Abdel-Rahman and co-workers extended these observations to

longer treatment periods in a subsequent publication (Abdel-Rahman
et al., 1984a). Again, only four animals were assigned to each
treatment group. At 7 and 9 months of treatment, red blood cells
appeared to become resistant to osmotic shock at all treatment
concentrations (1–1000 mg/litre). These data did not display a clear
dose–response despite the large variation in the dose administered.
The study of Abdel-Rahman et al. (1984a) also reported changes

in the incorporation of 3H-thymidine into the DNA of various organs.
Incorporation was significantly inhibited in testes and apparently
increased in the intestinal mucosa. The effect on apparent DNA
synthesis was particularly marked in the testes at 100 mg/litre,
amounting to about 60% inhibition. These data are difficult to
interpret for several reasons. First, rats were sacrificed 8 h after being
injected with 3H-thymidine. Ordinarily, sacrifices are made 30–60
min after injection because the blood is essentially depleted of 3H-
thymidine in an hour. Thus, it is not possible to determine if the
lowered amount of label is related to decreased synthesis or to
104
increased turnover of DNA. Second, the result was based on total

counts in DNA, which makes it impossible to determine what cell
type is affected or whether the change was associated with replicative
or repair synthesis. Third, only four animals were used per
experimental group.
In a rising-dose protocol study, Bercz et al. (1982) evaluated the

effects of chlorine dioxide on African green monkeys. These animals
were provided chlorine dioxide in drinking-water at concentrations of
0, 30, 100 or 200 mg/litre, corresponding to doses of 0, 3.5, 9.5 or
11 mg/kg of body weight per day. Each dose was maintained for
30–60 days. Animals showed signs of dehydration at the highest dose
(11 mg/kg of body weight per day), so exposure at that dose was dis-
continued. No effect was observed on any haematological parameter,
including methaemoglobinaemia. However, statistically significant
depressions in serum thyroxine levels were observed when animals
were dosed with chlorine dioxide at a concentration of 100 mg/litre.
No effect had been observed in a prior exposure of the same animals
to 30 mg/litre for 30 days. The NOAEL in this study was 3.5 mg/kg
of body weight per day.
The effects of chlorine dioxide on thyroid function were followed

up by Harrington et al. (1986). Thyroxine levels in African green
monkeys administered drinking-water containing 100 mg of chlorine
dioxide per litre (4.6 mg/kg of body weight per day) were again found
to be depressed at 4 weeks of treatment, but rebounded to above-
normal levels after 8 weeks of treatment. These investigators also
found significantly depressed thyroxine levels in rats treated with 100
or 200 mg of chlorine dioxide per litre in drinking-water (equivalent
to 14 and 28 mg/kg of body weight per day) for 8 weeks. This change
was dose-related. Lower doses were not examined in the rat study.
The authors indicated that the results were based on 12
determinations; it was not clear if these measurements were made on
individual animals.
A set of in vivo and in vitro experiments was conducted in an

attempt to explain the effects of chlorine dioxide on serum thyroid
hormone concentrations. The authors demonstrated that chlorine
dioxide oxidizes iodide to reactive iodine species that would bind to
the stomach and oesophageal epithelium. Rat chow that was treated
with chlorine dioxide at approximately 80 mg/litre was found to
increase the binding of iodine to chow constituents. This activation of
iodine resulted in retention of labelled iodine in the ileum and colon
105
and reduced uptake by the thyroid gland. Previous work demonstrated

that chlorine dioxide was more effective than chlorine in activating
iodide to a form that would covalently bind with a variety of natural
foodstuffs (Bercz et al., 1986). Based on these observations, the
authors concluded that the effects of chlorine dioxide were probably
due to altered gastrointestinal absorption of iodide and reduced uptake
into the thyroid gland.
A number of reproductive effects have been reported in studies

with laboratory animals, but the relevance for humans of these
findings remains uncertain. The reproductive effects of chlorine
dioxide in Long-Evans rats were studied by Carlton et al. (1991).
Chlorine dioxide was administered by gavage at doses of 2.5, 5 or 10
mg/kg of body weight per day to male rats (12 per group) for 56 days
prior to and through mating and to female rats (24 per group) from 14
days prior to mating and through pregnancy. Fertility measures were
not significantly different among the dose groups. There were no
dose-related changes in sperm parameters (i.e., concentration,
motility, progressive movement or morphology). Thyroid hormone
levels were altered significantly, but not in a consistent pattern. The
only significant difference was significantly depressed vaginal weights
in female pups whose dams had been treated with 10 mg/kg of body
weight per day.
An evaluation of the effects of chlorine dioxide on the fetal

development of Sprague-Dawley rats was conducted by Suh et al.
(1983). Chlorine dioxide was administered at 0, 1, 10 or 100 mg/litre
(0, 0.1, 1 or 10 mg/kg of body weight per day) for 2.5 months prior to
mating and throughout gestation. The total number of implants per
dam was significantly reduced at the highest concentration of chlorine
dioxide. The percentage of anomalous fetuses was increased in a dose-
related manner, but the response was not statistically significant.
These anomalies arose primarily as the percentage of abnormal or
incomplete sternebrae in treated rats relative to controls. The lack of
statistical significance was undoubtedly related to the relatively few
female rats that were included in the study (6–8 females per treatment
group). As a consequence, the results of this study must be considered
inconclusive.
106
Orme et al. (1985) found that chlorine dioxide administered in

the drinking-water of female Sprague-Dawley rats (13–16 per dose)
at concentrations of 0, 2, 20 or 100 mg/litre (0, 1, 3 or 14 mg/kg of
body weight per day) throughout pregnancy and through weaning
decreased thyroxine levels in the serum of the pups at 100 mg/litre.
This was associated with delayed development of exploratory
behaviour in the pups away from their dams, and this, in turn, was
probably due to an indirect effect on iodine uptake. In a second
experiment, pups given 14 mg of chlorine dioxide per kg of body
weight per day directly by gavage on postnatal days 5–20 showed
significantly depressed activity and a decrease in serum thyroxine
levels. Studies by the same group (Taylor & Pfohl, 1985) indicated
that cerebellar and forebrain cell counts (based on DNA
measurements) were depressed in 11-day-old pups that had been
treated with chlorine dioxide at 14 mg/kg of body weight per day by
gavage from 5 days of age. Cerebellar cell counts remained depressed
in rats at 21 days, but forebrain counts were essentially the same as in
controls. At 50–60 days of age, the locomotor activity (measured by
wheel-running) of these animals was depressed relative to control
animals.
The effects of chlorine dioxide on brain development were

examined further by Toth et al. (1990). These authors administered
chlorine dioxide by gavage at 14 mg/kg of body weight per day from
postnatal day 1 to 20. Body weight was reduced, but cerebellar weight
was unaltered at any age. Forebrain weight and protein content were
reduced on postnatal days 21 and 35. DNA content was depressed on
postnatal day 35, and the number of dendritic spines on cerebral
cortical pyramidal cells was significantly reduced. No histo-
pathological changes in the forebrain, cerebellum or brain stem were
observed. There were no consistent changes in serum thyroxine or
triiodothyronine levels in treated animals.
Collectively, these data suggest some effects of chlorine dioxide

on brain development. In most studies, there are suggestions of
modified thyroid function associated with these effects. It must be
pointed out that the changes in thyroid hormone levels are modest,
much less than are produced with classical antithyroid drugs such as
propylthiouracil (Toth et al., 1990).
107
The effects of chlorine dioxide were assessed in a two-phase

study in 10 healthy male volunteers. The first study was a rising-dose
tolerance study (Lubbers & Bianchine, 1984) in which doses of
chlorine dioxide were increased from 0.1 to 24 mg/litre, administered
in two 500-ml portions. The maximum dose for a 70-kg person was
0.34 mg/kg of body weight. The details of this study were described
in section 3.1.3. Some small changes in a variety of clinical chemistry
parameters were observed, but none was found to be outside the
accepted range of normal. The second phase of the experiment
involved the daily administration of a 500-ml portion of a solution
containing 5 mg/litre to 10 healthy volunteers for a period of 12 weeks
(Lubbers et al., 1984a). Again, measurement of a large battery of
clinical chemistry parameters and routine physical examination failed
to identify any effects of chlorine dioxide that fell outside of the
normal range. Parameters yielding significant differences appeared to
be primarily a result of parallel drift of values with the control group.
As with other disinfectants, it is important to recognize that

chlorine dioxide is a potent respiratory irritant. No quantitative data
can be used to construct a dose–response relationship for this effect.
The mutagenic or clastogenic effects of chlorine dioxide have

received little attention. Ishidate et al. (1984) found chlorine dioxide
to be positive in Salmonella typhimurium tester strain TA100. A
linear dose–response was observed at concentrations between 2 and
20 :g per plate. Chlorine dioxide was ineffective as a clastogenic
agent in a CHO system.
Meier et al. (1985b) evaluated the ability of chlorine dioxide to

induce chromosomal aberrations and micronuclei in bone marrow of
CD-1 mice or sperm head anomalies. Chlorine dioxide failed to
produce such damage following gavage doses of up to 16 mg/kg of
body weight for 5 days.
With the exception of a 1949 study by Haag (cited in TERA,

1998), which has serious limitations, no tests of the carcinogenic
108
activity of chlorine dioxide in experimental animals were identified

in the scientific literature.
There are significant differences in the pharmacokinetics of 36Cl

obtained from different disinfectants. The absorption rate for 36Cl-
labelled chlorine dioxide was at least 10 times that observed with
chlorine, chloramine or chloride. The relative amount of 36Cl that is
eliminated in the urine and faeces at 24 h has a distinct pattern from
that observed with other disinfectants and sodium chloride. However,
the terminal half-life of the 36Cl appears similar for all disinfectants.
These data suggest that the form of 36Cl that is being absorbed differs
chemically with the different disinfectants. In the case of chlorine
dioxide, this is supported by the observation that measurable amounts
of chlorite (about 3% of the original dose) are eliminated in the urine
during the first 24 h, and chlorite comprises about 20% of the label
present in blood 72 h after administration of the test dose (Abdel-
Rahman et al., 1980). However, this higher absorption rate is not
explained by the absorption rates of chlorite and chlorate, which are
about one-tenth as rapid (Abdel-Rahman et al., 1982b). This suggests
that some of the absorption could be as chlorine dioxide itself. On the
surface, this hypothesis would seem to be incompatible with the high
reactivity of this disinfectant. As with other disinfectants, the terminal
elimination phases observed for 36Cl from chlorine dioxide seem
compatible with the hypothesis that the bulk of the elimination is as
chloride ion.
109

216 Disinfectants Part 2

Uploaded by

Copyright:

Available Formats

216 Disinfectants Part 2

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

216 Disinfectants Part 2

Uploaded by

Copyright:

Available Formats

2.

CHEMISTRY OF DISINFECTANTS AND

The use of chlorine (Cl2) as a water disinfectant has come under

Chloramine presents the significant advantage of virtually elimi-

The use of chlorine dioxide, like chloramine, can reduce the

If used as a primary disinfectant followed by a chloramine residual

with NOM to produce organic DBPs such as aldehydes and increase

A thorough understanding of the mechanisms of DBP formation

2.2 Physical and chemical properties of common

The important physical and chemical properties of commonly used

Chlorine, a gas under normal pressure and temperature, can be

Chlorine is used in the form of gaseous chlorine or hypochlorite

Chlorine gas hydrolyses in water almost completely to form

Cl2 + H2O 6 HOCl + H+ + Cl–

The hypochlorous acid dissociates into hydrogen ions (H+) and

Hypochlorous acid is a weak acid with a pKa of approximately 7.5

The rates of the decomposition reactions of chlorine increase as

XO– + XO2– 6 X– + XO3–

Another reaction that occurs in waters containing bromide ion and

HOCl + Br– 6 HOBr + Cl–

This reaction is irreversible, and the product hypobromous acid is a

common analytical procedures for free chlorine. The presence of

Hypobromous acid is a weak acid (pKa = 8.7); like hypochlorite,

3OBr– 6 BrO3– + 2Br–

2.2.2 Chlorine dioxide

Chlorine dioxide is one of the few compounds that exists almost

Chlorine dioxide represents an oxidation state (+4) intermediate

2ClO2 + H2O 6 ClO2– + ClO3– + 2H+

Chlorine dioxide has an absorption spectrum with a maximum at

greenish-yellow solution. It can be involved in a variety of redox

Bromide (in the absence of sunlight) is not oxidized by chlorine

Ozone is a strong oxidizing agent (Eo = 2.07 V). Oxidation

Many oxidative applications of ozone have been developed,

Monochloramine has much higher CT values1 than free chlorine

Monochloramine is the only useful ammonia-chloramine disin-

2.3 Analytical methods for disinfectant by-products

2.3.1 Trihalomethanes, haloacetonitriles, chloral hydrate,

Gas chromatographic (GC) techniques are generally employed

DBPs Analytical method APHA a Minimum detection Major References

in the purge-and-trap technique; and (iv) low extraction efficiency for

The extraction of THMs can be accomplished using MTBE (EPA

The haloacetic acid (HAA) analytical method involves using an

2.3.2 Inorganic disinfectant by-products

An ion chromatography (IC) method (EPA Method 300) has been

involves measurement by an IC system using a separating column (e.g.,

2.3.3 Total organic carbon and UV absorbance at 254 nm

TOC is the primary surrogate parameter for the measurement of

Knowledge of the amine content of the water during water treat-

of many organic chloramines continue to challenge the analyst.

2.4 Mechanisms involved in the formation of disinfectant

2.4.1 Chlorine reactions

Chlorine reacts with humic substances (dissolved organic matter)

HOCl + DOC 6 DBPs

It is generally accepted that the reaction between chlorine and

Free chlorine reacts with water constituents by three general

this reaction is too slow to be of importance in water treatment. The

Most chlorine DBPs are formed through oxidation and substitu-

Rook (1977) proposed resorcinol structures to be the major

However, the production of chlorinated compounds such as