vxEPA
United States
Environmental Protection
Agency
Determining Active Oxidant
Species Reacting with
Organophosphate Pesticides
in Chlorinated Drinking Water
RESEARCH AND DEVELOPMENT
-------
EPA/600/R-06/103
September 2006
Determining Active Oxidant Species Reacting with
Organophosphate Pesticides in Chlorinated Drinking
Water
By
Stephen E. Duirk, Daniel P. Cherney^, Christopher J. Tarr*, and Timothy W. Collette
National Exposure Research Laboratory
Ecosystems Research Division
Athens, GA
^National Research Council
*Student Services Authority
U.S. Environmental Protection Agency
Office of Research and Development
Washington D.C. 20460
-------
NOTICE
This work has been subject to external peer and administrative review, and has been approved
for publication as an EPA document. Mention of trade names or commercial products does not
constitute endorsement or recommendation for use.
11
-------
ABSTRACT
Chlorpyrifos (CP) is an organophosphate (OP) pesticide that was used as a model
compound to investigate the transformation of OP pesticides at low pH and in the presence of
bromide and natural organic matter (NOM) under drinking water treatment conditions. Raman
spectroscopy was used to determine which active chlorine species was responsible for the rapid
oxidation of CP below neutral pH. Over the pH range of 2-11, three active chlorine species were
identified with Raman bands of 725, 711, and 538 cm"1. The first two bands were easily
assigned to HOC1 and OC1" respectively. The 538 cm"1 band was identified as CbCaq) after
bubbling chlorine gas through phosphate buffered water at pH 2. Either at low pH or in water
treatment plants that use direct injection of CbCg), molecular chlorine rapidly reacts with CP
transforming it to chlorpyrifos oxon (CPO). In the presence of bromide, the loss of CP was
found to be accelerated when aqueous chlorine was added. It was found that bromide acts as a
catalyst in the oxidation of CP to CPO via the formation of hypobromous acid (HOBr) at
concentrations relevant to drinking water treatment. Also, NOM was found to not inhibit the
oxidation of CP to CPO in the presence of free chlorine under the experimental conditions in this
study. A previously developed screening-level model used to predict OP pesticide degradation
pathways was modified to model the effects of both bromide and NOM. The model adequately
described the oxidation of CP to CPO by free chlorine as well as the stability of CPO over the
course of the experiment. This work demonstrated the applicability of screening-level models to
predict the fate of OP pesticides under drinking water treatment conditions.
in
-------
ACKNOWLEDGEMENT
The authors would like to thank Jimmy Avants his technical assistance. Also, we would like to
thank Dr. Wayne Garrison, Dr. Jackson Ellington, and Dr. John Kenneke for their consultation
and expertise.
IV
-------
TABLE OF CONTENTS
ABSTRACT iii
ACKNOWLEDGEMENT iv
LIST OF TABLES vi
LIST OF FIGURES vii
EXECUTIVE SUMMARY x
1 INTRODUCTION 1
2 EXPERIMENTAL PROCEDURES 7
2.1 Materials 7
2.2 Methods 8
2.2.1 Spectroscopic Studies of Active Chlorine Species 8
2.2.2 Chlorpyrifos Kinetic Experiments with Bromide andNOM 10
2.3 Instrumentation 11
3 RESULTS AND DISCUSSION 15
3.1 Spectroscopic Studies of Active Chlorine Species 15
3.2 Loss of CP in the Presence of Bromide and Free Chlorine 24
3.3 Loss of CP in the Presence of Free Chlorine andNOM 29
4 CONCLUSIONS 32
5 REFERENCES 34
TABLES 38
FIGURES 42
-------
LIST OF TABLES
Table 1 NOM and source water characteristics for the ACC water collected prior to
chlorination 39
Table 2 The slopes of kobs vs. [Br"] for each pH and [HOC1] concentration 40
Table 3 Stoichiometric equations and rate coefficients used in the chlorpyrifos degradation
pathway model 41
VI
-------
LIST OF FIGURES
Figure 1 Observed first order loss rate coefficients for chlorpyrifos over the pH range of 2-1 1
in the presence of ECS. [CP]0 = 0.5 (jM, [Buffer]T = 10 mM, Temperature =
25±1°C, and [HOC1]T= 10 nM [[[ 43
Figure 2 (A.) Speciation plot of the 26.2 mM commercial free chlorine solution (the ECS
solution) generated by monitoring the normalized areas of the Raman bands from
OC1" at 711 cm-1 (A), HOC1 at 725 cm-l(B) and the low pH species at 538 cm"1
(+). (B.) Background-subtracted Raman spectra of ECS solutions. The OC1" band
is observed in the top two spectra and the HOC1 band is observed in the bottom
three spectra. An additional band for the low pH form appears as the pH is lowered
below 5 [[[ 44
Figure 3 Speciation plots of commercial free chlorine solutions (ECS solutions) generated by
monitoring the normalized areas of the Raman bands from OC1" at 711 cm-1 (*•,
dash-dot model line), HOC1 at 725 cm-1 (•, dashed model line) and the low pH
species at 538 cm"1 (+, solid model line). Vertical lines indicate the apparent pKas
at 13.1 mM. The pKa for the protonation of OC1" to HOC1 occurs at 7.50 for all
three concentrations. The apparent pK at lower pH varies from (from top to
bottom): 2.11, 1.84, and 1.25 [[[ 45
Figure 4 Raman spectrum acquired from (A) a buffered aqueous solution at pH 1.45 through
which chlorine gas was bubbled (BCS) and (B) a pH 1.63 solution containing 13.1
mM commercial free chlorine (ECS). The band at 538 cm"1 is due to the presence
of Cb (aq) in each solution [[[ 46
Figure 5 Raman spectra comparing 13.1 mM free chlorine solutions with (A) equimolar
chloride ion at pH 0.75 (ECS) and (B) free chlorine solution with reduced chloride
ion at pH 0.80 (LCS). The spectrum of the solution containing equimolar chloride
ion has a much larger band at 538 cm"1 (due to aqueous Cb) than the solution with
greatly reduced chloride concentration [[[ 47
Figure 6 Speciation profile of the LCS free chlorine solution with reduced chloride ion
concentration generated by monitoring the normalized areas of the Raman bands
from HOC1 at 725 cm-1 (•) and OC1" at 71 1 cm-1 (±). The predominant species in
solution at low pH remains HOC1 throughout, and a significant drop in its total area
-------
Figure 8 Observed first order rate coefficients for CP loss in the presence of increasing
bromide concentrations at [HOCl]x = 10 |jM over the pH range of 7-9. [CP]0 = 0.5
[Buffer]T = 10 mM, and Temperature = 25 °C ................................................ 50
Figure 9 Observed first order rate coefficients for CP loss in the presence of increasing
bromide concentrations at [HOC1]T = 50 [iM at pH 8 and 9. [CP]0 = 0.5 [iM,
[Buffer]T = 10 mM, and Temperature = 25 °C [[[ 51
Figure 10 Two possible oxidation pathways for CP at high and low [HOCl]:[Bf ] ratios. The
pathway on the left shows the predominant pathway when [HOCl]:»[Bf ] and the
pathway on the right shows the predominant pathway when [HOCl]:[Bf ]
approaches 1 [[[ 52
Figure 1 1 Loss of CP in the presence of hypobromous acid at pH 6.5. [CP]0 = 0.42 |jM,
[HOBr]T = 10 nM, [Buffer]T = 10 mM, and Temperature = 25 °C ......................... 53
Figure 12 CP degradation in the presence of free chlorine at pH 8.0 in the absence of bromide.
[CP]0 = 0.39 nM, [HOC1]T = 10 |iM, [PO4]T = 10 mM, and Temperature = 25 °C.
Lines represent model results [[[ 54
Figure 13 CP degradation in the presence of free chlorine at pH 8.0 in the absence of bromide.
[CP]0 = 0.42 nM, [HOC1]T = 25 |iM, [PO4]T = 10 mM, and Temperature = 25 °C.
Lines represent model results [[[ 55
Figure 14 CP degradation in the presence of free chlorine at pH 8.0 in the absence of bromide.
[CP]0 = 0.42 nM, [HOC1]T = 50 |iM, [PO4]T = 10 mM, and Temperature = 25 °C.
Lines represent model results [[[ 56
Figure 15 CP degradation in the presence of free chlorine and bromide at pH 8.0. [CP]0 =
0.51 nM, [HOC1]T = 10 nM, [Bf ] = 1 |iM, [PO4]T = 10 mM, and Temperature = 25
°C. Lines represent model results [[[ 57
Figure 16 CP degradation in the presence of free chlorine and bromide at pH 8.0. [CP]0 =
0.48 |iM, [HOC1]T = 25 |iM, [Bf] = 1 |iM, [PO4]T = 10 mM, and Temperature = 25
°C. Lines represent model results [[[ 58
-------
Figure 19 CP degradation in the presence of free chlorine and bromide at pH 8.0. [CP]0 =
0.43 nM, [HOC1]T = 50 |iM, [Bf] = 5 |iM, [PO4]T = 10 mM, and Temperature = 25
°C. Lines represent model results 61
Figure 20 CP degradation in the presence of free chlorine and bromide at pH 8.0. [CP]0 =
0.39 nM, [HOC1]T = 50 |iM, [Bf] = 10 |iM, [PO4]T = 10 mM, and Temperature =
25 °C. Lines represent model results 62
Figure 21 Chlorine demand of ACC water over the pH range of 7-9. [HOCl]x = 50 |jM,
[Buffer]T = 10 mM, [DOC] = 1.1 mg-C/L, and Temperature = 25 °C 63
Figure 22 CP degradation in the presence of free chlorine and ACC water over the pH range
of 7-9. [CP]0 ~ 0.5 nM, [HOC1]T = 50 |iM, [DOC] = 1.1 mg-C/L, [Buffer]T = 10
mM, and Temperature = 25 °C. Lines represent model results 64
Figure 23 CP degradation in the presence of free chlorine and ACC water at pH 7. [CP]0 =
0.46 nM, [HOC1]T = 50 (iM, [DOC] = 1.1 mg-C/L, [Buffer]T = 10 mM, and
Temperature = 25 °C. Lines represent model results 65
Figure 24 CP degradation in the presence of free chlorine and ACC water at pH 7.5. [CP]0 =
0.35 nM, [HOC1]T = 50 |iM, [DOC] = 1.1 mg-C/L, [Buffer]T = 10 mM, and
Temperature = 25 °C. Lines represent model results 66
Figure 25 CP degradation in the presence of free chlorine and ACC water at pH 8.0. [CP]0 =
0.42 |iM, [HOC1]T = 50 |iM, [DOC] = 1.1 mg-C/L, [Buffer]T = 10 mM, and
Temperature = 25 °C. Lines represent model results 67
Figure 26 CP degradation in the presence of free chlorine and ACC water at pH 8.5. [CP]0 =
0.35 nM, [HOC1]T = 50 (iM, [DOC] = 1.1 mg-C/L, [Buffer]T = 10 mM, and
Temperature = 25 °C. Lines represent model results 68
Figure 27 CP degradation in the presence of free chlorine and ACC water at pH 9.0. [CP]0 =
0.5 nM, [HOC1]T = 50 |iM, [DOC] = 1.1 mg-C/L, [Buffer]T = 10 mM, and
Temperature = 25 °C. Lines represent model results 69
IX
-------
EXECUTIVE SUMMARY
Environmental regulations require that all relevant routes of human exposure to
anthropogenic chemicals be considered in risk assessments. Community water systems (CWSs)
serve approximately 95% of the US population and potable water is considered a relevant route
of exposure to anthropogenic chemicals. There is available monitoring data for important
pesticides and toxic chemicals in drinking water sources (both surface and ground water).
However, there is very little monitoring data for these chemicals or their degradates in finished
drinking water. Limited experimental studies show that some chemicals are partially removed
by physical water treatment processes (e.g., filtration, flocculation, etc.), and some are
transformed by reactions that occur during chemical treatment (e.g., disinfection and softening).
Transformation products of some contaminants have been shown to be more toxic than the
parent compound.
This report is in partial fulfillment of the National Exposure Research Laboratory Task #
16608, "Fate of Pesticides and Toxic Chemicals During Drinking Water Treatment", under Goal
4, GPRA Objective 4.5, and GPRA Sub-objective 4.5.2. The goals of this research task are to: 1)
provide chemical-specific information on the effects of water treatment for high-priority
pollutants, 2) provide physicochemical parameters for transformation products, and 3) develop
predictive models for forecasting treatment effects that cross chemical class and treatment
conditions.
Drinking water sources are a complex matrix of inorganic anions and cations and natural
organic matter (NOM). Because of this complexity, it is exceedingly difficult to predicting the
potential transformation of anthropogenic compounds under conditions that simulate drinking
water treatment. Therefore, it is necessary to identify specific components in drinking water
matrices that could significantly influence the transformation of anthropogenic chemicals (e.g.,
-------
organophosphate (OP) pesticides). Also, the effect of these native aqueous constituents could be
dependent on other drinking water plant operational parameters such as pH adjustment or
application of disinfectants.
Results of experiments reported here demonstrate that the change in risk associated with
the transformation of OP pesticides due to drinking water chlorination practices is a very
complex issue. For example, pH of the finished water can significantly affect the rate of OP
pesticide transformation. Previous work has shown that oxidation of OP pesticides by free
chlorine near neutral pH leads very rapidly to a more toxic transformation product (i.e., oxons).
Also, OP pesticides are susceptible to chlorine-assisted hydrolysis above neutral pH, which leads
to a less-toxic transformation product. Below pH of 6, the oxidation of OPs can proceed much
more rapidly than expected. This phenomenon has been observed by others in the drinking water
community examining the loss of other anthropogenic compounds in the presence of free
chlorine. Therefore, we chose to investigate this phenomenon in order to determine its relevance
to drinking water treatment. Additionally, two native aquatic constituents (i.e., bromide and
NOM) were investigated to determine their effect on the transformation of OP pesticides. It was
hypothesized that bromide could accelerate the rate of OP transformation, while the presence of
NOM was expected to decrease the rate of OP transformation due to the fact that NOM could act
as a sink for active oxidants commonly used in drinking water treatment.
Raman spectroscopy was used to determine the presence of active chlorine species over
the pH range of 1-11. Over the pH range 5-11, unique Raman bands were identified and
associated with hypochlorous acid (HOC1) and hypochlorite ion (OC1") with Raman shifts of 725
and 711 cm"1 respectively. As the pH was lowered below pH 5, in-growth of a band at 538 cm"1
appeared while the band at 725 cm"1 disappeared. This low-pH active chlorine species was
XI
-------
identified as CbCaq). Molecular chlorine in the aqueous phase is a more potent oxidant than
HOC1. The relevance of this to the drinking water industry is that many plants use chlorine gas
to chlorinate potable water. CbCg) does hydrolyze to HOC1 and OC1" quickly near neutral pH;
however, its selective and rapid reactivity with not only OP pesticides but other anthropogenic
chemicals is a factor in determining the risk associated with drinking potable water contaminated
with anthropogenic chemicals.
Bromide and NOM were investigated to determine their effect on the transformation rate
of OP pesticides. Bromide was found to significantly increase the rate of chlorpyrifos (CP)
transformation. This is due to the oxidation of Br" by HOC1 resulting in the formation of
hypobromous acid (HOBr), which is a more potent oxidant than HOC1. Since the formation of
HOBr is pH dependent, a screening-level model that forecasts the concentrations of all reaction
products as a function of pH, chlorine dose, OP pesticide concentration, bromide concentration,
and time after chlorine dosing was developed. Using this simple model, the concentrations of
CP, chlorpyrifos oxon (CPO), and 3,5,6-trichloro-2-pyridinol (TCP) can be adequately described
under a variety of scenarios that are similar to drinking water treatment conditions using an
apparent rate coefficient describing the reaction between HOBr and CP. In the presence of
NOM, the model was found to predict the transformation of CP to CPO as well as the
degradation of CPO to TCP in the presence of free chlorine.
The work reported here demonstrates the "proof-of-concept" towards the development of
a comprehensive modeling tool for OP fate in drinking water treatment plants and distribution
systems. Following the reported work, we have now begun to apply the experiments and models
to other OP pesticides that were judiciously selected to reflect the full range of chemical
structure variability in this class. Our objective is to develop predictive models that associate
xn
-------
relative rates of reactivity to structural variability. This will allow decision makers to rank and
prioritize chemicals found in drinking water sources according to potential risk. Also, we are
currently in the process of investigating chloramination disinfection practices on the
transformation of OP pesticides.
Xlll
-------
1 INTRODUCTION
The Food Quality Protection Act of 1996 (FQPA) requires that all pesticide chemical
residuals in or on food be considered for anticipated human exposure. Drinking water is
considered a potential pathway for dietary exposure, but there is reliable monitoring data for only
the source water. For example, the United States Geological Survey (USGS) completed a
national reconnaissance survey known as NAWQA (National Water Quality Assessment) to help
define human exposure to various contaminants (1). For the NAWQA survey, 90 pesticides (and
some selected metabolites) were chosen as target chemicals to monitor in US drinking water
sources. However, there is a relative dearth of information on occurrence of pesticide residuals
and pesticide metabolites in finished drinking water. Two surveys have been conducted for a
few community water systems examining pesticide concentrations in the source and finished
drinking water (2,3). Neither of these studies thoroughly examined the effect of each treatment
process on a single slug of water, hence only the influent and effluent of each treatment facility
could be qualitatively discussed with respect to overall removal efficacy. Also, these studies did
not account for the treatment plant hydraulic retention time, thus it was not possible to ensure
that influent and effluent samples were properly paired. Clearly, more thorough occurrence
studies that include monitoring drinking water for both pesticides and their expected
transformation products are needed.
Chlorination is the most commonly used chemical disinfection process for community
water systems (4) and is known to react with numerous pesticides. For example, four s-triazines
were found to degrade in the presence of free chlorine (HOC1 + OC1") (5,6). Atrazine was also
found to be significantly degraded by ozone (7); however, subsequent chlorination of the
ozonated effluent had very little affect on the concentration of residual atrazine or its ozone
degradation products (2). Also, some carbamate pesticides have been shown to react with free
1
-------
chlorine while other members of this pesticide class were found to be stable in chlorinated water.
For example, carbaryl and propoxur do not react with free chlorine; but aldicarb, methomyl, and
thiobencarb do exhibit significant reactivity (8-10). These findings demonstrate that free
chlorine reactivity with different members in a specific class of pesticides can vary significantly
due to chemical structure variations. Therefore, it is prudent to study the fate and transformation
pathways of entire chemical classes, using class members that exhibit systematic structural
variations and employing carefully selected experimentation and numerical modeling.
When chlorine reacts with the phosphorothioate subgroup of organophosphate (OP)
pesticides, the thiophosphate functionality (P=S) can be oxidized to its corresponding oxon
(P=O) (11-13). The resulting oxons are typically more potent than the parent as an inhibitor of
acetlycholinesterase, an enzyme necessary for proper function of the nervous system (13). Duirk
and Collette, (14) elucidated the fate of chlorpyrifos (CP) and its transformation products over
the pH range of 6-11. They were able to model the loss of CP and chlorpyrifos oxon (CPO) to
the stable end-product of 3,5,6-trichloro-2-pyridinol (TCP) over this pH range in buffered
deionized water systems. However, other factors such as pH below 6, bromide ion, and natural
organic matter can significantly influence the transformation rate of CP.
Several recent papers in drinking water literature have reported free chlorine oxidation
phenomena that stray from expected results as the pH was lowered in the course of kinetic
experiments with various anthropogenic chemicals (15-17). Specifically, a dramatic increase in
reaction rate was observed in solutions when the pH was lowered below approximately 5. The
same trend has been observed while studying the free chlorine oxidation of CP, and a kinetic plot
depicting this phenomenon is displayed in Figure 1. Initially, the observed reaction rate was
assumed to be relatively constant as the pH was lowered below the point where HOC1 was the
-------
principal species and the OC1" concentration was insignificant. This result was expected because
HOC1 is generally assumed to be the predominate oxidant in free chlorine solutions. However,
since the observed rate coefficient increased dramatically as pH was dropped below 5, it was
clear that there was another chlorine species in solution, more potent as an oxidant than HOC1.
At least two competing hypotheses have been put forward by various researchers
regarding the identity of the species responsible for this rapid oxidation. Specifically, some
researchers have proposed H2OC1+ as the oxidant (18), and other researchers have proposed
Q2(aq) (17). In literature, there are no thorough studies designed to rigorously determine the
identity and speciation profile of this potent oxidant. It appears that the H2OC1+ species was first
proposed as an active species at low pH in 1962 by Arotsky and Symons (19) in their discussion
of halogen cations. Although they did not measure the pKa for H2OC1+, it was estimated to be
between -3 and - 4 (19). Some publications have discussed the existence of the H2OC1+ cation
(20,21) as part of a reverse hydrolysis reaction. Despite the low proposed pKa, many
chlorination (15,16,18,22,23) and other oxidation (24) studies have either proposed or discussed
H2OC1+ as the reactive, oxidative species in low pH aqueous solutions. There have been a few
publications disagreeing with the hypotheses involving this species, but these are much fewer in
number and are found in the inorganic chemistry literature (25,26).
The presence of bromide ion (Br") in drinking water sources has been of interest to the
drinking water industry for several reasons. Bromide ion can be oxidized by ozone (Os)
resulting in the formation of bromate ion (Br(V) (27). Bromate is a suspected carcinogen and is
regulated under Stage 1 of the Disinfection/Disinfection Byproducts (D/DBP) Rule at a
maximum contaminant level (MCL) of 0.01 mg/L. Bromide can also be easily oxidized to
hypobromous acid in the presence of free chlorine (28), it is a stronger oxidant that HOC1,
-------
6-1-1
kHOci!Br = 5.6xl0M-h- (1)
HOBr +H2O - OBr' + H3O+ pKa = 8.8 (2)
and the pKa of hypobromous acid is 8.8 (28). In the presence of excess chlorine, all bromide
present will be oxidized to active bromine (29). Over the pH range of 6.5-9, which is the pH
range of the National Secondary Drinking Water Standards, most of the active bromine present
will be in the HOBr form (i.e., the active oxidant). Therefore, it is believed that the presence of
bromide under drinking water treatment conditions could accelerate the transformation of
anthropogenic chemicals. For example, the rate coefficient for HOBr reacting with pyrene was
found to be 2 orders of magnitude faster than HOC1 in aqueous solutions (30). In the presence of
natural organic matter (NOM), both brominated and chlorinated oxidants react with NOM
resulting in the formation of disinfection byproducts (DBFs) (3 1,32). The brominated DBFs are
believed to be more carcinogenic than their chlorinated analogs (33). Westerhoff et al., (34)
conducted a nationwide survey of 101 drinking water sources and found that the average
bromide concentration was « 1.25 |jM. Since bromide is a relatively ubiquitous and 90% of
community drinking water systems use chlorine for both primary and secondary disinfection, the
presence of bromide ion in drinking water sources could affect the rate of organophosphorus
(OP) pesticide transformation in chlorinated drinking water.
The presence of NOM in drinking water sources has also been a primary concern for the
drinking water industry. The physical-chemical properties of NOM are a function of decaying
organic matter from both terrestrial and aquatic sources, environmental conditions, and
microorganisms present in the watershed (35). Hence, the composition and concentration of
NOM in source waters fluctuates from source to source and often varies within each source
depending on conditions such as rainfall, change in seasons, and temperature. When chlorine
-------
reacts with NOM, it often results in the formation of DBFs (33). The electron density (i.e.,
aromaticity) of NOM is often expressed as specific ultraviolet absorbance (SUVA), which is the
ratio of ultraviolet absorbance of the NOM at a specific wavelength divided by the total organic
carbon concentration (TOC) in mg-C/L. This specific physical-chemical property of NOM has
been found to correlate to reactivity of NOM with free chlorine and serve as a reliable surrogate
for DBF formation (36,37). Since both chlorinated and brominated oxidants react with NOM
(38), the presence of NOM can potentially act as a sink for active oxidants in potable water.
Thus, its presence can impact the rate of OP pesticide transformation under drinking water
treatment conditions.
One purpose of this study was to identify active oxidants under drinking water treatment
conditions. Toward this end, Raman spectroscopy was used to determine active chlorine species
present over the pH range of 1-11 in a solution where active chlorine to chloride ion molar ratios
were equivalent (equal-molar chlorine solution: ECS). Also, a low chloride ion active chlorine
solution (low-chloride chlorine solution: LCS) was prepared to assist in the identification of
active chlorine species as a function of pH. Kinetic studies with CP in the presence of both ECS
and LCS were conducted to investigate CP transformation below pH 6. Bromide ion was
investigated to determine the effect it has on transformation of CP in the pH range of 7-9. Also,
the influence of NOM on the transformation of CP was investigated over the pH range of 7-9.
The work reported here serves to further elucidate the OP transformation pathways under
drinking water treatment conditions. The long-term objective is the development of a
comprehensive modeling tool for OP fate in drinking water treatment plants and distribution
systems. This will allow decision makers to rank and prioritize chemicals found in drinking
water sources according to potential risk. The goal of these experiments was to incorporate these
-------
important matrix effects into our models so that the models can be applied to actual drinking
water treatment scenarios.
-------
2 EXPERIMENTAL PROCEDURES
2.1 Materials
Chlorpyrifos (99.5%), chlorpyrifos oxon (98.7%), and 3,5,6-trichloro-2-pyridinol (99%)
were purchased from ChemService (West Chester, PA). Commercial 10-13% sodium
hypochlorite (NaOCl), purchased from Aldrich (Milwaukee, WI), contained equal-molar
amounts of OC1" and Cl". Aqueous stock solutions and experiments utilized laboratory prepared
deionized water (18 MQ cm"1) from a Barnstead ROPure Infinity™/NANOPure ™ system
(Barnstead-Thermolyne Corp., Dubuque, IA). Chlorine gas (99.9% purity) was purchased from
Matheson Tri-Gas (Montgomeryville, PA). Filters with pore size of 0.45 |j,m were purchased
from Millipore (Billerica, MA). Phosphate and carbonate salts used for buffer solutions were
dissolved in deionized water and filtered through a 0.45 |j,m filter, which was pre-rinsed with
deionized water. The pH for the experiments was adjusted with either 1 N H2SO4 or NaOH
unless otherwise noted. All other organic and inorganic chemicals were certified ACS reagent
grade and used without further purification. The glassware and polytetrafluoroethylene (PTFE)
septa used in this study were soaked in a concentrated free chlorine solution for 24 hours, rinsed
with deionized water, and dried at 105 °C prior to use. All chlorination experiments were
conducted at constant temperature (25±1°C).
Water was collected form the Athens-Clarke County (ACC) water treatment plant and
used in some experiments. The ACC water treatment facility uses conventional physical-
chemical surface water treatment (i.e., coagulation, flocculation, sedimentation, and dual media
filtration) prior to chlorination. Water was collected prior to chlorination and the NOM and
source water characteristics are shown in Table 1. The SUVA254 of the ACC effluent was 1.61
L/mg-m, which indicates a low aquatic humic substance concentration in this water (36).
-------
2.2 Methods
2.2.1 Spectroscopic Studies of Active Chlorine Species
For speciation studies, Raman spectra were collected of free-chlorine solutions as a
function of pH (at least 20 points) over the range of approximately 1 to 12. The commercial
sodium hypochlorite solution (ECS) was diluted with deionized water to a stock solution with a
free chlorine concentration of 262.0 mM to more closely match that of the low chloride solution
(LCS) that was prepared in our lab. Total free chlorine concentration was again determined via
Standard Method 4500-C1 F DPD-FAS titrimetric method (39). The final concentrations of free
chlorine in the ECS solutions monitored with Raman spectroscopy were 26.2, 13.1 and 6.6 mM.
The pH of 50 mM phosphate buffer was raised with 5 N sodium hydroxide and lowered with a 2
N sulfuric acid solution. The cuvette used for Raman work was relieved of chlorine demand by
soaking it in a concentrated free-chlorine solution overnight and rinsing it several times with
deionized water. Samples for Raman analysis were prepared by adding an appropriate amount of
buffer directly to the cuvette, followed by the addition of free-chlorine solution. The cuvette was
then rapidly capped, and inverted several times for mixing. Minimal headspace was left in the
capped cuvette to minimize the possible loss of a volatile species in a low pH solution. (There
was ~ 100 |j,L of gas volume between the solution and the cap.) The filled cuvette was then
placed into a sample holder and the Raman spectra were acquired. Unless otherwise noted, the
time for each acquisition was 5 s. To improve the signal-to-noise ratio in the spectrum, 200 of
these 5 s acquisitions were co-added (i.e., averaged). After acquisition of the Raman spectra, the
samples were transferred into a vial that had been relieved of chlorine demand and the final
solution pH was recorded.
Some of our experiments were carried out with a free-chlorine solution containing a low
concentration of chloride ion, LCS. Our method for preparing this solution, which was adapted
-------
from Reinhard and Stumm (40), involves complexing and removing excess chloride ion from
hypochlorous acid/hypochlorite preparations. First, all glassware was soaked in a concentrated
free-chlorine solution to satisfy any chlorine demand. The glassware was rinsed with deionized
water and dried at 105 °C prior to use. To a 250 mL graduated cylinder, sodium hydroxide
(2.5mL of a 1 N solution) was added to deionized water for a final volume of 250 mL. Mercuric
oxide (8.33g) was added to the graduated cylinder and the solution was stirred for 10 min.
Chlorine gas was then bubbled through the HgO/NaOH/water solution for 3 min and this
solution was stirred for 1 h. After the stirring time had elapsed, the solution was filtered through
a 0.45 |j,m filter that had been pre-rinsed with 1.5 L of deionized water. The filtrate was then
distilled at 90°C into sodium hydroxide (2.5 mL of a 0.1 N solution). The distillate was stored at
4°C until further used.
The amount of chloride ion in the LCS was determined in a multi-step process. First,
using the Standard Method 4500-C1 F DPD-FAS titrimetric procedure (39), we determined that
the total free-chlorine concentration was 202.5 mM. (At the alkaline pH of this solution, the
total free-chlorine concentration is essentially equal to the OC1" concentration.) Next, an aliquot
of the LCS was reduced with excess sodium sulfite, thereby converting all species of chlorine to
chloride ion. Then, using ion chromatograph with chemical suppression, we determined that the
chloride ion concentration in the reduced aliquot was 220.0 mM. Therefore, by difference, we
determined that the chloride ion concentration in the LCS was 17.5 mM. This method of
producing a low chloride ion solution was effective - the free chlorine to chloride ion molar ratio
in our prepared LCS was 11.6:1, whereas in the commercial ECS the ratio was 1:1.
Many studies have been published on the absorption of chlorine into water (41) and on
the kinetics of HOC1 production at various temperatures (42-44) and ionic strengths (44). This
-------
reaction has been shown to proceed more rapidly at both higher temperature and higher ionic
strength, (42-47) and experimental evidence has been obtained with several buffers arguing that
the reaction involves base-assisted hydrolysis (48-50). A low pH free-chlorine solution was
prepared by bubbling chlorine gas through a solution of 50 mM phosphate buffer for
approximately 20 sec. The initial pH of the solution was 1.8 and the final pH was 1.4 and the
color of the solution changed from clear to yellow. We report experiments here using this
solution, which was assigned the acronym BCS to denote Bubbled Chlorine Solution
2.2.2 Chlorpyrifos Kinetic Experiments with Bromide andNOM
For all CP oxidation experiments, CP was spiked by adding 0.5 mL of a 4 mM stock
solution in ethyl acetate into an empty 4 L borosilicate glass Erlenmeyer flask. A gentle flow of
nitrogen gas was used to evaporate the ethyl acetate and then 4 L of deionized or ACC water was
added to the flask. The solution was slowly stirred and allowed to dissolve for 12 hours resulting
in an aqueous concentration of 0.5 |jM. Experiments studying the effect of bromide, bromide
was added to the aqueous system prior to chlorination. CP chlorination kinetic experiments were
conducted under pseudo-first-order conditions: total chlorine to chlorpyrifos molar ratios of
20:1, 50:1, and 100:1 were used. Chlorine was added to solutions under rapid mix conditions
achieved with a magnetic stir plate and a PTFE coated stir bar. All kinetic experiments used to
estimate rate coefficients measured at least 87% loss of parent compound.
Above pH 8, 10 mM carbonate [CO3]x buffer was used to maintain pH. The purchased
free chlorine solution was first diluted to 250 mM then added to the aqueous system containing
0.5 |jM chlorpyrifos and carbonate buffer in a 2 L Erlenmeyer flask. Fourteen aliquots from the
large 2 L reactor were then placed into 128 mL amber reaction vessels with PTFE septa and
stored in the dark. In the pH range of 6.0-7.5, the rate of CP loss in the presence of free chlorine
10
-------
was very fast. Therefore, fourteen 100 mL aliquots of the 2 L aqueous system containing 10 mM
phosphate buffer, [PO^x, and 0.5 |jM CP were placed in 250 mL amber Erlenmeyer flasks.
Then each flask was individually dosed with the appropriate amount of chlorine.
At each discrete sampling interval, two reaction vessels were sacrificed in their entirety.
One vessel was used to determine total free chlorine concentration ([HOCl]x = [HOC1] +[OC1"])
via Standard Method 4500-C1 F DPD-FAS titrimetric method (39). In the other vessel, free
chlorine residuals were quenched and the pH of a 100 mL sample was then adjusted to 2 for
analysis of CP and its degradation products.
Chlorine residuals were quenched with sodium sulfite in 20% excess of the initial free
chlorine concentration. Both sodium sulfite and ascorbic acid were tested as quenching agents to
determine if they affected the stability or recovery of CP, CPO, or TCP. Control studies
compared the recoveries of all three analytes in water without chlorine present after two hours at
room temperature with and without a quenching agent present. Sodium sulfite had no apparent
effect on the recovery of any of the three analytes. However, ascorbic acid reduced the recovery
of CP by about 5.1% and increased the recovery of CPO by about 7.6%, as compared to both the
sodium sulfite quenched samples and the samples with no quenching agent present.
Furthermore, ascorbic acid has previously been found to react with organophosphate pesticides
(51). Therefore, ascorbic acid was not used to quench chlorine residuals in this work.
2.3 Instrumentation
The Raman spectra were acquired with a Kaiser Optical Systems (Ann Arbor, MI)
HoloProbe, using 532 nm laser excitation from a Coherent™ (Santa Clara, CA) Verdi V Series
frequency-doubled YVO4 laser. This type of Raman instrument has been fully described
elsewhere (52,53). Briefly, laser light was coupled to a remote probe head via a 1.9 m long
fiber-optic cable. Raman scattering and luminescence were removed from the excitation optical
11
-------
train via a holographic grating and spatial filter within the probe head. This filtered laser light
was brought to a focus with a lOx microscope objective (Olympus) approximately in the center
of a standard quartz cuvette with a screw cap. The power of the laser light at the cuvette was
approximately 220 mW. Raman scattered photons from the solution were collected by the
microscope objective along the same path as the excitation laser beam (i.e. 180° backscattering
geometry), and were coupled to a separate 1.9 m fiber-optic cable for delivery to the/1.8 axial
transmissive-type spectrograph. The probe head contained a holographic notch filter placed
along the collection path such that only inelastically scattered light was inputted into the fiber.
Inside the spectrograph the scattered light was passed through another holographic notch filter to
further remove elastically scattered light returning from the probe head. The Raman scattered
light was then focused through a 50 |j,m slit and directed through a holographic grating with
double transmission layers. These layers permit two regions of the dispersed Raman spectrum to
be stacked and simultaneously focused onto the back-illuminated CCD detector (Andor
Technology model DU420). This configuration permitted the acquisition of the entire Raman
spectrum with a useable Stokes Raman shift of about 4450 to 100 cm"1 in a single exposure with
about 5 cm"1 spectral resolution. The CCD detector was thermoelectrically cooled to -65°C and
contained 1024x 256 pixels.
A Kaiser Optical Systems, Inc. HoloLab calibration accessory was used to correct both
the frequency and intensity of the Raman spectra. Briefly, each spectrum was frequency-
corrected based on previously measured lines from a neon bulb. Also, each spectrum was
intensity-corrected based on a previously measured NIST-traceable, broadband light source (i.e.,
a tungsten bulb) that was powered by a current regulated power source for consistent spectral
output.
12
-------
Analysis of a Raman spectrum required the subtraction of a 50 mM phosphate buffer
spectra that was the same pH of the chlorine-containing solutions. The pH of the buffer was no
more than a few hundredths of a pH unit different from that of the associated solution. The
bands in the spectrum were then analyzed by fitting a Gaussian-Lorentzian mixed curve in
GRAMS/32 (Thermo Galactic). Since the HOC1 and OC1" bands overlapped with each other, the
center frequency and band width for these two species were determined with spectra that were
not near the pKa of hypochlorous acid. These two parameters were then held constant
throughout the analysis and the band areas were determined for each spectral acquisition. In
addition, the band associated with atmospheric nitrogen was averaged and used, as has been
previously described (54) to correct the band areas corresponding to the chlorine species for any
changes in instrumental conditions (e.g., alignment, laser power fluctuations, etc.).
We sought to use these individual Raman band areas to construct chemical speciation
plots. For our purpose, a speciation plot is useful when it depicts the fractions of the total
analytical concentration that belong to the various species in solution as a function of pH.
Unfortunately, individual band areas could not be used directly to construct such plots because
we could not assume that Raman scattering cross sections of the various species were equal.
However, at the highest analytical concentration of free chlorine in ECS solutions examined by
Raman spectroscopy (26.2 mM), it was possible by varying pH to obtain three individual Raman
spectra, each of which uniquely contained appreciable intensity for only one of the three
observed species. From these three spectra, we estimated an apparent Raman scattering cross
section for each of the three species. Then, using these apparent cross sections, we calculated the
intensity that we would observe for a given species at a given analytical concentration if 100% of
the analytical concentration existed in that form. Individual Raman band area collected at the
13
-------
various pH's for a given analytical concentration were then divided by this quantity in order to
"normalize" them. These normalized band intensities, which ranged from nominal values of 0 to
1 were used in speciation plots to approximate the fractions of the total analytical concentration
that belonged to the various species in solution as a function of pH.
CP and its degradation products were extracted from water using C-18 solid phase
extraction cartridges purchased from Supleco (Bellefonte, PA). The 100 mL sample was pre-
adjusted to pH < 2 to increase the recovery of TCP (pKa = 4.55) on the solid phase adsorbent
(55). Then, the sample was spiked with 1 |jM of phenthorate (internal standard), mixed
thoroughly by hand for two minutes, passed through the SPE cartridge at an approximate flow-
rate of 7 mL/min, and eluted with 3 mL of ethyl acetate. Quantification for each analyte was
compared to eight extracted standards over the concentration range of 0.01 to 1 |jM. A Hewlett-
Packard 6890 GC equipped with a 5973 MSD was used to analyze CP, CPO, and TCP. GC
conditions were as follows: 30-m Restek Rtx-200 column with a 0.25-mm ID and 0.5-|j,m film
thickness. The temperature profile was: 100 °C for 5 minutes, 100 to 250 °C at 10 °C/minute,
and then held at 250 °C for 25 minutes. Mass balances of 80% or greater were obtained for each
experiment.
UV-Vis spectra were acquired with a Shimadzu 1700 UV-Vis spectrometer (Shimadzu
Scientific, Columbia, MD). Total organic carbon (TOC) was measured using a Shimadzu TOC
5000 (Shimadzu Scientific, Columbia, MD) and calibrated according to Standard Methods 505A
(39). Ion chromatography was performed on a Metrohm (Houston, TX) MIC-2 ion
chromatograph with chemical suppression. All pH measurements were obtained with an Orion
940 pH meter using a Ross combination electrode from Fisher Scientific (Pittsburgh, PA).
14
-------
3 RESULTS AND DISCUSSION
3.1 Spectroscopic Studies of Active Chlorine Species
As discussed earlier, there was a large increase in the rate of aqueous free chlorine
oxidation for chlorpyrifos when the pH was decreased below ~ 5. This behavior, which is
similar to that recently reported with other anthropogenic chemicals (15,17,56,57), suggests the
presence of an oxidant other than OC1" and HOC1, and more potent than either, that is formed as
the pH decreases below 5. To firmly establish the existence of this third species, and to
determine whether it is Cb (aq) or H2OC1+, Raman spectra of the aqueous free chlorine ECS
solution (26.2 mM) were collected over a pH range from approximately 1 to 12. Across this pH
range, three Raman bands appeared in the spectra due to various chlorine species in solution.
Specifically, a band at 711 cm"1 was observed to increase in intensity as the pH was increased
above 6, and its intensity was relatively constant above pH of ~ 9.5. A second band at 725 cm"1
was observed to increase in intensity as the pH was lowered from 9 to 6, and its intensity
decreased as the pH was lowered below ~ 5. Finally, a third band at 538 cm"1 was observed to
increase in intensity as the pH was lowered below 5. The normalized areas of these bands are
plotted over the entire pH range in Figure 2A.
Figure 2B shows spectra of chlorine acid species in solution at four discrete pHs (11.96,
7.46, 4.59, and 2.73) with the aqueous phosphate buffer subtracted. Bands in the figure are
labeled according to the species to which they are attributed. The band at 711 cm"1 is confidently
attributed uniquely to the OC1" species and the band at 725 cm"1 uniquely to the HOC1 species, in
agreement with an earlier study (58). The in-growth of a Raman band at 538 cm"1 as the pH is
lowered below 5 supports the presence of a third chlorine species. We believe this band at 538
cm"1 can be attributed uniquely to the species that is responsible for the unexpectedly rapid
15
-------
oxidation of anthropogenic chemicals that is observed in free chlorine solutions as pH decreases
below 5.
Raman spectroscopy has been demonstrated as an effective method for determining the
equilibrium constants of organic species in aqueous solution (54). By monitoring monotonic
changes in spectra, the amount of each species can be quantified at each pH studied. This
general method is applied to the current experiment to model chlorine species in aqueous
solution. The pKa for the protonation of OC1", described by:
HOC1 ^ OCr + H3O+ (3)
can be readily approximated by monitoring the Raman-active vibrational bands of the species
HOC1 and OC1" (at 725 and 711 cm"1, respectively) as a function of pH. Specifically, for a
simple isolated protonation event of this type, the pKa can be approximated as the pH at which
the normalized intensities of the HOC1 and OC1" bands are equal because, according to the
Henderson-Hasslebach equation, when [HOC1] = [OC1"], the
PH = -logK1=pK1 (4)
and
Ki = [OCI-] [H3O+]/[HOCl] (5)
(Although it would be accurate in this case to label the equilibrium constant as Ka, for this work
the more generic KI and K2 to denote constants for the equilibria discussed at high and low pH,
respectively. For simplicity, concentration-based equilibrium constants will be presented. Using
the extended Debye-Hiickel equation (59) the calculated activity coefficients for all charged
species discussed herein and have determined that, at the concentrations used in this work,
activity-based equilibrium constants do not differ significantly from those based on
concentration. Specifically, for determination of pKs, the differences were calculated to be
16
-------
between 0.7 and 1.2% for the high pH event and 1.3 to 6.4% for the low pH event for all
equilibrium cases and concentrations considered. For these calculations, activity coefficients of
all neutral species were assumed to be 1.
Furthermore, the species arising at a lower pH can be described by either equation 6 or 7,
as discussed in the Introduction:
H2OC1+ . H3O+ + HOC1 (6)
H2O + C12 - H3O+ + HOC1 + Cr (7)
If equation 6 is correct, then the low-pH species is formed from a simple protonation event like
that described in equation 3. On the other hand, if equation 7 is correct, the equilibrium is more
complicated, involving species other than a proton and a protonated/deprotonated pair.
A convenient way to distinguish between the two situations described by equations 6 and
7 was to monitor the relative concentrations of HOC1 and the low-pH form as a function of pH as
the absolute concentration of total chlorine was varied. This was achieved by preparing
speciation plots such as that shown in Figure 2A for several chlorine concentrations. If equation
6 was the proper description, the pH at which the normalized intensities of the bands for the low-
pH form and HOC1 are equal would approximate the pKa and this observed pKa would not vary
as a function of total free chlorine. Specifically, for equation 6 (as with equation 3), when
[HOC1] = [H2OC1+], the
where
H2oa+ = [HOa] [H3Q+] /fH20Cfj (9)
17
-------
Alternately, if equation 7 was the proper description, the pH at which the normalized
intensity of the bands were equal would change as a function of total free chlorine. Specifically,
for equation 7, when [HOC1] = [C12], the
pH = -log f£ + log fcrj = p f£ + log fcrj (10)
where
K"2 = [Hoci] [cr] [H3o+] / [Ci2] (ii)
The commercial reagents used to prepare free-chlorine solutions (ECS) were equimolar in HOCI
and Cl". So, as the free chlorine concentration is increased, [Cl"] will increase. Thus, from
equation 10, we would expect the pH at which the normalized intensity of the bands are equal to
increase as the free chlorine concentration increases in ECS solutions if equation 7 is the proper
description of the low pH behavior.
Full speciation plots, using normalized Raman band intensities, for three different
commercial total free chlorine solutions (ECS solutions) are displayed in Figure 3. In order to
determine if the apparent pK for either equilibrium changed as a function of chlorine
concentration, the normalized band intensities were fit using standard equations that describe the
concentration profiles of polyprotic acids (60). To present these equations, we assigned K2 as
the equilibrium constant for the low pH event described by equation 9 or 11 and KI for the
higher pH event described by equation 5. For these generalized equations, the amount of each
component (a;) is described in terms of the total amount of the acid or base present. In this case:
(60)
_ [lowpHform]
18
-------
03)
where CT = [low pH form] + [HOC1] + [OC1"]. It is important to note that the sum of the a;
components must be equal to 1 under every condition. The numerical value of each a;
component was obtained by utilizing the following equations: (60)
a.- - i- - 04)
06)
The model was fit to the data by using a least-squares analysis and allowing the values
for K2 and KI to vary in order to minimize the sum of the squared residuals. The fit of the model
in the top graph of Figure 3 most closely matches the experimental data because it was used to
determine the apparent Raman scattering cross-section for each species. Subsequent plots of
experimental data for 13.1 mM and 6.6 mM solutions deviate slightly from the model because
the spectral correction with the nitrogen band (as described in the Experimental section) cannot
perfectly encompass all variations in the instrument over long periods of time. (Note that all of
the data in Figure 3 was collected over a time period of 60 days). Furthermore, the greatest
variation between experimental data and the model is seen in the amounts of HOC1 and OC1".
19
-------
This is likely due to these bands not being completely resolved from one another, limiting the
accuracy of the curve fitting routine.
The vertical line on the right in Figure 3 indicates the apparent pKa (i.e. pKi) for the
protonation of OC1" that occurs between pH 7 and 8 for each concentration of free chlorine. This
line indicates that the pKa of hypochlorous acid is approximately 7.50, which agrees very well
with the previously published result of a pKa near 7.5 (44). Indeed, the pKa for each
concentration was identical to the significant digits that we can accurately report. On the other
hand, it is evident, as shown by the line on the left in Figure 3, that the apparent pK for the low
pH equilibrium does not occur at the same pH regardless of concentration. The apparent pK
increases from 1.25 to 2.11 as the concentration of free chlorine in solution is increased. The
inconsistent location of the pK in this equilibrium strongly suggests that a protonation event is
not the source of the change and that equation 7 is the proper description of the low-pH behavior.
Indeed, the direction of the change - an increase in apparent pK as the free chlorine
concentration is increased - was predicted (in equation 8) when equation 7 was the proper
description. Furthermore, the magnitude of the change in apparent pK on going from 6.6mM to
26.2 mM - over 65% - was much larger than could be explained by changes in activity
coefficients over this concentration range. Indeed, changes in activity coefficients alone would
result in a change in apparent pK over this concentration range of less than 5%.
To further investigate the hypothesis that Cb was the correct assignment for the low-pH
form of aqueous free chlorine, a Raman spectrum was acquired of a solution (BCS) that had been
prepared by bubbling chlorine gas through an acidic 50mM phosphate buffered solution for
approximately 20 s. The pH of the BCS solution was intentionally kept low during preparation
for comparison to the diluted, acidic ECS solutions that were monitored earlier. The Raman
20
-------
spectrum of the acidic BCS solution is shown as trace A in Figure 4. The band at 538 cm"1, as
previously observed in low-pH ECS solutions prepared from HOC1, was very prominent in the
BCS spectrum and is much larger than the observed HOC1 vibration (at 725 cm"1). The Raman
spectrum of a diluted ECS solution at approximately the same pH is also shown, as trace B, in
Figure 4. Our hypothesis that Cb is the correct assignment of the low pH form of chlorine in
solution is strengthened because the band at 538 cm"1 is coincident in both spectra. The observed
disparity in relative band areas is simply indicative that Cb in the BCS solution is much more
concentrated relative to HOC1 than in the ECS solution.
If equation 7 were, indeed, the correct description of the low-pH behavior of aqueous free
chlorine, then minimizing the chloride ion (Cl") concentration in solution would diminish the
formation of the potent oxidant as the pH is reduced. As described above, our commercial free-
chlorine solutions (ECS) contained an equimolar amount of HOC1 and Cl". However, we were
able to prepare a stock low chloride solution (LCS) at alkaline pH using the method adapted
from Reinhard and Stumm (40) as described in the Experimental Section. This LCS stock
solution was diluted to 13.1 mM in free chlorine, which resulted in a Cl" concentration of 1.12
mM. For the ECS solution at this free-chlorine concentration, the concentration of Cl" (i.e. 13.1
mM) was more than ten times higher.
Figure 5 displays the Raman spectrum acquired from the LCS at pH 0.80 (trace B) and of
the ECS at a similar pH (trace A) with the same concentration of free chlorine (i.e., 13.1 mM). It
is easily seen that the ratio of band intensities in the Raman spectra at 538 to 725 cm"1 is much
larger for the ECS solution, which contained equimolar chloride ion. Figure 6 displays the
normalized band areas for the Raman bands located at 711 cm"1 (OC1") and 725 cm"1 (HOC1) as a
function of pH for the 13.1 mM LCS sample The normalized band area of the HOC1 Raman band
21
-------
at 725 cm"1 does not exhibit a clear and consistent decrease as the pH of the solution is dropped
below 5 as was observed in Figure 3 with the ECS solutions. Even at a pH below 0, there was
not a significant decrease in the HOC1 Raman band at 725 cm"1, nor is there any significant
Raman scattering observed in the 538 cm"1 region of the spectrum. (Note that a normalization
procedure that was simpler than that described in the Experimental Section was used to construct
the speciation plot in Figure 6. Because only two species were observed and both species could
be completely isolated one from the other within the pH range sampled, it was acceptable to
normalize each individual band area by the maximum area observed for that band.) Without
chloride ion present in solution, a reverse hydrolysis reaction cannot take place if equation 7 is
the correct description of the low-pH behavior. However, if equation 6 were the correct
description, then lowering the chloride ion concentration should have no effect on the
equilibrium towards H2OC1+ (equation 5). Hence all of the spectral observations lead to the
conclusion that the Raman band at 538 cm"1 results from scattering by Cb(aq).
Having established that equation 7 is the correct description of the low pH event, it is
possible to use data from Figure 3 to estimate the concentration-based equilibrium constant
K2°12. By rearranging equation 10, we establish that, when [HOC1] = [Cb]:
pKT =PH logfcrj (17)
The three speciation plots in Figure 3, and the assumed chloride ion concentrations, provide three
independent estimates of K2C12. The mean (± sample standard deviation) of these three estimates
was 2.56 (± 1.01) x 10"4 M2. However, note that this equilibrium constant should be considered
an estimate only. For example, it was based on only three determinations. Also, the Cl"
concentration was not confirmed by measurements in these ECS solutions. Instead, it was
assumed to be equal (as stated for the commercial product) to the known free-chlorine
22
-------
concentration. Furthermore, it is a concentration-based equilibrium constant. However, we note
that our value is in reasonable agreement with the concentration-based value previously reported
(42) of 6 x 10"4M2 for solutions with an ionic strength nearly 10 times greater than present in the
solutions that we used in these experiments.
CP loss was observed in the presence of both ECS and LCS chlorine solutions over the
pH range of 2-11. In the presence of ECS (Figure 1), the observed first order loss of CP
increased as pH decreased from 11 to 2. The insert in Figure 1 shows the loss of CP over the pH
range of 6-11. Duirk and Collette, (14) found over this pH range, that HOC1 was responsible for
the oxidation of CP to CPO and that OC1" accelerates the hydrolysis of both CP and CPO
chlorine assisted hydrolysis. Below pH 5, the observed CP loss rate greatly increased as pH
decreased from 5 to 2. This increase in the observed loss rate of CP was then attributed to the
formation of Cb(aq). In the presence of LCS (Figure 7), the observed first order rate of CP loss
did increase as pH decreased from 11 to 2. However, the observed rate of CP loss in the
presence of LCS at low pH was not nearly as significant as it was in the presence of ECS. At pH
of 2, the rate of loss of CP in LCS was about 2 orders of magnitude smaller than that observed
for the ECS. However, this increase in the rate of CP loss with LCS is significant, and, in part,
could be explained via the following proposed pathway. When HOC1 reacts with CP, chloride
ion is one of the products as well as a reduced sulfur species (S), equation 18. In the presence of
excess free chlorine,
HOC1 + CP »- CPO + (T + S + H+ (18)
4HOC1 + S » SO42" + 4Cr + 4H+ (19)
HOC1 + H+ + Cr . Cl2(aq) + H2O (20)
Cl2(aq) + CP »- products (21)
23
-------
S will be rapidly oxidized to sulfate (SC>42") resulting in four more moles of chloride for each
mole of S oxidized (61), equation 19. With the presence of additional chloride ions in solution at
low pH, the formation potential of Cb(aq) is increased greatly. Therefore, the observed rate of
CP loss in the presence of the LCS free chlorine solution increases due to the fact that aqueous
molecular chlorine can be formed under these reaction conditions.
3.2 Loss of CP in the Presence of Bromide and Free Chlorine
The presence of bromide in drinking water sources has long been a concern of the water
treatment industry because it can result in enhanced formation of potentially-hazardous DBFs
under drinking water treatment conditions. Since free chlorine can rapidly oxidize bromide to
hypobromous acid (HOBr) (28), it was thought that its presence might act as a catalyst in the
transformation of CP during drinking water chlorination.
Initial experiments to determine if bromide does accelerate the rate of CP loss were
conducted over the pH range of 7-9, [CP] = 0.5 |jM, [HOCl]x = 10 and 50 |jM, and increasing
bromide concentrations from 0-10 |jM. At [HOCl]x =10 |jM (Figure 8), plotting kobs versus the
bromide concentration was found to be linearly proportional to increasing bromide
concentration. This was also observed at pH 8 and 9 at the higher chlorine concentration of 50
|jM (Figure 9). Experiments were also conducted at pH 7 and 50 |jM chlorine concentration;
however, the reaction rate was too rapid to accurately quantify an observed rate of CP loss.
By comparing the slope of the regression lines at constant chlorine dose and varying pH
(Table 2), the k0bs in the presence of bromide and free chlorine was found to be pH dependent.
The pH of the solution determines the speciation of [HOCl]x between HOC1 and OC1".
Experiments conducted at lower pH have a greater percentage of [HOCl]x present as HOC1
species. Since the pKa of HOC1 (7.5) is close to the pH range over which experiments are
conducted, there is a large variation in the speciation of HOC1 as pH increased from 7 to 9. At
24
-------
pH 7, 8, and 9, the percentage of free chlorine present as HOC1 species is approximately 75%,
23%, and 3% respectively. Since HOC1 is primarily responsible for the formation of HOBr,
lowering the pH increased the rate of HOBr formation and subsequently the k0bs for CP loss.
The formation of HOBr, due to HOC1 oxidizing Br", could be responsible for the increase
in k0bs with respect to increasing bromide concentration. At the lower bromide concentrations
(0.1 and 1.0 jiM) and chlorine concentration of 10 |jM, CP is competitive with bromide for
oxidation by HOC1. Table 3 shows the rate coefficients for HOC1 reacting with both CP and
bromide. The reaction rate coefficient for HOC1 with bromide is approximately 3 times faster
than HOC1 with CP. As bromide concentrations increased to 5.0 and 10.0 jiM, active bromine
becomes present in greater abundance than CP. Therefore, HOC1 is more likely to oxidize
bromide than CP when chlorine to bromide molar ratios approach 1. At low bromide
concentrations and in the presence of excess free chlorine, the oxidation of CP is most likely
carried out by both HOC1 and HOBr. Therefore, bromide appears to act as a catalyst in the
oxidation of CP by free chlorine.
While all observed rates were faster with a 50 jiM chlorine concentration as compared to
the analogous experiment with [HOCl]x = 10.0 jiM, the magnitude of bromide-catalyzed CP
oxidation on k0bs compared to the control experiments was greater at the lower chlorine dose for
pH 8 and 9. At the 10 jiM chlorine concentration, addition of 10 jiM of bromide induced a -20
fold and -30 fold increase in k0bs for pH 9 and 8 respectively, whereas addition of 10 jiM of
bromide at the 50 jiM chlorine dose only increased the k0bs -4 fold and 10 fold at pH 9 and 8
respectively. This indicates that the magnitude of bromide catalyzed oxidation of CP is
controlled by the ratio of HOC1 to bromide as well as pH. When a large excess of HOC1 exists,
it is the primary oxidant responsible for the loss of CP. However, when [HOC1]: [bromide] molar
25
-------
ratios approach unity, HOBr becomes the primary oxidant. These two reaction scenarios are
depicted schematically in Figure 10.
The relationship between k0bs and bromide concentration shows that increasing bromide
is linearly proportional to the observed rate of CP loss in the presence of free chlorine. This
result indicates that HOBr is likely the dominant oxidant responsible for CP degradation at
higher bromide concentrations. The linearity of the trend also indicates that bromide acts as a
catalyst in the oxidation reaction between HOC1 and CP. Bromide is a catalyst in this pathway
because it does not appear in the balanced equation for the reaction and after being reduced by
chlorpyrifos, HOBr yields H+ and Br". The regenerated bromide ion can then be re-oxidized by
HOC1 and participate in additional reactions with CP. The slope of the regression line for k0bs
vs. [Br"] concentration was observed to be dependent on the concentration of free chlorine, pH,
and the concentration of bromide (Table 2). All three of these parameters affect the rate of
HOBr formation.
A model was developed to predict the loss of CP in the presence of free chlorine and
bromide. From the stoichiometric equations in Table 3, a system of ordinary differential
equations (ODEs), equations 22-28, were incorporated into a model developed to predict CP loss
pathways in buffered deionized water (14). Key assumptions in this model are that HOBr will
only react with CP resulting in CPO, and that all HOBr that reacts with CP results in
d[HOCl]T
- 4 * kHOC1 s [HOC1][S] - kHOC1 Br [HOCl][Br
-kHOC1 cp [HOC1][CP] - kocl cp [OC1 ][CP] - kocl cpo [OC1 ][CPO]
dt ' ' (22)
d[HOBr]T
dt
d[Br-]
= kHOC1Br[HOCl][Br-]-kHOBrCP[HOBr][CP] (23)
dt
= -kHOClBr[HOCl][Br-]-kHOBr,cp[HOBr][CP] (24)
26
-------
kHOBrCP[HOBr][CP]-kHOCls[HOCl][S] (25)
^T^ = -kHOC1GP[HOCl][CP] - khcp[CP] - koclcp[OCr ][CP]
dt ' (26)
-kHOBrCP[HOBr][CP]
dt ' (27)
+ kHOBrCP[HOBr][CP]
= kh cp [CP] + koacp [OC1 ][CP] + kh cpo [CPO] + kocl cpo [OC1- ][CPO] (28)
the regeneration of bromide in the aqueous system. Also, that oxidation of the reduced sulfur
species (S) can be adequately predicted by the rate coefficient determined for HOC1 reaction
with sulfite (SO32") (61). Scientist™, an ODE solver by Micromath (Salt Lake City, UT), was
used to fit the rate coefficient for CP oxidation by hypobromous acid (kHoBr,cp). Scientist™ uses
a modified Powell algorithm to minimize the unweighted sum of the squares of the residual error
between the predicted and experimentally observed values to estimate specific parameters in the
model.
To aid in parameter estimation, the loss of CP in the presence of HOBr was examined at
pH 6.5 (Figure 1 1). At this pH, over 99% of the active bromine present will be in the HOBr
form due to its pKa of 8.8 (28). The reaction of HOBr with CP was expected to be 1 to 3 orders
of magnitude faster than HOC1 with CP. This has been observed by others examining the
reactivity of HOC1 and HOBr with pyrene, NOM, and sulfite (30,38,61,62). The reaction was
essentially complete after only 10 seconds. The second order rate coefficient for this reaction
was determined by nonlinear regression analysis using Sigma Plot version 8.0 (Point Richmond,
CA). The kHoBr,cp was found to be 1.14 x 109 M^h"1, which is three orders of magnitude greater
than HOC1 oxidation of CP (kHOci,cp = 1.72 x 106 M'V).
27
-------
The system of ODEs in equations 22-28 were used to model the loss of CP in the
presence of free chlorine and bromide. In Figures 12-14, model results are shown for CP loss as
well as the formation of CPO and TCP at pH 8 and [HOC1]T = 10, 25 and 50 |iM in the absence
of bromide. The model adequately predicts loss of CP and the formation of CPO and TCP. The
only significant change from the model presented here verses the original model by Duirk and
Collette (14) is the incorporation of oxidation of the reduced sulfur species, equation 25. Since
no rate coefficients exist in literature for the oxidation of S by HOC1, it was assumed that the rate
coefficient for HOC1 reacting with SOs2" was sufficiently large enough to predict the loss of free
chlorine due to the reaction with S. Under these experimental conditions, the model appears
quite capable of predicting CP degradation pathways in the presence of free chlorine.
At pH of 8, two different sets of experiments were conducted to determine if the
experimentally determined rate coefficient for kHoBr,cp was adequate to describe the loss of CP in
the presence of bromide and free chlorine. The first set of experiments was conducted at a
constant bromide concentration of 1 (jM, with increasing free chlorine concentrations from 10 to
50 (jM. The second set of experiments was conducted at constant free chlorine concentration of
50 (jM, with increasing bromide concentrations from 1 to 10 (jM. Using the experimentally
determined kHoBr,cp, CP loss was found to be significantly over-predicted for both sets of
experiments. This could have been due to the fact that all active bromine being formed was
assumed to only react with CP. However, HOBr does react with reduced sulfur species, such as
sulfite, more rapidly than HOC1 (61,62).
In the current model, all reduced sulfur species are assumed to be oxidized by HOC1 and
all active bromine reacts with CP. Since reaction rate coefficients for HOC1 and HOBr with all
the reduced sulfur species from S to SOs2" are not available in literature, it is not possible to
28
-------
proportion the amount of HOBr that could potentially react with either CP or one of the reduced
sulfur species. Therefore, an apparent second order rate coefficient (kapp HOBr,cp) was determined
by pooling these data sets to predict the loss of CP in the presence of free chlorine and bromide
for these experimental conditions (Figures 15-20). The kapp HOBr,cp was found to be 6.01(+0.40)
x 107 M^h"1 (95% confidence interval shown in parentheses), which is two orders of magnitude
less than the experimentally determined knoBr,cp. Although the apparent rate coefficient is
significantly smaller than the experimentally determined rate coefficient, it was able to
adequately predict the loss of CP and the formation of CPO and TCP under these experimental
conditions using this value for the apparent rate coefficient. Also, it still falls within the
expected range of 1 to 3 orders of magnitude faster than its the chlorinated analog (61,62). The
reaction with the HOBr and CP may have been too fast to be adequately captured by the
experimental procedure used to investigate HOC1 reactions with CP below pH of 8. This might
explain, in part, the discrepancy between the experimentally determined and the apparent rate
coefficients. In any event, these results indicate that active bromine pathways in the presence of
OP pesticides under drinking water treatment conditions still needs to be investigated further.
3.3 Loss of CP in the Presence of Free Chlorine and NOM
The presence of NOM could potentially decrease the rate of CP degradation by acting as
a sink for oxidants used in drinking water treatment. Conventional physical-chemical water
treatment processes such as coagulation and lime softening remove a portion of the hydrophobic
fraction of the NOM (63,64). More advanced water treatment processes like granular activated
carbon (GAC) and membranes are capable of removing more hydrophilic NOM fractions (65).
However, few community water systems employ these more advanced technologies. The ACC
water treatment plant uses only conventional surface water treatment processes. Therefore, only
a portion of the hydrophobic NOM fraction will be removed. Figure 21 shows the chlorine
29
-------
demand studies conducted over the pH range of 7-9. Over the first half hour, approximately 12%
of the free chlorine initially present is consumed by the NOM. After the initial chlorine demand
of the NOM is satisfied, the loss of free chlorine attributable to reaction with NOM is relatively
slow (k0bs = 0.012 h"1) over the pH range of 7-9. These results suggest that the NOM present in
the ACC water may affect the rate of CP transformation over the first half hour in the presence of
free chlorine.
To examine the effect of NOM on the transformation of CP, experiments were conducted
over the pH range of 7-9 and in the presence of 50 |jM free chlorine. Figure 22 shows the
observed first order rate of CP loss with and without the presence of NOM. In the presence of
ACC NOM, the observed loss of CP was very similar to the control experiments. It appears that
the conventional treatment performed at the ACC removed a portion of the hydrophobic NOM
fraction that is usually associated with significant chlorine demand (36). Therefore, it might
possible for the current model (developed with experiments in laboratory water) to accurately
predict the transformation of CP in this low humic water.
Figures 23-27 show the experimental and model results for the loss of CP and formation
of CPO and TCP in ACC water for a variety of experimental conditions. At pH 7 and 7.5
(Figures 23 and 24), the model adequately predicted CP loss and the formation of CPO and TCP.
However, at these pHs, the model did slightly over-predict CP loss. This could have been due to
bromide being present at 0.35 |jM, but this low bromide concentration was not expected to
significantly affect the rate of CP transformation (See Figures 8 and 9). On the other hand, the
model performed very well predicting the formation and stability of CPO, the more toxic
transformation product, as well as TCP. Over the pH range of 8-9 (Figures 25-27), the model
adequately predicted the loss of CP over the experimental time frame. At pH values greater than
30
-------
7.5, CPO becomes susceptible to chlorine-assisted hydrolysis due to the greater abundance of the
hypochlorite ion. The model not only predicted CPO formation, but it also showed that CPO
was slowly being transformed to TCP over the course of the experiment at pH 8-9. This appears
to validate the ability of the model to adequately predict the loss of CP in the presence of free
chlorine and NOM, as well as the ability of the model to predict the formation and stability of
CPO in the presence of free chlorine.
-------
4 CONCLUSIONS
The goal of this work was to determine the active oxidant species responsible for the
degradation of CP under drinking water treatment conditions. Using Raman spectroscopy, the
third species in free chlorine solutions at low pH was identified, which is responsible for the very
rapid oxidation of CP at low pH. The Raman band at 538 cm"1 appeared as the pH of the ECS
free chlorine solution was lowered and indicated the presence of a third species in addition to the
hypochlorite ion and hypochlorous acid. The speciation profiles of ECS solutions at different
concentrations showing variation in the low-pH "apparent pKa," and disappearance of the band
at 538 cm"1 in LCS solutions indicated that the third species in solution is Cl2(aq) and not
H2OC1+ as has been frequently assumed.
In the presence of bromide, the transformation of CP was found to be accelerated upon
chlorination. HOC1 rapidly oxidizes bromide to HOBr, which is a stronger oxidant. Once HOBr
reacts with CP, bromide is then regenerated in the aqueous system (i.e., bromide is a catalyst
increasing the rate of CP transformation). In the presence of HOBr, the reaction rate coefficient
of CP with HOBr was found to be three orders of magnitude greater than the rate coefficient
found for HOC1 with CP. However, the experimentally determined rate coefficient was found to
be too large to accurately predict the loss of CP and the formation of CPO and TCP under the
experimental conditions used in this study. An apparent second order rate coefficient
(kapp HOBr,cp), which was two orders of magnitude less than the experimentally determined rate
coefficient, was found to adequately fit the experimental results. This was assumed to be due to
HOBr reacting with not only CP but with reduced sulfur species evolving from the initial
oxidation of CP by either HOC1 or HOBr. Future work will be done to further elucidate active
bromine pathways in the presence of free chlorine and CP.
32
-------
The role of NOM on the transformation of CP was investigated using Athens-Clarke
County water, which was collected prior to chlorination. Under the experimental conditions
used in this study, the established model, initially developed to predict the transformation
pathways of CP in laboratory water, was found to adequately predict the loss of CP and the
formation of CPO and TCP in the ACC water. It was assumed that due to the low humic content
of the ACC water, the chlorine demand that was exerted by the NOM was not sufficient to
significantly affect the rate of CP transformation. Also, the model was able to predict the
formation and stability of CPO over the entire pH range. In waters that are low in aquatic humic
substances, the model appears to adequately predict CP transformation pathways.
The work presented here shows the application of a model, developed to predict the loss
of CP in buffered aqueous systems, to predict the loss of CP over a greater range of conditions
that are more closely associated with drinking water treatment. However, the need still exists to
systematically investigate chlorination of OP pesticides as a class and determine if reactivity
with aqueous chlorine is related in a predictable way to OP chemical structure. If a relationship
between reactivity and pesticide structure exists, the ability to predict the fate of OP pesticides in
the presence of free chlorine would greatly aid regulators in assuring that tolerances to OP
pesticide exposure from drinking water are not exceeded. After developing predictive tools for
the entire class of OP pesticides, we plan to address other important classes of pesticides and
toxic chemicals.
33
-------
5 REFERENCES
(1) Kolpin, D. W.; Barbash, J. E.; Gilliom, R. J. Pesticides in ground water of the United
States, 1992-1996. Ground Water 2000, 38, 858-863.
(2) Verstraeten, I. M.; Thurman, E. M.; Lindsey, M. E.; Lee, E. C.; Smith, R. D. Changes in
concentrations of triazine and acetamide herbicides by bank filtration, ozonation, and
chlorination in a public water supply. Journal of Hydrology 2003, 276, 287-288.
(3) Coupe, R. H.; Blomquist, J. D. Water-soluble pesticides in finished water of community
water supplies. Journal American Water Works Association 2004, 96, 56-68.
(4) Community Water System Survey, United States Environmental Protection Agency, 1997;
Vol. l,EPA815-R-97-011a.
(5) Lopez, A.; Mascolo, G.; Tiravanti, G.; Santori, M.; Passino, R. Oxidation of Sulfur-
Containing S-Triazines During Groundwater Hypochlorination. Water Science and
Technology 1994, 30, 53-59.
(6) Mascolo, G.; Lopez, A.; Passino, R.; Ricco, G.; Tiravanti, G. Degradation of Sulfur-
Containing S-Triazines During Water Chlorination. Water Research 1994, 28, 2499-
2506.
(7) Adams, C. D.; Randtke, S. J. Removal of Atrazine from Drinking-Water by Ozonation.
Journal American Water Works Association 1992, 84, 91-102.
(8) Miles, C. J. Degradation of Aldicarb, Aldicarb Sulfoxide, and Aldicarb Sulfone in
Chlorinated Water. Environmental Science & Technology 1991, 25, 1774-1779.
(9) Mason, Y. Z.; Choshen, E.; Ravacha, C. Carbamate Insecticides - Removal from Water
by Chlorination and Ozonation. Water Research 1990, 24, 11-21.
(10) Kodama, S.; Yamamoto, A.; Matsunaga, A. S-Oxygenation of thiobencarb in tap water
processed by chlorination. Journal of Agricultural and Food Chemistry 1997, 45, 990-
994.
(11) Zhang, Q.; Pehkonen, S. O. Oxidation of diazinon by aqueous chlorine: Kinetics,
mechanisms, and product studies. Journal of Agricultural and Food Chemistry 1999, 47,
1760-1766.
(12) Magara, Y.; Aizawa, T.; Matumoto, N.; Souna, F. Degradation of Pesticides by
Chlorination During Water-Purification. Water Science and Technology 1994, 30, 119-
128.
(13) Wu, J. G.; Laird, D. A. Abiotic transformation of chlorpyrifos to chlorpyrifos oxon in
chlorinated water. Environmental Toxicology and Chemistry 2003, 22, 261-264.
(14) Duirk, S. E.; Collette, T. W. Degradation of Chlorpyrifos in Aqueous Chlorine Solutions:
Pathways, Kinetics, and Modeling. Environmental Science and Technology 2006, 40,
546-551.
(15) Deborde, M.; Rabouan, S.; Gallard, H.; Legube, B. Aqueous chlorination kinetics of
some endocrine disrupters. Environmental Science & Technology 2004, 38, 5577-5583.
(16) Gallard, H.; Von Gunten, U. Chlorination of Phenols: Kinetics and Formation of
Chloroform. Environmental Science & Technology 2002, 36, 884-890.
(17) Rule, K. L.; Ebbert, V. R.; Vikesland, P. J. Formation of Chloroform and Chlorinated
Organics by Free-Chlorine-Mediated Oxidation of Triclosan. Environmental Science and
Technology 2005, 39, 3176-3185.
(18) Rebenne, L. M.; Gonzalez, A. C.; Olson, T. M. Aqueous chlorination kinetics and
mechanism of substituted dihydroxybenzenes. Environmental Science & Technology
1996, 30, 2235-2242.
34
-------
(19) Arotsky, J.; Symons, M. C. R. Halogen Cations. Quarterly Reviews of the Chemical
Society 1962, 16, 282-297.
(20) Donaldson, D. J.; Ravishankara, A. R.; Hanson, D. R. Detailed Study of HOC1 + HC1 ->
C12 + H2O in Sulfuric Acid. Journal of Physical Chemistry A 1997, 101, 4717-4725.
(21) Liu, D. K.; Chang, S.-G. Removal of Nitric Oxide from Flue Gas Using Water-Soluble
Iron(II) Dithiocarbamates. Environmental Science Technology 1988, 22, 1196-2000.
(22) Swain, C. G.; Crist, D. R. Mechanisms of Chlorination by Hypochlorous Acid. The Last
of the Chloronium Ion, C1+. Journal of the American Chemical Society 1972, 94, 3195-
3200.
(23) Swain, C. G.; Ketley, A. D. The Chloronium Ion as an Intermediate in Chlorination of
Aromatic Compounds by Hypochlorous Acid. Journal of the American Chemical Society
1955, 77, 3410.
(24) Saha, B.; Stanbury, D. M. Oxidation of [Ru(NH3)5isn](BF4)2 by Hypochlorous Acid and
Chlorine in Aqueous Acidic Media. Inorganic Chemistry 2001, 40, 5139-5146.
(25) Cooper, J. N.; Margerum, D. W. Kinetics and mechanism of the oxidation of
hydroxylamine by aqueous chlorine Inorganic Chemistry 1993, 32, 5905-5910.
(26) Dowell, C. T. The Action of Chlorine Upon Hydrazine, Hydroxylamine and Urea.
Journal of the American Chemical Society 1919, 41, 124-125.
(27) Siddiqui, M. S.; Amy, G. L.; Rice, R. G. Bromate Ion Formation - a Critical-Review.
Journal American Water Works Association 1995, 87, 58-70.
(28) Kumar, K.; Margerum, D. W. Kinetics and Mechanism of General-Acid-Assisted
Oxidation of Bromide by Hypochlorite and Hypochlorous Acid. Inorganic Chemistry
1987,26,2706-2711.
(29) Bousher, A.; Brimblecombe, P.; Midgley, D. Rate of Hypobromite Formation in
Chlorinated Seawater. Water Research 1986, 20, 865-870.
(30) Hu, J. Y.; Jin, X. H.; Kunikane, S.; Terao, Y.; Aizawa, T. Transformation of pyrene in
aqueous chlorination in the presence and absence of bromide ion: Kinetics, products, and
their aryl hydrocarbon receptor-mediated activities. Environmental Science &
Technology 2006, 40, 487-493.
(31) Pourmoghaddas, H.; Stevens, A. A. Relationship between Trihalomethanes and
Haloacetic Acids with Total Organic Halogen During Chlorination. Water Research
1995, 29, 2059-2062.
(32) Pourmoghaddas, H.; Stevens, A. A.; Kinman, R. N.; Dressman, R. C.; Moore, L. A.;
Ireland, J. C. Effect of Bromide Ion on Formation of Haas During Chlorination. Journal
American Water Works Association 1993, 85, 82-87.
(33) Singer, P. C. Humic substances as precursors for potentially harmful disinfection by-
products. Water Science and Technology 1999, 40, 25-30.
(34) Westerhoff, P.; Siddiqui, M. S.; Debroux, J.; Zhai, W.; Ozekin, K.; Amy, G. Nation-wide
Bromide Occurence and Bromate Formation Potential in Drinking Water Supplies;
Amreican Society of Civil Engineers: Salem, MA, 1994.
(35) Thurman, E. M. Organic geochemistry of natural waters; Junk Publishers: Boston, MA,
1985.
(36) Najm, I. N.; Patania, N. L.; Jacangelo, J. G.; Krasner, S. W. Evaluating Surrogates for
Disinfection by-Products. Journal American Water Works Association 1994, 86, 98-106.
35
-------
(37) Reckhow, D. A.; Singer, P. C.; Malcolm, R. L. Chlorination of Humic Materials - by-
Product Formation and Chemical Interpretations. Environmental Science & Technology
1990, 24, 1655-1664.
(38) Westerhoff, P.; Chao, P.; Mash, H. Reactivity of natural organic matter with aqueous
chlorine and bromine. Water Research 2004, 38, 1502-1513.
(39) Standard Methods for the Examination of Water and Wastewater; 20th ed.; APHA,
AWWA and WEF: Washington D. C., 1998.
(40) Reinhard, M.; Stumm, W. Kinetics of Chlorination of p-Xylene in Aqueous Solution. In
Water Chlorination: Chemistry, Environmental Impact, and Health Effects; Jolley, R. L.,
Ed.; Lewis Publishers: Ann Arbor, 1980; Vol. 3, pp 209-218.
(41) Leaist, D. G. Absorption of Chlorine into Water. Journal of Solution Chemistry 1986, 15,
827-838.
(42) Eigen, M.; Kustin, K. The Kinetics of Halogen Hydrolysis. Journal of the American
Chemical Society 1962, 84, 1355-1361.
(43) Lifshitz, A.; Perlmutter-Hayman, B. The Kinetics of the Hydrolysis of Chlorine. I.
Reinvestigation of the Hydrolysis in Pure Drinking Water. Journal of Physical Chemistry
1960, 64, 1663-1665.
(44) Wang, T. X.; Margerum, D. W. Kinetics of Reversible Chlorine Hydrolysis -
Temperature-Dependence and General Acid Base-Assisted Mechanisms. Inorganic
Chemistry 1994, 33, 1050-1055.
(45) Aieta, E. M.; Roberts, P. V. Application of Mass-Transfer Theory to the Kinetics of a
Fast Gas-Liquid Reaction: Chlorine Hydrolysis. Environmental Science & Technology
1986, 20, 44-50.
(46) Connick, R. E. The Interaction of Hydrogen Peroxide and Hypochlorous Acid in Acidic
Solutions Containing Chloride Ion. Journal of the American Chemical Society 1947, 69,
1509-1513.
(47) Connick, R. E.; Chia, Y.-T. The Hydrolysis of Chlorine and its Variation with
Temperature. Journal of the American Chemical Society 1959, 81, 1280-1283.
(48) Gershenzon, M.; Davidovits, P.; Jayne, J. T.; Kolb, C. E.; Worsnop, D. R. Rate Constant
for the Reaction of C12 (aq) with OH-. Journal of Physical Chemistry A 2002, 106, 7748-
7754.
(49) Lifshitz, A.; Perlmutter-Hayman, B. The Kinetics of the Hydrolysis of Chlorine. II. The
Hydrolysis in the Presence of Acetate. Journal of Physical Chemistry 1961, 65, 753-757.
(50) Lifshitz, A.; Perlmutter-Hayman, B. The Kinetics of the Hydrolysis of Chlorine. III. The
Reaction in the Presence of Various Bases and a Discussion of the Mechanism. Journal
of Physical Chemistry 1962, 66, 701-705.
(51) Rehfeld, B. M.; Pratt, D. E. Influence of Organophosphate Insecticides on Ascorbic Acid
Oxidation in Aqueous Systems. Journal of Food Science 1969, 34, 551-&.
(52) Everall, N.; Owen, H.; Slater, J. Performance Analysis of an Integrated Process Raman
Analyzer Using a Multiplexed Transmission Holographic Grating, CCD Detection, and
Confocal Fiberoptic Sampling. AppliedSpectroscopy 1995, 49, 610-615.
(53) Williams, T. L.; Martin, R. B.; Collette, T. W. Raman Spectroscopic Analysis of
Fertilizers and Plant Tissue for Perchlroate. Applied Spectroscopy 2001, 55, 967-988.
(54) Collette, T. W.; Williams, T. L.; D'Angelo, J. C. Optimization of Raman Spectroscopy
for Speciation of Organics in Water. Applied Spectroscopy 2001, 55, 750 - 766.
36
-------
(55) Liu, B.; McConnell, L. L.; Torrents, A. Hydrolysis of chlorpyrifos in natural waters of
the Chesapeake Bay. Chemosphere 2001, 44, 1315-1323.
(56) Hu, J. Y.; Morita, T.; Magara, Y.; Aizawa, T. Evaluation of reactivity of pesticides with
ozone in water using the energies of frontier molecular orbitals. Water Research 2000,
34, 2215-2222.
(57) Kolpin, D. W.; Furlong, E. T.; Meyer, M. T.; Thurman, E. M.; Zaugg, S. D.; Barber, L.
B.; Buxton, H. T. Pharmaceuticals, hormones, and other organic wastewater
contaminants in US streams, 1999-2000: A national reconnaissance. Environmental
Science & Technology 2002, 36, 1202-1211.
(58) Nakagawara, S.; Goto, T.; Nara, M.; Ozawa, Y.; Hotta, K.; Arata, Y. Spectroscopic
Characterization and the pH Dependence of Bactericidal Activity of the Aqueous
Chlorine Solution. Analytical Sciences 1998, 14, 691-698.
(59) Snoeyink, V. L.; Jenkins, D. A. Water Chemistry, John Wiley & Sons: New York, NY,
1980.
(60) Skoog, D. A.; West, D. M. In Fundamentals of Analytical Chemistry, 2nd ed.; Holt,
Reinhart and Winston, Inc.: New York, 1969; pp 294-297.
(61) Fogelman, K. D.; Walker, D. M.; Margerum, D. W. Non-Metal Redox Kinetics -
Hypochlorite and Hypochlorous Acid Reactions with Sulfite. Inorganic Chemistry 1989,
28, 986-993.
(62) Troy, R. C.; Margerum, D. W. Nonmetal Redox Kinetics - Hypobromite and
Hypobromous Acid Reactions with Iodide and with Sulfite and the Hydrolysis of
Bromosulfate. Inorganic Chemistry 1991, 30, 3538-3543.
(63) Bob, M.; Walker, H. W. Lime-soda softening process modifications for enhanced NOM
removal. Journal of Environmental Engineer'ing-Asce 2006, 732, 158-165.
(64) Crozes, G.; White, P.; Marshall, M. Enhance Coagulation - Its Effect on Nom Removal
and Chemical Costs. Journal American Water Works Association 1995, 87, 78-89.
(65) Jacangelo, J. G.; Demarco, J.; Owen, D. M.; Randtke, S. J. Selected Processes for
Removing Nom - an Overview. Journal American Water Works Association 1995, 87,
64-77.
(66) Macalady, D. L.; Wolfe, N. L. New Perspectives on the Hydrolytic Degradation of the
Organophosphorothioate Insecticide Chlorpyrifos. Journal of Agricultural and Food
Chemistry 1983, 31, 1139-1147.
37
-------
TABLES
38
-------
Table 1 NOM and source water characteristics for the ACC water collected prior to
chlorination.
Source
ACC
SUVA254
(L/m-mg)
1.61
SUVA280
(L/m-mg)
1.07
a[NH3]T
(HM)
2.06
[C1-]
(HM)
93.46
[N02-]
(HM)
0.04
[Brl
(HM)
0.35
b[P04]T
(HM)
CND
[S042-]
(HM)
8.30
a[NH3]T = [NH4 ] + [NH3]
b[P04]T=[H3P04]
CND = Non-detect
b[P04]T = [H3P04] + [H2P04-] + [HP042-] + [P043-]
39
-------
Table 2 The slopes of k0bs vs. [Br"] for each pH and [HOC1] concentration.
pH [HOC1] (|iM) kobs (h'1) vs. [Br']
7 10 10.3
8 10 2.7
9 10 0.4
8 50 5.3
9 50 0.6
40
-------
Table 3 Stoichiometric equations and rate coefficients used in the chlorpyrifos degradation
pathway model.
1
2
O
4
5
6
7
8
9
Reaction Stoichiometry Rate/Equilibrium Coefficient
(25 °C)
HOC1 + CP kr >CPO + H++C1 +S kHOci,cp= 1.72x10 M' h'
4HOri i S HOCI, s v /ITT+ i /if]- , en2" HOC1'S
HOC1 i Rr kHoci,Br >uoRr i C1 kHoci,Br=5.6 x 10 M h
HORr i CP kHoBr,cp >rpr) , H+ , r] kHoBr,cp = see text
/-r> khcP ^^ kh;Cp = kN;Cp + kB,Cp[OH"]
cpo khcPO kh;Cpo = kN,Cpo + kB!cpo[OH-]
kB,cpo = 230 M^h"1
rp , on- k°^cp , TCP , rr koclcp = "° M"lh"1
^^ , ^,- koci CPO . T™ , ^,- koci,cpo = 1340 M^h-1
MUUl ^ M + UU1 pKa — /.5
Reference
(14)
(61)
(28)
This work
(66)
(14)
(14)
(14)
(59)
41
-------
FIGURES
42
-------
zuuu
1500 -
1000 -
500 -
n -
10 i
8 •
• "e 6 •
J 4-
2 •
0 •
(
•
5 7 8 9 10 11 12
pH
• . •. . m . ••• . m , m .
10
12
Figure 1 Observed first order loss rate coefficients for chlorpyrifos over the pH range of 2-
11 in the presence of ECS. [CP]0 = 0.5 |iM, [Buffer]T = 10 mM, Temperature =
25±1°C, and[HOCl]T= 10
43
-------
CO
C
CO
CD
"S 0.5
:=
"(0
A
0
6
PH
8
10
12
c
0)
-I—•
eh
11.96
B
7.46
4.59
2.73
•low pH
form
800 750 700 650 600 550 500 450
Raman Shift (cm-1)
Figure 2 (A.) Speciation plot of the 26.2 mM commercial free chlorine solution (the ECS
solution) generated by monitoring the normalized areas of the Raman bands from
OC1" at 711 cm-1 (A), HOCI at 725 cm-l(B) and the low pH species at 538 cm"1
(4). (B.) Background-subtracted Raman spectra of ECS solutions. The OC1" band
is observed in the top two spectra and the HOCI band is observed in the bottom
three spectra. An additional band for the low pH form appears as the pH is lowered
below 5.
44
-------
N
"(0
0.5
0
(0 -I
I
T3
0.5
0
1
0.5
0
*'•
iS.
26.2 mM
13.1 mM
/ A
6.6 mM
/*
0 2 4 6 8 10 12
PH
Figure 3 Speciation plots of commercial free chlorine solutions (ECS solutions) generated
by monitoring the normalized areas of the Raman bands from OC1" at 711 cm-1 (±,
dash-dot model line), HOC1 at 725 cm-1 (•, dashed model line) and the low pH
species at 538 cm"1 (4, solid model line). Vertical lines indicate the apparent pKas
at 13.1 mM. The pKa for the protonation of OC1" to HOC1 occurs at 7.50 for all
three concentrations. The apparent pK at lower pH varies from (from top to
bottom): 2.11, 1.84, and 1.25.
45
-------
800 750
700 650 600 550
Raman Shift (cnv1)
500 450
Figure 4 Raman spectrum acquired from (A) a buffered aqueous solution at pH 1.45 through
which chlorine gas was bubbled (BCS) and (B) a pH 1.63 solution containing 13.1
mM commercial free chlorine (ECS).
of C\2 (aq) in each solution.
-i
The band at 538 cm" is due to the presence
46
-------
800 750
700 650 600 550
Raman Shift (cnv1)
500 450
Figure 5 Raman spectra comparing 13.1 mM free chlorine solutions with (A) equimolar
chloride ion at pH 0.75 (ECS) and (B) free chlorine solution with reduced chloride
ion at pH 0.80 (LCS). The spectrum of the solution containing equimolar chloride
ion has a much larger band at 538 cm" (due to aqueous Cb) than the solution with
greatly reduced chloride concentration.
47
-------
CO
0
A A
CD
0
Q_
-Q
0
"CD
E
o
AAA A A
0
-2
2468
PH
10
12
Figure 6 Speciation profile of the LCS free chlorine solution with reduced chloride ion
concentration generated by monitoring the normalized areas of the Raman bands
from HOC1 at 725 cm-1 (•) and OC1" at 711 cm-1 (±). The predominant species in
solution at low pH remains HOC1 throughout, and a significant drop in its total area
is not observed as pH decreases.
48
-------
40
30 -
20 -
10 -
6
PH
10
12
Figure 7 Observed first order loss rate coefficients for chlorpyrifos over the pH range of 2-
11 in the presence of LCS. [CP]0 = 0.5 |iM, [Buffer]T = 10 mM, Temperature =
25±1°C, and [HOC1]T = 10
49
-------
125
100 -
75 -
50 -
25 -
10
12
Bromide (|JM)
Figure 8 Observed first order rate coefficients for CP loss in the presence of increasing
bromide concentrations at [HOCl]x = 10 |jM over the pH range of 7-9. [CP]0 = 0.5
[Buffer]T = 10 mM, and Temperature = 25 °C.
50
-------
10
12
Bromide (|JM)
Figure 9
Observed first order rate coefficients for CP loss in the presence of increasing
bromide concentrations at [HOC1]T = 50 [iM at pH 8 and 9. [CP]0 = 0.5
[Buffer]T = 10 mM, and Temperature = 25 °C.
51
-------
Br-
Figure 10 Two possible oxidation pathways for CP at high and low [HOCl]:[Br"J ratios. The
pathway on the left shows the predominant pathway when [HOCl]:»[Br"] and the
pathway on the right shows the predominant pathway when [HOCl]:[Br"]
approaches 1.
52
-------
fin
o
0.5
0.4 -
0.3 -
0.2 -
0.1 -
0.0
4 6
Time (seconds)
10 12
Figure 11 Loss of CP in the presence of hypobromous acid at pH 6.5. [CP]0 = 0.42 |jM,
[HOBr]T = 10 nM, [Buffer]T = 10 mM, and Temperature = 25 °C.
53
-------
0.0
0.5
1.0 1.5
Time (hours)
2.0
2.5
Figure 12
CP degradation in the presence of free chlorine at pH 8.0 in the absence of
bromide. [CP]0 = 0.39 |iM, [HOC1]T = 10 |iM, [PO4]T = 10 mM, and Temperature
= 25 °C. Lines represent model results.
54
-------
0.00
0.25
0.50 0.75
Time (hours)
1.00
1.25
Figure 13
CP degradation in the presence of free chlorine at pH 8.0 in the absence of
bromide. [CP]0 = 0.42 |iM, [HOC1]T = 25 |iM, [PO4]T = 10 mM, and Temperature
= 25 °C. Lines represent model results.
55
-------
0.00
0.25 0.50
Time (hours)
0.75
Figure 14
CP degradation in the presence of free chlorine at pH 8.0 in the absence of
bromide. [CP]0 = 0.42 |iM, [HOC1]T = 50 |iM, [PO4]T = 10 mM, and Temperature
= 25 °C. Lines represent model results.
56
-------
o
'I
-4—»
(L>
O
O
O
0.0
0.1
0.2
Time (hours)
0.3
0.4
Figure 15
CP degradation in the presence of free chlorine and bromide at pH 8.0. [CP]0 =
0.51 nM, [HOC1]T = 10 nM, [Br ] = 1 |iM, [PO4]T = 10 mM, and Temperature = 25
°C. Lines represent model results.
57
-------
o
'I
-4—»
(L>
O
O
O
0.2 -
0.1 -
0.0
0.00 0.02 0.04 0.06 0.08
Time (hours)
0.10
0.12
0.14
Figure 16
CP degradation in the presence of free chlorine and bromide at pH 8.0. [CP]0 =
0.48 nM, [HOC1]T = 25 |iM, [Br ] = 1 |iM, [PO4]T = 10 mM, and Temperature = 25
°C. Lines represent model results.
58
-------
o
'I
-4—»
(L>
O
O
O
0.2 -
0.1 -
0.0
0.00
0.02
0.04
Time (hours)
0.06
0.08
Figure 17
CP degradation in the presence of free chlorine and bromide at pH 8.0. [CP]0 =
0.49 nM, [HOC1]T = 50 |iM, [Br ] = 1 |iM, [PO4]T = 10 mM, and Temperature = 25
°C. Lines represent model results.
59
-------
0.6
0.5 -
0.0
0.0
CP
CPO
TCP
0.1
0.2
Time (hours)
0.3
0.4
Figure 18
CP degradation in the presence of free chlorine and bromide at pH 8.0. [CP]0 =
0.35 nM, [HOC1]T = 50 |iM, [Br ] = 1 |iM, [PO4]T = 10 mM, and Temperature = 25
°C. Lines represent model results.
60
-------
0.000 0.025 0.050 0.075 0.100
Time (hours)
0.125
0.150
Figure 19
CP degradation in the presence of free chlorine and bromide at pH 8.0. [CP]0 =
0.43 nM, [HOC1]T = 50 |iM, [Br ] = 5 |iM, [PO4]T = 10 mM, and Temperature = 25
°C. Lines represent model results.
61
-------
0.0
0.00
0.02
0.04
Time (hours)
0.06
0.08
Figure 20 CP degradation in the presence of free chlorine and bromide at pH 8.0. [CP]0 =
0.39 [iM, [HOC1]T = 50 [iM, [Br"] = 10 [iM, [PO4]T = 10 mM, and Temperature
25 °C. Lines represent model results.
62
-------
0.06
0.05
0.04 -
£ 0.03 -
o
o
£
0.02 -
0.01 -
0.00
0.0 0.5 1.0 1.5 2.0
Time (hours)
2.5
3.0
3.5
Figure 21 Chlorine demand of ACC water over the pH range of 7-9. [HOCl]x = 50 |jM,
[Buffer]T = 10 mM, [DOC] = 1.1 mg-C/L, and Temperature = 25 °C.
63
-------
jz
M
35
30 -
25 :
20 :
15 :
10 :
5 :
6.5
• Control
O ACCF
$
n
7.0
7.5
8.0
PH
8.5
9.0
9.5
Figure 22
CP degradation in the presence of free chlorine and ACC water over the pH range
of 7-9. [CP]0 « 0.5 nM, [HOC1]T = 50 |iM, [DOC] = 1.1 mg-C/L, [Buffer]T = 10
mM, and Temperature = 25 °C. Lines represent model results.
64
-------
0.0
0.0
CP
CPO
TCP
0.1
0.2 0.3 0.4
Time (hours)
0.5
0.6
Figure 23 CP degradation in the presence of free chlorine and ACC water at pH 7. [CP]C
0.46 nM, [HOC1]T = 50 nM, [DOC] = 1.1 mg-C/L, [Buffer]T = 10 mM, and
Temperature = 25 °C. Lines represent model results.
65
-------
0.6
0.5 -
0.4 -
CP
CPO
TCP
0.00 0.05 0.10 0.15 0.20
Time (hours)
0.25
0.30
0.35
Figure 24 CP degradation in the presence of free chlorine and ACC water at pH 7.5. [CP]0
0.35 nM, [HOC1]T = 50 |iM, [DOC] = 1.1 mg-C/L, [Buffer]T = 10 mM, and
Temperature = 25 °C. Lines represent model results.
66
-------
0.00
0.25
0.50 0.75
Time (hours)
1.00
1.25
Figure 25
CP degradation in the presence of free chlorine and ACC water at pH 8.0. [CP]0
0.42 nM, [HOC1]T = 50 nM, [DOC] = 1.1 mg-C/L, [Buffer]T = 10 mM, and
Temperature = 25 °C. Lines represent model results.
67
-------
0.6
0.5 -
CP
CPO
TCP
0.0
0.5
1.0 1.5
Time (hours)
2.0
2.5
Figure 26
CP degradation in the presence of free chlorine and ACC water at pH 8.5. [CP]0
0.35 nM, [HOC1]T = 50 nM, [DOC] = 1.1 mg-C/L, [Buffer]T = 10 mM, and
Temperature = 25 °C. Lines represent model results.
68
-------
0.0
0.0
0.5
1.0
1.5 2.0
Time (hours)
2.5
i.O
i.5
Figure 27
CP degradation in the presence of free chlorine and ACC water at pH 9.0. [CP]0
0.5 nM, [HOC1]T = 50 nM, [DOC] = 1.1 mg-C/L, [Buffer]T = 10 mM, and
Temperature = 25 °C. Lines represent model results.
69
------- |